Partisan Mortality Cycles

Daniel L. Millimet∗
Southern Methodist University & IZA

Travis Whitacre†
Yale University

August 9, 2024

Abstract
Geographic inequities in mortality rates in the US are pronounced and growing. The Black-White mortality
gap is volatile but persistent, while the Rich-Poor mortality gap is increasing dramatically. Yet, the causes
of these inequalities are not understood. Recently, the focus has turned to the role of place-specific
factors. Here, we explore the importance of politics as a place-specific factor contributing to mortality
inequality. Specifically, we test for the existence of partisan mortality cycles using panel data on counties
from 1968-2016 and information on the political ideology of state and federal political officials. We confirm
the existence of partisan mortality cycles, finding lower mortality in counties governed by more liberal
political regimes; the evidence regarding political party is mixed. Several sources of heterogeneity are also
uncovered. While additional research is needed to understand the mechanisms involved, the analysis here
suggests that analyses of spatial, racial, and income differences in mortality ought to start with the political
system.

JEL: I10, I18, J10

Keywords: Mortality, Political Cycles

∗
The authors wish to thank Tatyana Deryugina, John Francois, John Holbein, Kandice Kapinos, Jason Lindo, John
Mullahy, Orgul Ozturk, Alicia Plemmons, Hannes Schwandt for useful comments, as well as conference participants at the
Southern Economic Association and seminar participants at SMU. All errors are our own. Corresponding author. Daniel
Millimet, Department of Economics, Box 0496, Southern Methodist University, Dallas, TX 75275-0496. Tel: (214) 768-3269.
E-mail: [email protected]. ORCID: 0000-0001-5178-5993.
†
Travis Whitacre, Department of Health Policy and Management, Box 208034, Yale School of Public Health, New Haven,
CT 06520-8034. E-mail: [email protected].
 “[A]ll manner of domestic-policy issues have been recast as matters of life and death – urgent,
uncompromising, and absolute.”
- The New Yorker, 22 November 20091

1 Introduction
The age-adjusted mortality rate in the US has fallen sharply over the past century. Cutler et al. (2006,
p. 97) state, “Through the twentieth century in the United States and other high-income countries, growth
in real incomes was accompanied by a historically unprecedented decline in mortality rates that caused
life expectancy at birth to grow by nearly 30 years.” As shown in Figure 1, the age-adjusted all-cause
mortality rate has fallen from roughly 1,300 per 100,000 people in 1968 to below 800 in 2016. Yet, the
improvement has not been shared equally as large disparities exist in the US based on geographic location.
The interquartile range across counties follows a U-shaped pattern, falling from a high of roughly 250 per
100,000 people in 1968 to a low of less than 160 in the early 1980s and rising to nearly 210 in 2016. Figure
A1 in Appendix A shows the variation across counties in more detail at six different points in time.2 The
figure reveals that the southeastern part of the US, along with isolated areas in the upper Midwest and
Southwest, have higher mortality rates than much of the remainder of the country.

Figure 1: Historical Age-Adjusted Mortality Rate

Notes: The age-adjusted mortality rate in each year is computed as the population-weighted average of
county-level age-adjusted mortality rates. The interquartile range is the difference in the 75th and 25th
percentiles of the age-adjusted mortality rates at the county-level.

1
https://www.newyorker.com/magazine/2009/11/30/the-politics-of-death
2
The Supplemental Appendix is available at https://smu.box.com/s/114p6kflefc2j1ul5ixsff633g1xmfc3.

1
 Figure 2: Historical Racial & Income Differences in Age-Adjusted Mortality Rates

(A) Racial Differences (B) Income Differences

Notes: The age-adjusted mortality rate in each year used to compute the mortality ratio is obtained as the
population-weighted average of county-level age-adjusted mortality rates by race and sex in Panel (A) and
by whether the county-level per capita income is above or below the year-specific median county-level per
capita income in Panel (B).

Inequality in mortality by race and income has also varied tremendously over this time period. Panel
A in Figure 2 shows the ratio of Black-to-White age-adjusted all-cause mortality rates separately by sex.3
The mortality rate for Black females fluctuated between 25% and 35% higher than for White females over
the period from 1968 until about 2005. The ratio has declined since 2005 and was about 15% higher in
2016. The mortality rate for Black males increased from about 20% higher than for White males in the
early 1970s to more than 40% higher in the early to mid-1990s. Since the mid-1990s, the ratio has declined,
hovering around 20% since 2011. Panel B in Figure 2 shows the ratio of age-adjusted all-cause mortality
rates in counties with per capita income below the year-specific median to counties above the median.
The mortality rate in low-income counties was roughly 2% higher than in high-income counties in 1980.
However, since that time the gap has been widening and reached about 20% in 2016.
The decline in mortality, rise in geographic inequality in mortality, rise in the income-mortality gradient,
and persistent but volatile inequality in the Black-White mortality gap, are now the focus of much attention.
A growing part of this literature has examined potential origins of geographic mortality inequality “through
the lens of place” (Schwandt et al., 2021, p. 1). Despite this attention, much remains unknown. Couillard
et al. (2021, p. 142) conclude that “these widening geographic disparities in state-level mortality cannot be
attributed to changing spatial patterns in education levels, income inequality, or rising deaths of despair.”
In this paper, we build on this literature by taking a step back and assessing the role of politics. We ask
what role, if any, do partisan politics and, specifically, the political ideology of those in power contribute
to the spatial and temporal trends in mortality, both overall as well as by race and sex.
Specifically, we test for the presence of what we refer to as partisan mortality cycles. Since Nordhaus
(1975) and Hibbs (1977), there has been an extensive literature assessing, both theoretically and empirically,
3
Historically, the recorded gender on a death certificate has reflected the recorded sex on the corresponding birth certificate
and/or the decedent’s anatomy at death. Of late, several states have made it possible for death certificates to reflect gender
identity. As most of these legislative changes have occurred after our sample, we use the term ‘sex’ to reflect the way coroners
documented deaths over the majority of our sample.

2
so-called political and partisan business cycles (e.g., Alesina, 1987; Alesina and Sachs, 1988; Rogoff, 1990;
Alesina and Roubini, 1992; Hibbs, 1992). While the definitions of these terms are not exact, political
business cycles generally refer to changes in economic conditions driven by the timing of elections and
re-election concerns; they are not tied to a specific political party or ideology. This literature is usually
attributed to Nordhaus (1975). In contrast, partisan business cycles occur when politicians choose policies
aligned with the preferences of their constituents. As a result, political regimes beholden to different
constituents with different preferences will lead to economic fluctuations. Hibbs (1977) is often credited
for initiating this literature; Price (1997) offers an excellent review.
Our analysis is in line with the partisan view of fluctuations as we test for spatial and temporal fluc-
tuations in mortality depending on the political ideology of those in power. Just as political regimes drive
economic fluctuations due to distinct policy preferences, political regimes may have important consequences
on both short- and long-term public health. While our hypothesis that partisan mortality cycles may be
a salient factor underlying the rise in geographic inequality stems from the role politics played during the
COVID-19 pandemic, the collision between partisan politics and public health began long ago. For decades,
the US has experienced ideological clashes over health care reform, public mental health institutions, pol-
lution, gun control, the social safety net, cigarette taxes, and more. There are also ideological differences in
regulatory oversight, enforcement of public health laws, and preferences for economic redistribution. Not
only might these ideological differences impact overall public health, but the partisan model also suggests
that political regimes may focus on public health issues which primarily affect their own voter base. As
Hibbs (1977, p. 1487) states, “The real winners of elections are perhaps best determined by examining the
policy consequences of partisan change rather than by simply tallying the votes.”
To proceed, we build on Chetty et al. (2016) and Finkelstein et al. (2021) to construct a very simple
theoretical model showing that mortality depends on local individual health capital and location- and
time-specific health capital. Our approach takes a high level view and assesses whether political regimes
are the origin of the black box that is the place effect. In particular, we assess whether partisan ideology is
the ultimate, rather than the proximate, cause of geographic variation in mortality. As noted in Torche and
Rauf (2021, p. 378), “how changes in political contexts within countries over time can shape health has
been relatively less explored.”4 If partisan mortality cycles are present, a full investigation of proximate
causes are beyond the scope of the current paper. We do offer some preliminary thoughts in the discussion.
Nonetheless, there is significant value-added in knowing whether geographic inequities in mortality are
related to political regimes.
Motivated by the theoretical framework, we use county-level panel data on age-adjusted mortality from
the US Centers for Disease Control and Prevention (CDC) over the period 1968 to 2016 to estimate event-
study-like specifications in the sense that we include both lags and leads of political variables. This allows
us to say something about whether any associations are likely causal as inclusion of leads allows us to test
the crucial strict exogeneity assumption (Wooldridge, 2010, p. 325). This strategy has been used elsewhere
to establish causal relationships, such as in Heyes and Saberian (2019) who analyze the causal effect of
temperature on court decisions.
As political regimes can impact public health through both short-term actions (e.g., gun control or
enforcement of safety regulations) and long-term actions (e.g., pollution control or safety net generosity
4
Italics in original.

3
and efficiency), we analyze two measures of mortality, all-cause mortality and short-term mortality, where
the latter includes mortality attributable to causes that operate within a short window. As political
regimes in the US often turn over frequently, it may be easier for one to identify partisan mortality cycles,
if they exist, in short-term mortality where mortality is attributable to more immediate factors.5 However,
one should not dismiss the possibility of an event having an immediate impact on all-cause mortality. For
example, Borgschulte and Vogler (2020) found that Medicaid expansion in 2014 reduced all-cause mortality
by more than 11 deaths per 100,000 individuals within two years in treated counties.
To characterize the local political regime, our focus is on the political ideology of state and federal
politicians. While we also explore the role of political party, we focus on ideology for two reasons. First,
there is significant temporal and spatial variation in what each political party represents. In particular,
the Democratic party, especially at the state level, has not always been tied to liberal ideology (Strong,
1971). The timing of the shift of white, conservative southerners from the Democratic to the Republican
party varied by state (Kuziemko and Washington, 2018). In addition, as we discuss below, today there
remains significant ideological differences within political party, particularly at the state level. Second,
Dynes and Holbein (2020) provide an exhaustive examination of the effects of state-level executive and
legislative party control on 28 policy outcomes from 1960 to 2016, finding precisely estimated null effects.
Our analysis yields several important findings. First, and most importantly, we confirm the existence
of partisan political mortality cycles as political ideology is associated with county-level mortality. We find
a consistent pattern of lower all-cause and short-term mortality under more liberal regimes over the full
sample period. A one standard deviation increase in liberalism is associated with 1-2 fewer all-cause deaths
and 0.7-1.0 fewer short-term deaths per 100,000 in an average county-year. Moreover, the results suggest
that these findings are causal.
Second, we uncover heterogeneity in partisan mortality cycles by race and sex, consistent with the
argument in Hibbs (1977) that political regimes may focus more on issues which primarily affecting their
own voter base. Over the full sample period, we find a negative association between liberalism and short-
term mortality for all demographic groups except Black males. However, further analysis reveals a negative,
statistically meaningful association between liberalism and short-term mortality for both Black females or
males only during the period covering the crack epidemic (mid-1980s to the early 1990s). Ideology is also
associated with the short-term mortality of White females and males from 1968–1984. Most importantly,
we find no statistically meaningful evidence of partisan mortality cycles after 2000 for any group.
That we find no statistically meaningful evidence of partisan mortality cycles in the 21st century
before COVID-19 is remarkable. We present evidence that this may be attributable to the rise in political
polarization and greater legislative gridlock experienced since 2000. This suggests that policymakers in
the current political climate are ineffective in reducing mortality. As such, amelioration of the geographic
inequality in mortality may require solutions outside the political system. However, as noted above, we
leave a careful investigation into mechanisms for future research.
The remainder of the paper is organized as follows. Section 2 provides a brief literature review. Section 3
develops a motivating theory of county-specific mortality rates. Section 4 presents our empirical framework
and data. Section 5 discusses the results. Section 6 offers a brief discussion and concludes.
5
The analysis of short-term mortality also avoids concerns over locational differences in place of death versus place of
residence throughout one’s lifetime (e.g., Fletcher et al., 2022).

4
2 Literature Review
Our examination of partisan mortality cycles contributes to four strands of literature. The first liter-
ature, as mentioned above, focuses on geographic, racial, and income inequality in mortality and the role
of economic aspects of place in understanding these inequities (e.g., Cutler et al., 2006; Peltzman, 2009;
Currie and Schwandt, 2016; Case and Deaton, 2017; Vierboom et al., 2019; Benjamins et al., 2021; Baker
et al., 2021; Deryugina and Molitor, 2021; Couillard et al., 2021). Place-based factors that have been
examined include education (Montez and Berkman, 2014), income (Chetty et al., 2016), income inequality
(Deaton and Paxson, 2004; Avendano, 2012), poverty (Currie and Schwandt, 2016; Schwandt et al., 2021),
public health interventions (Anderson et al., 2022), and social policy (Kim, 2016; Reynolds and Avendano,
2018). We add to this literature by finding an important role of political aspects of place in understanding
mortality inequality.
The second literature explores the connection between politics and health. Several cross-country studies
document associations between political regime and health (e.g., Navarro and Shi, 2001; Navarro et al.,
2006; Blum et al., 2021). These studies find that democratic, social democratic, and egalitarian political
regimes are associated with better health outcomes. Other studies focus on a single country, often assessing
associations between political variables and suicide or infant mortality (e.g., Classen and Dunn, 2010;
Torche and Rauf, 2021). These studies point to some association between more liberal regimes and better
health outcomes. We add to this literature by examining county-level panel data over nearly 50 years and
focusing on the political ideology of those in power at both the state and federal levels.
The final two literatures uncover possible linkages between political regimes and mortality. Specifically,
the third literature examines the effect of political ideology on economic outcomes and policy choices, while
the fourth literature relates the economic environment to mortality. Examples in the former include Alesina
and Roubini (1992); Kelly and Witko (2014); Magkonis et al. (2021), while some in the latter include Ruhm
(2000); Miller et al. (2009); Coile et al. (2014). Overall, these studies find statistically meaningful linkages
between political regimes and economic outcomes, as well as economic outcomes and mortality. We add
to this literature by examining the direct association between political ideology and mortality.

3 Theory
To motivate our empirical specification, we follow Chetty et al. (2016) and Finkelstein et al. (2021) and
specify the log mortality hazard rate of individual i at age a residing in location c at time t as

log[mict (a)] = θi + βa + γct , (1)

where θ is individual-level, time invariant health capital and γ is location- and time-specific health capital
(referred to as the place effect in Finkelstein et al. (2021) and Couillard et al. (2021)). Let θct denote the
average individual-specific health capital of residents of location c at time t. Note, θct will generally vary
over time despite θi being time invariant due to changes in the local population through migration, fertility,
and past mortality (Deryugina and Molitor, 2021).

5
 We define the mortality rate of an average individual residing in location c at time t at age a as


mct (a) = exp θct + βa + γct . (2)

The age-adjusted mortality rate in county c at time t is given by

Z


mct = exp θct + βa + γct fs (a)da (3)
a

where fs (a) is the age distribution of the standard population.

Suppose age in the standard population is distributed as χ2 (k), a chi-squared distribution with k degrees
of freedom, and A is the upper limit on age. As shown in Appendix B, this implies

log mct = βe + θct + γct , (4)

where βe is a constant that depends on k and A. Thus, the log age-adjusted mortality rate in county c at
time t depends on a constant term, the average health capital of residents, θct , and the place effect, γct .

4 Empirics
4.1 Model
The empirical model for the age-adjusted mortality rate is based on (4). Our baseline specification is

log mct = αc + λt + fc (t) + xhct β h + x̃lct β l + εct , (5)

where mct is age-adjusted mortality rate in county c in year t, αc and λt are location and time fixed effects,
fc (t) is a county-specific, flexible control for time, xhct is a vector of factors capturing the individual health
capital, θi , of the local population (entering via θct ), x̃lct is a vector of factors capturing the health capital
of a location (entering via the place effect, γct ), and εct is a mean zero error term capturing differences
between the right-hand sides of (4) and (5).
Importantly, we envision political variables entering via x̃lct . To explore the dynamics of the interplay,
if any, between the politics and mortality, we include the political variables in an event-study-like manner.
Specifically, we model the health capital of a location as

4
X 3
X
x̃lct β l = l
pc,t−s βlag,s + pct β0l + l
pc,t+s βlead,s + xlct βxl , (6)
s=1 s=1

l
where pct is the vector of location-specific political variables, βlag,· l
(βlead,· ) represents the association
between lagged (future) political variables and current mortality, β0l represents the contemporaneous asso-
ciation, and xlct denotes the remaining covariates entering via the place effect, γct . pct includes measures
of state liberalism, state liberalism interacted with presidential and US Congressional ideology, and the
triple interaction. Presidential and US Congressional ideology alone are absorbed by the time fixed effects.
We refer to the estimating equation obtained by substituting (6) into (5) as event-study-like as we expect

6
 l
βlag,· and/or β0l to be non-zero if partisan mortality cycles are present.6
A few comments are warranted. First, despite the political variables being measured at the state and
federal levels, effects at the county-level are identified through the temporal variation in these variables
given the inclusion of county fixed effects. We do not aggregate the analysis to the state-level as this
discards useful variation in both mortality as well as other controls that are available at the county level.
Moreover, recent research finds that local politics can affect economic outcomes that are less visible to the
l
electorate (Dippel, 2022). Second, to increase the precision of the estimates, we restrict βlag,s , s = 1, ..., 4,
l
to be equal. Similarly, we restrict βlead,s , s = 1, ..., 3, to be equal as well.7 Third, the temporal variation
in the ideology variables comes from two sources: cross and within changes in political preferences. Cross
changes occur when elections, resignations, or deaths cause one politician to be replaced by another.
Within changes occur when the preferences of an elected official change over time. Fourth, we consider
the combined effects of state and federal ideology as actions at one level of government may amplify or
undermine actions at the other level.
Finally, while we believe there to be considerable value in documenting associations between political
ideology and mortality across US counties given the dearth of prior research, we understand the desire
for causal estimates as well. To that end, it is common in empirical models focusing on political party
to exploit close elections in a regression discontinuity design (e.g., Dynes and Holbein, 2020). We do not
pursue this strategy here given that our focus is on variation in ideology at the state and federal levels and
across legislative and executive branches. As such, the overall political climate affecting a given county in
a given year is measured using several continuous variables and their interactions, arising from numerous
elections across many time periods.8
Instead, we aim to identify the causal effects of ideology through the use of county fixed effects and rich
controls for unobserved time effects and observed attributes of mortality. The key assumption required
for our estimates to be unbiased is strict exogeneity. This assumption is testable by augmenting the
l
model with leads of the covariates. In our case, βlead,· being statistically indistinguishable from zero is
consistent with the political variables being strictly exogenous and thus the estimated coefficients having
a causal interpretation (Wooldridge, 2010, p. 325). It is important to note that testing whether the
coefficients on the leads are statistically different from zero not only tests whether political variables at
time t are (conditionally) correlated with the idiosyncratic error term in prior periods, εc,t−s , s > 0, but
also whether they are (conditionally) correlated with the idiosyncratic error term in any time period, εc,t−s ,
s = T − t, ..., −1, 0, 1, ..., t − 1, since the model is estimated using Pooled Ordinary Least Squares (POLS)
applied to the mean-differenced equation. This also means that the presence of the leads tests for the
possibility of reverse causation as this would lead to violation of strict exogeneity.
In our baseline model xlct also includes controls for the age distribution. In particular, we include
the share of the population in detailed age categories. While age-adjusted mortality controls for the
age distribution to some extent, we allow for additional effects of the age distribution that may operate
6
In the literature assessing the effects of political regimes on outcomes, there is ambiguity regarding how long it should
take for effects to materialize, if they materialize at all. Dynes and Holbein (2020) focus on two to four years after a regime
comes into power when examining a plethora of different outcomes. In studies that specifically examine health outcomes, the
focus is often on a one year lag (e.g., Finkelstein et al., 2012). Here, we include four lags and three leads.
7
Unrestricted estimates for most specifications are provided graphically in Appendix A.
8
Regression discontinuity approaches have also been criticized in the political party context. Marshall (forthcoming) shows
that regression discontinuity approaches that exploit close elections to isolate the effects of a particular attribute of a politician
(such as their ideology) can be misleading if the attribute directly impacts the vote margin.

7
through other channels. We also include the share of the population in different sex-race categories. xhct
includes the share of the population in different education categories, given the importance of education
in explaining mortality (e.g., Case and Deaton, 2021; Couillard et al., 2021). Lastly, we model fc (t) using
state-specific linear or quadratic time trends or county-specific linear time trends to control for changes in
local population health capital and place-specific factors affecting mortality other than politics.9 We allow
for quadratic time trends since mortality has a natural lower bound.
Finally, we consider an extended version of our baseline specification where xlct also includes controls
for income inequality. Specifically, we control for state-level income shares of the top 1% and 10%. We
then interact each of these variables with a time invariant indicator if the county is urban or non-urban.
Income inequality has recently received particular attention as a possible source of the place effect on
mortality. However, we do not include these controls in our baseline specification as income inequality may
be a mechanism by which partisan mortality cycles arise.
We cluster standard errors at the county level. We explored clustering at the state-year level as well
as at the state-level. Clustering at the state level leads to an effective cluster size of roughly ten given the
cross-cluster heterogeneity in cluster size and covariates, well below the number considered appropriate
(Carter et al., 2017). Clustering at the state-year level results in 180 effective clusters. While this is
sufficient for hypothesis testing, the clustering is at a level such that the unit of observation (county) is not
nested within clusters.10 For this reason, we cluster at the county-level, which yields 735 effective clusters.

4.2 Data
Data on age-adjusted all-cause mortality and age-adjusted cause-specific mortality come from the US
CDC’s WONDER database.11 We use county-level data for the period 1968-2016. This time period is
optimal for our study as it begins after the point in which the literature indicates modern platforms and
ideologies of the Democratic and Republican parties beginning to take shape at the Federal level (Kuziemko
and Washington, 2018). Further, it ends before the beginning of the COVID-19 pandemic.
We analyze two measures of county-level mortality: all-cause mortality and short-term mortality. Each
measure is constructed for the full county-year population, as well as separately by race (Black and
White) and sex (male and female). The distinction between all-cause and short-term mortality follows
the World Health Organization (WHO) assignment of causes of death to three categories: communica-
ble, non-communicable, and injuries (World Health Organization, 2020). Communicable causes of deaths
are attributable to a non-chronic health condition, such as infectious and parasitic diseases, nutritional
conditions, respiratory conditions, and maternal and perinatal conditions. Non-communicable causes of
death are attributable to diseases and conditions stemming from long-term, chronic conditions. Injuries
capture all non-natural causes of death, such as deaths of despair (suicide, gun violence, etc.) and car
accidents. We define short-term mortality as causes of death which fall under the communicable or injuries
9
Anderson et al. (2022) similarly use time fixed effects and municipality-level linear time trends when modeling urban
mortality rates. Dynes and Holbein (2020) control for state-specific linear time trends while also including time fixed effects
in their analysis of policy outcomes.
10
Clustering at the state-year level leads to p-values that are roughly three times as large when county-level clustering
leads to a p-value below 0.10. Thus, for tests with p-values around 0.02 when clustering at the county-level, the p-value is
approximately 0.05 when clustering at the state-year level.
11
See https://wonder.cdc.gov/mortSQL.html.

8
category.12 Figure A2 in Appendix A shows the variation in short-term mortality across counties at six
points in time. As with all-cause mortality, the Southeast faces higher short-term mortality. However,
many counties in the Southwest and Northwest/Midwest also have higher short-term mortality rates than
much of the remainder of the country.
County-level data on population by age, sex, race are from the National Institute of Health.13 County-
level data on educational attainment are from the USDA Economic Research Service.14 We classify counties
as urban or non-urban using the 2013 Rural-Urban Continuum Codes (RUCC) from the US Department
of Agriculture.15 The RUCC classifies counties into nine categories on the basis of metropolitan status,
population size, and adjacency of nonmetropolitan areas to a metropolitan area. We define urban counties
as those classified as metropolitan in 2013; thus, our definition is time invariant. State income shares come
from the Frank-Sommeiller-Price series.16
To characterize the local political regime, our primary focus is on the political ideology of state and
federal politicians. The federal ideological political variables are the DW-Nominate (Dynamic Weighted
NOMINAl Three-step Estimation) scores available from UCLA Social Science Division’s Voteview.17 The
Nominate scores vary from -1 to 1, with higher values corresponding to greater liberalism. Each politician
receives two scores per two-year Congressional session. These scores are referred to as the first and second
dimensions. Dimension 1 is interpreted as representing economic ideology. Dimension 2 is interpreted as
representing social ideology. Our baseline analysis uses Dimension 1. We consider Dimension 2 to assess
sensitivity. We form an overall ideology score for the US Congress by computing the median ideology score
separately for the US House and the US Senate in each year and then averaging. For the state level, we
use a similar ideological score capturing overall state policy liberalism (Caughey and Warshaw, 2016).18
The scores vary from roughly -3 to 3, again with higher scores corresponding to greater liberalism.
Figure 3 is similar to Figure 1, but shows trends in mortality separately by ideology for both all-cause
(Panel A) and short-term (Panel B) mortality. Liberal states are defined as those with a liberalism score
above the sample median; years under Republican presidents are shaded in grey. The figure reveals a
persistent gap favoring liberal states. This gap becomes even more pronounced when focusing on short-
term mortality. For all-cause mortality, the gap has been steadily widening since the mid-1980s.
Figure 4 is identical to Figure 3 except now each panel shows mortality separately by race (Black or
White) and sex (male or female). When we decompose mortality rates based on race and sex, several
important stylized facts emerge. First, all-cause mortality is highest for Black males, followed by White
males, then Black females, and finally White females. This ranking persists over the entire sample period,
although mortality has fallen for all groups over time. There is not much difference in short-term mortality
12
The term short-term mortality is used to represent different aspects of mortality in different contexts. In the health
literature, it often refers to death within 90 days of entering the hospital. In economics, it has been used to refer to mortality
within days or weeks of some intervention. Here, for lack of a better term, we use it to refer to causes of death that do not
take long to be realized. Note, also, while the CDC censors cause-specific mortality with fewer than 10 deaths, we extract
aggregate age-adjusted mortality for all categories we label as “short-term”. As a result, censoring in our data only occurs if
there are fewer than 10 total short-term deaths in a county-year cell, which almost never happens.
13
See https://seer.cancer.gov/popdata/download.html.
14
See https://www.ers.usda.gov/data-products/county-level-data-sets. The data are available every ten years; we
use linear interpolation for the remainder.
15
See https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/.
16
See https://www.shsu.edu/eco mwf/inequality.html.
17
See https://www.voteview.com.
18
Available at http://www.chriswarshaw.com/data.php.

9
by race and sex. Second, the gap in all-cause mortality favoring liberal states emerges by the early 1990s
and widens over the remainder of the sample for all groups. Finally, the gap in short-term mortality
favoring liberal states exists over the entire sample with the exception of the early to mid 1990s for Blacks.

Figure 3: Historical Age-Adjusted Mortality Rates by Political Ideology

(A) All-Cause (B) “Short-Term”-Cause

Notes: The age-adjusted mortality rate in each year is computed as the population-weighted average
of county-level age-adjusted mortality rates. Gray shaded regions denotes periods with a Republican
president.

Figure 4: Historical Age-Adjusted Mortality Rates by Political Ideology Across Race and Sex

(A) All-Cause (B) “Short-Term”-Cause

Notes: The age-adjusted mortality rate in each year is computed as the population-weighted average
of county-level age-adjusted mortality rates. Gray shaded regions denotes periods with a Republican
president.

10
5 Results
5.1 Baseline Specifications
Tables 1 and 2 present the marginal effects from the baseline specification in (5). Table 1 contains the
results for all-cause mortality, while Table 2 presents the results for short-term mortality. The tables display
the estimates (and statistical significance) of a one standard deviation increase in state and presidential
liberalism. We refer to this as tests of Executive Liberalism. Similarly, we report marginal effects of a one
standard deviation increase in state and federal legislative liberalism. We refer to this as tests of Legislative
Liberalism. Finally, we report marginal effects of a one standard deviation increase in state, presidential,
and federal legislative liberalism. We refer to this as tests of Full Liberalism.19
We consider various sets of controls across the columns in each table, but our preferred specifications in-
clude location and population health controls, county and year fixed effects, and county-specific linear time
trends (Column 6 in each table). Column 7 in each table includes additional controls for income inequality.
We focus on Columns 6 and 7 in our discussion as these include the most exhaustive set of controls. Figure
5 graphically displays the results, including both the restricted and unrestricted specifications. The actual
coefficient estimates from the estimation of (5) are relegated to Tables C1–C2 in Appendix C.
The assessment of all-cause mortality in Table 1 yields two findings. First, we find a statistically
significant, negative association at the p < 0.05 level between lagged Executive Liberalism and all-cause
mortality in our preferred specification (Column 6). Controlling for income inequality does not change this
result (Column 7). The effect size is modest; a one standard deviation increase in Executive Liberalism
in the prior four years is associated with a 0.2% decline in all-cause mortality. This translates to roughly
two fewer deaths per 100,000 for the average county-year; 5,000 fewer deaths in the average year. To put
this in context, a one standard deviation increase in Executive Liberalism is roughly moving from a county
in Kansas in 1979 under a Republican governor and President Jimmy Carter (median of the distribution)
to a county in Oregon in 2000 under a Democratic governor and President Bill Clinton (85th percentile of
the distribution). The contemporaneous and lead effects of Executive Liberalism are relatively precisely
estimated zeros. This is consistent with a negative causal effect of Executive Liberalism on future all-cause
mortality. The top left panel in Figure 5 shows this visually.
Second, we find a statistically significant, negative association at the p < 0.10 level between lagged
Legislative Liberalism in our preferred specification (Column 6). While the point estimate is unchanged,
the estimate is no longer statistically significant at conventional levels after controlling for income inequality
(Column 7). The effect size is small; a one standard deviation increase in Legislative Liberalism in the
prior four years is associated with a 0.1% decrease in all-cause mortality. This is roughly equivalent to
moving from a county in Florida in 1991 with a 58% (61%) Democratic share of the Senate (House of
Representatives) in the 102nd Congress (15th percentile of the distribution) to a county in Ohio in 2009
with a 58% (59%) Democratic share of the US Senate (House of Representatives) in the 111th Congress
(median of the distribution). The contemporaneous effects of Legislative Liberalism are precisely estimated
zeros; the lead effect is statistically significant, but positive, at the p < 0.10 level in Column 6. Because
19
Note, the marginal effects of Full Liberalism should not be interpreted as the effects holding the other marginal effects
constant. In other words, the marginal effect of Full Liberalism is the effect of a one standard deviation in liberalism at
all levels included in the model holding constant only the non-political covariates. In addition, while a standard deviation
increase in each measure of ideology is not trivial, it is much smaller than the interquartile range for each variable (presidential,
congressional, and state).

11
the lead effect is of the opposite sign of the lagged effect, we cautiously interpret this as modest evidence
of a small, negative causal effect of Legislative Liberalism on future all-cause mortality. The middle left
panel in Figure 5 shows this visually. The effects of Full Liberalism are never statistically significant at
conventional levels.

Table 1: Marginal Effects of Ideology on All-Cause Mortality

(1) (2) (3) (4) (5) (6) (7)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0004 -0.0014 -0.0022 0.0005 -0.0021 -0.0016 -0.0015
P-value 0.8332 0.1258 0.0060 0.5519 0.0071 0.0322 0.0480
1σ ↑ State & Federal Legislative Liberalism 0.0010 -0.0010 -0.0017 0.0010 -0.0016 -0.0012 -0.0011
P-value 0.5944 0.2685 0.0276 0.2189 0.0321 0.0919 0.1337
1σ ↑ Full Liberalism 0.0002 -0.0001 -0.0016 0.0013 -0.0015 -0.0009 -0.0007
P-value 0.8905 0.9038 0.0506 0.1389 0.0564 0.2356 0.3353
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism 0.0020 0.0041 0.0007 -0.0039 0.0005 0.0002 -0.0003
P-value 0.5962 0.1748 0.8170 0.1837 0.8691 0.9547 0.9159
1σ ↑ State & Federal Legislative Liberalism 0.0032 0.0062 0.0014 -0.0029 0.0012 0.0009 0.0007
P-value 0.4038 0.0385 0.6403 0.3288 0.6897 0.7393 0.7901
1σ ↑ Full Liberalism 0.0006 0.0052 0.0002 -0.0040 -0.0000 0.0000 -0.0002
P-value 0.8748 0.0844 0.9529 0.1831 0.9919 0.9989 0.9307
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0185 -0.0071 -0.0017 -0.0023 -0.0019 0.0009 0.0009
P-value 0.0000 0.0000 0.1024 0.0474 0.0649 0.3395 0.3513
1σ ↑ State & Federal Legislative Liberalism 0.0164 -0.0022 -0.0008 -0.0021 -0.0011 0.0016 0.0015
P-value 0.0000 0.0425 0.4394 0.0640 0.2994 0.0930 0.1298
1σ ↑ Full Liberalism 0.0156 -0.0041 -0.0017 -0.0026 -0.0019 0.0012 0.0011
P-value 0.0000 0.0002 0.1116 0.0234 0.0665 0.2157 0.2537

Location Health Controls N N N N N Y Y

Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be
equal). Lead values correspond to the effects over the future three years (constrained to be equal).

12
 Figure 5: Marginal Effects of Ideology on Mortality

(A) Executive

(B) Legislative

(C) Full
Notes: All-cause mortality results are shown in the left column; Short-Term mortality results are
shown in the right column. Mortality is measured in period 0. Periods -1 to -4 capture the
associations between lagged political variables and current mortality; periods 1 to 3 capture the
associations between future political variables and current mortality.

13
 Table 2: Marginal Effects of Ideology on Short-Term Mortality

(1) (2) (3) (4) (5) (6) (7)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0124 -0.0136 -0.0076 -0.0034 -0.0083 -0.0080 -0.0076
P-value 0.0000 0.0000 0.0002 0.1196 0.0000 0.0001 0.0002
1σ ↑ State & Federal Legislative Liberalism -0.0099 -0.0111 -0.0049 -0.0003 -0.0056 -0.0054 -0.0050
P-value 0.0001 0.0000 0.0110 0.8963 0.0038 0.0046 0.0099
1σ ↑ Full Liberalism -0.0080 -0.0111 -0.0055 -0.0011 -0.0062 -0.0064 -0.0058
P-value 0.0013 0.0000 0.0063 0.6239 0.0021 0.0016 0.0042
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism 0.0231 -0.0000 -0.0041 -0.0097 -0.0052 -0.0026 -0.0042
P-value 0.0043 0.9991 0.6085 0.2243 0.5172 0.7399 0.5862
1σ ↑ State & Federal Legislative Liberalism 0.0289 0.0052 0.0007 -0.0047 -0.0005 0.0016 0.0008
P-value 0.0004 0.5181 0.9277 0.5564 0.9532 0.8394 0.9134
1σ ↑ Full Liberalism 0.0232 0.0004 -0.0043 -0.0101 -0.0055 -0.0029 -0.0038
P-value 0.0044 0.9576 0.5901 0.2081 0.4969 0.7099 0.6262
Lead Effects
1σ ↑ State & Federal Executive Liberalism -0.0043 -0.0060 -0.0003 -0.0023 0.0006 0.0027 0.0027
P-value 0.0777 0.0107 0.9072 0.4227 0.8079 0.3052 0.3010
1σ ↑ State & Federal Legislative Liberalism -0.0065 -0.0064 -0.0009 -0.0029 0.0002 0.0018 0.0015
P-value 0.0117 0.0090 0.7316 0.3064 0.9350 0.4944 0.5770
1σ ↑ Full Liberalism -0.0097 -0.0073 -0.0021 -0.0045 -0.0010 0.0009 0.0007
P-value 0.0002 0.0035 0.4258 0.1321 0.7006 0.7455 0.7992

Location Health Controls N N N N N Y Y

Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be
equal). Lead values correspond to the effects over the future three years (constrained to be equal).

Turning to the short-term mortality in Tables 2, we find statistically significant, negative associations
between lagged Executive, Legislative, and Full Liberalism and mortality at the p < 0.01 level in the
majority of specifications, including our preferred specification (Column 6) and after controlling for income
inequality (Column 7). Moreover, as seen in Figure 5, the estimated contemporaneous and lead effects
of each dimension of liberalism on short-term mortality are precise zeros in all specifications that include
year fixed effects and at least state-specific linear time trends (Columns 3-7). This is consistent with our
estimates having a causal interpretation. The magnitude of the effects are larger than in the case of all-cause
mortality in percentage terms. Specifically, we find that a one standard deviation increase in Executive
(Legislative) Liberalism leads to a 0.8% (0.5%) reduction in short-term mortality; a one standard deviation
increase in Full Liberalism leads to a 0.6% reduction. In absolute terms, this corresponds to roughly 0.7-1.0
fewer short-term deaths per 100,000 in the average county-year.

14
 In sum, the baseline results point to the existence of partisan mortality cycles of modest magnitudes
as all-cause and short-term mortality vary according to the ideology of the political regime. Moreover,
the results are consistent with these partisan mortality cycles representing causal relationships given the
statistically insignificant or contrasting estimates of the lead effects. Finally, controlling for state-level
income inequality does not alter these findings.

5.2 Assessing Heterogeneity

We examine several possible sources of heterogeneity. First, we assess heterogeneity across demographic
groups and time periods. As shown in Figures 2 and 3, the Black-White mortality gap is large and volatile
over time. Moreover, the gap in life expectancy fell by 48% over this time period (Schwandt et al.,
2021). Prior research has found that overall geographic inequality in mortality is strongly linked to racial
differences in mortality (e.g., Schwandt et al., 2021). In addition, we allow for heterogeneity across time
periods. Allowing for temporal heterogeneity explores the role of ideology during the early part of the
sample period when southern Democrats were primarily conservative. These politicians were very different
from their northern Democratic counterparts (Strong, 1971). Moreover, the timing of the shift of White,
conservative southerners from the Democratic party to the Republican party varied by state (Kuziemko
and Washington, 2018). Additional sources of temporal heterogeneity include the deinstitutionalization
movement during the early part of the sample period whereby capacity in public mental health institutions
was severely curtailed, the crack epidemic during the middle part of the sample period, and the opioid
epidemic and expanding political polarization during the final part of the sample period. Second, we
examine the robustness of our findings to an alternative measure of political ideology. Finally, we examine
heterogeneity across the distribution of mortality using quantile regression.

5.2.1 Race and Sex

We now turn to exploring heterogeneity across demographic groups. Specifically, the dependent vari-
ables are now the mortality rates of a specific race (Black or White) and sex (male or female) pair. In
addition to estimating models for the full sample, we also analyze heterogeneity across different time peri-
ods: (i) 1968–1984, (ii) 1985-2000, and (iii) 2001-2016. Our choice of the three time periods is motivated
by historical developments. As discussed previously, the transformation of the South from Democrat to
Republican occurred predominantly during the early time period (Kuziemko and Washington, 2018). The
early time period also coincides with the deinstitutionalization period, when states drastically reduced the
capacity of public mental health facilities (Lerman, 1985).
The second period, spanning 1985 to 2000, contains the crack epidemic. Fryer et al. (2013, p. 1652-
4) state that crack cocaine was “virtually unheard of prior to the mid-1980s,” but then it “rose sharply
beginning in 1985, peaked in 1989, and slowly declined thereafter.” This era saw a profound increase in
the homicide rate of young black males. In addition, usage by females led to a rise in mortality due to
AIDS and a decline in fetal health. This led to a dramatic increase in all-cause mortality among young
black males in urban areas which returned to pre-crack cocaine levels by 2000 (Evans et al., 2016).
The final period is characterized by a significant increase in political polarization in the US, leading
to a growing chasm in the policy choices across locations. Grumbach (2018, p. 417) states that “[policy]
variation is increasingly related to party control of government; prior to 2000, whether a state was controlled

15
by Democrats or Republicans said little about the policies it would adopt, but the parties have implemented
highly divergent policy agendas after 2000.” The widening policy differences across the political parties
is also mirrored in the preferences of individuals. Autor et al. (2020, p. 1340) write: “The twenty-
first century has also seen greater polarization in the policy preferences and media viewing habits of the
American public. In the 1990s and early 2000s, roughly one-half of respondents took moderate positions
on prominent political issues; by the late 2000s the centrist share had shrunk to under 40 percent, as
individuals adopted more strident views on the left or right.”
Marginal effects are reported in Tables 3–10. Each presents the results for the full sample period and
the three sub-periods. For brevity, we only present the results from our preferred specifications; those that
include county-specific linear time trends with and without controls for inequality (analogous to Columns
6 and 7 in Tables 1 and 2). Figures A4–A11 in Appendix A display the results graphically for the full
sample period. The coefficient estimates from (5) are in Tables C3–C10 in Appendix C.
Because many models are being estimated (4 demographic groups × 4 time periods × 2 outcomes ×
2 specifications), we follow the advice in Benjamin et al. (2018) and focus on estimates of lagged and
contemporaneous effects that are statistically significant at at least the p < 0.005 level and where the
lead effects are either statistically insignificant at the p < 0.10 level or statistically significant but of the
opposite sign of the lagged and/or contemporaneous effects. Using this very conservative approach, if such
cases arise we are confident that the effects are not due to Type I errors and have a causal interpretation.
Even adopting this conservative approach, several important findings arise. To start, we find no sta-
tistically meaningful effects of ideology on all-cause mortality for any demographic group or time period.
However, we find statistically and economically meaningful estimates for short-term mortality for all demo-
graphic groups. Over the full sample period, contemporaneous Executive, Legislative, and Full Liberalism
lower the age-adjusted short-term mortality rate for Black females and White males and females. The
magnitudes range from 3.7% for Black females to 4.1% for White females to 5.5% for White males for a
one standard deviation increase in liberalism.
Splitting the sample by time period, we find statistically and economically significant effects for all
demographic groups in the period covering the crack epidemic. A one standard deviation increase in Full
Liberalism leads to an 11% and 12% decrease in the age-adjusted mortality rate for Black males and
females, respectively; a 5% and 4% decline for White males and females, respectively. Lagged liberalism is
also a statistically and economically significant determinant of short-term mortality for all groups except
Black females during this time period. A one standard deviation increase in Full Liberalism over the prior
four years leads to a 22% decrease in the age-adjusted mortality rate for Black males and a 9% and 10%
for White males and females, respectively. This translates to roughly six fewer short-term deaths per
100,000 for Black males in the average county-year during this era. This increases to 12 (29) per 100,000
in the county-year observation at the 95th (99th ) percentile of the Black male age-adjusted mortality rate
distribution. We also find sizeable negative effects of lagged liberalism on White males and females in the
first time period, covering 1968–1984.
Overall, we find no effect of ideology on all-cause mortality by demographic group. Thus, the effects
found in the baseline model do not materialize in a statistically meaningful way when the sample is confined
to a particular demographic group. However, liberalism consistently lowers short-term mortality during the
20th century, especially during the crack epidemic. Moreover, liberalism benefits all demographic groups.

16
 Table 3: Marginal Effects of Ideology on All-Cause Mortality for Black Males

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0171 -0.0165 -0.2074 -0.2201 -0.0186 -0.0190 0.0159 0.0169
P-value 0.3708 0.3859 0.2302 0.2073 0.4506 0.4385 0.6612 0.6424
1σ ↑ State & Federal Legislative Liberalism -0.0222 -0.0217 -0.1055 -0.1082 -0.0223 -0.0234 0.0297 0.0306
P-value 0.2479 0.2578 0.1269 0.1186 0.2943 0.2704 0.3810 0.3670
1σ ↑ Full Liberalism -0.0199 -0.0194 -0.1440 -0.1510 -0.0172 -0.0173 0.0238 0.0249
P-value 0.3052 0.3162 0.1797 0.1622 0.4788 0.4734 0.5029 0.4850
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0067 -0.0068 0.1003 0.0976 -0.0248 -0.0253 -0.0153 -0.0132
P-value 0.1134 0.1081 0.3543 0.3700 0.0241 0.0240 0.3838 0.4527
1σ ↑ State & Federal Legislative Liberalism -0.0062 -0.0063 0.0286 0.0269 -0.0102 -0.0109 -0.0177 -0.0150
P-value 0.1359 0.1283 0.3820 0.4128 0.1824 0.1694 0.2875 0.3653
1σ ↑ Full Liberalism -0.0078 -0.0079 0.0456 0.0425 -0.0139 -0.0143 -0.0195 -0.0156
P-value 0.0771 0.0716 0.4621 0.4950 0.2083 0.2042 0.2868 0.3966
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0075 0.0074 -0.2942 -0.3130 0.0044 0.0046 -0.0411 -0.0381
P-value 0.2229 0.2311 0.0166 0.0118 0.7091 0.7005 0.0212 0.0328

17
1σ ↑ State & Federal Legislative Liberalism 0.0086 0.0084 -0.0985 -0.1041 0.0071 0.0073 -0.0314 -0.0281
P-value 0.1512 0.1623 0.0281 0.0219 0.5273 0.5235 0.0769 0.1118
1σ ↑ Full Liberalism 0.0082 0.0081 -0.1816 -0.1927 0.0098 0.0101 -0.0357 -0.0329
P-value 0.1851 0.1951 0.0172 0.0125 0.4157 0.4049 0.0549 0.0762

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table 4: Marginal Effects of Ideology on “Short-Term”-Cause Mortality for Black Males

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0487 -0.0577 0.0860 0.1845 -0.2032 -0.1992 0.0728 0.0803
P-value 0.2547 0.1759 0.7625 0.5265 0.0014 0.0018 0.3858 0.3409
1σ ↑ State & Federal Legislative Liberalism -0.0323 -0.0410 0.0457 0.0579 -0.1993 -0.1883 0.1349 0.1405
P-value 0.4469 0.3332 0.6882 0.6149 0.0002 0.0005 0.0948 0.0840
1σ ↑ Full Liberalism -0.0453 -0.0544 0.0558 0.1012 -0.2201 -0.2161 0.1093 0.1174
P-value 0.2913 0.2031 0.7566 0.5813 0.0004 0.0005 0.1825 0.1556
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0377 -0.0358 -0.0734 0.0176 -0.1325 -0.1395 0.0370 0.0372
P-value 0.0014 0.0027 0.6822 0.9219 0.0000 0.0000 0.2696 0.2743
1σ ↑ State & Federal Legislative Liberalism -0.0315 -0.0291 0.0191 0.0411 -0.0583 -0.0635 0.0461 0.0443
P-value 0.0059 0.0117 0.7257 0.4523 0.0114 0.0054 0.1524 0.1767
1σ ↑ Full Liberalism -0.0415 -0.0389 -0.0153 0.0352 -0.1087 -0.1152 0.0483 0.0476
P-value 0.0008 0.0018 0.8816 0.7332 0.0007 0.0003 0.1790 0.1986
Lead Effects
1σ ↑ State & Federal Executive Liberalism -0.0632 -0.0644 0.2869 0.3037 -0.0098 -0.0322 -0.0316 -0.0359
P-value 0.0003 0.0003 0.1981 0.1788 0.7212 0.2363 0.3779 0.3160

18
1σ ↑ State & Federal Legislative Liberalism -0.0666 -0.0670 0.0934 0.0877 -0.0193 -0.0347 -0.0243 -0.0288
P-value 0.0001 0.0001 0.2554 0.2905 0.4738 0.1968 0.4629 0.3849
1σ ↑ Full Liberalism -0.0647 -0.0659 0.1802 0.1814 -0.0227 -0.0441 -0.0282 -0.0307
P-value 0.0002 0.0002 0.2055 0.2079 0.4245 0.1126 0.4268 0.3871

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table 5: Marginal Effects of Ideology on All-Cause Mortality for Black Females

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0242 -0.0261 -0.0661 -0.0705 -0.0306 -0.0314 0.0208 0.0176
P-value 0.1810 0.1493 0.6959 0.6791 0.2019 0.1867 0.5040 0.5709
1σ ↑ State & Federal Legislative Liberalism -0.0272 -0.0280 -0.0406 -0.0446 -0.0251 -0.0250 -0.0006 -0.0041
P-value 0.1296 0.1176 0.5416 0.5033 0.2374 0.2295 0.9844 0.8891
1σ ↑ Full Liberalism -0.0274 -0.0284 -0.0571 -0.0616 -0.0263 -0.0267 0.0148 0.0107
P-value 0.1349 0.1196 0.5848 0.5579 0.2738 0.2607 0.6253 0.7242
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism 0.0012 0.0014 0.0590 0.0611 0.0150 0.0157 -0.0080 -0.0093
P-value 0.7631 0.7311 0.5763 0.5653 0.1350 0.1239 0.5734 0.5147
1σ ↑ State & Federal Legislative Liberalism 0.0008 0.0012 0.0211 0.0205 0.0080 0.0085 -0.0068 -0.0079
P-value 0.8301 0.7627 0.5150 0.5286 0.2615 0.2430 0.6120 0.5606
1σ ↑ Full Liberalism 0.0015 0.0019 0.0281 0.0280 0.0150 0.0155 0.0086 0.0067
P-value 0.7136 0.6427 0.6431 0.6462 0.1503 0.1418 0.5698 0.6657
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0032 0.0034 -0.1913 -0.2018 0.0138 0.0139 0.0078 0.0071
P-value 0.5815 0.5500 0.0962 0.0820 0.2144 0.2127 0.6097 0.6424

19
1σ ↑ State & Federal Legislative Liberalism 0.0044 0.0044 -0.0590 -0.0635 0.0172 0.0175 -0.0021 -0.0029
P-value 0.4184 0.4215 0.1599 0.1329 0.1089 0.1059 0.8877 0.8450
1σ ↑ Full Liberalism 0.0043 0.0044 -0.1123 -0.1194 0.0129 0.0130 0.0093 0.0080
P-value 0.4589 0.4508 0.1166 0.0977 0.2624 0.2629 0.5555 0.6107

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table 6: Marginal Effects of Ideology on “Short-Term”-Cause Mortality for Black Females

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism 0.0194 0.0105 -0.5977 -0.6522 -0.0014 -0.0059 -0.0034 -0.0151
P-value 0.6752 0.8196 0.1049 0.0834 0.9831 0.9290 0.9692 0.8662
1σ ↑ State & Federal Legislative Liberalism 0.0320 0.0248 -0.1646 -0.1862 -0.0017 -0.0029 0.0148 0.0064
P-value 0.4862 0.5879 0.2531 0.2025 0.9766 0.9605 0.8625 0.9408
1σ ↑ Full Liberalism 0.0195 0.0123 -0.3442 -0.3800 -0.0147 -0.0191 0.0092 0.0029
P-value 0.6752 0.7907 0.1374 0.1076 0.8151 0.7635 0.9167 0.9745
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0352 -0.0350 0.0453 0.0667 -0.1341 -0.1352 0.0844 0.0815
P-value 0.0097 0.0104 0.8411 0.7688 0.0000 0.0000 0.0671 0.0824
1σ ↑ State & Federal Legislative Liberalism -0.0306 -0.0299 0.0224 0.0225 -0.0670 -0.0677 0.0815 0.0764
P-value 0.0220 0.0256 0.7510 0.7502 0.0031 0.0037 0.0609 0.0848
1σ ↑ Full Liberalism -0.0373 -0.0366 0.0222 0.0270 -0.1149 -0.1168 0.0944 0.0898
P-value 0.0068 0.0081 0.8646 0.8360 0.0002 0.0002 0.0497 0.0710
Lead Effects
1σ ↑ State & Federal Executive Liberalism -0.0160 -0.0147 -0.0689 -0.1466 0.0148 0.0104 -0.0012 0.0004
P-value 0.3682 0.4126 0.8044 0.6000 0.5902 0.7036 0.9795 0.9936

20
1σ ↑ State & Federal Legislative Liberalism -0.0163 -0.0167 -0.0057 -0.0368 0.0008 -0.0030 0.0266 0.0265
P-value 0.3459 0.3354 0.9564 0.7238 0.9769 0.9096 0.5642 0.5647
1σ ↑ Full Liberalism -0.0127 -0.0129 -0.0413 -0.0925 0.0086 0.0039 0.0096 0.0141
P-value 0.4763 0.4719 0.8165 0.6049 0.7663 0.8925 0.8435 0.7728

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table 7: Marginal Effects of Ideology on All-Cause Mortality for White Males

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism 0.0006 0.0002 -0.0229 -0.0304 -0.0057 -0.0058 -0.0019 -0.0037
P-value 0.8880 0.9579 0.5530 0.4312 0.4265 0.4174 0.8572 0.7285
1σ ↑ State & Federal Legislative Liberalism 0.0018 0.0017 -0.0128 -0.0150 -0.0024 -0.0025 0.0015 -0.0003
P-value 0.6484 0.6663 0.3727 0.2979 0.7120 0.6951 0.8772 0.9707
1σ ↑ Full Liberalism 0.0008 0.0006 -0.0160 -0.0204 -0.0043 -0.0044 -0.0001 -0.0022
P-value 0.8491 0.8792 0.4923 0.3804 0.5493 0.5434 0.9903 0.8246
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0004 -0.0004 -0.0046 -0.0045 -0.0003 -0.0003 0.0018 0.0003
P-value 0.6739 0.7311 0.8574 0.8622 0.9253 0.9055 0.6889 0.9417
1σ ↑ State & Federal Legislative Liberalism -0.0001 -0.0000 -0.0008 -0.0011 0.0027 0.0027 0.0009 -0.0006
P-value 0.9013 0.9715 0.9233 0.8922 0.2511 0.2526 0.8394 0.8902
1σ ↑ Full Liberalism -0.0001 0.0000 -0.0040 -0.0046 -0.0004 -0.0004 0.0023 0.0001
P-value 0.9342 0.9784 0.7925 0.7650 0.9121 0.9162 0.6694 0.9916
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0013 0.0013 -0.0345 -0.0410 -0.0006 -0.0006 -0.0054 -0.0058
P-value 0.3137 0.3138 0.1769 0.1135 0.8411 0.8405 0.2521 0.2143

21
1σ ↑ State & Federal Legislative Liberalism 0.0013 0.0012 -0.0141 -0.0159 -0.0021 -0.0022 -0.0041 -0.0051
P-value 0.2930 0.3536 0.1269 0.0917 0.4762 0.4478 0.3653 0.2602
1σ ↑ Full Liberalism 0.0011 0.0010 -0.0217 -0.0256 -0.0010 -0.0010 -0.0035 -0.0041
P-value 0.3977 0.4336 0.1665 0.1093 0.7442 0.7533 0.4819 0.4084

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table 8: Marginal Effects of Ideology on “Short-Term”-Cause Mortality for White Males

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism 0.0142 0.0198 -0.2890 -0.3213 -0.0816 -0.0722 0.0055 0.0056
P-value 0.3356 0.1833 0.0098 0.0040 0.0033 0.0083 0.8166 0.8138
1σ ↑ State & Federal Legislative Liberalism 0.0192 0.0228 -0.1027 -0.1328 -0.0481 -0.0424 -0.0104 -0.0109
P-value 0.1939 0.1224 0.0163 0.0019 0.0472 0.0750 0.6357 0.6191
1σ ↑ Full Liberalism 0.0177 0.0217 -0.1660 -0.1982 -0.0900 -0.0817 -0.0044 -0.0051
P-value 0.2311 0.1435 0.0151 0.0037 0.0010 0.0024 0.8478 0.8255
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0448 -0.0457 0.0919 0.1008 -0.0431 -0.0475 0.0076 0.0056
P-value 0.0000 0.0000 0.2215 0.1808 0.0014 0.0005 0.4431 0.5777
1σ ↑ State & Federal Legislative Liberalism -0.0508 -0.0520 0.0484 0.0466 -0.0105 -0.0149 0.0035 0.0017
P-value 0.0000 0.0000 0.0383 0.0475 0.3275 0.1663 0.7110 0.8579
1σ ↑ Full Liberalism -0.0545 -0.0559 0.0720 0.0710 -0.0487 -0.0519 0.0076 0.0052
P-value 0.0000 0.0000 0.0967 0.1029 0.0006 0.0003 0.5244 0.6717
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0059 0.0057 0.2373 0.1603 0.0421 0.0364 0.0071 0.0071
P-value 0.3701 0.3905 0.0018 0.0387 0.0008 0.0038 0.5102 0.5065

22
1σ ↑ State & Federal Legislative Liberalism -0.0032 -0.0027 0.0652 0.0301 0.0325 0.0297 0.0036 0.0031
P-value 0.6106 0.6688 0.0161 0.2785 0.0096 0.0186 0.7224 0.7611
1σ ↑ Full Liberalism 0.0080 0.0083 0.1370 0.0838 0.0271 0.0232 0.0085 0.0082
P-value 0.2371 0.2224 0.0038 0.0837 0.0395 0.0799 0.4479 0.4671

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table 9: Marginal Effects of Ideology on All-Cause Mortality for White Females

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism 0.0008 0.0007 0.0065 0.0040 -0.0042 -0.0049 0.0158 0.0149
P-value 0.8491 0.8681 0.8675 0.9181 0.5489 0.4835 0.1464 0.1708
1σ ↑ State & Federal Legislative Liberalism 0.0010 0.0010 -0.0048 -0.0029 -0.0041 -0.0041 0.0144 0.0135
P-value 0.8081 0.8045 0.7418 0.8424 0.5330 0.5274 0.1348 0.1632
1σ ↑ Full Liberalism -0.0001 -0.0002 -0.0003 -0.0003 -0.0073 -0.0078 0.0178 0.0167
P-value 0.9739 0.9710 0.9897 0.9916 0.3078 0.2780 0.0825 0.1047
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0016 -0.0016 0.0486 0.0494 -0.0063 -0.0059 0.0101 0.0094
P-value 0.1483 0.1523 0.0766 0.0719 0.0381 0.0561 0.0273 0.0411
1σ ↑ State & Federal Legislative Liberalism -0.0010 -0.0010 0.0151 0.0156 -0.0030 -0.0024 0.0079 0.0073
P-value 0.3382 0.3485 0.0796 0.0693 0.2137 0.3077 0.0678 0.0929
1σ ↑ Full Liberalism -0.0011 -0.0011 0.0275 0.0281 -0.0065 -0.0060 0.0096 0.0086
P-value 0.2898 0.3006 0.0865 0.0796 0.0626 0.0864 0.0842 0.1292
Lead Effects
1σ ↑ State & Federal Executive Liberalism -0.0022 -0.0023 -0.0018 0.0002 0.0055 0.0050 0.0023 0.0022
P-value 0.1164 0.1155 0.9483 0.9957 0.0590 0.0849 0.6430 0.6560

23
1σ ↑ State & Federal Legislative Liberalism -0.0013 -0.0014 -0.0084 -0.0062 0.0035 0.0028 0.0066 0.0064
P-value 0.3766 0.3418 0.4133 0.5542 0.2267 0.3354 0.1585 0.1747
1σ ↑ Full Liberalism -0.0022 -0.0022 -0.0068 -0.0047 0.0052 0.0047 0.0054 0.0052
P-value 0.1367 0.1275 0.6978 0.7912 0.1036 0.1439 0.2927 0.3141

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table 10: Marginal Effects of Ideology on “Short-Term”-Cause Mortality for White Females

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism 0.0054 0.0039 -0.5127 -0.5288 -0.0944 -0.0892 -0.0013 -0.0024
P-value 0.7268 0.7995 0.0000 0.0000 0.0002 0.0004 0.9647 0.9337
1σ ↑ State & Federal Legislative Liberalism 0.0131 0.0113 -0.1687 -0.1850 -0.0561 -0.0520 -0.0009 -0.0029
P-value 0.3947 0.4619 0.0003 0.0001 0.0132 0.0206 0.9740 0.9135
1σ ↑ Full Liberalism 0.0053 0.0035 -0.3017 -0.3184 -0.0960 -0.0913 -0.0002 -0.0023
P-value 0.7356 0.8226 0.0001 0.0000 0.0001 0.0003 0.9945 0.9322
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0365 -0.0357 0.0799 0.0834 -0.0399 -0.0383 -0.0073 -0.0100
P-value 0.0000 0.0000 0.3259 0.3059 0.0021 0.0035 0.5708 0.4398
1σ ↑ State & Federal Legislative Liberalism -0.0372 -0.0366 0.0260 0.0247 -0.0129 -0.0125 -0.0100 -0.0126
P-value 0.0000 0.0000 0.3130 0.3395 0.1884 0.2099 0.4077 0.3052
1σ ↑ Full Liberalism -0.0410 -0.0403 0.0435 0.0425 -0.0385 -0.0368 -0.0042 -0.0079
P-value 0.0000 0.0000 0.3584 0.3712 0.0043 0.0068 0.7713 0.5934
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0046 0.0040 -0.1208 -0.1618 0.0520 0.0488 -0.0078 -0.0080
P-value 0.4410 0.5013 0.1850 0.0811 0.0000 0.0000 0.5581 0.5496

24
1σ ↑ State & Federal Legislative Liberalism -0.0029 -0.0031 -0.0508 -0.0699 0.0436 0.0427 0.0018 0.0006
P-value 0.6156 0.5920 0.1252 0.0386 0.0002 0.0002 0.8889 0.9609
1σ ↑ Full Liberalism 0.0041 0.0037 -0.0857 -0.1141 0.0422 0.0399 -0.0065 -0.0071
P-value 0.5017 0.5504 0.1298 0.0476 0.0005 0.0011 0.6412 0.6097

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and popu-
lation shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income
shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (con-
strained to be equal). Lead values correspond to the effects over the future three years (constrained to be equal).
5.2.2 Ideology Scores

Our next extension explores heterogeneity across the measure of ideology. As discussed in Section 4.2,
we measure ideology using the first dimension of the DW-NOMINATE scores, corresponding to economic
ideology. Tables D1–D2 in Appendix D are identical to Tables 1–2 except now ideology is measured using
the second dimension, corresponding to social ideology. The results are a bit weaker, but qualitatively
similar to the baseline results. Specifically, we continue to find negative effects of liberalism on all-cause
mortality, but the point estimates are smaller (in absolute value) and no longer statistically significant at
conventional levels. For short-term mortality, the negative effects of lagged Legislative and Full Liberalism
continue to be negative and statistically significant at the p < 0.05 level. The effect of lagged Executive
Liberalism is negative, but smaller (in absolute value) and no longer statistically significant at conventional
levels. Overall, the results suggest that ideology on economic issues matters more for mortality than
ideology on social issues. This is consistent with the relationship between economic outcomes and mortality
discussed in Section 2.
The greater importance of economic liberalism is also critical for understanding future mortality trends.
The Trump presidency symbolized a change in US politics as social issues came to the forefront. A 2022
article in The Atlantic states:

“Politics seems more existential, not less. Pundits and partisans cast everything as a culture
war, even those things that have little to do with culture... Elites in both parties enjoy a certain
privilege – one appropriate to a rich, advanced democracy – that allows them to emphasize
culture while deprioritizing economic well-being. Civilizational concerns gain more political
resonance precisely as perceptions of civilizational decline intensify on right and left alike.”20

This shift away from economic issues in general, and a more liberal economic agenda in particular, may
portend future public health crises.

5.2.3 Quantile Regression

Our final extension explores heterogeneity across the mortality distribution via quantile regression. For
brevity the results are relegated to Appendix E and reveal contrasting results for all-cause and short-term
mortality. For all-cause mortality, liberalism is more beneficial (or less harmful) at higher quantiles. For
short-term mortality, liberalism is most beneficial at low quantiles. This implies that liberalism amelio-
rates (conditional) inequities in all-cause mortality across locations due to larger relative improvements in
locations with high (conditional) all-cause mortality, but exacerbates (conditional) inequities in short-term
mortality due to larger relative improvements in locations with low (conditional) short-term mortality. As
the greater inequity in short-term mortality arises from greater improvements in locations that are rela-
tively healthy, as opposed to deterioration in locations that are relatively unhealthy, this is not necessarily a
bad outcome. However, it is consistent with more liberal regimes being ineffective in reducing disparities in
short-term mortality. Yet, such regimes do benefit the least healthy places in terms of all-cause mortality.
20
See https://www.theatlantic.com/ideas/archive/2022/01/republicans-democrats-forever-culture-war/621184
/.

25
5.3 Role of Political Party
Our focus has been on partisan mortality cycles as driven by the ideology of political regimes. We do so
since ideology scores are arguably more closely connected to actual policy decisions. Nonetheless, a large
literature explores the effect of political party on various economic and social outcomes. In light of this,
we briefly explore the role of political party.
To proceed, we replace our ideology measures with covariates indicating Democratic party control of
the executive branch at the state and federal levels, Democratic party control of the legislative branch at
the state and federal levels, and their interactions. We measure Executive Control with a dummy variable
equal to one if the state governor is a Democrat (zero otherwise) and the interaction of this variable with
a dummy variable equal to one if the President is a Democrat (zero otherwise). We measure Legislative
Control with a dummy variable equal to one if the state legislature (both House and Senate) is controlled
by a Democratic majority (zero otherwise) and the interaction of this variables with a dummy variable
equal to one if the federal legislature (both House and Senate) is controlled by a Democratic majority (zero
otherwise).21 Finally, we measure Full Control by interacting executive and legislative control. Executive
control enables the party in power the ability to enforce laws. Legislative control enables the party in
power the ability to enact laws. Full control enables the party in power the ability enact and enforce laws,
as well as appoint judges.
Overall, the results using political party are broadly consistent with those using ideology, particularly
after accounting for the ideology of Southern Democrats in the early part of the sample. This highlights
the importance of understanding the spatial and temporal variation in views within US political parties.
More comprehensive discussion and all results are provided in Appendix F.

6 Discussion
Mortality has been declining for decades in the US prior to COVID-19. However, the overall trend
masks important heterogeneity. Geographic differences in mortality rates are on the rise and racial and
income disparities in mortality remain significant. Couillard et al. (2021, p. 123) notes that “disparities
in mortality have become increasingly apparent among different groups in the population.” Such inequity
is now the “subject of intense interest” (Baker et al., 2021, p. 26). Cutler et al. (2006, p. 112) state that
“the elimination or at least reduction of differences in health by income, race or geography has become a
major focus of health policy in many countries...”
Despite attempts to understand the sources of geographic inequality in mortality, observed attributes
of localities examined to date only explain so much. Peltzman (2009, p. 189) writes:

“To place the intra-U.S. geographic dispersion in life expectancy in international context, the
top decile of the U.S. population is a year or two below the very top of the international
distribution, which might be exemplified by Japan. The bottom decile of the U.S. population in
life expectancy is at roughly the average life expectancy level of Mexico or Argentina. In terms of
characteristics, while the places in the bottom decile of the U.S. life expectancy distribution tend
to have low income and education, substantial variety remains uncaptured by such observables.”
21
We do not include a dummy variable for either the party of the President or federal legislative control as this is absorbed
by the time fixed effects; only the interaction term is identified.

26
Here, we step back and test for the presence of partisan mortality cycles.
Specifically, we assess the relationship between the ideological attributes of political regimes and the
health capital of locations as reflected in age-adjusted mortality rates. Our empirical framework uses event-
study like specifications. In so doing, we have been careful when giving the results a causal interpretation.
In cases where the lagged and/or contemporaneous effects are statistically significant and the lead effects
are either statistically insignificant or significant but in the opposite direction, we interpret the results as
consistent with a causal effect of politics. But, the usual caveats apply. That said, even if we ‘only’ identify
associations, this is important. Baker et al. (2021, p.26) argue that “documenting these inequalities and
how they have evolved over time ... is key to addressing them.” Similarly, Torche and Rauf (2021, p. 378)
state that “understanding country trajectories is critical to examine how plausible variation in institutional,
political, and policy contexts shapes health outcomes” (italics in original).
Overall, we find evidence of partisan mortality cycles as the political ideology of political regimes is
associated with fluctuations in mortality. We consistently see that more liberal regimes are associated with
lower all-cause and short-term mortality over the full sample period. We also document some differences in
partisan mortality cycles across racial and sex groups and time period. In particular, the negative effects
of liberalism on short-term mortality are common to all demographic groups, with the exception of Black
males when using the full sample period. When allowing for heterogeneity over time, we find that any
beneficial effects liberalism on Black males or females is confined to the period covering the crack epidemic.
However, ideology also affects White males and females in early part of the sample. Finally, we find no
statistically meaningful evidence of partisan mortality cycles after 2000 either in the full sample or for any
demographic group.
Our last finding is that liberalism (and Democratic control) reduces geographic inequality in all-cause
mortality by being more beneficial (or less harmful) in locations with the highest (conditional) levels. The
fact that the timing of the decline in state-level liberalism starting around 1990 coincides with the rise
geographic inequality suggests that future work exploring the relationship between politics and mortality is
warranted. However, liberalism (and Democratic control) contribute to geographic inequality in short-term
mortality due to larger relative improvements in locations with the lowest (conditional) levels. While the
reduction in short-term mortality in locations with already low levels of (conditional) short-term mortality
is a good outcome, these political regimes are less effective in locations with high levels of (conditional)
short-term mortality.
In the end, our analysis indicates a complex relationship between attributes of the political regime
controlling a county and the dynamics of local mortality. However, it identifies several keys areas for
future research, such as an in-depth examination of the mechanisms behind the partisan mortality cycles
in the 20th century and the lack of such cycles in the 21st century. In Appendix G we undertake a
cursory analysis of some possible mechanisms. We document evidence suggesting that differences in safety
net participation and cigarette taxes may play an important role. But, growing legislative gridlock and
political polarization, along with the opioid epidemic, may have limited the impact of political regimes in
the 21st century.

27
References
Alesina, A. (1987). Macroeconomic policy in a two-party system as a repeated game. Quarterly Journal
of Economics, 102 (3), 651–678.

— and Roubini, N. (1992). Political cycles in OECD economies. Review of Economic Studies, 59 (4),
663–688.

— and Sachs, J. (1988). Political parties and the business cycle in the united states, 1948-1984. Journal
of Money, Credit and Banking, 20 (1), 63–82.

Anderson, D. M., Kerwin Kofi, C. and Rees, D. I. (2022). Re-examining the contribution of public
health efforts to the decline in urban mortality. American Economic Journal: Applied Economics, 14,
126–157.

Autor, D., Dorn, D., Hanson, G. and Majlesi, K. (2020). Importing political polarization? the
electoral consequences of rising trade exposure. American Economic Review, 110 (10), 3139–3183.

Avendano, M. (2012). Correlation or causation? Income inequality and infant mortality in fixed effects
models in the period 1960–2008 in 34 OECD countries. Social Science & Medicine, 75 (4), 754–760.

Baker, M., Currie, J., Miloucheva, B., Schwandt, H. and Thuilliez, J. (2021). Inequality in
mortality: Updated estimates for the United States, Canada and France. Fiscal Studies, 42 (1), 25–46.

Benjamin, D. J., Berger, J. O., Johannesson, M., Nosek, B. A., Wagenmakers, E.-J., Berk,
R., Bollen, K. A., Brembs, B., Brown, L., Camerer, C., Cesarini, D., Chambers, C. D.,
Clyde, M., Cook, T. D., De Boeck, P., Dienes, Z., Dreber, A., Easwaran, K., Efferson,
C., Fehr, E., Fidler, F., Field, A. P., Forster, M., George, E. I., Gonzalez, R., Goodman,
S., Green, E., Green, D. P., Greenwald, A. G., Hadfield, J. D., Hedges, L. V., Held, L.,
Hua Ho, T., Hoijtink, H., Hruschka, D. J., Imai, K., Imbens, G., Ioannidis, J. P. A., Jeon,
M., Jones, J. H., Kirchler, M., Laibson, D., List, J., Little, R., Lupia, A., Machery, E.,
Maxwell, S. E., McCarthy, M., Moore, D. A., Morgan, S. L., Munafó, M., Nakagawa, S.,
Nyhan, B., Parker, T. H., Pericchi, L., Perugini, M., Rouder, J., Rousseau, J., Savalei, V.,
Schönbrodt, F. D., Sellke, T., Sinclair, B., Tingley, D., Van Zandt, T., Vazire, S., Watts,
D. J., Winship, C., Wolpert, R. L., Xie, Y., Young, C., Zinman, J. and Johnson, V. E. (2018).
Redefine statistical significance. Nature Human Behavior, 2 (1), 6–10.

Benjamins, M. R., Silva, A., Saiyed, N. S. and De Maio, F. G. (2021). Comparison of all-cause
mortality rates and inequities between black and white populations across the 30 most populous us
cities. JAMA Network Open, 4 (1), e2032086–e2032086.

Binder, S. (2014). Polarized we govern? Tech. rep., Center for Effective Public Management, Brookings
Institution.

Binder, S. A. (1999). The dynamics of legislative gridlock, 1947-96. The American Political Science
Review, 93 (3), 519–533.

28
Blinder, A. S. and Watson, M. W. (2016). Presidents and the U.S. economy: An econometric explo-
ration. American Economic Review, 106 (4), 1015–1045.

Blum, J., Dorn, F. and Heuer, A. (2021). Political institutions and health expenditure. International
Tax and Public Finance, 28, 323–363.

Borgschulte, M. and Vogler, J. (2020). Did the aca medicaid expansion save lives? Journal of Health
Economics, 72, 1–17.

Carter, A. V., Schnepel, K. T. and Steigerwald, D. G. (2017). Asymptotic bheavior of a t-test

robust to cluster heterogeneity. Review of Economics and Statistics, 99 (4), 698–709.

Case, A. and Deaton, A. (2017). Mortality and morbidity in the 21st century. Brookings Papers on
Economic Activity, Spring, 397–476.

— and — (2021). The Great Divide: Education, Despair and Death. Tech. rep., NBER Working Paper
29241.

Caughey, D. and Warshaw, C. (2016). The dynamics of state policy liberalism, 1936–2014. American
Journal of Political Science, 60 (4), 899–913.

Chetty, R., Stepner, M., Abraham, S., Lin, S., Scuderi, B., Turner, N., Bergeron, A. and
Cutler, D. (2016). The association between income and life expectancy in the United States, 2001-2014.
Journal of the American Medical Association, 315, 1750–1766.

Classen, T. J. and Dunn, R. A. (2010). The politics of hope and despair: The effect of presidential
election outcomes on suicide rates. Social Science Quarterly, 91 (3), 593–612.

Coile, C. C., Levine, P. B. and McKnight, R. (2014). Recessions, older workers, and longevity: How
long are recessions good for your health? American Economic Journal: Economic Policy, 6 (3), 92–119.

Couillard, B., Foote, C., Gandhi, K., Meara, E. and Skinner, J. (2021). Rising geographic dis-
parities in US mortality. Journal of Economic Perspectives, 35, 123–146.

Currie, J. and Schwandt, H. (2016). Mortality inequality: The good news from a county-level approach.
Journal of Economic Perspectives, 30 (2), 29–52.

— and — (2021). The opioid epidemic was not caused by economic distress but by factors that could
be more rapidly addressed. The ANNALS of the American Academy of Political and Social Science,
695 (1), 276–291.

Cutler, D., Deaton, A. and Lleras-Muney, A. (2006). The determinants of mortality. Journal of
Economic Perspectives, 20 (3), 97–120.

Deaton, A. and Paxson, C. (2004). Perspectives on the Economics of Aging, Chicago: University of
Chicago Press, chap. Mortality, Income, and Income Inequality Over Time in Britain and the United
States.

29
Deryugina, T. and Molitor, D. (2021). The causal effects of place on health and longevity. Journal of
Economic Perspectives, 35, 147–170.

Dippel, C. (2022). Political parties do matter in us cities . . . for their unfunded pensions. American
Economic Journal: Economic Policy, 14 (3), 33–54.

Dynes, A. M. and Holbein, J. B. (2020). Noisy retrospection: The effect of party control on policy
outcomes. American Political Science Review, 114, 237–257.

Evans, W. N., Garthwaite, C. and Moore, T. J. (2016). The white/black educational gap, stalled
progress, and the long-term consequences of the emergence of crack cocaine markets. The Review of
Economics and Statistics, 98 (5), 832–847.

Finkelstein, A., Gentkow, M. and Williams, H. (2021). Place-based drivers of mortality: Evidence
from migration. American Economic Review, 111, 2697–2735.

—, Taubman, S., Wright, B., Bernstein, M., Gruber, J., Newhouse, J. P., Allen, H., Baicker,
K. and Group, O. H. S. (2012). The Oregon health insurance experiment: Evidence from the first year.
Quarterly Journal of Economics, 127 (3), 1057–1106.

Fletcher, J., Schwarz, H. G., Engelman, M., Johnson, N., Hakes, J. and Palloni, A. (2022).
Understanding Geographic Disparities in Mortality. Tech. rep., NBER Working Paper 30572.

Fryer, R. G., Heaton, P. S., Levitt, S. D. and Murphy, K. M. (2013). Measuring crack cocaine
and its impact. Economic Inquiry, 51 (3), 1651–1681.

Golden, S. D., Ribisl, K. M. and Perreira, K. M. (2014). Economic and political influence on tobacco
tax rates: a nationwide analysis of 31 years of state data. American Journal of Public Health, 104 (2),
350–357.

Grant, J. T. and Kelly, N. J. (2008). Legislative productivity of the U.S. congress, 1789–2004. Political
Analysis, 16 (3), 303–323.

Grumbach, J. M. (2018). From backwaters to major policymakers: Policy polarization in the states,
1970–2014. Perspectives on Politics, 16 (2), 416–435.

Heyes, A. and Saberian, S. (2019). Temperature and decisions: Evidence from 207,000 court cases.
American Economic Journal: Applied Economics, 11 (2), 238–65.

Hibbs, D. A. (1977). Political parties and macroeconomic policy. American Political Science Review,
71 (4), 1467–1487.

— (1992). Partisan theory after fifteen years. European Journal of Political Economy, 8 (3), 361–373.

Kelly, N. J. (2018). Legislative productivity of the U.S. congress. Data retrieved from Harvard
Dataverse at https://doi.org/10.7910/DVN/ILILUD.

— and Witko, C. (2014). Government ideology and unemployment in the U.S. states. State Politics &
Policy Quarterly, 14 (4), 389–413.

30
Kim, D. (2016). The associations between US state and local social spending, income inequality, and
individual all-cause and cause-specific mortality: The National Longitudinal Mortality Study. Preventive
Medicine, 84, 62–68.

Kuziemko, I. and Washington, E. (2018). Why did the democrats lose the south? Bringing new data
to an old debate. American Economic Review, 108 (10), 2830–2867.

Lerman, P. (1985). Deinstitutionalization and welfare policies. The Annals of the American Academy of
Political and Social Science, 479, 132–155.

Machado, J. A. F. and Santos Silva, J. M. C. (2019). Quantiles via moments. Journal of Economet-
rics, 213 (1), 145–173.

Magkonis, G., Zekente, K. and Logothetis, V. (2021). Does the left spend more? An econometric
survey of partisan politics. Oxford Bulletin of Economics and Statistics, 83, 1077–1099.

Marshall, J. (forthcoming). Can close election regression discontinuity designs identify effects of winning
politician characteristics? American Journal of Political Science.

Mehta, N. K., Abrams, L. R. and Myrskylä, M. (2020). US life expectancy stalls due to cardiovascular
disease, not drug deaths. Proceedings of the National Academy of Sciences, 117 (13), 6998–7000.

Miller, D. L., Page, M. E., Stevens, A. H. and Filipski, M. (2009). Why are recessions good for
your health? American Economic Review, 99 (2), 122–127.

Montez, J. K. and Berkman, L. F. (2014). Trends in the educational gradient of mortality among us
adults aged 45 to 84 years: Bringing regional context into the explanation. American Journal of Public
Health, 104 (1), e82–e90.

Navarro, V., Muntañer, C., Borrell, C., Benach, J. and Pasarı́n, M. (2006). Politics and health
outcomes. The Lancet, 368, 1033–1037.

— and Shi, L. (2001). The political context of social inequalities and health. Social Science & Medicine,
52 (3), 481–491.

Nordhaus, W. D. (1975). The political business cycle. Review of Economic Studies, 42 (2), 169–190.

Peltzman, S. (2009). Mortality inequality. Journal of Economic Perspectives, 23 (4), 175–190.

Price, S. (1997). Political business cycles and macroeconomic credibility: A survey. Public Choice, 92,
407–427.

Reynolds, M. and Avendano, M. (2018). Social policy expenditures and life expectancy in high-income
countries. American Journal of Preventive Medicine, 54, 72–79.

Rogoff, K. (1990). Equilibrium political budget cycles. American Economic Review, 80 (1), 21–36.

Ruhm, C. J. (2000). Are recessions good for your health? Quarterly Journal of Economics, 115 (2),
617–650.

31
Santa-Clara, P. and Valkanov, R. (2003). The presidential puzzle: Political cycles and the stock
market. Journal of Finance, 58 (5), 1841–1872.

Schwandt, H., Currie, J., Bär, M., Banks, J., Bertoli, P., Bütikofer, A., Cattan, S., Chao,
B. Z.-Y., Costa, C., González, L., Grembi, V., Huttunen, K., Karadakic, R., Kraftman,
L., Krutikova, S., Lombardi, S., Redler, P., Riumallo-Herl, C., Rodrı́guez-González, A.,
Salvanes, K. G., Santana, P., Thuilliez, J., van Doorslaer, E., Van Ourti, T., Winter,
J. K., Wouterse, B. and Wuppermann, A. (2021). Inequality in mortality between black and white
americans by age, place, and cause and in comparison to europe, 1990 to 2018. Proceedings of the National
Academy of Sciences, 118 (40), e2104684118.

Strong, D. S. (1971). Further reflections on southern politics. Journal of Politics, 33 (2), 239–256.

Torche, F. and Rauf, T. (2021). The political context and infant health in the United States. American
Sociological Review, 86 (3), 377–405.

Vierboom, Y. C., Preston, S. H. and Hendi, A. S. (2019). Rising geographic inequality in mortality
in the United States. SSM-Population Health, 9, 100478.

Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data. MIT Press, 2nd edn.

World Health Organization (2020). The top 10 causes of death. Tech. rep.

32
Partisan Mortality Cycles
Supplemental Appendices

Daniel L. Millimet & Travis Whitacre

August 9, 2024
Appendix A Supplemental Figures
Appendix A.1 Descriptive Figures

Figure A1: Geographic Variation in Historical Age-Adjusted All-Cause Mortality Rates

(A) 1970 (B) 1980

(C) 1990 (D) 2000

(E) 2010 (F) 2016

1
Figure A2: Geographic Variation in Historical Age-Adjusted “Short-Term”-Cause Mortality Rates

(A) 1970 (B) 1980

(C) 1990 (D) 2000

(E) 2010 (F) 2016

2
Appendix A.2 Omit Southern Region

3
Figure A3: Partisan Determinants of All-Cause Mortality Omitting the Southeast Region

(A) Executive

(B) Legislative

(C) Full

Notes: Mortality is measured in period 0. Periods -1 to -4 capture the associations between lagged
political variables and current mortality; periods 1 to 3 capture the associations between future
political variables and current mortality. Definition of the Southeast region comes from the US
Bureau of Economic Analysis. 4
Appendix A.3 By Race & Sex

5
Figure A4: Partisan and Ideological Determinants of All-Cause Mortality for Black Males

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

6
Figure A5: Partisan and Ideological Determinants of “Short-Term”-Cause Mortality for Black Males

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

7
Figure A6: Partisan and Ideological Determinants of All-Cause Mortality for Black Females

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

8
Figure A7: Partisan and Ideological Determinants of “Short-Term”-Cause Mortality for Black Females

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

9
Figure A8: Partisan and Ideological Determinants of All-Cause Mortality for White Males

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

10
Figure A9: Partisan and Ideological Determinants of “Short-Term”-Cause Mortality for White Males

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

11
Figure A10: Partisan and Ideological Determinants of All-Cause Mortality for White Females

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

12
Figure A11: Partisan and Ideological Determinants of “Short-Term”-Cause Mortality for White Females

(A) Executive

(B) Legislative

(C) Full
Notes: Results are for the full sample period, 1968–2016. Partisan results are shown in the left
column; ideological results are shown in the right column. Mortality is measured in period 0.
Periods -1 to -4 capture the associations between lagged political variables and current mortality;
periods 1 to 3 capture the associations between future political variables and current mortality.

13
Appendix A.4 By Political Party

14
 Figure A12: Partisan Determinants of Mortality

(A) Executive

(B) Legislative

(C) Full
Notes: All-cause mortality results are shown in the left column; “Short-Term”-cause results are
shown in the right column. Mortality is measured in period 0. Periods -1 to -4 capture the
associations between lagged political variables and current mortality; periods 1 to 3 capture the
associations between future political variables and current mortality.

15
Appendix B Theoretical Derivation
We follow Chetty et al. (2016) and Finkelstein et al. (2021) and specify the log mortality hazard rate
of individual i at age a residing in location c at time t as

log[mict (a)] = θi + βa + γct , (Appendix B.1)

where θ is individual-level, time invariant health capital and γ is location- and time-specific health capital
(referred to as the place effect in Finkelstein et al. (2021) and Couillard et al. (2021)). Let θct denote the
average individual-specific health capital of residents of location c at time t. Note, θct will generally vary
over time despite θi being time invariant due to changes in the local population through migration, fertility,
and past mortality (Deryugina and Molitor, 2021).
We define the mortality rate of an average individual residing in location c at time t at age a as


mct (a) = exp θct + βa + γct . (Appendix B.2)

The age-adjusted mortality rate in county c at time t is given by

Z


mct = exp θct + βa + γct fs (a)da (Appendix B.3)
a Z


= exp θct + γct exp (βa) fs (a)da
a

where fs (a) is the age distribution of the standard population.

Suppose age in the standard population is distributed as χ2 (k), a chi-squared distribution with k degrees
of freedom. This implies
"
# Z
exp θct + γct A   
k 1
mct = a( 2 −1) exp
β− a da (Appendix B.4)
2k/2 Γ(k/2) 1 2
k A
"
# 2
−1 k


−
   X
exp θct + γct 1 k 1 !
= exp β − a  (−1) 2 −1−j 2

k a
j
2k/2 Γ(k/2) 2
j=0 j! β − 2 1 2 −j
 " #1 
k k
−1 k
−1−j ( 2 −1)!
 exp β − 12 A
 
Aj

 P 2

#  j=0 (−1)
"  2


k
1 2 −j

exp θct + γct  j!(β− 2 )

= " #
2k/2 Γ(k/2)   1

 P k
−1 k
−1−j ( k
−1) ! 
− exp β − 2 2
j=0 (−1)
2

 2


 k −j 
j!(β− 12 ) 2
 
k 
"
# 2
−1 k


−1 !
   X  
exp θct + γct 1 k
−1−j 1
= exp β − A  (−1) 2 2

k Aj − 
2k/2 Γ(k/2) 2
j=0 j! β − 2 1 2 −j exp (A)

16
where Γ(·) is the gamma function. Taking logs yields
   
log 2 k 1
log mct = θct + γct − k − log Γ + β− A (Appendix B.5)
2 2 2
k 
2
−1 k


2 −1 !
 
X k 1
+ log  (−1) 2 −1−j k Aj − 
j! β − 21 2

−j exp (A)
j=0

= βe + θct + γct

where
   
1 log 2 k
βe ≡ β− A− k − log Γ
2 2 2
k 
2
−1 k


−
 
X k 1 ! 1
+ log  (−1) 2 −1−j 2

k Aj − .
j! β − 21 2 −j exp (A)
j=0

Thus, the log age-adjusted mortality rate in county c at time t depends on a constant term, the average
health capital of residents, θct , and the place effect, γct .

17
Appendix C Supplemental Estimation Results: Coefficients
Appendix C.1 Baseline Specifications

Table C1: Ideological Determinants of All-Cause Mortality

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
State Liberalism 0.0027 0.0065** 0.0015 -0.0037 0.0013 0.0010 0.0007
(0.0043) (0.0033) (0.0032) (0.0032) (0.0033) (0.0029) (0.0029)
State × Presidential Liberalism -0.0009 -0.0036*** -0.0014 -0.0013 -0.0015 -0.0015* -0.0020**
(0.0011) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009)
State × Federal Legislative Liberalism 0.0063* 0.0031 0.0000 0.0035 0.0001 -0.0000 0.0003
(0.0033) (0.0032) (0.0032) (0.0032) (0.0032) (0.0032) (0.0032)
State × Presidential × -0.0348*** 0.0119** -0.0082 -0.0072 -0.0085 -0.0023 0.0002
Federal Legislative Liberalism (0.0069) (0.0057) (0.0056) (0.0057) (0.0057) (0.0055) (0.0055)
Lagged Effects
State Liberalism 0.0010 -0.0040*** -0.0030*** 0.0002 -0.0029*** -0.0022*** -0.0021***
(0.0019) (0.0010) (0.0008) (0.0009) (0.0008) (0.0008) (0.0008)
State × Presidential Liberalism -0.0026*** 0.0045*** 0.0011*** 0.0006* 0.0010*** 0.0010*** 0.0009***
(0.0006) (0.0004) (0.0004) (0.0004) (0.0004) (0.0003) (0.0003)
State × Federal Legislative Liberalism 0.0009 0.0214*** 0.0083*** 0.0067*** 0.0082*** 0.0069*** 0.0069***
(0.0021) (0.0013) (0.0012) (0.0012) (0.0012) (0.0012) (0.0012)
State × Presidential × 0.0083** -0.0196*** -0.0063*** -0.0002 -0.0060*** -0.0023 -0.0017
Federal Legislative Liberalism (0.0040) (0.0021) (0.0020) (0.0019) (0.0020) (0.0020) (0.0020)
Lead Effects
State Liberalism 0.0207*** -0.0063*** -0.0014 -0.0024* -0.0016 0.0012 0.0011
(0.0021) (0.0012) (0.0011) (0.0012) (0.0011) (0.0010) (0.0010)
State × Presidential Liberalism -0.0001 -0.0031*** -0.0010*** -0.0003 -0.0010** -0.0003 -0.0001
(0.0006) (0.0004) (0.0004) (0.0004) (0.0004) (0.0004) (0.0004)
State × Federal Legislative Liberalism -0.0180*** 0.0286*** 0.0036** 0.0004 0.0033* 0.0043*** 0.0040**
(0.0042) (0.0019) (0.0018) (0.0018) (0.0018) (0.0016) (0.0016)
State × Presidential × -0.0126*** -0.0066*** -0.0066*** -0.0063*** -0.0067*** -0.0035 -0.0040*
Federal Legislative Liberalism (0.0026) (0.0023) (0.0022) (0.0022) (0.0023) (0.0022) (0.0022)

Observations 129300 129300 129300 129300 129300 117504 117504

R2 0.017 0.469 0.488 0.491 0.554 0.617 0.617
Location Health Controls N N N N N Y Y
Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal).
Lead values correspond to the effects over the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.

18
 Table C2: Ideological Determinants of “Short-Term”-Cause Mortality

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
State Liberalism 0.0250*** 0.0038 -0.0010 -0.0072 -0.0022 0.0005 -0.0005
(0.0088) (0.0088) (0.0087) (0.0087) (0.0088) (0.0083) (0.0082)
State × Presidential Liberalism 0.0005 -0.0072*** -0.0067*** -0.0065*** -0.0066*** -0.0062*** -0.0076***
(0.0023) (0.0023) (0.0023) (0.0023) (0.0023) (0.0022) (0.0022)
State × Federal Legislative Liberalism 0.0508*** 0.0144* 0.0133* 0.0154** 0.0129* 0.0088 0.0107
(0.0076) (0.0076) (0.0077) (0.0078) (0.0077) (0.0078) (0.0078)
State × Presidential × -0.0941*** -0.0203 -0.0286** -0.0352** -0.0285** -0.0214 -0.0134
Federal Legislative Liberalism (0.0146) (0.0140) (0.0139) (0.0140) (0.0139) (0.0137) (0.0138)
Lagged Effects
State Liberalism -0.0125*** -0.0151*** -0.0080*** -0.0031 -0.0088*** -0.0082*** -0.0077***
(0.0026) (0.0021) (0.0021) (0.0023) (0.0021) (0.0021) (0.0021)
State × Presidential Liberalism -0.0022** 0.0003 -0.0006 -0.0012 -0.0006 -0.0008 -0.0009
(0.0009) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009)
State × Federal Legislative Liberalism 0.0125*** 0.0220*** 0.0196*** 0.0213*** 0.0202*** 0.0177*** 0.0180***
(0.0029) (0.0028) (0.0030) (0.0030) (0.0030) (0.0030) (0.0030)
State × Presidential × 0.0457*** -0.0028 -0.0050 -0.0036 -0.0054 -0.0088* -0.0068
Federal Legislative Liberalism (0.0063) (0.0052) (0.0052) (0.0050) (0.0052) (0.0053) (0.0053)
Lead Effects
State Liberalism -0.0023 -0.0074*** -0.0008 -0.0026 0.0003 0.0023 0.0020
(0.0027) (0.0026) (0.0028) (0.0031) (0.0028) (0.0028) (0.0028)
State × Presidential Liberalism -0.0046*** 0.0014 0.0008 0.0002 0.0008 0.0012 0.0018*
(0.0009) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009)
State × Federal Legislative Liberalism -0.0365*** 0.0025 -0.0016 -0.0045 -0.0004 -0.0026 -0.0034
(0.0050) (0.0042) (0.0042) (0.0042) (0.0042) (0.0041) (0.0041)
State × Presidential × -0.0150*** -0.0251*** -0.0261*** -0.0257*** -0.0258*** -0.0233*** -0.0253***
Federal Legislative Liberalism (0.0058) (0.0057) (0.0057) (0.0057) (0.0057) (0.0055) (0.0055)

Observations 113261 113261 113261 113261 113261 105542 105542

R2 0.002 0.101 0.117 0.121 0.206 0.225 0.225
Location Health Controls N N N N N Y Y
Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal).
Lead values correspond to the effects over the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.

19
Appendix C.2 Race & Sex

20
 Table C3: Ideological Determinants of All-Cause Mortality for Black Males

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism -0.0191 -0.0185 -0.1536 -0.1590 -0.0190 -0.0213 0.0225 0.0228
(0.0194) (0.0194) (0.1212) (0.1219) (0.0220) (0.0220) (0.0286) (0.0287)
State × Presidential Liberalism 0.0034 0.0033 -0.1771 -0.1956 -0.0009 0.0026 -0.0175 -0.0164
(0.0046) (0.0047) (0.1685) (0.1713) (0.0369) (0.0377) (0.0167) (0.0168)
State × Federal Legislative Liberalism -0.0274* -0.0281* 0.3701 0.3985 -0.0352 -0.0249 0.0225 0.0265
(0.0149) (0.0151) (0.5766) (0.5827) (0.0893) (0.0910) (0.0401) (0.0400)
State × Presidential × 0.0090 0.0091 1.1239 1.2318 0.0876 0.0730 0.0601 0.0529
Federal Legislative Liberalism (0.0306) (0.0310) (0.9512) (0.9684) (0.1949) (0.1992) (0.0714) (0.0725)
Lagged Effects
State Liberalism -0.0070* -0.0071* 0.0590 0.0567 -0.0102 -0.0106 -0.0152 -0.0137
(0.0041) (0.0041) (0.0603) (0.0606) (0.0072) (0.0074) (0.0144) (0.0144)
State × Presidential Liberalism 0.0002 0.0002 0.1155 0.1136 -0.0293* -0.0296* 0.0042 0.0047
(0.0019) (0.0019) (0.1361) (0.1371) (0.0150) (0.0153) (0.0056) (0.0056)
State × Federal Legislative Liberalism 0.0055 0.0051 -0.3203 -0.3171 -0.0036 -0.0063 0.0007 0.0062
(0.0058) (0.0060) (0.3004) (0.3024) (0.0277) (0.0291) (0.0262) (0.0267)
State × Presidential × -0.0265** -0.0268** -0.9678 -0.9845 0.1758* 0.1831* -0.0567 -0.0439
Federal Legislative Liberalism (0.0108) (0.0108) (0.8469) (0.8530) (0.0951) (0.0972) (0.0763) (0.0773)
Lead Effects
State Liberalism 0.0072 0.0071 -0.1824** -0.1930** 0.0042 0.0042 -0.0302** -0.0274*
(0.0062) (0.0062) (0.0790) (0.0800) (0.0126) (0.0126) (0.0144) (0.0144)
State × Presidential Liberalism 0.0008 0.0009 -0.3202** -0.3426** 0.0007 0.0012 -0.0099 -0.0102

21
(0.0019) (0.0019) (0.1323) (0.1337) (0.0048) (0.0051) (0.0114) (0.0114)
State × Federal Legislative Liberalism 0.0118 0.0114 0.8396** 0.8902** 0.0262 0.0276 0.0261 0.0264
(0.0084) (0.0085) (0.3555) (0.3588) (0.0289) (0.0289) (0.0211) (0.0212)
State × Presidential × -0.0118 -0.0118 1.6368** 1.7614** 0.0354 0.0354 0.0198 0.0163
Federal Legislative Liberalism (0.0116) (0.0116) (0.7487) (0.7564) (0.0406) (0.0415) (0.0416) (0.0421)

Observations 54787 54787 22236 22236 18161 18161 14390 14390

R2 0.201 0.201 0.173 0.173 0.312 0.312 0.262 0.263
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status.
Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over
the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table C4: Ideological Determinants of “Short-Term”-Cause Mortality for Black Males

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism -0.0236 -0.0329 0.0800 0.1199 -0.1728*** -0.1595*** 0.0848 0.0896
(0.0401) (0.0399) (0.1892) (0.1921) (0.0490) (0.0490) (0.0584) (0.0588)
State × Presidential Liberalism -0.0434*** -0.0420*** 0.0346 0.1764 -0.0197 -0.0378 -0.0610** -0.0597**
(0.0098) (0.0101) (0.2792) (0.2882) (0.0631) (0.0636) (0.0293) (0.0294)
State × Federal Legislative Liberalism -0.0570* -0.0478* -0.3636 -0.7014 -0.0570 -0.0843 0.1113* 0.1061
(0.0292) (0.0287) (0.8940) (0.9123) (0.1277) (0.1268) (0.0668) (0.0675)
State × Presidential × 0.1621*** 0.1426** -0.1543 -1.0238 -0.1238 -0.0721 0.1867 0.2044
Federal Legislative Liberalism (0.0580) (0.0591) (1.5167) (1.5704) (0.3071) (0.3063) (0.1372) (0.1381)
Lagged Effects
State Liberalism -0.0352*** -0.0336*** -0.0060 0.0386 -0.0366* -0.0455** 0.0246 0.0239
(0.0106) (0.0107) (0.0955) (0.0959) (0.0188) (0.0187) (0.0240) (0.0243)
State × Presidential Liberalism -0.0009 -0.0004 -0.1526 -0.0368 -0.1545*** -0.1499*** 0.0052 0.0067
(0.0040) (0.0040) (0.2197) (0.2205) (0.0376) (0.0378) (0.0101) (0.0102)
State × Federal Legislative Liberalism 0.0447*** 0.0501*** 0.3566 0.1026 -0.1292* -0.0954 0.0722* 0.0670
(0.0149) (0.0153) (0.4691) (0.4702) (0.0695) (0.0717) (0.0435) (0.0444)
State × Presidential × -0.1432*** -0.1435*** 0.9770 0.3049 0.5646*** 0.5082** -0.0158 -0.0160
Federal Legislative Liberalism (0.0267) (0.0262) (1.4299) (1.4337) (0.2029) (0.2063) (0.1227) (0.1235)
Lead Effects
State Liberalism -0.0664*** -0.0668*** 0.1694 0.1657 -0.0094 -0.0239 -0.0208 -0.0242
(0.0160) (0.0160) (0.1369) (0.1385) (0.0251) (0.0250) (0.0240) (0.0240)
State × Presidential Liberalism 0.0129*** 0.0115** 0.3078 0.3527 0.0010 -0.0095 -0.0049 -0.0043

22
(0.0046) (0.0048) (0.2342) (0.2378) (0.0119) (0.0120) (0.0174) (0.0174)
State × Federal Legislative Liberalism 0.0285 0.0287 -0.8213 -0.8563 -0.0646 -0.0592 0.0223 0.0234
(0.0178) (0.0181) (0.6077) (0.6153) (0.0603) (0.0597) (0.0346) (0.0348)
State × Presidential × -0.0783*** -0.0784*** -1.4644 -1.8458 -0.0547 -0.0521 -0.0056 0.0128
Federal Legislative Liberalism (0.0204) (0.0207) (1.2447) (1.2663) (0.0779) (0.0813) (0.0578) (0.0586)

Observations 19738 19738 11197 11197 4413 4413 4128 4128

R2 0.387 0.388 0.448 0.449 0.398 0.402 0.317 0.318
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table C5: Ideological Determinants of All-Cause Mortality for Black Females

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism -0.0244 -0.0257 -0.0323 -0.0365 -0.0288 -0.0285 0.0054 0.0026
(0.0183) (0.0183) (0.1189) (0.1194) (0.0223) (0.0219) (0.0251) (0.0250)
State × Presidential Liberalism -0.0008 -0.0020 -0.0897 -0.0915 -0.0064 -0.0086 0.0249* 0.0251*
(0.0046) (0.0047) (0.1658) (0.1679) (0.0313) (0.0322) (0.0146) (0.0147)
State × Federal Legislative Liberalism -0.0277** -0.0241* -0.1997 -0.2061 0.0186 0.0163 -0.0464 -0.0480
(0.0133) (0.0137) (0.5708) (0.5757) (0.0715) (0.0733) (0.0334) (0.0339)
State × Presidential × 0.0043 0.0095 0.6528 0.6633 0.0327 0.0424 0.0107 0.0015
Federal Legislative Liberalism (0.0286) (0.0292) (0.9250) (0.9385) (0.1541) (0.1577) (0.0634) (0.0646)
Lagged Effects
State Liberalism -0.0002 -0.0000 0.0403 0.0404 -0.0007 -0.0002 -0.0108 -0.0115
(0.0038) (0.0038) (0.0598) (0.0601) (0.0069) (0.0071) (0.0117) (0.0118)
State × Presidential Liberalism 0.0027 0.0027 0.0577 0.0621 0.0309** 0.0312** 0.0075 0.0066
(0.0017) (0.0017) (0.1314) (0.1322) (0.0141) (0.0143) (0.0047) (0.0049)
State × Federal Legislative Liberalism 0.0082 0.0095 -0.1944 -0.2055 0.0731*** 0.0732*** 0.0376* 0.0360
(0.0056) (0.0058) (0.2961) (0.2976) (0.0261) (0.0273) (0.0218) (0.0223)
State × Presidential × -0.0111 -0.0099 -0.5259 -0.5671 -0.1363 -0.1392 0.1484** 0.1436**
Federal Legislative Liberalism (0.0090) (0.0091) (0.8175) (0.8221) (0.0869) (0.0889) (0.0603) (0.0610)
Lead Effects
State Liberalism 0.0030 0.0032 -0.1100 -0.1174 0.0188 0.0190 -0.0008 -0.0013
(0.0057) (0.0057) (0.0742) (0.0747) (0.0119) (0.0120) (0.0124) (0.0124)
State × Presidential Liberalism 0.0004 0.0007 -0.2268* -0.2367* -0.0080* -0.0079* 0.0150* 0.0148*

23
(0.0019) (0.0019) (0.1245) (0.1259) (0.0041) (0.0045) (0.0085) (0.0086)
State × Federal Legislative Liberalism 0.0115 0.0105 0.5085 0.5347 -0.0051 -0.0037 -0.0077 -0.0092
(0.0079) (0.0080) (0.3350) (0.3369) (0.0253) (0.0254) (0.0162) (0.0163)
State × Presidential × -0.0049 -0.0058 1.3159* 1.3631* -0.0044 -0.0082 0.0343 0.0302
Federal Legislative Liberalism (0.0113) (0.0113) (0.6964) (0.7062) (0.0391) (0.0397) (0.0309) (0.0314)

Observations 52634 52634 21214 21214 17507 17507 13913 13913

R2 0.177 0.177 0.175 0.175 0.303 0.303 0.264 0.264
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status.
Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over
the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table C6: Ideological Determinants of “Short-Term”-Cause Mortality for Black Females

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism 0.0330 0.0259 -0.3417 -0.3743 0.0091 0.0114 -0.0000 -0.0047
(0.0420) (0.0418) (0.2355) (0.2397) (0.0535) (0.0534) (0.0599) (0.0606)
State × Presidential Liberalism -0.0293** -0.0313*** -0.6188* -0.6724* -0.0192 -0.0309 -0.0047 -0.0119
(0.0114) (0.0115) (0.3505) (0.3589) (0.0667) (0.0666) (0.0310) (0.0309)
State × Federal Legislative Liberalism -0.0304 -0.0260 2.0539* 2.1663* -0.0862 -0.1136 0.0820 0.0719
(0.0294) (0.0296) (1.1007) (1.1258) (0.1375) (0.1369) (0.0742) (0.0755)
State × Presidential × 0.0594 0.0757 2.9767 3.2688* -0.0184 0.0339 -0.0231 0.0544
Federal Legislative Liberalism (0.0651) (0.0662) (1.9015) (1.9432) (0.3083) (0.3103) (0.1548) (0.1584)
Lagged Effects
State Liberalism -0.0324*** -0.0316*** 0.0238 0.0310 -0.0341* -0.0349* 0.0535* 0.0503
(0.0121) (0.0121) (0.1188) (0.1194) (0.0191) (0.0193) (0.0308) (0.0312)
State × Presidential Liberalism 0.0000 -0.0011 0.0509 0.0829 -0.1570*** -0.1573*** 0.0133 0.0154
(0.0043) (0.0044) (0.2636) (0.2642) (0.0363) (0.0365) (0.0115) (0.0120)
State × Federal Legislative Liberalism 0.0355*** 0.0339** 0.0105 -0.0788 -0.2050*** -0.2032*** 0.0368 0.0325
(0.0131) (0.0135) (0.5695) (0.5744) (0.0688) (0.0690) (0.0474) (0.0486)
State × Presidential × -0.0986*** -0.0892*** -0.6574 -0.9386 0.6369*** 0.6223*** 0.0365 0.0259
Federal Legislative Liberalism (0.0258) (0.0254) (1.6985) (1.7026) (0.2128) (0.2101) (0.1282) (0.1317)
Lead Effects
State Liberalism -0.0193 -0.0190 -0.0190 -0.0684 0.0004 -0.0035 0.0112 0.0103
(0.0160) (0.0161) (0.1678) (0.1686) (0.0244) (0.0243) (0.0314) (0.0312)
State × Presidential Liberalism 0.0089** 0.0105** -0.1110 -0.1818 0.0237** 0.0235** -0.0236 -0.0197

24
(0.0044) (0.0045) (0.2774) (0.2791) (0.0109) (0.0111) (0.0200) (0.0202)
State × Federal Legislative Liberalism 0.0364* 0.0299 0.1636 0.3570 0.0027 0.0062 0.0603 0.0664
(0.0191) (0.0193) (0.7292) (0.7350) (0.0555) (0.0554) (0.0399) (0.0405)
State × Presidential × -0.0217 -0.0315 0.4392 0.7915 -0.0878 -0.0983 -0.0009 0.0190
Federal Legislative Liberalism (0.0253) (0.0257) (1.4688) (1.4755) (0.0825) (0.0826) (0.0776) (0.0804)

Observations 13825 13825 8109 8109 3067 3067 2649 2649

R2 0.254 0.255 0.282 0.283 0.380 0.381 0.335 0.339
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table C7: Ideological Determinants of All-Cause Mortality for White Males

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism 0.0024 0.0022 -0.0156 -0.0199 0.0001 -0.0000 0.0014 0.0002
(0.0043) (0.0043) (0.0271) (0.0271) (0.0072) (0.0072) (0.0090) (0.0090)
State × Presidential Liberalism -0.0033** -0.0038*** -0.0242 -0.0339 -0.0122 -0.0123 -0.0063 -0.0074
(0.0014) (0.0014) (0.0411) (0.0411) (0.0093) (0.0094) (0.0059) (0.0059)
State × Federal Legislative Liberalism -0.0030 -0.0028 -0.0006 0.0194 -0.0209 -0.0217 0.0000 -0.0042
(0.0047) (0.0048) (0.1324) (0.1321) (0.0201) (0.0203) (0.0139) (0.0140)
State × Presidential × 0.0086 0.0114 0.2094 0.2622 0.0607 0.0629 0.0262 0.0300
Federal Legislative Liberalism (0.0084) (0.0084) (0.2327) (0.2326) (0.0418) (0.0423) (0.0285) (0.0288)
Lagged Effects
State Liberalism -0.0010 -0.0009 0.0005 0.0003 0.0050** 0.0051** 0.0003 -0.0008
(0.0010) (0.0010) (0.0157) (0.0157) (0.0025) (0.0025) (0.0040) (0.0040)
State × Presidential Liberalism 0.0009* 0.0009* -0.0121 -0.0113 -0.0099** -0.0103** 0.0028 0.0022
(0.0005) (0.0005) (0.0318) (0.0317) (0.0043) (0.0043) (0.0023) (0.0023)
State × Federal Legislative Liberalism 0.0062*** 0.0062*** -0.0188 -0.0213 -0.0144 -0.0155 0.0040 0.0016
(0.0017) (0.0018) (0.0740) (0.0739) (0.0093) (0.0095) (0.0084) (0.0084)
State × Presidential × -0.0064** -0.0058** 0.0555 0.0389 0.0253 0.0289 -0.0006 -0.0072
Federal Legislative Liberalism (0.0027) (0.0028) (0.1889) (0.1886) (0.0260) (0.0263) (0.0231) (0.0233)
Lead Effects
State Liberalism 0.0013 0.0011 -0.0226 -0.0263 -0.0024 -0.0025 -0.0046 -0.0055
(0.0014) (0.0014) (0.0169) (0.0172) (0.0033) (0.0033) (0.0042) (0.0042)
State × Presidential Liberalism 0.0003 0.0005 -0.0382 -0.0467 0.0032** 0.0034** -0.0010 -0.0002

25
(0.0005) (0.0005) (0.0289) (0.0293) (0.0016) (0.0016) (0.0031) (0.0031)
State × Federal Legislative Liberalism 0.0015 0.0011 0.0668 0.0857 0.0002 0.0000 0.0054 0.0043
(0.0024) (0.0024) (0.0751) (0.0761) (0.0090) (0.0090) (0.0058) (0.0058)
State × Presidential × -0.0056* -0.0061* 0.2535 0.2983* -0.0073 -0.0064 0.0164 0.0162
Federal Legislative Liberalism (0.0032) (0.0032) (0.1664) (0.1681) (0.0124) (0.0124) (0.0147) (0.0148)

Observations 112829 112829 33595 33595 40008 40008 39226 39226

R2 0.582 0.582 0.326 0.326 0.264 0.264 0.241 0.241
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table C8: Ideological Determinants of “Short-Term”-Cause Mortality for White Males

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism 0.0277* 0.0320** -0.2275*** -0.2605*** -0.0148 -0.0135 -0.0010 -0.0010
(0.0150) (0.0150) (0.0794) (0.0792) (0.0247) (0.0244) (0.0188) (0.0189)
State × Presidential Liberalism -0.0251*** -0.0225*** -0.2419** -0.2543** -0.1255*** -0.1103*** 0.0112 0.0115
(0.0040) (0.0041) (0.1168) (0.1167) (0.0276) (0.0276) (0.0109) (0.0110)
State × Federal Legislative Liberalism -0.0628*** -0.0671*** 1.3016*** 1.2570*** -0.2660*** -0.2312*** -0.0619** -0.0651**
(0.0131) (0.0130) (0.3913) (0.3884) (0.0584) (0.0588) (0.0255) (0.0258)
State × Presidential × 0.1736*** 0.1585*** 1.1662* 1.2564* 0.3705*** 0.2894** -0.0087 -0.0125
Federal Legislative Liberalism (0.0229) (0.0232) (0.6446) (0.6437) (0.1281) (0.1279) (0.0545) (0.0547)
Lagged Effects
State Liberalism -0.0487*** -0.0496*** 0.0775* 0.0793* 0.0069 0.0010 0.0046 0.0032
(0.0045) (0.0045) (0.0441) (0.0443) (0.0102) (0.0102) (0.0079) (0.0080)
State × Presidential Liberalism 0.0052*** 0.0055*** 0.0668 0.0846 -0.0937*** -0.0908*** 0.0040 0.0032
(0.0017) (0.0017) (0.0948) (0.0949) (0.0153) (0.0154) (0.0044) (0.0045)
State × Federal Legislative Liberalism -0.0263*** -0.0282*** -0.2358 -0.2869 -0.1383*** -0.1263*** -0.0116 -0.0131
(0.0080) (0.0081) (0.2116) (0.2119) (0.0284) (0.0287) (0.0168) (0.0170)
State × Presidential × -0.0974*** -0.1023*** -0.1375 -0.3389 0.1744* 0.1686* 0.0230 0.0202
Federal Legislative Liberalism (0.0117) (0.0116) (0.5961) (0.5979) (0.0927) (0.0941) (0.0479) (0.0481)
Lead Effects
State Liberalism -0.0074 -0.0070 0.1356*** 0.0785 0.0368*** 0.0339*** 0.0033 0.0031
(0.0064) (0.0064) (0.0494) (0.0506) (0.0127) (0.0127) (0.0084) (0.0085)
State × Presidential Liberalism 0.0258*** 0.0247*** 0.2985*** 0.2273*** 0.0105** 0.0053 0.0055 0.0060

26
(0.0021) (0.0022) (0.0863) (0.0879) (0.0047) (0.0047) (0.0062) (0.0062)
State × Federal Legislative Liberalism 0.0317*** 0.0332*** -0.7139*** -0.5309** -0.0324 -0.0311 -0.0016 -0.0035
(0.0075) (0.0076) (0.2254) (0.2289) (0.0263) (0.0263) (0.0116) (0.0117)
State × Presidential × -0.0282*** -0.0237** -1.6269*** -1.2700*** -0.1618*** -0.1379*** 0.0221 0.0204
Federal Legislative Liberalism (0.0094) (0.0096) (0.4715) (0.4798) (0.0353) (0.0358) (0.0280) (0.0280)

Observations 71277 71277 31751 31751 17966 17966 21560 21560

R2 0.505 0.505 0.622 0.624 0.388 0.390 0.152 0.152
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three
years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table C9: Ideological Determinants of All-Cause Mortality for White Females

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism 0.0012 0.0012 -0.0038 -0.0033 -0.0001 0.0006 0.0131 0.0124
(0.0045) (0.0045) (0.0276) (0.0278) (0.0075) (0.0075) (0.0092) (0.0092)
State × Presidential Liberalism -0.0006 -0.0008 0.0230 0.0161 -0.0088 -0.0116 0.0037 0.0035
(0.0014) (0.0014) (0.0415) (0.0417) (0.0101) (0.0101) (0.0057) (0.0057)
State × Federal Legislative Liberalism -0.0007 -0.0006 -0.0259 -0.0037 -0.0348 -0.0402* 0.0047 0.0033
(0.0048) (0.0048) (0.1380) (0.1382) (0.0224) (0.0227) (0.0141) (0.0143)
State × Presidential × -0.0133 -0.0121 -0.1554 -0.1224 0.0144 0.0285 0.0207 0.0198
Federal Legislative Liberalism (0.0083) (0.0084) (0.2346) (0.2361) (0.0464) (0.0469) (0.0282) (0.0284)
Lagged Effects
State Liberalism -0.0019* -0.0019* 0.0325** 0.0331** -0.0013 -0.0007 0.0078** 0.0074*
(0.0011) (0.0011) (0.0165) (0.0165) (0.0025) (0.0025) (0.0039) (0.0040)
State × Presidential Liberalism 0.0005 0.0005 0.0528 0.0535 -0.0110** -0.0112** 0.0035 0.0031
(0.0005) (0.0005) (0.0343) (0.0343) (0.0044) (0.0045) (0.0023) (0.0024)
State × Federal Legislative Liberalism 0.0066*** 0.0066*** -0.1769** -0.1757** -0.0156* -0.0157* -0.0021 -0.0030
(0.0018) (0.0018) (0.0793) (0.0792) (0.0091) (0.0092) (0.0088) (0.0089)
State × Presidential × -0.0061** -0.0059** -0.2918 -0.2982 0.0268 0.0278 -0.0032 -0.0065
Federal Legislative Liberalism (0.0030) (0.0030) (0.2065) (0.2064) (0.0284) (0.0289) (0.0242) (0.0244)
Lead Effects
State Liberalism -0.0019 -0.0019 -0.0065 -0.0038 0.0042 0.0035 0.0042 0.0040
(0.0015) (0.0015) (0.0191) (0.0194) (0.0033) (0.0033) (0.0043) (0.0043)
State × Presidential Liberalism -0.0011** -0.0010* 0.0083 0.0078 0.0039*** 0.0042*** -0.0039 -0.0038

27
(0.0005) (0.0005) (0.0318) (0.0323) (0.0015) (0.0015) (0.0031) (0.0031)
State × Federal Legislative Liberalism 0.0037 0.0034 -0.0479 -0.0479 -0.0011 -0.0020 0.0168*** 0.0162***
(0.0025) (0.0025) (0.0877) (0.0888) (0.0088) (0.0088) (0.0061) (0.0062)
State × Presidential × -0.0061* -0.0063* -0.0557 -0.0532 -0.0021 -0.0012 0.0137 0.0127
Federal Legislative Liberalism (0.0032) (0.0033) (0.1808) (0.1834) (0.0137) (0.0137) (0.0143) (0.0144)

Observations 112594 112594 33590 33590 39912 39912 39092 39092

R2 0.298 0.298 0.289 0.289 0.168 0.169 0.192 0.192
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status.
Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over
the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table C10: Ideological Determinants of “Short-Term”-Cause Mortality for White Females

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
State Liberalism 0.0172 0.0152 -0.3642*** -0.3819*** -0.0293 -0.0274 -0.0050 -0.0059
(0.0158) (0.0157) (0.0880) (0.0883) (0.0237) (0.0236) (0.0222) (0.0222)
State × Presidential Liberalism -0.0221*** -0.0213*** -0.5035*** -0.5070*** -0.1278*** -0.1212*** 0.0076 0.0075
(0.0042) (0.0043) (0.1320) (0.1326) (0.0285) (0.0287) (0.0138) (0.0139)
State × Federal Legislative Liberalism -0.0285** -0.0276** 2.0179*** 1.9913*** -0.2288*** -0.2103*** 0.0332 0.0267
(0.0139) (0.0138) (0.4371) (0.4371) (0.0613) (0.0618) (0.0326) (0.0333)
State × Presidential × 0.0524** 0.0460* 2.3953*** 2.4273*** 0.4006*** 0.3586*** -0.0507 -0.0514
Federal Legislative Liberalism (0.0241) (0.0246) (0.7337) (0.7376) (0.1320) (0.1324) (0.0645) (0.0652)
Lagged Effects
State Liberalism -0.0379*** -0.0372*** 0.0459 0.0461 -0.0027 -0.0042 -0.0090 -0.0107
(0.0043) (0.0044) (0.0480) (0.0481) (0.0094) (0.0094) (0.0101) (0.0101)
State × Presidential Liberalism 0.0004 0.0006 0.0995 0.1074 -0.0719*** -0.0661*** 0.0055 0.0044
(0.0016) (0.0016) (0.1019) (0.1020) (0.0161) (0.0162) (0.0050) (0.0051)
State × Federal Legislative Liberalism -0.0044 -0.0042 -0.1819 -0.2041 -0.0846*** -0.0692** 0.0033 0.0003
(0.0068) (0.0069) (0.2278) (0.2280) (0.0302) (0.0305) (0.0204) (0.0207)
State × Presidential × -0.0617*** -0.0631*** -0.7355 -0.8281 0.1781* 0.1505 0.0326 0.0265
Federal Legislative Liberalism (0.0109) (0.0108) (0.6317) (0.6329) (0.0924) (0.0933) (0.0568) (0.0571)
Lead Effects
State Liberalism -0.0039 -0.0042 -0.0875 -0.1184* 0.0459*** 0.0449*** -0.0004 -0.0010
(0.0059) (0.0059) (0.0603) (0.0614) (0.0120) (0.0121) (0.0103) (0.0103)
State × Presidential Liberalism 0.0164*** 0.0160*** -0.1154 -0.1522 0.0150*** 0.0108** -0.0122 -0.0112

28
(0.0020) (0.0020) (0.1011) (0.1028) (0.0047) (0.0048) (0.0081) (0.0081)
State × Federal Legislative Liberalism 0.0066 0.0079 0.3263 0.4245 -0.0046 -0.0041 0.0146 0.0121
(0.0078) (0.0078) (0.2708) (0.2750) (0.0272) (0.0271) (0.0147) (0.0149)
State × Presidential × -0.0198** -0.0204** 0.4182 0.6039 -0.1399*** -0.1281*** -0.0112 -0.0120
Federal Legislative Liberalism (0.0095) (0.0096) (0.5642) (0.5729) (0.0369) (0.0376) (0.0320) (0.0320)

Observations 63715 63715 29624 29624 16845 16845 17246 17246

R2 0.389 0.389 0.439 0.440 0.339 0.340 0.166 0.166
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three
years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
Appendix D Supplemental Estimation Results: Alternative Measure
of Ideology
Appendix D.1 Marginal Effects

Table D1: Marginal Effects of Ideology on All-Cause Mortality

(1) (2) (3) (4) (5) (6) (7)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism 0.0016 0.0019 -0.0006 -0.0010 -0.0006 -0.0000 0.0001
P-value 0.4306 0.0649 0.5016 0.3152 0.5050 0.9722 0.8753
1σ ↑ State & Federal Legislative Liberalism 0.0029 -0.0048 -0.0016 -0.0005 -0.0015 -0.0011 -0.0010
P-value 0.0937 0.0000 0.0321 0.5704 0.0444 0.1061 0.1528
1σ ↑ Full Liberalism 0.0020 -0.0030 -0.0017 -0.0007 -0.0016 -0.0011 -0.0010
P-value 0.2785 0.0010 0.0275 0.3798 0.0342 0.1083 0.1531
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism 0.0019 -0.0003 -0.0030 -0.0065 -0.0031 -0.0020 -0.0019
P-value 0.6948 0.9429 0.3841 0.0587 0.3709 0.5054 0.5418
1σ ↑ State & Federal Legislative Liberalism 0.0011 -0.0022 -0.0031 -0.0053 -0.0032 -0.0026 -0.0026
P-value 0.7928 0.4883 0.3123 0.0814 0.3025 0.3391 0.3362
1σ ↑ Full Liberalism 0.0018 0.0026 -0.0016 -0.0045 -0.0018 -0.0011 -0.0010
P-value 0.6555 0.3966 0.5895 0.1347 0.5644 0.6740 0.7060
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0169 -0.0196 -0.0037 -0.0080 -0.0037 -0.0015 -0.0017
P-value 0.0000 0.0000 0.0041 0.0000 0.0037 0.2050 0.1422
1σ ↑ State & Federal Legislative Liberalism 0.0208 -0.0115 -0.0004 -0.0044 -0.0006 0.0010 0.0009
P-value 0.0000 0.0000 0.6818 0.0004 0.6062 0.3042 0.3695
1σ ↑ Full Liberalism 0.0189 -0.0111 -0.0011 -0.0053 -0.0012 0.0005 0.0003
P-value 0.0000 0.0000 0.3216 0.0000 0.2587 0.6091 0.7492

Location Health Controls N N N N N Y Y

Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be
equal). Lead values correspond to the effects over the future three years (constrained to be equal).

29
 Table D2: Marginal Effects of Ideology on “Short-Term”-Cause Mortality

(1) (2) (3) (4) (5) (6) (7)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0116 -0.0068 -0.0027 -0.0071 -0.0031 -0.0023 -0.0016
P-value 0.0000 0.0027 0.2369 0.0093 0.1828 0.3027 0.4821
1σ ↑ State & Federal Legislative Liberalism -0.0068 -0.0116 -0.0049 -0.0053 -0.0056 -0.0048 -0.0043
P-value 0.0062 0.0000 0.0151 0.0185 0.0053 0.0142 0.0302
1σ ↑ Full Liberalism -0.0111 -0.0108 -0.0047 -0.0056 -0.0053 -0.0045 -0.0039
P-value 0.0000 0.0000 0.0199 0.0143 0.0093 0.0248 0.0497
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism 0.0322 -0.0021 -0.0042 -0.0123 -0.0059 -0.0032 -0.0028
P-value 0.0008 0.8244 0.6490 0.1849 0.5264 0.7206 0.7529
1σ ↑ State & Federal Legislative Liberalism 0.0288 -0.0047 -0.0074 -0.0107 -0.0091 -0.0061 -0.0063
P-value 0.0007 0.5712 0.3703 0.1942 0.2744 0.4454 0.4243
1σ ↑ Full Liberalism 0.0291 0.0003 -0.0038 -0.0085 -0.0053 -0.0029 -0.0027
P-value 0.0006 0.9680 0.6432 0.3010 0.5205 0.7187 0.7328
Lead Effects
1σ ↑ State & Federal Executive Liberalism -0.0048 -0.0164 -0.0045 -0.0178 -0.0043 -0.0031 -0.0038
P-value 0.1344 0.0000 0.1868 0.0000 0.2128 0.3608 0.2606
1σ ↑ State & Federal Legislative Liberalism 0.0024 -0.0076 0.0020 -0.0075 0.0026 0.0032 0.0027
P-value 0.3678 0.0026 0.4806 0.0194 0.3660 0.2681 0.3419
1σ ↑ Full Liberalism -0.0028 -0.0109 -0.0011 -0.0112 -0.0005 0.0007 0.0002
P-value 0.2919 0.0000 0.6880 0.0003 0.8535 0.7904 0.9473

Location Health Controls N N N N N Y Y

Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be
equal). Lead values correspond to the effects over the future three years (constrained to be equal).

30
Appendix D.2 Coefficients

Table D3: Ideological Determinants of All-Cause Mortality

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
State Liberalism 0.0025 -0.0083** -0.0060 -0.0086** -0.0061 -0.0048 -0.0047
(0.0053) (0.0040) (0.0039) (0.0039) (0.0039) (0.0034) (0.0034)
State × Presidential Liberalism -0.0012 0.0246*** 0.0081*** 0.0041* 0.0080*** 0.0079*** 0.0083***
(0.0035) (0.0023) (0.0023) (0.0024) (0.0023) (0.0021) (0.0021)
State × Federal Legislative Liberalism -0.0334 0.1583*** 0.0679** 0.0706** 0.0677** 0.0522* 0.0501*
(0.0356) (0.0308) (0.0302) (0.0301) (0.0303) (0.0267) (0.0267)
State × Presidential × 0.0992** -0.2227*** -0.0845** -0.0356 -0.0853** -0.0800** -0.0785**
Federal Legislative Liberalism (0.0452) (0.0369) (0.0362) (0.0366) (0.0367) (0.0340) (0.0342)
Lagged Effects
State Liberalism 0.0036* -0.0023** -0.0009 -0.0011 -0.0008 -0.0002 -0.0000
(0.0019) (0.0011) (0.0009) (0.0011) (0.0009) (0.0008) (0.0008)
State × Presidential Liberalism -0.0056** 0.0134*** 0.0009 -0.0002 0.0005 0.0006 0.0005
(0.0025) (0.0010) (0.0010) (0.0010) (0.0010) (0.0009) (0.0009)
State × Federal Legislative Liberalism -0.0119 -0.0805*** -0.0241*** 0.0138 -0.0239*** -0.0279*** -0.0292***
(0.0081) (0.0067) (0.0066) (0.0091) (0.0066) (0.0063) (0.0065)
State × Presidential × 0.0666** -0.1915*** -0.0251* -0.0190 -0.0204 -0.0144 -0.0135
Federal Legislative Liberalism (0.0339) (0.0150) (0.0150) (0.0151) (0.0150) (0.0137) (0.0138)
Lead Effects
State Liberalism 0.0219*** -0.0231*** -0.0033** -0.0073*** -0.0033** -0.0011 -0.0013
(0.0034) (0.0016) (0.0015) (0.0017) (0.0015) (0.0014) (0.0014)
State × Presidential Liberalism -0.0092*** 0.0036*** -0.0026** -0.0051*** -0.0027** -0.0018* -0.0020**
(0.0020) (0.0011) (0.0011) (0.0011) (0.0011) (0.0010) (0.0010)
State × Federal Legislative Liberalism 0.0362 0.2740*** 0.0752*** 0.0639*** 0.0711*** 0.0596*** 0.0612***
(0.0360) (0.0168) (0.0157) (0.0159) (0.0157) (0.0141) (0.0141)
State × Presidential × 0.0658*** -0.0630*** 0.0130 0.0538*** 0.0134 0.0006 0.0006
Federal Legislative Liberalism (0.0242) (0.0151) (0.0148) (0.0150) (0.0149) (0.0145) (0.0145)

Observations 129300 129300 129300 129300 129300 117504 117504

R2 0.021 0.468 0.488 0.491 0.555 0.617 0.617
Location Health Controls N N N N N Y Y
Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal).
Lead values correspond to the effects over the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.

31
 Table D4: Ideological Determinants of “Short-Term”-Cause Mortality

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
State Liberalism 0.0352*** -0.0113 -0.0117 -0.0173* -0.0138 -0.0094 -0.0095
(0.0104) (0.0101) (0.0101) (0.0101) (0.0101) (0.0096) (0.0095)
State × Presidential Liberalism 0.0003 0.0276*** 0.0217*** 0.0117* 0.0224*** 0.0186*** 0.0202***
(0.0070) (0.0069) (0.0069) (0.0070) (0.0070) (0.0068) (0.0068)
State × Federal Legislative Liberalism -0.0984 0.1698** 0.0989 0.1515** 0.1054 0.0817 0.0757
(0.0742) (0.0727) (0.0727) (0.0728) (0.0729) (0.0708) (0.0708)
State × Presidential × 0.0131 -0.2983*** -0.2672** -0.1179 -0.2713** -0.2239** -0.2270**
Federal Legislative Liberalism (0.1056) (0.1049) (0.1047) (0.1065) (0.1061) (0.1032) (0.1034)
Lagged Effects
State Liberalism -0.0044 -0.0078*** -0.0020 -0.0069** -0.0027 -0.0022 -0.0013
(0.0028) (0.0024) (0.0024) (0.0029) (0.0024) (0.0023) (0.0023)
State × Presidential Liberalism -0.0254*** 0.0011 -0.0031 -0.0025 -0.0021 -0.0010 -0.0012
(0.0032) (0.0024) (0.0028) (0.0030) (0.0028) (0.0028) (0.0028)
State × Federal Legislative Liberalism -0.0843*** -0.1352*** -0.0922*** 0.0306 -0.0947*** -0.0860*** -0.0930***
(0.0158) (0.0156) (0.0160) (0.0226) (0.0161) (0.0167) (0.0171)
State × Presidential × 0.3054*** 0.0384 0.0973** 0.0407 0.0860** 0.0636 0.0688*
Federal Legislative Liberalism (0.0420) (0.0345) (0.0394) (0.0412) (0.0392) (0.0391) (0.0394)
Lead Effects
State Liberalism 0.0055 -0.0139*** -0.0008 -0.0136*** -0.0006 -0.0009 -0.0016
(0.0040) (0.0035) (0.0041) (0.0044) (0.0041) (0.0039) (0.0040)
State × Presidential Liberalism -0.0331*** -0.0125*** -0.0127*** -0.0180*** -0.0124*** -0.0075** -0.0078**
(0.0035) (0.0032) (0.0033) (0.0033) (0.0033) (0.0032) (0.0032)
State × Federal Legislative Liberalism -0.0790* 0.1541*** 0.0839** 0.1468*** 0.0968** 0.1227*** 0.1296***
(0.0430) (0.0358) (0.0402) (0.0409) (0.0401) (0.0389) (0.0391)
State × Presidential × 0.4334*** 0.0386 0.0587 0.1555*** 0.0516 -0.0173 -0.0218
Federal Legislative Liberalism (0.0482) (0.0429) (0.0435) (0.0442) (0.0437) (0.0440) (0.0440)

Observations 113261 113261 113261 113261 113261 105542 105542

R2 0.007 0.101 0.117 0.122 0.205 0.225 0.225
Location Health Controls N N N N N Y Y
Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population
shares across 18 different age categories. Population health controls include population shares by education (HS degree, some
college, and at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male,
black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal).
Lead values correspond to the effects over the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.

32
Appendix E Supplemental Estimation Results: Quantile Regression
Figure 1 documents a stark increase in the interquartile range of county-level all-cause mortality since
around the mid 1980s. Figure 2 highlights a dramatic increase in the mortality gap between counties with
per capita income above and below the median. To explore the role of politics in these gaps, we do not
wish to split the sample into high and low per capita income counties as income is likely endogenous to
politics given the prior literature on partisan business cycles discussed in Section 2. Instead, we explore
heterogeneity in the associations between ideology and mortality across the distribution of mortality. To
do so, we estimate (5) using fixed effect quantile regression (Machado and Santos Silva, 2019).22 Tables
E1–E2 present the marginal effects of liberalism at the 0.10, 0.25, 0.50, 0.75, and 0.90 quantiles, controlling
for state-specific linear time trends given the added complexity of the model. Tables E3–E4 presents the
coefficient estimates. As before, we chose to be conservative and focus on specifications where the lagged
and contemporaneous effects are statistically significant and where the lead effects are either statistically
insignificant at the p < 0.10 level or statistically significant but of the opposite sign of the lagged and/or
contemporaneous effects.
Two interesting results emerge. First, the statistically significant, negative effect of lagged Executive
Liberalism on all-cause mortality documented in Table 1 holds for the upper quantiles at the p < 0.10 level.
Here, the magnitudes increase (in absolute value) as one moves to higher quantiles; the effects are precise
zeros at the 0.10 quantile, -0.1% for the second and third quantiles, and -0.2% for two highest quantiles.
Second, the statistically significant, negative effects of lagged Executive, Legislative, and Full Liberalism
on short-term mortality documented in Table 2 holds across nearly all quantiles. The lone exception is
the statistically insignificant effects at the p < 0.10 level of Legislative Liberalism at the 0.90 quantile.
However, the magnitudes decline (in absolute value) as one moves to higher quantiles. For example, a one
standard deviation increase in Full Liberalism reduces the short-term mortality rate by 0.7% at the 0.10
quantile, 0.6% at the median, and 0.5% at the 0.90 quantile.
In sum, the quantile regression reveal contrasting results for all-cause and short-term mortality. For
all-cause mortality, liberalism is more beneficial (or less harmful) at higher quantiles. For short-term mor-
tality, the opposite occurs. Here, liberalism is most beneficial at low quantiles. This implies that liberalism
ameliorates (conditional) inequities in all-cause mortality across locations due to larger relative improve-
ments in locations with high (conditional) all-cause mortality, but exacerbates (conditional) inequities in
short-term mortality due to larger relative improvements in locations with low (conditional) short-term
mortality. As the greater inequity in short-term mortality arises from greater improvements in locations
that are relatively healthy, as opposed to deterioration in locations that are relatively unhealthy, this is
not necessarily a bad outcome. However, it is consistent with more liberal regimes being ineffective in
reducing disparities in short-term mortality. Yet, such regimes do benefit the least healthy places in terms
of all-cause mortality.

22
Estimation performed in Stata using xtqreg.

33
 Table E1: Marginal Effects of Ideology on All-Cause Mortality Across Quantiles

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0002 -0.0002 -0.0007 -0.0006 -0.0012 -0.0012 -0.0018 -0.0017 -0.0022 -0.0021
P-value 0.8732 0.9079 0.4449 0.4856 0.0662 0.0814 0.0439 0.0524 0.0750 0.0843
1σ ↑ State & Federal Legislative Liberalism -0.0002 -0.0001 -0.0005 -0.0004 -0.0008 -0.0008 -0.0012 -0.0011 -0.0015 -0.0014
P-value 0.8866 0.9377 0.5650 0.6317 0.1823 0.2298 0.1491 0.1810 0.2053 0.2336
1σ ↑ Full Liberalism 0.0004 0.0006 -0.0001 0.0000 -0.0007 -0.0006 -0.0012 -0.0011 -0.0017 -0.0016
P-value 0.7294 0.6731 0.9332 0.9756 0.3148 0.3983 0.1523 0.1922 0.1598 0.1892
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism -0.0031 -0.0033 -0.0025 -0.0028 -0.0018 -0.0021 -0.0011 -0.0015 -0.0005 -0.0010
P-value 0.5797 0.5561 0.5215 0.4778 0.5401 0.4633 0.7705 0.6851 0.9238 0.8505
1σ ↑ State & Federal Legislative Liberalism -0.0016 -0.0016 -0.0014 -0.0014 -0.0011 -0.0011 -0.0008 -0.0009 -0.0005 -0.0007
P-value 0.7689 0.7765 0.7217 0.7216 0.7103 0.6938 0.8345 0.8087 0.9204 0.8940
1σ ↑ Full Liberalism -0.0027 -0.0027 -0.0025 -0.0025 -0.0022 -0.0023 -0.0020 -0.0021 -0.0017 -0.0020
P-value 0.6336 0.6335 0.5266 0.5166 0.4468 0.4206 0.6012 0.5658 0.7422 0.7069
Lead Effects
1σ ↑ State & Federal Executive Liberalism 0.0002 0.0001 -0.0000 -0.0001 -0.0003 -0.0003 -0.0006 -0.0006 -0.0008 -0.0008
P-value 0.9042 0.9360 0.9750 0.9421 0.7218 0.6998 0.6132 0.6102 0.6200 0.6259

34
1σ ↑ State & Federal Legislative Liberalism 0.0009 0.0007 0.0006 0.0004 0.0003 0.0001 -0.0001 -0.0002 -0.0004 -0.0005
P-value 0.5855 0.6610 0.6003 0.7021 0.7672 0.8989 0.9443 0.8525 0.8180 0.7621
1σ ↑ Full Liberalism 0.0004 0.0003 0.0001 -0.0000 -0.0003 -0.0004 -0.0007 -0.0008 -0.0011 -0.0011
P-value 0.7979 0.8518 0.9464 0.9806 0.7088 0.6296 0.5295 0.4818 0.5162 0.4865

Location Health Controls Y Y Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and population
shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income shares for
the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be
equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 Table E2: Marginal Effects of Ideology on “Short Term”-Cause Mortality Across Quantiles

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Lagged Effects
1σ ↑ State & Federal Executive Liberalism -0.0094 -0.0084 -0.0086 -0.0079 -0.0077 -0.0072 -0.0069 -0.0067 -0.0062 -0.0062
P-value 0.0085 0.0188 0.0004 0.0012 0.0000 0.0000 0.0021 0.0029 0.0486 0.0493
1σ ↑ State & Federal Legislative Liberalism -0.0067 -0.0057 -0.0060 -0.0053 -0.0053 -0.0048 -0.0047 -0.0044 -0.0041 -0.0040
P-value 0.0507 0.0955 0.0091 0.0223 0.0016 0.0042 0.0290 0.0393 0.1692 0.1772
1σ ↑ Full Liberalism -0.0072 -0.0059 -0.0067 -0.0058 -0.0062 -0.0056 -0.0057 -0.0054 -0.0053 -0.0053
P-value 0.0430 0.0968 0.0052 0.0170 0.0004 0.0016 0.0103 0.0156 0.0890 0.0938
Contemporaneous Effects
1σ ↑ State & Federal Executive Liberalism 0.0001 -0.0027 -0.0026 -0.0048 -0.0055 -0.0071 -0.0083 -0.0092 -0.0106 -0.0110
P-value 0.9952 0.8637 0.8016 0.6476 0.4705 0.3572 0.3941 0.3428 0.4386 0.4201
1σ ↑ State & Federal Legislative Liberalism 0.0050 0.0037 0.0020 0.0009 -0.0013 -0.0020 -0.0044 -0.0048 -0.0070 -0.0071
P-value 0.7437 0.8091 0.8513 0.9277 0.8607 0.7914 0.6450 0.6175 0.6038 0.5976
1σ ↑ Full Liberalism 0.0013 -0.0002 -0.0021 -0.0033 -0.0058 -0.0066 -0.0092 -0.0097 -0.0121 -0.0122
P-value 0.9325 0.9894 0.8403 0.7543 0.4527 0.3925 0.3433 0.3197 0.3758 0.3694
Lead Effects
1σ ↑ State & Federal Executive Liberalism -0.0001 -0.0003 0.0012 0.0011 0.0026 0.0026 0.0040 0.0040 0.0051 0.0052
P-value 0.9750 0.9563 0.7048 0.7207 0.2559 0.2596 0.1743 0.1713 0.2144 0.2087

35
1σ ↑ State & Federal Legislative Liberalism -0.0011 -0.0017 0.0002 -0.0003 0.0015 0.0011 0.0027 0.0025 0.0037 0.0036
P-value 0.8205 0.7209 0.9562 0.9213 0.5196 0.6237 0.3509 0.3923 0.3596 0.3754
1σ ↑ Full Liberalism -0.0020 -0.0024 -0.0007 -0.0010 0.0007 0.0005 0.0020 0.0019 0.0031 0.0030
P-value 0.6803 0.6220 0.8365 0.7622 0.7602 0.8352 0.4969 0.5293 0.4572 0.4693

Location Health Controls Y Y Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age
categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and population
shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income shares for
the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be
equal). Lead values correspond to the effects over the future three years (constrained to be equal).
 To link these results to the current rise in geographic mortality disparities, Figure E1 shows trends in
state policy liberalism (left panel) and median liberalism of US Senators and Representatives (right panel)
over time. Panel A shows a persistent trend away from liberalism at the state level starting around 1990
and lasting until about 2005. It is striking that the rapid decline in state policy liberalism during the
1990s roughly coincides with the dramatic increase in the interquartile range seen in Figure 1. The fact
that we find liberal regimes to reduce inequities in all-cause mortality by having larger effects at higher
(conditional) quantiles points to partisan mortality cycles as a potentially salient factor contributing to
the geographic inequality in mortality. Panel B shows that the ideology of the median US Senator and
Representative is quite volatile and shows more of a trend toward extremes than toward any particular
ideology when using the first dimension of DW-NOMINATE scores.

Figure E1: State & Federal Ideology Over Time

(A) State (B) Federal

36
 Table E3: Ideological Determinants of All-Cause Mortality Across Quantiles

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Lagged Effects
State Liberalism -0.0020 -0.0019 -0.0016 -0.0016 -0.0011 -0.0013 -0.0007 -0.0009 -0.0003 -0.0006
(0.0060) (0.0060) (0.0042) (0.0042) (0.0031) (0.0031) (0.0040) (0.0040) (0.0057) (0.0057)
State × Presidential Liberalism -0.0027 -0.0032* -0.0021 -0.0026** -0.0015 -0.0020** -0.0009 -0.0014 -0.0004 -0.0009
(0.0019) (0.0019) (0.0013) (0.0013) (0.0010) (0.0010) (0.0013) (0.0013) (0.0018) (0.0018)
State × Federal Legislative Liberalism 0.0015 0.0017 0.0008 0.0010 -0.0001 0.0003 -0.0010 -0.0004 -0.0017 -0.0010
(0.0065) (0.0065) (0.0045) (0.0045) (0.0033) (0.0033) (0.0043) (0.0043) (0.0061) (0.0061)
State × Presidential × 0.0041 0.0069 -0.0007 0.0020 -0.0062 -0.0036 -0.0114 -0.0089 -0.0158 -0.0134
Federal Legislative Liberalism (0.0116) (0.0116) (0.0080) (0.0080) (0.0059) (0.0060) (0.0077) (0.0077) (0.0109) (0.0110)
Contemporaneous Effects
State Liberalism -0.0011 -0.0011 -0.0014 -0.0014 -0.0018*** -0.0017** -0.0021** -0.0021** -0.0024* -0.0023*
(0.0013) (0.0013) (0.0009) (0.0009) (0.0007) (0.0007) (0.0009) (0.0009) (0.0012) (0.0013)
State × Presidential Liberalism 0.0017** 0.0017** 0.0013*** 0.0013*** 0.0008** 0.0008** 0.0004 0.0004 0.0000 0.0000
(0.0007) (0.0007) (0.0005) (0.0005) (0.0004) (0.0004) (0.0005) (0.0005) (0.0007) (0.0007)
State × Federal Legislative Liberalism 0.0070*** 0.0071*** 0.0068*** 0.0069*** 0.0066*** 0.0067*** 0.0064*** 0.0065*** 0.0062*** 0.0063***
(0.0023) (0.0023) (0.0016) (0.0016) (0.0012) (0.0012) (0.0015) (0.0015) (0.0021) (0.0021)
State × Presidential × -0.0031 -0.0026 -0.0033 -0.0029 -0.0036** -0.0032* -0.0039* -0.0035 -0.0041 -0.0037
Federal Legislative Liberalism (0.0035) (0.0036) (0.0024) (0.0025) (0.0018) (0.0018) (0.0023) (0.0024) (0.0033) (0.0033)
Lead Effects
State Liberalism 0.0004 0.0002 0.0001 -0.0000 -0.0001 -0.0002 -0.0003 -0.0004 -0.0005 -0.0006
(0.0018) (0.0018) (0.0013) (0.0013) (0.0009) (0.0009) (0.0012) (0.0012) (0.0017) (0.0017)

37
State × Presidential Liberalism -0.0003 -0.0001 -0.0004 -0.0002 -0.0005 -0.0003 -0.0006 -0.0004 -0.0006 -0.0005
(0.0007) (0.0007) (0.0005) (0.0005) (0.0003) (0.0003) (0.0004) (0.0004) (0.0006) (0.0006)
State × Federal Legislative Liberalism 0.0048 0.0047 0.0039* 0.0037 0.0028* 0.0025 0.0018 0.0015 0.0010 0.0005
(0.0033) (0.0033) (0.0023) (0.0023) (0.0017) (0.0017) (0.0022) (0.0022) (0.0031) (0.0031)
State × Presidential × -0.0054 -0.0060 -0.0055* -0.0061** -0.0057*** -0.0062*** -0.0058** -0.0063** -0.0059 -0.0064
Federal Legislative Liberalism (0.0042) (0.0042) (0.0029) (0.0029) (0.0022) (0.0022) (0.0028) (0.0028) (0.0040) (0.0040)

Observations 117504 117504 117504 117504 117504 117504 117504 117504 117504 117504
Location Health Controls Y Y Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age cate-
gories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and population shares
by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income shares for the top
10% and top 1% and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead
values correspond to the effects over the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table E4: Ideological Determinants of “Short Term”-Cause Mortality Across Quantiles

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Lagged Effects
State Liberalism 0.0037 0.0019 0.0006 -0.0007 -0.0026 -0.0036 -0.0057 -0.0063 -0.0083 -0.0085
(0.0165) (0.0165) (0.0111) (0.0111) (0.0081) (0.0081) (0.0103) (0.0103) (0.0144) (0.0144)
State × Presidential Liberalism -0.0068 -0.0090* -0.0065** -0.0083** -0.0063*** -0.0076*** -0.0060** -0.0069** -0.0058 -0.0063
(0.0049) (0.0049) (0.0033) (0.0033) (0.0024) (0.0024) (0.0030) (0.0031) (0.0043) (0.0043)
State × Federal Legislative Liberalism 0.0132 0.0159 0.0112 0.0134 0.0091 0.0107 0.0071 0.0082 0.0055 0.0060
(0.0160) (0.0160) (0.0108) (0.0108) (0.0079) (0.0079) (0.0100) (0.0100) (0.0140) (0.0141)
State × Presidential × -0.0060 0.0079 -0.0131 -0.0019 -0.0207 -0.0124 -0.0279 -0.0223 -0.0339 -0.0306
Federal Legislative Liberalism (0.0295) (0.0297) (0.0199) (0.0201) (0.0146) (0.0147) (0.0185) (0.0185) (0.0259) (0.0261)
Contemporaneous Effects
State Liberalism -0.0099*** -0.0089** -0.0090*** -0.0082*** -0.0080*** -0.0075*** -0.0071*** -0.0068*** -0.0063** -0.0062*
(0.0036) (0.0036) (0.0025) (0.0025) (0.0018) (0.0018) (0.0023) (0.0023) (0.0032) (0.0032)
State × Presidential Liberalism -0.0005 -0.0004 -0.0006 -0.0005 -0.0006 -0.0006 -0.0007 -0.0008 -0.0008 -0.0009
(0.0018) (0.0018) (0.0012) (0.0012) (0.0009) (0.0009) (0.0011) (0.0011) (0.0016) (0.0016)
State × Federal Legislative Liberalism 0.0206*** 0.0208*** 0.0189*** 0.0191*** 0.0171*** 0.0172*** 0.0154*** 0.0155*** 0.0140*** 0.0140***
(0.0057) (0.0058) (0.0039) (0.0039) (0.0028) (0.0029) (0.0036) (0.0036) (0.0050) (0.0051)
State × Presidential × -0.0046 -0.0010 -0.0067 -0.0039 -0.0090** -0.0069 -0.0111* -0.0098* -0.0128 -0.0121
Federal Legislative Liberalism (0.0091) (0.0092) (0.0061) (0.0062) (0.0045) (0.0045) (0.0057) (0.0057) (0.0080) (0.0080)
Lead Effects
State Liberalism -0.0007 -0.0013 0.0006 0.0001 0.0020 0.0017 0.0033 0.0031 0.0044 0.0044
(0.0050) (0.0050) (0.0033) (0.0034) (0.0024) (0.0025) (0.0031) (0.0031) (0.0043) (0.0044)

38
State × Presidential Liberalism 0.0011 0.0020 0.0014 0.0021* 0.0016* 0.0022** 0.0019* 0.0023** 0.0021 0.0024
(0.0018) (0.0018) (0.0012) (0.0012) (0.0009) (0.0009) (0.0011) (0.0011) (0.0015) (0.0016)
State × Federal Legislative Liberalism -0.0031 -0.0035 -0.0030 -0.0035 -0.0030 -0.0035 -0.0029 -0.0035 -0.0028 -0.0035
(0.0084) (0.0084) (0.0057) (0.0057) (0.0042) (0.0042) (0.0053) (0.0053) (0.0074) (0.0074)
State × Presidential × -0.0224** -0.0258** -0.0232*** -0.0260*** -0.0241*** -0.0263*** -0.0249*** -0.0265*** -0.0256*** -0.0267***
Federal Legislative Liberalism (0.0110) (0.0110) (0.0074) (0.0075) (0.0054) (0.0055) (0.0069) (0.0069) (0.0097) (0.0097)

Observations 105542 105542 105542 105542 105542 105542 105542 105542 105542 105542
Location Health Controls Y Y Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age categories.
Population health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and population shares by gender
and race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1%
and each interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond
to the effects over the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
Appendix F Supplemental Estimation Results: Role of Political Party
Our focus has been on partisan mortality cycles as driven by the ideology of political regimes. We do so
since ideology scores are arguably more closely connected to actual policy decisions. Nonetheless, a large
literature explores the effect of political party on various economic and social outcomes. One subset of this
literature examines associations between the political party of the US president and various outcomes. For
example, Alesina and Sachs (1988) analyze data from 1948 to 1984, finding that macroeconomic outcomes
differ by presidential party. Santa-Clara and Valkanov (2003) use monthly data from 1927 to 1998 to assess
the differences in several financial variables by presidential party. The authors find excess stock market
returns to be 9-16% higher under Democratic presidents arising higher real stock returns and lower interest
rates with no difference in risk. Blinder and Watson (2016) assess data from President Truman through
President Obama, documenting a robust 1.8% gap in GDP growth favoring Democratic presidencies,
not attributable to reverse causality, congressional control, and fiscal or monetary policy performances.
Another subset of this literature focuses on state-level partisanship. Most recently, Dynes and Holbein
(2020) examine 28 policy outcomes using state-level panel data from 1960 to 2016 and both fixed effects
and regression discontinuity methods. The measures of partisanship reflect the party of the governor and
the party controlling the state House and Senate. Overall, the authors find precisely estimated, null effects.
In light of this, we briefly explore the role of political party. To do so, we replace our ideology measures
with covariates indicating Democratic party control of the executive branch at the state and federal levels,
Democratic party control of the legislative branch at the state and federal levels, and their interactions.
We measure Executive Control with a dummy variable equal to one if the state governor is a Democrat
(zero otherwise) and the interaction of this variable with a dummy variable equal to one if the President
is a Democrat (zero otherwise). We measure Legislative Control with a dummy variable equal to one if
the state legislature (both House and Senate) is controlled by a Democratic majority (zero otherwise) and
the interaction of this variables with a dummy variable equal to one if the federal legislature (both House
and Senate) is controlled by a Democratic majority (zero otherwise).23 Finally, we measure Full Control
by interacting executive and legislative control. Executive control enables the party in power the ability to
enforce laws. Legislative control enables the party in power the ability to enact laws. Full control enables
the party in power the ability enact and enforce laws, as well as appoint judges.24
Estimates are provided in Appendix F.1 – Appendix F.3; corresponding figures are shown in Appendix
A.4. For the tables reporting marginal effects, we report tests of significance for the sum of the coefficients
on Democratic governor and the interaction between Democratic governor and president. We refer to
this as tests of Democratic Executive Control. Similarly, we report tests of significance for the sum of
the coefficients on Democratic majority in the state legislature and the interaction between Democratic
majority in the state and federal legislature. We refer to this as tests of Democratic Legislative Control.
Finally, we report tests of significance of the sum of all partisan variables. We refer to this as tests
of Democratic Full Control. As in the analysis of political ideology, our preferred specifications include
23
We do not include a dummy variable for either the party of the President or federal legislative control as this is absorbed
by the time fixed effects; only the interaction term is identified.
24
Note, our measures of partisanship are defined as one under Democratic control and zero otherwise. Thus, the measures
equal zero under either Republican control or in divided governments. The state data are from Dynes and Holbein (2020);
available at https://doi.org/10.7910/DVN/VGWNP9. The federal Congressional data are from https://www.senate.gov/his
tory/partydiv.htm and https://history.house.gov/Institution/Party-Divisions/Party-Divisions/.

39
location and population health controls, county and year fixed effects, and county-specific time trends.
A few interesting results emerge. First, in the baseline specifications, we find statistically significant,
positive associations at the p < 0.10 level between lagged Democratic Legislative and Full Control and
all-cause mortality in our preferred specification with or without controlling for income inequality. The
effects are modest, as Democratic Legislative (Full) Control over the prior four years is associated with a
0.1% (0.2%) increase in current all-cause mortality. This corresponds to roughly 1 (2) additional deaths
per 100,000 in the average county-year; 2,500 (5,000) additional deaths in the average year. The fact that
the estimated contemporaneous and lead effects of Democratic Legislative and Full Control are precise
zeros is consistent with these estimates having a causal interpretation. For short-term mortality, there is
some evidence of a negative association with contemporaneous Democratic Executive, Legislative, and Full
Control. However, the results for the lead effects suggest this should be interpreted with caution.
The positive effect of lagged Democratic control on all-cause mortality is in stark contract to the
negative effects of lagged liberalism on all-cause (and short-term) mortality. A possible explanation is that
the overall effect of political party masks considerable heterogeneity in what the parties have stood for
over time as the early part of our sample covers the period when many Southern Democrats transitioned
to the Republican party in response to the pro-Civil Rights stance adopted by some Democratic presidents
(Strong, 1971). As such, Democratic affiliation is not always synonymous with liberalism, particularly
during the 20th century. Nor is the Republican party always synonymous with conservatism.
To explore this, Figure F1 plots ideology scores by government branch and partisan control both pre-
and post-2000. Panel A plots state policy liberalism by time period and party of the governor. Governor
political party is not indicative of ideology, particularly under Democratic governors. There is even a mass
of liberal states with Republican governors since 2001. The same holds for Democratic control of state
legislatures in Panel B. Prior to 2001, the distribution of state policy liberalism is bimodal for states under
Democratic control of the state legislature. Since 2001, state policy liberalism is highly correlated with the
party controlling the state legislature, yet each distribution contains a long tail in the opposite direction.
Panels C and D show similar plots at the national level. Here, political party and ideology scores of
individual representatives and senators align with current perceptions of both parties. Thus, it is critical
not to naı̈vely ascribe the ideology and policy preferences of the current national party platforms to state
and local politicians. This is consistent with our view of partisan mortality cycles based on the partisan
business cycle framework originating in Hibbs (1977). Under this framework, politicians are beholden
to the policy preferences of the constituents that voted them into office. As such, if the ideology of the
national party diverges from the ideology of a politician’s local base, politicians should reflect the latter
more than former. For example, California has an extremely high state policy liberalism score in the early
2000s despite Republican Arnold Schwarzenegger being governor. Conversely, Mississippi historically has
some of the lowest state policy liberalism scores despite having Democratic governors during most of our
sample period. Similarly, Alabama and Mississippi, until recently, were primarily under Democratic state
legislative control, yet have some of the lowest state policy liberalism scores.

40
 Figure F1: Historical Ideology Scores by Government Branch, Partisan Control, and Time Period

(A) Governor (B) State Congress

(C) US House of Representatives (D) US Senate

Notes: Higher values of all ideology scores correspond to greater liberalism.

To further examine the potential role of Southern Democrats last century, we re-estimate the baseline
specification for the effects of political party on all-cause mortality omitting the Southeast region of the US
(as defined by the US BEA).25 The results are displayed in Figure A3 in Appendix A. Now, all estimates
are statistically insignificant at the p < 0.05 level.
The second interesting finding in the analysis of political party effects is that there is a lot of hetero-
geneity across demographic groups and over time. To begin, the positive association between Democratic
control and all-cause mortality found in the baseline specification is confined to White females when using
the full sample. However, when splitting the sample by time period, we only find a negative association
with Black females during the period 1985–2000. In contrast, the negative association between Democratic
control and short-term mortality is common to all demographic groups when using the full sample period.
Moreover, the effects are not trivial. For example, Democratic Legislative Control over the prior four
25
This includes Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina,
Tennessee, West Virginia, and Virginia.

41
years reduces the short-term mortality rate by 3.9% (2.7%) for Black males (females). This corresponds
to roughly 1 fewer short-term death per 100,000 in the average county-year for Black males. We also
find negative effects of contemporaneous Democratic Executive (and Legislative) Control for White males
(females). Here, Democratic Control leads to roughly a 2–3% reduction in the short-term mortality rates.
Splitting the sample by time period, we find that these negative associations are only statistically mean-
ingful for White males and females during the period 1985–2000, and White females during the period
1968–1984. There is a positive association between Democratic control and the short-term mortality of
White males during the period 1968–1984. Finally, there is no statistically meaningful evidence of partisan
mortality cycles after 2000, consistent with the prior results based on political ideology.
Finally, quantile regression estimates indicate that Democratic control is associated with lower geo-
graphic inequality in all-cause mortality due to larger negative and smaller positive associations in loca-
tions with the highest (conditional) mortality levels. However, consistent with the effects of liberalism,
Democratic control contributes to geographic inequality in short-term mortality due to larger (in absolute
value) negative associations in locations with the lowest (conditional) levels.
Overall, the results using political party are broadly consistent with those using ideology, particularly
after accounting for the ideology of Southern Democrats in the early part of the sample. This highlights
the importance of understanding the spatial and temporal variation in views within US political parties.

Appendix F.1 Baseline Specification

42
 Table F1: Political Party Determinants of All-Cause Mortality: Marginal Effects

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
Executive Control -0.0003 0.0035 0.0023 0.0017 0.0023 0.0022 0.0021
P-value 0.6999 0.0000 0.0000 0.0042 0.0000 0.0000 0.0000
Legislative Control 0.0211 -0.0054 0.0009 0.0013 0.0010 0.0009 0.0010
P-value 0.0000 0.0000 0.0803 0.0176 0.0656 0.0587 0.0440
Full Control 0.0230 -0.0115 -0.0006 -0.0011 -0.0008 0.0015 0.0015
P-value 0.0000 0.0000 0.4806 0.2298 0.3467 0.0709 0.0570
Lagged Effects
Executive Control -0.0044 0.0018 -0.0018 -0.0016 -0.0017 -0.0018 -0.0019
P-value 0.0045 0.2323 0.2464 0.2931 0.2788 0.1760 0.1652
Legislative Control -0.0116 0.0007 0.0005 0.0012 0.0007 0.0004 0.0003
P-value 0.0000 0.6236 0.7283 0.4298 0.6256 0.7966 0.8474
Full Control 0.0004 -0.0025 -0.0029 -0.0028 -0.0026 -0.0013 -0.0012
P-value 0.8369 0.2487 0.1676 0.1752 0.2190 0.5227 0.5400
Lead Effects
Executive Control -0.0174 -0.0007 -0.0001 0.0004 -0.0001 0.0011 0.0010
P-value 0.0000 0.3293 0.9250 0.5860 0.8354 0.0598 0.0761
Legislative Control 0.0376 -0.0116 -0.0019 -0.0010 -0.0018 0.0001 0.0002
P-value 0.0000 0.0000 0.0120 0.1875 0.0205 0.9410 0.7985
Full Control 0.0181 -0.0141 -0.0028 -0.0019 -0.0027 -0.0002 0.0000
P-value 0.0000 0.0000 0.0058 0.0650 0.0094 0.7936 0.9962

Location Health Controls N N N N N Y Y

Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include
population shares across 18 different age categories. Population health controls include population shares by
education (HS degree, some college, and at least a 4-year college degree) and population shares by gender and
race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls in-
clude state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond
to the effects over the future three years (constrained to be equal).

43
Table F2: Political Party Determinants of “Short-Term”-Cause Mortality: Marginal Effects

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
Executive Control 0.0032 0.0075 0.0061 0.0047 0.0058 0.0055 0.0054
P-value 0.0055 0.0000 0.0000 0.0004 0.0000 0.0000 0.0000
Legislative Control 0.0179 -0.0026 -0.0005 0.0008 -0.0001 0.0004 0.0007
P-value 0.0000 0.0411 0.7206 0.5463 0.9387 0.7509 0.5971
Full Control 0.0064 -0.0037 -0.0024 -0.0022 -0.0019 0.0001 0.0004
P-value 0.0002 0.0922 0.2823 0.3554 0.4050 0.9670 0.8471
Lagged Effects
Executive Control -0.0051 -0.0105 -0.0142 -0.0128 -0.0143 -0.0113 -0.0115
P-value 0.1157 0.0028 0.0000 0.0003 0.0001 0.0007 0.0006
Legislative Control -0.0069 -0.0057 -0.0095 -0.0071 -0.0105 -0.0091 -0.0095
P-value 0.0511 0.1623 0.0201 0.0785 0.0100 0.0183 0.0132
Full Control -0.0158 -0.0137 -0.0203 -0.0168 -0.0206 -0.0173 -0.0172
P-value 0.0003 0.0125 0.0002 0.0024 0.0002 0.0011 0.0011
Lead Effects
Executive Control 0.0038 0.0060 0.0058 0.0054 0.0057 0.0047 0.0046
P-value 0.0065 0.0001 0.0001 0.0004 0.0001 0.0006 0.0010
Legislative Control 0.0319 -0.0024 -0.0024 -0.0015 -0.0022 -0.0026 -0.0023
P-value 0.0000 0.2017 0.1930 0.4170 0.2178 0.1319 0.1857
Full Control 0.0077 -0.0084 -0.0089 -0.0074 -0.0091 -0.0068 -0.0062
P-value 0.0005 0.0012 0.0006 0.0046 0.0005 0.0064 0.0122

Location Health Controls N N N N N Y Y

Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include
population shares across 18 different age categories. Population health controls include population shares by
education (HS degree, some college, and at least a 4-year college degree) and population shares by gender and
race (white/male, white/female, black/male, black/female, and other/male). Income inequality controls in-
clude state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond
to the effects over the future three years (constrained to be equal).

44
 Table F3: Political Party Determinants of All-Cause Mortality: Coefficients

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
Democrat Governor (1 = Yes) 0.0115*** -0.0014 -0.0011 -0.0013 -0.0010 -0.0014 -0.0013
(0.0012) (0.0011) (0.0011) (0.0011) (0.0011) (0.0010) (0.0010)
Democrat Governor & President (1 = Yes) -0.0158*** 0.0032** -0.0007 -0.0003 -0.0006 -0.0004 -0.0006
(0.0015) (0.0016) (0.0016) (0.0016) (0.0016) (0.0014) (0.0014)
Democrat Majority, State Legislature (1 = Yes) 0.0008 -0.0021 -0.0019 -0.0010 -0.0020 -0.0022* -0.0021*
(0.0013) (0.0014) (0.0013) (0.0013) (0.0013) (0.0013) (0.0013)
Democrat Majority, Federal -0.0124*** 0.0028 0.0024 0.0021 0.0028 0.0025 0.0024
& State Legislature (1 = Yes) (0.0014) (0.0017) (0.0017) (0.0017) (0.0017) (0.0016) (0.0016)
Democrat Governor & President & Federal 0.0164*** -0.0051** -0.0017 -0.0024 -0.0017 0.0002 0.0004
& State Legislature Majority (1 = Yes) (0.0020) (0.0021) (0.0020) (0.0020) (0.0020) (0.0017) (0.0017)
Lagged Effects
Democrat Governor (1 = Yes) 0.0031*** -0.0009** 0.0017*** 0.0010*** 0.0017*** 0.0001 0.0001
(0.0006) (0.0004) (0.0004) (0.0004) (0.0004) (0.0003) (0.0003)
Democrat Governor & President (1 = Yes) -0.0034*** 0.0045*** 0.0006 0.0007 0.0006 0.0021*** 0.0020***
(0.0007) (0.0007) (0.0007) (0.0007) (0.0007) (0.0006) (0.0006)
Democrat Majority, State Legislature (1 = Yes) -0.0075*** 0.0046*** 0.0026*** 0.0020*** 0.0024*** 0.0030*** 0.0030***
(0.0007) (0.0006) (0.0005) (0.0005) (0.0005) (0.0005) (0.0005)
Democrat Majority, Federal 0.0287*** -0.0100*** -0.0017*** -0.0007 -0.0015** -0.0021*** -0.0020***
& State Legislature (1 = Yes) (0.0006) (0.0007) (0.0006) (0.0006) (0.0006) (0.0006) (0.0006)
Democrat Governor & President & Federal 0.0021* -0.0096*** -0.0038*** -0.0040*** -0.0041*** -0.0016* -0.0015*
& State Legislature Majority (1 = Yes) (0.0012) (0.0011) (0.0009) (0.0010) (0.0009) (0.0009) (0.0009)
Lead Effects
Democrat Governor (1 = Yes) 0.0179*** -0.0008 0.0011** 0.0007 0.0011** -0.0003 -0.0004
(0.0006) (0.0005) (0.0005) (0.0005) (0.0005) (0.0004) (0.0004)
Democrat Governor & President (1 = Yes) -0.0353*** 0.0001 -0.0011 -0.0003 -0.0013* 0.0014** 0.0014**
(0.0008) (0.0009) (0.0008) (0.0008) (0.0008) (0.0007) (0.0007)
Democrat Majority, State Legislature (1 = Yes) 0.0029*** -0.0031*** -0.0010* -0.0007 -0.0011* -0.0002 -0.0001
(0.0008) (0.0006) (0.0006) (0.0006) (0.0006) (0.0006) (0.0005)
Democrat Majority, Federal 0.0347*** -0.0086*** -0.0009 -0.0003 -0.0007 0.0002 0.0003
& State Legislature (1 = Yes) (0.0008) (0.0009) (0.0008) (0.0008) (0.0008) (0.0007) (0.0007)
Democrat Governor & President & Federal -0.0022 -0.0017 -0.0009 -0.0013 -0.0008 -0.0014 -0.0012
& State Legislature Majority (1 = Yes) (0.0014) (0.0011) (0.0010) (0.0010) (0.0010) (0.0009) (0.0009)

Observations 129300 129300 129300 129300 129300 117504 117504

R2 0.166 0.462 0.488 0.491 0.554 0.617 0.617
Location Health Controls N N N N N Y Y
Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.

45
 Table F4: Political Party Determinants of “Short-Term”-Cause Mortality: Coefficients

(1) (2) (3) (4) (5) (6) (7)

Contemporaneous Effects
Democrat Governor (1 = Yes) 0.0078*** -0.0055* -0.0048* -0.0041 -0.0045 -0.0058** -0.0055**
(0.0027) (0.0028) (0.0028) (0.0028) (0.0029) (0.0027) (0.0027)
Democrat Governor & President (1 = Yes) -0.0129*** -0.0050 -0.0094** -0.0087** -0.0097** -0.0055 -0.0060*
(0.0033) (0.0038) (0.0038) (0.0038) (0.0038) (0.0036) (0.0036)
Democrat Majority, State Legislature (1 = Yes) -0.0106*** -0.0066* -0.0089** -0.0060* -0.0093*** -0.0067** -0.0066**
(0.0030) (0.0034) (0.0035) (0.0035) (0.0035) (0.0032) (0.0032)
Democrat Majority, Federal 0.0037 0.0009 -0.0006 -0.0011 -0.0012 -0.0023 -0.0029
& State Legislature (1 = Yes) (0.0035) (0.0045) (0.0045) (0.0045) (0.0045) (0.0042) (0.0042)
Democrat Governor & President & Federal -0.0039 0.0025 0.0033 0.0032 0.0042 0.0031 0.0037
& State Legislature Majority (1 = Yes) (0.0046) (0.0050) (0.0050) (0.0050) (0.0050) (0.0048) (0.0048)
Lagged Effects
Democrat Governor (1 = Yes) 0.0041*** 0.0008 0.0025*** 0.0010 0.0024*** 0.0011 0.0011
(0.0009) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009) (0.0009)
Democrat Governor & President (1 = Yes) -0.0010 0.0067*** 0.0036** 0.0037** 0.0035** 0.0044*** 0.0043***
(0.0012) (0.0016) (0.0015) (0.0016) (0.0015) (0.0015) (0.0015)
Democrat Majority, State Legislature (1 = Yes) -0.0115*** 0.0058*** 0.0046*** 0.0048*** 0.0052*** 0.0046*** 0.0046***
(0.0011) (0.0012) (0.0011) (0.0012) (0.0011) (0.0011) (0.0011)
Democrat Majority, Federal 0.0294*** -0.0084*** -0.0051*** -0.0039*** -0.0053*** -0.0042*** -0.0039***
& State Legislature (1 = Yes) (0.0011) (0.0014) (0.0015) (0.0015) (0.0015) (0.0014) (0.0014)
Democrat Governor & President & Federal -0.0146*** -0.0085*** -0.0080*** -0.0077*** -0.0076*** -0.0059*** -0.0057***
& State Legislature Majority (1 = Yes) (0.0021) (0.0023) (0.0023) (0.0024) (0.0023) (0.0022) (0.0022)
Lead Effects
Democrat Governor (1 = Yes) 0.0058*** 0.0029** 0.0026** 0.0017 0.0029** 0.0011 0.0009
(0.0012) (0.0012) (0.0012) (0.0012) (0.0012) (0.0011) (0.0011)
Democrat Governor & President (1 = Yes) -0.0020 0.0031* 0.0031* 0.0037** 0.0027 0.0037** 0.0037**
(0.0015) (0.0018) (0.0017) (0.0019) (0.0018) (0.0017) (0.0017)
Democrat Majority, State Legislature (1 = Yes) -0.0062*** 0.0003 0.0005 0.0004 0.0007 0.0002 0.0003
(0.0013) (0.0014) (0.0014) (0.0014) (0.0014) (0.0013) (0.0013)
Democrat Majority, Federal 0.0381*** -0.0027 -0.0029 -0.0019 -0.0029 -0.0028 -0.0026
& State Legislature (1 = Yes) (0.0017) (0.0020) (0.0019) (0.0020) (0.0019) (0.0019) (0.0019)
Democrat Governor & President & Federal -0.0280*** -0.0119*** -0.0123*** -0.0113*** -0.0125*** -0.0089*** -0.0085***
& State Legislature Majority (1 = Yes) (0.0023) (0.0026) (0.0026) (0.0026) (0.0026) (0.0025) (0.0025)

Observations 113261 113261 113261 113261 113261 105542 105542

R2 0.034 0.095 0.116 0.120 0.205 0.224 0.224
Location Health Controls N N N N N Y Y
Population Health Controls N N N N N Y Y
Income Inequality Controls N N N N N N Y
County FEs Y Y Y Y Y Y Y
Year FEs N Y Y Y Y Y Y
State Time Trends (Linear) N N Y Y N N N
State Time Trends (Quadratic) N N N Y N N N
County Time Trends (Linear) N N N N Y Y Y

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.

46
Appendix F.2 Race & Sex

47
 Table F5: Political Party Determinants of All-Cause Mortality for Black Males: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0031 0.0031 0.0271 0.0256 -0.0122 -0.0120 0.0021 0.0009
P-value 0.2473 0.2533 0.2565 0.2854 0.0509 0.0549 0.6948 0.8655
Legislative Control -0.0020 -0.0020 -0.0023 -0.0021 0.0038 0.0039 -0.0077 -0.0068
P-value 0.5465 0.5400 0.8602 0.8728 0.4678 0.4609 0.2865 0.3504
Full Control 0.0012 0.0011 0.0004 -0.0012 -0.0056 -0.0072 0.0046 0.0089
P-value 0.8001 0.8215 0.9822 0.9416 0.5089 0.4125 0.8531 0.7223
Contemporaneous Effects
Executive Control -0.0077 -0.0076 -0.0021 -0.0027 -0.0097 -0.0093 -0.0003 0.0016
P-value 0.3368 0.3403 0.9657 0.9572 0.2940 0.3233 0.9833 0.8975
Legislative Control 0.0118 0.0115 0.0056 0.0053 0.0029 0.0017 0.0037 0.0085
P-value 0.2569 0.2681 0.8553 0.8613 0.8638 0.9237 0.7330 0.4559
Full Control -0.0036 -0.0037 -0.0213 -0.0212 -0.0094 -0.0126 0.0195 0.0272
P-value 0.7975 0.7909 0.5615 0.5662 0.6165 0.5098 0.4106 0.2699
Lead Effects
Executive Control 0.0029 0.0029 0.0155 0.0155 -0.0042 -0.0041 0.0024 0.0015
P-value 0.3196 0.3320 0.5434 0.5501 0.4014 0.4195 0.6986 0.8141
Legislative Control -0.0027 -0.0027 -0.0139 -0.0131 0.0071 0.0066 -0.0055 -0.0044

48
P-value 0.5237 0.5240 0.3952 0.4247 0.5198 0.5532 0.3724 0.4875
Full Control 0.0024 0.0024 -0.0115 -0.0115 0.0008 -0.0004 0.0088 0.0099
P-value 0.6856 0.6878 0.5604 0.5623 0.9438 0.9758 0.5340 0.4868

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F6: Political Party Determinants of “Short-Term”-Cause Mortality for Black Males: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0206 0.0211 0.1090 0.1155 0.0098 0.0062 -0.0155 -0.0143
P-value 0.0017 0.0012 0.0090 0.0057 0.5514 0.7013 0.1391 0.1737
Legislative Control -0.0394 -0.0402 -0.0222 -0.0226 0.0009 -0.0000 0.0041 0.0028
P-value 0.0000 0.0000 0.2042 0.2060 0.9501 0.9978 0.7948 0.8563
Full Control 0.0026 0.0024 0.0066 0.0135 -0.0142 -0.0027 0.0706 0.0647
P-value 0.8253 0.8406 0.8117 0.6269 0.5486 0.9100 0.1093 0.1449
Contemporaneous Effects
Executive Control -0.0340 -0.0357 -0.1030 -0.1113 0.0314 0.0289 0.0015 0.0040
P-value 0.0339 0.0256 0.1275 0.1076 0.1643 0.1960 0.9419 0.8435
Legislative Control -0.0347 -0.0323 -0.0000 -0.0018 -0.0112 -0.0128 -0.0071 -0.0087
P-value 0.0317 0.0456 0.9992 0.9588 0.7152 0.6822 0.7642 0.7129
Full Control -0.0230 -0.0264 -0.0019 -0.0071 -0.0637 -0.0544 0.0932 0.0858
P-value 0.3302 0.2650 0.9669 0.8766 0.0897 0.1445 0.0365 0.0590
Lead Effects
Executive Control -0.0022 -0.0019 -0.0544 -0.0553 0.0272 0.0253 -0.0129 -0.0112
P-value 0.7719 0.7997 0.1506 0.1648 0.0523 0.0723 0.2344 0.3077
Legislative Control -0.0075 -0.0091 0.0031 0.0015 -0.0093 -0.0116 0.0007 -0.0018

49
P-value 0.4526 0.3559 0.8988 0.9514 0.6441 0.5663 0.9547 0.8820
Full Control -0.0422 -0.0429 -0.0327 -0.0329 0.0135 0.0140 0.0058 0.0042
P-value 0.0021 0.0017 0.2690 0.2637 0.6226 0.6071 0.8184 0.8694

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F7: Political Party Determinants of All-Cause Mortality for Black Males: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) -0.0142** -0.0141** -0.0124 -0.0122 -0.0073 -0.0086 -0.0068 -0.0061
(0.0061) (0.0061) (0.0150) (0.0153) (0.0105) (0.0106) (0.0075) (0.0075)
Democrat Governor 0.0065 0.0065 0.0103 0.0095 -0.0023 -0.0007 0.0066 0.0076
& President (1 = Yes) (0.0086) (0.0086) (0.0484) (0.0488) (0.0138) (0.0140) (0.0142) (0.0142)
Democrat Majority, 0.0147 0.0147 0.0153 0.0176 0.0083 0.0083 0.0011 0.0032
State Legislature (1 = Yes) (0.0093) (0.0093) (0.0326) (0.0328) (0.0153) (0.0153) (0.0106) (0.0108)
Democrat Majority, Federal -0.0029 -0.0031 -0.0097 -0.0122 -0.0055 -0.0066 0.0026 0.0053
& State Legislature (1 = Yes) (0.0115) (0.0115) (0.0397) (0.0399) (0.0197) (0.0199) (0.0118) (0.0121)
Democrat Governor & President & Federal -0.0077 -0.0076 -0.0247 -0.0239 -0.0026 -0.0050 0.0161 0.0171
& State Legislature Majority (1 = Yes) (0.0111) (0.0111) (0.0473) (0.0474) (0.0145) (0.0149) (0.0259) (0.0261)
Lagged Effects
Democrat Governor (1 = Yes) 0.0017 0.0017 0.0056 0.0059 0.0032 0.0032 0.0021 0.0019
(0.0019) (0.0020) (0.0069) (0.0069) (0.0031) (0.0031) (0.0034) (0.0034)
Democrat Governor 0.0014 0.0014 0.0214 0.0198 -0.0153** -0.0153** 0.0000 -0.0010
& President (1 = Yes) (0.0032) (0.0032) (0.0244) (0.0245) (0.0076) (0.0076) (0.0060) (0.0061)
Democrat Majority, 0.0012 0.0012 -0.0222 -0.0213 0.0015 0.0004 -0.0078 -0.0072
State Legislature (1 = Yes) (0.0025) (0.0025) (0.0240) (0.0243) (0.0063) (0.0065) (0.0058) (0.0058)
Democrat Majority, Federal -0.0032 -0.0032 0.0199 0.0192 0.0023 0.0035 0.0001 0.0004
& State Legislature (1 = Yes) (0.0035) (0.0035) (0.0249) (0.0251) (0.0059) (0.0061) (0.0075) (0.0075)
Democrat Governor & President & Federal 0.0001 0.0001 -0.0244 -0.0248 0.0028 0.0010 0.0102 0.0148
& State Legislature Majority (1 = Yes) (0.0049) (0.0049) (0.0243) (0.0243) (0.0089) (0.0093) (0.0260) (0.0265)
Lead Effects
Democrat Governor (1 = Yes) 0.0046* 0.0045* 0.0116 0.0131 0.0040 0.0040 0.0097* 0.0095*
(0.0027) (0.0027) (0.0085) (0.0087) (0.0048) (0.0048) (0.0051) (0.0051)

50
Democrat Governor -0.0016 -0.0016 0.0038 0.0024 -0.0082 -0.0080 -0.0074 -0.0080
& President (1 = Yes) (0.0038) (0.0038) (0.0272) (0.0278) (0.0067) (0.0067) (0.0075) (0.0076)
Democrat Majority, 0.0013 0.0012 -0.0105 -0.0089 -0.0049 -0.0052 -0.0119** -0.0103*
State Legislature (1 = Yes) (0.0033) (0.0033) (0.0195) (0.0198) (0.0070) (0.0070) (0.0059) (0.0060)
Democrat Majority, Federal -0.0040 -0.0039 -0.0033 -0.0042 0.0120 0.0117 0.0064 0.0059
& State Legislature (1 = Yes) (0.0047) (0.0047) (0.0251) (0.0251) (0.0107) (0.0107) (0.0065) (0.0066)
Democrat Governor & President & Federal 0.0022 0.0023 -0.0131 -0.0139 -0.0021 -0.0028 0.0120 0.0129
& State Legislature Majority (1 = Yes) (0.0057) (0.0057) (0.0249) (0.0253) (0.0087) (0.0087) (0.0152) (0.0152)

Observations 54787 54787 22236 22236 18161 18161 14390 14390

R2 0.200 0.200 0.173 0.173 0.312 0.312 0.262 0.263
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three
years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table F8: Political Party Determinants of “Short-Term”-Cause Mortality for Black Males: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) -0.0100 -0.0112 -0.0147 -0.0169 -0.0456* -0.0399* 0.0123 0.0089
(0.0125) (0.0125) (0.0231) (0.0238) (0.0235) (0.0232) (0.0151) (0.0151)
Democrat Governor -0.0241 -0.0246 -0.0883 -0.0944 0.0770** 0.0689** -0.0109 -0.0049
& President (1 = Yes) (0.0180) (0.0181) (0.0671) (0.0677) (0.0316) (0.0309) (0.0235) (0.0238)
Democrat Majority, -0.0623*** -0.0605*** -0.0849 -0.0910 0.0010 -0.0051 -0.0002 -0.0034
State Legislature (1 = Yes) (0.0183) (0.0184) (0.0740) (0.0746) (0.0255) (0.0254) (0.0208) (0.0213)
Democrat Majority, Federal 0.0276 0.0282 0.0849 0.0892 -0.0122 -0.0077 -0.0069 -0.0054
& State Legislature (1 = Yes) (0.0203) (0.0203) (0.0780) (0.0785) (0.0327) (0.0324) (0.0228) (0.0229)
Democrat Governor & President & Federal 0.0457** 0.0416** 0.1012 0.1060 -0.0840** -0.0705* 0.0988** 0.0905*
& State Legislature Majority (1 = Yes) (0.0203) (0.0205) (0.0659) (0.0664) (0.0376) (0.0377) (0.0492) (0.0493)
Lagged Effects
Democrat Governor (1 = Yes) 0.0062 0.0068 -0.0184 -0.0183 -0.0143* -0.0139* -0.0122* -0.0130*
(0.0054) (0.0053) (0.0116) (0.0118) (0.0084) (0.0082) (0.0066) (0.0067)
Democrat Governor 0.0144* 0.0143* 0.1275*** 0.1337*** 0.0241 0.0202 -0.0032 -0.0012
& President (1 = Yes) (0.0085) (0.0084) (0.0446) (0.0446) (0.0203) (0.0201) (0.0112) (0.0112)
Democrat Majority, -0.0113* -0.0109* 0.0659 0.0611 -0.0078 -0.0032 0.0248** 0.0259**
State Legislature (1 = Yes) (0.0061) (0.0061) (0.0473) (0.0483) (0.0164) (0.0167) (0.0102) (0.0102)
Democrat Majority, Federal -0.0281*** -0.0294*** -0.0881* -0.0838* 0.0086 0.0032 -0.0207 -0.0231
& State Legislature (1 = Yes) (0.0085) (0.0084) (0.0475) (0.0479) (0.0136) (0.0141) (0.0157) (0.0157)
Democrat Governor & President & Federal 0.0215* 0.0215* -0.0802* -0.0793* -0.0249 -0.0089 0.0820* 0.0761
& State Legislature Majority (1 = Yes) (0.0123) (0.0123) (0.0413) (0.0412) (0.0262) (0.0264) (0.0479) (0.0484)
Lead Effects
Democrat Governor (1 = Yes) 0.0344*** 0.0334*** 0.0513*** 0.0451*** -0.0077 -0.0074 -0.0101 -0.0112
(0.0065) (0.0064) (0.0148) (0.0147) (0.0118) (0.0120) (0.0098) (0.0099)

51
Democrat Governor -0.0366*** -0.0353*** -0.1056*** -0.1004** 0.0349* 0.0326* -0.0028 0.0000
& President (1 = Yes) (0.0102) (0.0102) (0.0398) (0.0419) (0.0191) (0.0190) (0.0150) (0.0153)
Democrat Majority, 0.0006 0.0019 0.0861** 0.0791** 0.0107 0.0105 -0.0061 -0.0072
State Legislature (1 = Yes) (0.0078) (0.0078) (0.0362) (0.0371) (0.0143) (0.0148) (0.0118) (0.0123)
Democrat Majority, Federal -0.0081 -0.0109 -0.0829* -0.0776* -0.0200 -0.0221 0.0068 0.0053
& State Legislature (1 = Yes) (0.0108) (0.0108) (0.0451) (0.0455) (0.0207) (0.0208) (0.0113) (0.0112)
Democrat Governor & President & Federal -0.0325*** -0.0319*** 0.0186 0.0209 -0.0044 0.0003 0.0181 0.0172
& State Legislature Majority (1 = Yes) (0.0114) (0.0114) (0.0388) (0.0402) (0.0219) (0.0218) (0.0288) (0.0289)

Observations 19738 19738 11197 11197 4413 4413 4128 4128

R2 0.388 0.388 0.453 0.453 0.391 0.394 0.318 0.319
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three years
(constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table F9: Political Party Determinants of All-Cause Mortality for Black Females: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0030 0.0033 -0.0326 -0.0325 0.0003 0.0003 -0.0019 -0.0007
P-value 0.2117 0.1770 0.1592 0.1609 0.9570 0.9659 0.6746 0.8776
Legislative Control -0.0013 -0.0011 -0.0177 -0.0186 -0.0021 -0.0020 -0.0058 -0.0064
P-value 0.6765 0.7290 0.1552 0.1374 0.6881 0.7025 0.3311 0.2851
Full Control -0.0031 -0.0026 -0.0370 -0.0371 0.0011 0.0010 0.0195 0.0145
P-value 0.4886 0.5671 0.0152 0.0152 0.8938 0.9000 0.3305 0.4754
Contemporaneous Effects
Executive Control -0.0068 -0.0076 0.0548 0.0573 -0.0255 -0.0255 -0.0235 -0.0263
P-value 0.3530 0.3011 0.1983 0.1806 0.0047 0.0053 0.0221 0.0120
Legislative Control -0.0171 -0.0171 -0.0195 -0.0200 -0.0133 -0.0129 -0.0190 -0.0242
P-value 0.0731 0.0741 0.4840 0.4714 0.4201 0.4441 0.0385 0.0115
Full Control -0.0305 -0.0310 -0.0497 -0.0482 -0.0216 -0.0222 -0.0193 -0.0278
P-value 0.0155 0.0143 0.1346 0.1491 0.2267 0.2209 0.3204 0.1706
Lead Effects
Executive Control 0.0023 0.0027 -0.0300 -0.0275 0.0002 0.0003 0.0044 0.0051
P-value 0.3985 0.3352 0.1991 0.2459 0.9696 0.9453 0.3269 0.2678
Legislative Control 0.0016 0.0020 -0.0054 -0.0062 0.0136 0.0137 0.0007 -0.0003

52
P-value 0.6927 0.6199 0.7249 0.6870 0.2542 0.2480 0.8890 0.9554
Full Control 0.0073 0.0079 -0.0036 -0.0040 0.0076 0.0070 0.0032 0.0013
P-value 0.1751 0.1407 0.8462 0.8320 0.5298 0.5602 0.7792 0.9094

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F10: Political Party Determinants of “Short-Term”-Cause Mortality for Black Females: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0088 0.0101 -0.0246 -0.0274 0.0137 0.0132 -0.0275 -0.0231
P-value 0.1613 0.1125 0.5373 0.4942 0.3214 0.3361 0.0073 0.0247
Legislative Control -0.0273 -0.0272 -0.0567 -0.0610 0.0010 0.0015 -0.0054 -0.0071
P-value 0.0008 0.0009 0.0083 0.0053 0.9376 0.9037 0.7485 0.6742
Full Control -0.0083 -0.0063 -0.0425 -0.0459 -0.0162 -0.0134 0.0429 0.0267
P-value 0.5239 0.6286 0.1860 0.1556 0.4189 0.5168 0.4296 0.6234
Contemporaneous Effects
Executive Control -0.0088 -0.0120 -0.0401 -0.0297 0.0180 0.0169 -0.0068 -0.0228
P-value 0.6231 0.5020 0.5633 0.6741 0.3838 0.4195 0.7614 0.3298
Legislative Control -0.0159 -0.0169 -0.0034 -0.0060 -0.0023 -0.0026 -0.0080 -0.0305
P-value 0.4416 0.4134 0.9389 0.8922 0.9394 0.9312 0.7603 0.2443
Full Control -0.0531 -0.0521 -0.0399 -0.0352 -0.0350 -0.0352 -0.0116 -0.0368
P-value 0.0797 0.0853 0.4991 0.5510 0.3667 0.3629 0.7955 0.4212
Lead Effects
Executive Control 0.0027 0.0036 -0.0555 -0.0470 0.0179 0.0184 -0.0067 -0.0062
P-value 0.7328 0.6481 0.1568 0.2407 0.1422 0.1380 0.5693 0.5989
Legislative Control 0.0021 0.0028 0.0075 0.0054 -0.0112 -0.0110 0.0219 0.0214

53
P-value 0.8432 0.7868 0.7969 0.8522 0.5537 0.5628 0.1211 0.1328
Full Control -0.0152 -0.0125 -0.0043 -0.0050 0.0319 0.0302 0.0153 0.0043
P-value 0.2745 0.3713 0.9011 0.8863 0.1768 0.1854 0.6264 0.8901

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F11: Political Party Determinants of All-Cause Mortality for Black Females: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) -0.0044 -0.0042 0.0088 0.0102 0.0018 0.0018 0.0133** 0.0131**
(0.0059) (0.0059) (0.0148) (0.0149) (0.0114) (0.0114) (0.0066) (0.0066)
Democrat Governor -0.0024 -0.0034 0.0459 0.0471 -0.0273* -0.0273* -0.0367*** -0.0395***
& President (1 = Yes) (0.0081) (0.0081) (0.0423) (0.0424) (0.0142) (0.0143) (0.0113) (0.0116)
Democrat Majority, -0.0031 -0.0029 -0.0424 -0.0446 0.0013 0.0014 -0.0114 -0.0134
State Legislature (1 = Yes) (0.0089) (0.0090) (0.0332) (0.0335) (0.0159) (0.0159) (0.0086) (0.0087)
Democrat Majority, Federal -0.0140 -0.0143 0.0229 0.0246 -0.0145 -0.0142 -0.0076 -0.0108
& State Legislature (1 = Yes) (0.0107) (0.0107) (0.0385) (0.0388) (0.0190) (0.0191) (0.0103) (0.0104)
Democrat Governor & President & Federal -0.0067 -0.0062 -0.0850** -0.0855** 0.0172 0.0162 0.0231 0.0227
& State Legislature Majority (1 = Yes) (0.0105) (0.0106) (0.0406) (0.0406) (0.0139) (0.0141) (0.0198) (0.0201)
Lagged Effects
Democrat Governor (1 = Yes) 0.0032* 0.0035* 0.0141** 0.0145** 0.0027 0.0029 0.0022 0.0024
(0.0018) (0.0018) (0.0068) (0.0068) (0.0031) (0.0031) (0.0028) (0.0028)
Democrat Governor -0.0002 -0.0003 -0.0468* -0.0470* -0.0024 -0.0026 -0.0041 -0.0031
& President (1 = Yes) (0.0029) (0.0029) (0.0246) (0.0246) (0.0074) (0.0074) (0.0049) (0.0049)
Democrat Majority, 0.0022 0.0024 -0.0159 -0.0180 -0.0001 -0.0005 -0.0031 -0.0039
State Legislature (1 = Yes) (0.0023) (0.0023) (0.0216) (0.0218) (0.0058) (0.0059) (0.0045) (0.0045)
Democrat Majority, Federal -0.0035 -0.0035 -0.0017 -0.0006 -0.0019 -0.0015 -0.0027 -0.0025
& State Legislature (1 = Yes) (0.0032) (0.0032) (0.0231) (0.0232) (0.0055) (0.0057) (0.0065) (0.0065)
Democrat Governor & President & Federal -0.0048 -0.0048 0.0133 0.0140 0.0028 0.0027 0.0271 0.0216
& State Legislature Majority (1 = Yes) (0.0044) (0.0044) (0.0234) (0.0235) (0.0084) (0.0088) (0.0212) (0.0216)
Lead Effects
Democrat Governor (1 = Yes) 0.0009 0.0010 0.0120 0.0117 0.0011 0.0010 0.0034 0.0038
(0.0025) (0.0025) (0.0079) (0.0081) (0.0050) (0.0050) (0.0046) (0.0047)

54
Democrat Governor 0.0014 0.0017 -0.0420* -0.0392 -0.0009 -0.0007 0.0010 0.0013
& President (1 = Yes) (0.0034) (0.0034) (0.0241) (0.0245) (0.0065) (0.0065) (0.0063) (0.0063)
Democrat Majority, 0.0036 0.0039 0.0018 -0.0005 -0.0041 -0.0036 -0.0080 -0.0097*
State Legislature (1 = Yes) (0.0031) (0.0032) (0.0189) (0.0193) (0.0074) (0.0073) (0.0049) (0.0050)
Democrat Majority, Federal -0.0020 -0.0018 -0.0072 -0.0057 0.0178 0.0173 0.0087 0.0094*
& State Legislature (1 = Yes) (0.0044) (0.0045) (0.0225) (0.0227) (0.0111) (0.0111) (0.0056) (0.0057)
Democrat Governor & President & Federal 0.0033 0.0032 0.0318 0.0298 -0.0062 -0.0070 -0.0019 -0.0035
& State Legislature Majority (1 = Yes) (0.0053) (0.0053) (0.0233) (0.0235) (0.0081) (0.0082) (0.0126) (0.0127)

Observations 52634 52634 21214 21214 17507 17507 13913 13913

R2 0.177 0.177 0.176 0.176 0.303 0.303 0.265 0.265
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three years
(constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table F12: Political Party Determinants of “Short-Term”-Cause Mortality for Black Females: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) 0.0031 0.0055 0.0068 0.0124 -0.0194 -0.0156 0.0027 0.0037
(0.0144) (0.0143) (0.0276) (0.0277) (0.0236) (0.0236) (0.0200) (0.0198)
Democrat Governor -0.0119 -0.0175 -0.0469 -0.0421 0.0374 0.0324 -0.0095 -0.0265
& President (1 = Yes) (0.0190) (0.0190) (0.0710) (0.0715) (0.0288) (0.0284) (0.0275) (0.0275)
Democrat Majority, -0.0106 -0.0091 -0.0036 -0.0107 0.0112 0.0090 0.0118 0.0056
State Legislature (1 = Yes) (0.0203) (0.0204) (0.0721) (0.0725) (0.0221) (0.0223) (0.0215) (0.0211)
Democrat Majority, Federal -0.0053 -0.0078 0.0002 0.0047 -0.0134 -0.0115 -0.0199 -0.0361
& State Legislature (1 = Yes) (0.0236) (0.0235) (0.0791) (0.0792) (0.0317) (0.0314) (0.0272) (0.0273)
Democrat Governor & President & Federal -0.0284 -0.0232 0.0035 0.0005 -0.0507 -0.0494 0.0033 0.0165
& State Legislature Majority (1 = Yes) (0.0237) (0.0235) (0.0651) (0.0653) (0.0372) (0.0365) (0.0464) (0.0462)
Lagged Effects
Democrat Governor (1 = Yes) -0.0031 -0.0025 -0.0270** -0.0257* -0.0031 -0.0027 -0.0019 0.0016
(0.0050) (0.0051) (0.0131) (0.0133) (0.0082) (0.0084) (0.0079) (0.0081)
Democrat Governor 0.0119 0.0126 0.0024 -0.0017 0.0168 0.0159 -0.0257** -0.0246**
& President (1 = Yes) (0.0081) (0.0081) (0.0429) (0.0433) (0.0181) (0.0179) (0.0122) (0.0121)
Democrat Majority, -0.0118* -0.0126** 0.0769 0.0707 -0.0041 -0.0027 0.0059 0.0040
State Legislature (1 = Yes) (0.0064) (0.0064) (0.0527) (0.0540) (0.0153) (0.0154) (0.0129) (0.0130)
Democrat Majority, Federal -0.0155* -0.0146* -0.1335** -0.1317** 0.0051 0.0042 -0.0113 -0.0111
& State Legislature (1 = Yes) (0.0086) (0.0086) (0.0560) (0.0564) (0.0126) (0.0124) (0.0160) (0.0162)
Democrat Governor & President & Federal 0.0101 0.0108 0.0387 0.0425 -0.0309 -0.0281 0.0758 0.0569
& State Legislature Majority (1 = Yes) (0.0130) (0.0130) (0.0421) (0.0423) (0.0225) (0.0227) (0.0564) (0.0567)
Lead Effects
Democrat Governor (1 = Yes) 0.0133** 0.0114* 0.0119 0.0111 0.0180 0.0180 0.0013 0.0067
(0.0061) (0.0061) (0.0160) (0.0164) (0.0126) (0.0127) (0.0123) (0.0126)

55
Democrat Governor -0.0106 -0.0078 -0.0674 -0.0581 -0.0001 0.0003 -0.0080 -0.0130
& President (1 = Yes) (0.0096) (0.0097) (0.0421) (0.0439) (0.0147) (0.0147) (0.0186) (0.0188)
Democrat Majority, -0.0012 -0.0014 0.0555 0.0488 0.0196 0.0200 0.0084 0.0014
State Legislature (1 = Yes) (0.0082) (0.0082) (0.0372) (0.0382) (0.0143) (0.0141) (0.0123) (0.0123)
Democrat Majority, Federal 0.0033 0.0042 -0.0481 -0.0433 -0.0308* -0.0311* 0.0136 0.0200
& State Legislature (1 = Yes) (0.0109) (0.0110) (0.0465) (0.0469) (0.0186) (0.0185) (0.0145) (0.0147)
Democrat Governor & President & Federal -0.0199 -0.0189 0.0437 0.0366 0.0252 0.0228 0.0001 -0.0108
& State Legislature Majority (1 = Yes) (0.0132) (0.0132) (0.0396) (0.0399) (0.0248) (0.0244) (0.0334) (0.0335)

Observations 13825 13825 8109 8109 3067 3067 2649 2649

R2 0.254 0.255 0.284 0.284 0.372 0.374 0.336 0.341
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three
years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table F13: Political Party Determinants of All-Cause Mortality for White Males: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0007 0.0006 0.0031 0.0031 0.0003 0.0003 -0.0022 -0.0020
P-value 0.2949 0.3848 0.3080 0.3098 0.8805 0.8635 0.2002 0.2330
Legislative Control 0.0014 0.0015 -0.0002 -0.0003 0.0030 0.0030 -0.0001 -0.0004
P-value 0.0338 0.0270 0.9352 0.8974 0.0356 0.0368 0.9723 0.8786
Full Control 0.0020 0.0021 -0.0006 -0.0011 0.0007 0.0008 -0.0039 -0.0064
P-value 0.0771 0.0765 0.8489 0.7161 0.7812 0.7558 0.6286 0.4245
Contemporaneous Effects
Executive Control -0.0015 -0.0015 0.0090 0.0097 0.0016 0.0017 -0.0008 -0.0021
P-value 0.4349 0.4215 0.1565 0.1301 0.6360 0.6243 0.8200 0.5602
Legislative Control 0.0015 0.0014 0.0014 0.0007 0.0055 0.0054 0.0023 -0.0001
P-value 0.4604 0.5132 0.7201 0.8691 0.1623 0.1705 0.6021 0.9780
Full Control -0.0021 -0.0022 0.0037 0.0039 -0.0014 -0.0011 -0.0038 -0.0085
P-value 0.4498 0.4389 0.5120 0.4899 0.8114 0.8467 0.6186 0.2740
Lead Effects
Executive Control -0.0008 -0.0008 -0.0019 -0.0012 0.0001 0.0000 0.0032 0.0032
P-value 0.2929 0.2542 0.5489 0.7055 0.9621 0.9990 0.0634 0.0619
Legislative Control -0.0005 -0.0003 -0.0002 -0.0002 0.0003 0.0003 -0.0017 -0.0017

56
P-value 0.6299 0.7286 0.9467 0.9229 0.8848 0.9012 0.5290 0.5231
Full Control -0.0009 -0.0007 0.0006 0.0004 -0.0000 0.0001 -0.0007 -0.0011
P-value 0.4887 0.6046 0.8514 0.8949 0.9878 0.9612 0.8847 0.8089

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F14: Political Party Determinants of “Short-Term”-Cause Mortality for White Males: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0012 0.0013 0.0307 0.0308 -0.0140 -0.0140 -0.0056 -0.0052
P-value 0.6320 0.6233 0.0096 0.0100 0.0454 0.0459 0.1322 0.1642
Legislative Control -0.0302 -0.0308 -0.0037 -0.0113 -0.0066 -0.0068 -0.0134 -0.0139
P-value 0.0000 0.0000 0.5762 0.0920 0.2002 0.1886 0.0157 0.0123
Full Control 0.0080 0.0071 0.0117 0.0124 -0.0141 -0.0086 0.0180 0.0143
P-value 0.1182 0.1658 0.2482 0.2235 0.2325 0.4625 0.2634 0.3753
Contemporaneous Effects
Executive Control -0.0171 -0.0160 -0.0219 -0.0120 0.0019 0.0018 -0.0149 -0.0154
P-value 0.0029 0.0053 0.2472 0.5290 0.8569 0.8603 0.0384 0.0330
Legislative Control -0.0165 -0.0155 0.0143 0.0063 -0.0002 -0.0054 -0.0090 -0.0115
P-value 0.0112 0.0163 0.2385 0.6008 0.9852 0.6546 0.3296 0.2116
Full Control 0.0129 0.0132 0.0551 0.0582 -0.0666 -0.0644 -0.0083 -0.0142
P-value 0.1891 0.1797 0.0021 0.0011 0.0005 0.0007 0.6000 0.3767
Lead Effects
Executive Control 0.0019 0.0019 -0.0307 -0.0244 0.0187 0.0165 -0.0005 -0.0004
P-value 0.5213 0.5030 0.0024 0.0170 0.0010 0.0034 0.8890 0.9094
Legislative Control 0.0135 0.0125 0.0063 0.0014 -0.0106 -0.0111 0.0009 0.0008

57
P-value 0.0001 0.0003 0.3794 0.8455 0.1894 0.1658 0.8453 0.8644
Full Control -0.0019 -0.0030 -0.0154 -0.0162 -0.0115 -0.0110 0.0106 0.0091
P-value 0.7211 0.5655 0.1088 0.0911 0.3306 0.3510 0.2673 0.3436

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F15: Political Party Determinants of All-Cause Mortality for White Males: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) -0.0017 -0.0016 0.0051* 0.0057* -0.0044 -0.0045 -0.0012 -0.0009
(0.0015) (0.0015) (0.0030) (0.0030) (0.0029) (0.0029) (0.0031) (0.0032)
Democrat Governor 0.0002 0.0000 0.0040 0.0040 0.0060 0.0061 0.0004 -0.0011
& President (1 = Yes) (0.0021) (0.0021) (0.0067) (0.0067) (0.0042) (0.0042) (0.0043) (0.0043)
Democrat Majority, -0.0006 -0.0006 -0.0005 -0.0004 0.0022 0.0022 -0.0014 -0.0025
State Legislature (1 = Yes) (0.0019) (0.0019) (0.0054) (0.0053) (0.0037) (0.0037) (0.0037) (0.0037)
Democrat Majority, Federal 0.0021 0.0019 0.0019 0.0010 0.0033 0.0032 0.0038 0.0024
& State Legislature (1 = Yes) (0.0023) (0.0023) (0.0058) (0.0058) (0.0041) (0.0041) (0.0048) (0.0047)
Democrat Governor & President & Federal -0.0021 -0.0020 -0.0068 -0.0065 -0.0085 -0.0081 -0.0053 -0.0062
& State Legislature Majority (1 = Yes) (0.0026) (0.0026) (0.0068) (0.0068) (0.0054) (0.0054) (0.0085) (0.0085)
Lagged Effects
Democrat Governor (1 = Yes) 0.0004 0.0004 -0.0000 0.0001 0.0006 0.0005 -0.0010 -0.0010
(0.0005) (0.0005) (0.0014) (0.0014) (0.0009) (0.0009) (0.0012) (0.0012)
Democrat Governor 0.0003 0.0002 0.0031 0.0029 -0.0003 -0.0002 -0.0011 -0.0010
& President (1 = Yes) (0.0008) (0.0008) (0.0034) (0.0034) (0.0020) (0.0020) (0.0019) (0.0019)
Democrat Majority, 0.0025*** 0.0025*** -0.0016 -0.0021 0.0022 0.0022 0.0006 0.0003
State Legislature (1 = Yes) (0.0006) (0.0006) (0.0039) (0.0039) (0.0014) (0.0014) (0.0017) (0.0017)
Democrat Majority, Federal -0.0011 -0.0010 0.0014 0.0018 0.0008 0.0008 -0.0007 -0.0007
& State Legislature (1 = Yes) (0.0008) (0.0008) (0.0039) (0.0039) (0.0014) (0.0014) (0.0031) (0.0031)
Democrat Governor & President & Federal -0.0001 -0.0000 -0.0035 -0.0039 -0.0025 -0.0025 -0.0016 -0.0039
& State Legislature Majority (1 = Yes) (0.0012) (0.0012) (0.0036) (0.0037) (0.0029) (0.0030) (0.0088) (0.0088)
Lead Effects
Democrat Governor (1 = Yes) -0.0004 -0.0006 -0.0015 -0.0012 0.0020 0.0020 0.0000 0.0001
(0.0006) (0.0006) (0.0016) (0.0016) (0.0014) (0.0014) (0.0019) (0.0019)

58
Democrat Governor -0.0003 -0.0003 -0.0004 -0.0000 -0.0020 -0.0020 0.0032 0.0032
& President (1 = Yes) (0.0009) (0.0009) (0.0033) (0.0033) (0.0019) (0.0019) (0.0026) (0.0027)
Democrat Majority, -0.0002 -0.0001 0.0013 0.0010 0.0019 0.0019 -0.0021 -0.0020
State Legislature (1 = Yes) (0.0008) (0.0008) (0.0027) (0.0028) (0.0020) (0.0020) (0.0021) (0.0021)
Democrat Majority, Federal -0.0003 -0.0002 -0.0014 -0.0012 -0.0016 -0.0017 0.0003 0.0002
& State Legislature (1 = Yes) (0.0010) (0.0010) (0.0033) (0.0034) (0.0025) (0.0025) (0.0024) (0.0024)
Democrat Governor & President & Federal 0.0003 0.0005 0.0026 0.0019 -0.0004 -0.0001 -0.0022 -0.0026
& State Legislature Majority (1 = Yes) (0.0013) (0.0013) (0.0034) (0.0034) (0.0029) (0.0029) (0.0053) (0.0054)

Observations 112829 112829 33595 33595 40008 40008 39226 39226

R2 0.582 0.582 0.326 0.326 0.264 0.264 0.241 0.241
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three
years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table F16: Political Party Determinants of “Short-Term”-Cause Mortality for White Males: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) 0.0026 0.0016 0.0123 0.0197** -0.0146 -0.0133 0.0049 0.0054
(0.0048) (0.0047) (0.0095) (0.0095) (0.0094) (0.0094) (0.0058) (0.0058)
Democrat Governor -0.0197*** -0.0176*** -0.0342* -0.0316* 0.0165 0.0151 -0.0198** -0.0208**
& President (1 = Yes) (0.0062) (0.0061) (0.0192) (0.0191) (0.0135) (0.0134) (0.0088) (0.0089)
Democrat Majority, -0.0105 -0.0106 0.0401** 0.0251 -0.0204* -0.0224* -0.0032 -0.0048
State Legislature (1 = Yes) (0.0068) (0.0068) (0.0203) (0.0204) (0.0123) (0.0122) (0.0075) (0.0075)
Democrat Majority, Federal -0.0060 -0.0049 -0.0258 -0.0187 0.0202 0.0170 -0.0058 -0.0066
& State Legislature (1 = Yes) (0.0073) (0.0073) (0.0208) (0.0208) (0.0134) (0.0134) (0.0090) (0.0090)
Democrat Governor & President & Federal 0.0465*** 0.0447*** 0.0628*** 0.0638*** -0.0683*** -0.0609*** 0.0156 0.0127
& State Legislature Majority (1 = Yes) (0.0084) (0.0084) (0.0189) (0.0189) (0.0187) (0.0187) (0.0175) (0.0177)
Lagged Effects
Democrat Governor (1 = Yes) 0.0048** 0.0046** -0.0000 0.0021 -0.0071** -0.0076** -0.0005 -0.0006
(0.0021) (0.0021) (0.0046) (0.0046) (0.0033) (0.0033) (0.0025) (0.0025)
Democrat Governor -0.0036 -0.0034 0.0307** 0.0287** -0.0069 -0.0064 -0.0051 -0.0045
& President (1 = Yes) (0.0031) (0.0031) (0.0128) (0.0128) (0.0079) (0.0079) (0.0040) (0.0040)
Democrat Majority, -0.0049* -0.0049* 0.0264* 0.0132 -0.0216*** -0.0206*** 0.0021 0.0018
State Legislature (1 = Yes) (0.0025) (0.0025) (0.0135) (0.0136) (0.0062) (0.0062) (0.0036) (0.0036)
Democrat Majority, Federal -0.0253*** -0.0259*** -0.0301** -0.0244* 0.0150*** 0.0138** -0.0155*** -0.0157***
& State Legislature (1 = Yes) (0.0034) (0.0034) (0.0136) (0.0137) (0.0056) (0.0057) (0.0056) (0.0056)
Democrat Governor & President & Federal 0.0370*** 0.0367*** -0.0153 -0.0071 0.0065 0.0122 0.0370** 0.0334*
& State Legislature Majority (1 = Yes) (0.0055) (0.0056) (0.0130) (0.0130) (0.0125) (0.0125) (0.0179) (0.0179)
Lead Effects
Democrat Governor (1 = Yes) 0.0120*** 0.0121*** 0.0292*** 0.0268*** -0.0099* -0.0100* -0.0078* -0.0081**
(0.0027) (0.0027) (0.0056) (0.0055) (0.0055) (0.0055) (0.0041) (0.0040)

59
Democrat Governor -0.0102*** -0.0102*** -0.0599*** -0.0512*** 0.0286*** 0.0265*** 0.0073 0.0077
& President (1 = Yes) (0.0039) (0.0039) (0.0110) (0.0112) (0.0071) (0.0070) (0.0054) (0.0054)
Democrat Majority, 0.0197*** 0.0196*** 0.0634*** 0.0454*** 0.0098 0.0112 -0.0005 -0.0005
State Legislature (1 = Yes) (0.0032) (0.0032) (0.0102) (0.0105) (0.0073) (0.0073) (0.0042) (0.0042)
Democrat Majority, Federal -0.0062* -0.0071* -0.0572*** -0.0440*** -0.0203** -0.0223*** 0.0014 0.0013
& State Legislature (1 = Yes) (0.0037) (0.0038) (0.0120) (0.0121) (0.0084) (0.0084) (0.0045) (0.0045)
Democrat Governor & President & Federal -0.0172*** -0.0175*** 0.0091 0.0068 -0.0196* -0.0163 0.0102 0.0087
& State Legislature Majority (1 = Yes) (0.0045) (0.0045) (0.0109) (0.0109) (0.0110) (0.0109) (0.0105) (0.0106)

Observations 71277 71277 31751 31751 17966 17966 21560 21560

R2 0.505 0.505 0.623 0.624 0.385 0.387 0.152 0.153
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 dif-
ferent age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college
degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality
controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to the
effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three years (constrained to be
equal). * p <.10, ** p< .05, *** p<.01.
 Table F17: Political Party Determinants of All-Cause Mortality for White Females: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0025 0.0024 0.0019 0.0015 0.0036 0.0034 0.0009 0.0011
P-value 0.0003 0.0005 0.5991 0.6747 0.0477 0.0615 0.5777 0.5039
Legislative Control 0.0015 0.0015 -0.0012 -0.0002 0.0031 0.0030 0.0011 0.0011
P-value 0.0409 0.0367 0.5881 0.9433 0.0214 0.0302 0.6935 0.6911
Full Control 0.0011 0.0010 0.0013 0.0006 0.0010 0.0003 0.0095 0.0080
P-value 0.3793 0.3994 0.6761 0.8405 0.6998 0.8989 0.2409 0.3283
Contemporaneous Effects
Executive Control -0.0020 -0.0020 -0.0007 -0.0023 -0.0033 -0.0029 0.0036 0.0034
P-value 0.3228 0.3318 0.9128 0.7332 0.3242 0.3951 0.3064 0.3372
Legislative Control -0.0028 -0.0029 -0.0027 -0.0025 -0.0039 -0.0033 0.0001 -0.0009
P-value 0.2016 0.1778 0.5316 0.5579 0.3531 0.4245 0.9868 0.8331
Full Control -0.0029 -0.0029 0.0055 0.0051 -0.0108 -0.0103 0.0107 0.0087
P-value 0.3482 0.3412 0.3681 0.3968 0.0729 0.0866 0.1659 0.2682
Lead Effects
Executive Control 0.0006 0.0005 0.0011 0.0005 0.0049 0.0050 -0.0012 -0.0012
P-value 0.4342 0.4970 0.7787 0.8866 0.0026 0.0019 0.5212 0.5111
Legislative Control -0.0003 -0.0003 -0.0024 -0.0017 -0.0004 -0.0005 -0.0006 -0.0007

60
P-value 0.7590 0.7877 0.3740 0.5157 0.8586 0.8483 0.8010 0.7837
Full Control -0.0024 -0.0024 -0.0010 -0.0013 0.0009 0.0006 0.0040 0.0033
P-value 0.0925 0.0982 0.7773 0.7302 0.7820 0.8573 0.4059 0.4866

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F18: Political Party Determinants of “Short-Term”-Cause Mortality for White Females: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged Effects
Executive Control 0.0052 0.0054 0.0182 0.0181 0.0124 0.0119 -0.0023 -0.0018
P-value 0.0343 0.0262 0.1117 0.1144 0.0722 0.0828 0.6045 0.6784
Legislative Control -0.0236 -0.0240 -0.0072 -0.0109 0.0066 0.0061 0.0007 0.0001
P-value 0.0000 0.0000 0.2980 0.1213 0.2021 0.2380 0.9199 0.9864
Full Control -0.0028 -0.0026 0.0136 0.0148 -0.0224 -0.0203 0.0255 0.0210
P-value 0.5619 0.5865 0.1885 0.1525 0.0377 0.0611 0.2050 0.2985
Contemporaneous Effects
Executive Control -0.0292 -0.0293 -0.0614 -0.0581 0.0074 0.0063 -0.0123 -0.0145
P-value 0.0000 0.0000 0.0035 0.0060 0.4535 0.5202 0.1574 0.0988
Legislative Control -0.0205 -0.0194 -0.0011 -0.0040 -0.0138 -0.0163 -0.0096 -0.0133
P-value 0.0035 0.0057 0.9371 0.7704 0.2522 0.1791 0.3712 0.2178
Full Control -0.0240 -0.0237 -0.0077 -0.0063 -0.0713 -0.0704 -0.0051 -0.0133
P-value 0.0225 0.0245 0.6884 0.7458 0.0003 0.0004 0.7876 0.4902
Lead Effects
Executive Control 0.0036 0.0042 0.0087 0.0110 0.0154 0.0152 -0.0050 -0.0050
P-value 0.1972 0.1405 0.4457 0.3399 0.0080 0.0088 0.2530 0.2494
Legislative Control 0.0043 0.0036 0.0071 0.0048 -0.0057 -0.0059 0.0046 0.0048

61
P-value 0.2133 0.2975 0.3931 0.5622 0.4593 0.4478 0.4435 0.4174
Full Control -0.0117 -0.0116 -0.0007 -0.0006 -0.0179 -0.0183 0.0051 0.0032
P-value 0.0227 0.0241 0.9513 0.9576 0.1096 0.0997 0.6672 0.7875

Location Health Controls Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016
Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across
18 different age categories. Population health controls include population shares by education (HS degree, some college, and at least a
4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male).
Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged
values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future
three years (constrained to be equal).
 Table F19: Political Party Determinants of All-Cause Mortality for White Females: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) -0.0013 -0.0012 0.0029 0.0026 -0.0063** -0.0060** 0.0039 0.0041
(0.0015) (0.0015) (0.0034) (0.0034) (0.0028) (0.0028) (0.0028) (0.0028)
Democrat Governor -0.0008 -0.0008 -0.0037 -0.0049 0.0030 0.0032 -0.0003 -0.0007
& President (1 = Yes) (0.0021) (0.0021) (0.0070) (0.0070) (0.0041) (0.0041) (0.0042) (0.0042)
Democrat Majority, -0.0038* -0.0038* -0.0031 -0.0001 -0.0089** -0.0086** 0.0002 -0.0001
State Legislature (1 = Yes) (0.0019) (0.0019) (0.0055) (0.0056) (0.0041) (0.0041) (0.0036) (0.0037)
Democrat Majority, Federal 0.0010 0.0009 0.0003 -0.0024 0.0050 0.0053 -0.0001 -0.0008
& State Legislature (1 = Yes) (0.0023) (0.0023) (0.0064) (0.0064) (0.0043) (0.0043) (0.0047) (0.0047)
Democrat Governor & President & Federal 0.0019 0.0020 0.0089 0.0100 -0.0036 -0.0041 0.0071 0.0063
& State Legislature Majority (1 = Yes) (0.0027) (0.0027) (0.0071) (0.0071) (0.0056) (0.0056) (0.0084) (0.0085)
Lagged Effects
Democrat Governor (1 = Yes) -0.0000 -0.0000 -0.0015 -0.0015 -0.0003 -0.0002 0.0008 0.0008
(0.0005) (0.0005) (0.0015) (0.0015) (0.0010) (0.0010) (0.0012) (0.0012)
Democrat Governor 0.0025*** 0.0024*** 0.0034 0.0030 0.0038* 0.0036* 0.0001 0.0003
& President (1 = Yes) (0.0008) (0.0008) (0.0039) (0.0039) (0.0021) (0.0021) (0.0019) (0.0019)
Democrat Majority, 0.0021*** 0.0020*** -0.0024 0.0000 0.0035** 0.0035** 0.0006 0.0005
State Legislature (1 = Yes) (0.0006) (0.0006) (0.0045) (0.0046) (0.0015) (0.0015) (0.0018) (0.0018)
Democrat Majority, Federal -0.0006 -0.0006 0.0011 -0.0002 -0.0004 -0.0005 0.0005 0.0006
& State Legislature (1 = Yes) (0.0008) (0.0008) (0.0046) (0.0046) (0.0014) (0.0014) (0.0030) (0.0030)
Democrat Governor & President & Federal -0.0029** -0.0029** 0.0007 -0.0007 -0.0057* -0.0060** 0.0075 0.0058
& State Legislature Majority (1 = Yes) (0.0012) (0.0012) (0.0039) (0.0040) (0.0030) (0.0030) (0.0087) (0.0088)
Lead Effects
Democrat Governor (1 = Yes) 0.0003 0.0002 -0.0037** -0.0029 0.0017 0.0017 0.0020 0.0019
(0.0007) (0.0007) (0.0018) (0.0018) (0.0014) (0.0014) (0.0019) (0.0019)

62
Democrat Governor 0.0003 0.0003 0.0048 0.0035 0.0032* 0.0033* -0.0032 -0.0031
& President (1 = Yes) (0.0010) (0.0010) (0.0040) (0.0040) (0.0019) (0.0019) (0.0027) (0.0027)
Democrat Majority, -0.0008 -0.0008 -0.0015 0.0012 0.0012 0.0007 -0.0020 -0.0019
State Legislature (1 = Yes) (0.0008) (0.0008) (0.0031) (0.0032) (0.0019) (0.0020) (0.0021) (0.0021)
Democrat Majority, Federal 0.0005 0.0005 -0.0009 -0.0029 -0.0017 -0.0012 0.0013 0.0012
& State Legislature (1 = Yes) (0.0010) (0.0010) (0.0038) (0.0039) (0.0025) (0.0025) (0.0023) (0.0023)
Democrat Governor & President & Federal -0.0027** -0.0026* 0.0003 -0.0001 -0.0036 -0.0040 0.0058 0.0052
& State Legislature Majority (1 = Yes) (0.0014) (0.0014) (0.0040) (0.0040) (0.0031) (0.0031) (0.0051) (0.0051)

Observations 112594 112594 33590 33590 39912 39912 39092 39092

R2 0.297 0.298 0.288 0.289 0.168 0.169 0.192 0.192
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18
different age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year
college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income
inequality controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values
correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three
years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table F20: Political Party Determinants of “Short-Term”-Cause Mortality for White Females: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8)

Contemporaneous Effects
Democrat Governor (1 = Yes) -0.0089* -0.0099* -0.0198* -0.0169 -0.0054 -0.0044 0.0134* 0.0141**
(0.0052) (0.0052) (0.0105) (0.0106) (0.0092) (0.0092) (0.0071) (0.0071)
Democrat Governor -0.0203*** -0.0194*** -0.0416* -0.0412* 0.0127 0.0108 -0.0257** -0.0287***
& President (1 = Yes) (0.0069) (0.0069) (0.0215) (0.0216) (0.0131) (0.0132) (0.0102) (0.0104)
Democrat Majority, -0.0154** -0.0151** -0.0021 -0.0093 -0.0040 -0.0045 -0.0026 -0.0048
State Legislature (1 = Yes) (0.0069) (0.0069) (0.0209) (0.0211) (0.0122) (0.0122) (0.0088) (0.0089)
Democrat Majority, Federal -0.0051 -0.0043 0.0010 0.0052 -0.0098 -0.0118 -0.0070 -0.0084
& State Legislature (1 = Yes) (0.0079) (0.0079) (0.0223) (0.0223) (0.0132) (0.0132) (0.0112) (0.0112)
Democrat Governor & President & Federal 0.0257*** 0.0250*** 0.0547** 0.0558*** -0.0648*** -0.0604*** 0.0168 0.0145
& State Legislature Majority (1 = Yes) (0.0090) (0.0090) (0.0215) (0.0215) (0.0185) (0.0185) (0.0206) (0.0208)
Lagged Effects
Democrat Governor (1 = Yes) 0.0039** 0.0040** 0.0094** 0.0103** 0.0005 0.0007 -0.0016 -0.0014
(0.0019) (0.0019) (0.0047) (0.0047) (0.0033) (0.0033) (0.0030) (0.0030)
Democrat Governor 0.0012 0.0014 0.0088 0.0078 0.0119 0.0112 -0.0007 -0.0004
& President (1 = Yes) (0.0031) (0.0031) (0.0124) (0.0124) (0.0079) (0.0079) (0.0047) (0.0048)
Democrat Majority, -0.0022 -0.0020 0.0231 0.0179 -0.0169*** -0.0152*** 0.0120*** 0.0114**
State Legislature (1 = Yes) (0.0023) (0.0023) (0.0142) (0.0145) (0.0058) (0.0058) (0.0046) (0.0046)
Democrat Majority, Federal -0.0214*** -0.0221*** -0.0303** -0.0287* 0.0235*** 0.0214*** -0.0113 -0.0113
& State Legislature (1 = Yes) (0.0032) (0.0032) (0.0147) (0.0149) (0.0051) (0.0052) (0.0069) (0.0069)
Democrat Governor & President & Federal 0.0157*** 0.0160*** 0.0026 0.0076 -0.0414*** -0.0383*** 0.0271 0.0227
& State Legislature Majority (1 = Yes) (0.0048) (0.0049) (0.0126) (0.0127) (0.0115) (0.0115) (0.0217) (0.0217)
Lead Effects
Democrat Governor (1 = Yes) 0.0133*** 0.0133*** 0.0335*** 0.0321*** -0.0120** -0.0110** 0.0061 0.0063
(0.0025) (0.0025) (0.0060) (0.0060) (0.0052) (0.0053) (0.0048) (0.0049)

63
Democrat Governor -0.0097*** -0.0092** -0.0248** -0.0211* 0.0273*** 0.0262*** -0.0111* -0.0113*
& President (1 = Yes) (0.0037) (0.0038) (0.0125) (0.0127) (0.0069) (0.0068) (0.0064) (0.0065)
Democrat Majority, 0.0116*** 0.0116*** 0.0246** 0.0161 0.0043 0.0056 -0.0013 -0.0015
State Legislature (1 = Yes) (0.0029) (0.0029) (0.0106) (0.0110) (0.0065) (0.0066) (0.0051) (0.0051)
Democrat Majority, Federal -0.0073* -0.0080** -0.0175 -0.0113 -0.0100 -0.0114 0.0058 0.0063
& State Legislature (1 = Yes) (0.0038) (0.0038) (0.0129) (0.0130) (0.0084) (0.0084) (0.0055) (0.0055)
Democrat Governor & President & Federal -0.0196*** -0.0194*** -0.0164 -0.0164 -0.0275** -0.0276** 0.0055 0.0034
& State Legislature Majority (1 = Yes) (0.0049) (0.0049) (0.0125) (0.0125) (0.0107) (0.0108) (0.0129) (0.0130)

Observations 63715 63715 29624 29624 16845 16845 17246 17246

R2 0.389 0.389 0.440 0.440 0.336 0.337 0.167 0.168
Location Health Controls Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y
State Time Trends (Linear) N N N N N N N N
State Time Trends (Quadratic) N N N N N N N N
County Time Trends (Linear) Y Y Y Y Y Y Y Y
Period 1968-2016 1968-2016 1968-1984 1968-1984 1985-2000 1985-2000 2001-2016 2001-2016

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 dif-
ferent age categories. Population health controls include population shares by education (HS degree, some college, and at least a 4-year college
degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and other/male). Income inequality
controls include state income shares for the top 10% and top 1% and each interacted with county urban status. Lagged values correspond to
the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over the future three years (constrained to
be equal). * p <.10, ** p< .05, *** p<.01.
Appendix F.3 Quantile Regression

Table F21: Political Party Determinants of All-Cause Mortality Across Quantiles: Marginal Effects

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Contemporaneous Effects
Executive Control 0.0027 0.0026 0.0023 0.0022 0.0019 0.0019 0.0015 0.0015 0.0012 0.0012
P-value 0.0022 0.0036 0.0001 0.0002 0.0000 0.0000 0.0081 0.0100 0.1412 0.1471
Legislative Control 0.0016 0.0017 0.0013 0.0014 0.0010 0.0011 0.0007 0.0008 0.0005 0.0006
P-value 0.0573 0.0434 0.0219 0.0142 0.0179 0.0102 0.1908 0.1477 0.5436 0.4845
Full Control -0.0001 0.0000 0.0002 0.0004 0.0006 0.0007 0.0010 0.0011 0.0013 0.0014
P-value 0.9414 0.9844 0.8040 0.7130 0.3817 0.3114 0.2831 0.2446 0.3188 0.2934
Lagged Effects
Executive Control -0.0014 -0.0013 -0.0016 -0.0016 -0.0018 -0.0018 -0.0020 -0.0021 -0.0022 -0.0023
P-value 0.6112 0.6217 0.4019 0.4032 0.2007 0.1912 0.2708 0.2521 0.3982 0.3737
Legislative Control 0.0003 0.0002 0.0003 0.0002 0.0004 0.0003 0.0005 0.0003 0.0006 0.0004
P-value 0.9286 0.9532 0.8598 0.9096 0.7578 0.8447 0.7714 0.8546 0.8131 0.8820
Full Control -0.0018 -0.0016 -0.0017 -0.0016 -0.0015 -0.0015 -0.0014 -0.0014 -0.0012 -0.0013
P-value 0.6434 0.6727 0.5352 0.5619 0.4506 0.4657 0.5994 0.5984 0.7372 0.7267
Lead Effects
Executive Control 0.0012 0.0011 0.0008 0.0007 0.0005 0.0004 0.0001 0.0000 -0.0002 -0.0002
P-value 0.2307 0.2807 0.2075 0.2710 0.3354 0.4407 0.8348 0.9410 0.8599 0.7936
Legislative Control -0.0017 -0.0015 -0.0012 -0.0010 -0.0006 -0.0005 -0.0001 0.0001 0.0004 0.0005
P-value 0.1501 0.2036 0.1440 0.2121 0.3144 0.4432 0.9401 0.9379 0.7106 0.6467
Full Control -0.0010 -0.0007 -0.0011 -0.0008 -0.0011 -0.0009 -0.0012 -0.0011 -0.0012 -0.0012
P-value 0.5541 0.6852 0.3631 0.4939 0.1937 0.2865 0.2884 0.3516 0.4335 0.4701

Location Health Controls Y Y Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares
across 18 different age categories. Population health controls include population shares by education (HS degree, some college, and
at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and
other/male). Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county
urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to
the effects over the future three years (constrained to be equal).

64
Table F22: Political Party Determinants of “Short-Term”-Cause Mortality Across Quantiles: Marginal
Effects

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Contemporaneous Effects
Executive Control 0.0049 0.0046 0.0051 0.0049 0.0054 0.0052 0.0056 0.0055 0.0058 0.0057
P-value 0.0244 0.0334 0.0005 0.0009 0.0000 0.0000 0.0000 0.0001 0.0025 0.0029
Legislative Control -0.0000 0.0005 -0.0002 0.0002 -0.0004 -0.0001 -0.0005 -0.0003 -0.0007 -0.0006
P-value 0.9861 0.8401 0.9013 0.8942 0.7541 0.9519 0.7228 0.8194 0.7520 0.7875
Full Control -0.0015 -0.0008 -0.0013 -0.0008 -0.0012 -0.0009 -0.0011 -0.0009 -0.0010 -0.0010
P-value 0.7024 0.8346 0.6043 0.7457 0.5219 0.6405 0.6495 0.6996 0.7695 0.7758
Lagged Effects
Executive Control -0.0117 -0.0120 -0.0113 -0.0114 -0.0108 -0.0108 -0.0103 -0.0102 -0.0099 -0.0098
P-value 0.0940 0.0864 0.0174 0.0158 0.0019 0.0018 0.0190 0.0195 0.1084 0.1132
Legislative Control -0.0135 -0.0140 -0.0111 -0.0116 -0.0085 -0.0090 -0.0061 -0.0066 -0.0041 -0.0046
P-value 0.0878 0.0762 0.0378 0.0299 0.0286 0.0208 0.2150 0.1825 0.5533 0.5109
Full Control -0.0236 -0.0233 -0.0210 -0.0209 -0.0183 -0.0183 -0.0157 -0.0158 -0.0136 -0.0137
P-value 0.0251 0.0264 0.0031 0.0033 0.0004 0.0004 0.0169 0.0163 0.1421 0.1372
Lead Effects
Executive Control 0.0037 0.0035 0.0044 0.0042 0.0052 0.0050 0.0059 0.0057 0.0065 0.0063
P-value 0.1433 0.1589 0.0094 0.0126 0.0000 0.0001 0.0002 0.0003 0.0030 0.0044
Legislative Control -0.0010 -0.0003 -0.0019 -0.0014 -0.0029 -0.0026 -0.0039 -0.0037 -0.0047 -0.0046
P-value 0.7638 0.9170 0.3827 0.5220 0.0687 0.1120 0.0567 0.0740 0.1024 0.1126
Full Control -0.0104 -0.0092 -0.0090 -0.0082 -0.0075 -0.0070 -0.0062 -0.0059 -0.0050 -0.0050
P-value 0.0255 0.0468 0.0041 0.0095 0.0010 0.0023 0.0328 0.0426 0.2152 0.2224

Location Health Controls Y Y Y Y Y Y Y Y Y Y

Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares
across 18 different age categories. Population health controls include population shares by education (HS degree, some college, and
at least a 4-year college degree) and population shares by gender and race (white/male, white/female, black/male, black/female, and
other/male). Income inequality controls include state income shares for the top 10% and top 1% and each interacted with county
urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to
the effects over the future three years (constrained to be equal).

65
 Table F23: Political Party Determinants of All-Cause Mortality Across Quantiles: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Contemporaneous Effects
Democrat Governor (1 = Yes) 0.0003 0.0005 -0.0001 0.0001 -0.0006 -0.0004 -0.0010 -0.0009 -0.0014 -0.0013
(0.0020) (0.0020) (0.0013) (0.0014) (0.0010) (0.0010) (0.0013) (0.0013) (0.0018) (0.0018)
Democrat Governor & President (1 = Yes) -0.0016 -0.0018 -0.0014 -0.0016 -0.0012 -0.0014 -0.0009 -0.0012 -0.0007 -0.0010
(0.0029) (0.0029) (0.0020) (0.0020) (0.0015) (0.0015) (0.0019) (0.0019) (0.0027) (0.0027)
Democrat Majority, State Legislature (1 = Yes) -0.0019 -0.0018 -0.0019 -0.0018 -0.0019 -0.0018 -0.0019 -0.0019 -0.0019 -0.0019
(0.0026) (0.0026) (0.0018) (0.0018) (0.0013) (0.0013) (0.0017) (0.0017) (0.0024) (0.0024)
Democrat Majority, Federal 0.0021 0.0019 0.0022 0.0020 0.0023 0.0021 0.0024 0.0022 0.0025 0.0023
& State Legislature (1 = Yes) (0.0032) (0.0032) (0.0022) (0.0022) (0.0016) (0.0016) (0.0021) (0.0021) (0.0030) (0.0030)
Democrat Governor & President & Federal -0.0007 -0.0005 -0.0004 -0.0002 -0.0002 0.0001 0.0001 0.0004 0.0003 0.0006
& State Legislature Majority (1 = Yes) (0.0035) (0.0035) (0.0024) (0.0024) (0.0018) (0.0018) (0.0023) (0.0023) (0.0033) (0.0033)
Lagged Effects
Democrat Governor (1 = Yes) 0.0006 0.0006 0.0006 0.0006 0.0006** 0.0006** 0.0006* 0.0006* 0.0006 0.0006
(0.0006) (0.0006) (0.0004) (0.0004) (0.0003) (0.0003) (0.0004) (0.0004) (0.0005) (0.0005)
Democrat Governor & President (1 = Yes) 0.0021* 0.0019* 0.0017** 0.0016** 0.0013** 0.0012** 0.0009 0.0009 0.0006 0.0006
(0.0010) (0.0010) (0.0007) (0.0007) (0.0005) (0.0005) (0.0007) (0.0007) (0.0010) (0.0010)
Democrat Majority, State Legislature (1 = Yes) 0.0044*** 0.0043*** 0.0036*** 0.0036*** 0.0028*** 0.0027*** 0.0020*** 0.0019*** 0.0013* 0.0013*
(0.0008) (0.0008) (0.0005) (0.0005) (0.0004) (0.0004) (0.0005) (0.0005) (0.0007) (0.0007)
Democrat Majority, Federal -0.0028*** -0.0026*** -0.0023*** -0.0021*** -0.0018*** -0.0016*** -0.0013** -0.0011* -0.0009 -0.0007
& State Legislature (1 = Yes) (0.0010) (0.0010) (0.0007) (0.0007) (0.0005) (0.0005) (0.0006) (0.0006) (0.0009) (0.0009)
Democrat Governor & President & Federal -0.0044*** -0.0043*** -0.0034*** -0.0033*** -0.0023*** -0.0022*** -0.0013 -0.0012 -0.0004 -0.0004
& State Legislature Majority (1 = Yes) (0.0015) (0.0015) (0.0010) (0.0010) (0.0007) (0.0007) (0.0010) (0.0010) (0.0014) (0.0014)
Lead Effects
Democrat Governor (1 = Yes) 0.0002 0.0001 0.0003 0.0002 0.0004 0.0003 0.0005 0.0004 0.0006 0.0005

66
(0.0008) (0.0008) (0.0005) (0.0005) (0.0004) (0.0004) (0.0005) (0.0005) (0.0007) (0.0007)
Democrat Governor & President (1 = Yes) 0.0010 0.0009 0.0005 0.0005 0.0001 0.0001 -0.0004 -0.0004 -0.0008 -0.0008
(0.0012) (0.0012) (0.0008) (0.0008) (0.0006) (0.0006) (0.0008) (0.0008) (0.0011) (0.0011)
Democrat Majority, State Legislature (1 = Yes) -0.0009 -0.0008 -0.0006 -0.0005 -0.0002 -0.0002 0.0002 0.0002 0.0005 0.0004
(0.0009) (0.0009) (0.0006) (0.0006) (0.0005) (0.0005) (0.0006) (0.0006) (0.0009) (0.0009)
Democrat Majority, Federal -0.0008 -0.0007 -0.0006 -0.0005 -0.0004 -0.0003 -0.0002 -0.0001 -0.0001 0.0001
& State Legislature (1 = Yes) (0.0013) (0.0013) (0.0009) (0.0009) (0.0007) (0.0007) (0.0009) (0.0009) (0.0012) (0.0012)
Democrat Governor & President & Federal -0.0005 -0.0003 -0.0007 -0.0005 -0.0010 -0.0009 -0.0013 -0.0012 -0.0015 -0.0014
& State Legislature Majority (1 = Yes) (0.0017) (0.0017) (0.0012) (0.0012) (0.0009) (0.0009) (0.0011) (0.0011) (0.0016) (0.0016)

Observations 117504 117504 117504 117504 117504 117504 117504 117504 117504 117504
Location Health Controls Y Y Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age categories. Pop-
ulation health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and population shares by gender and race
(white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each
interacted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects
over the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
 Table F24: Political Party Determinants of “Short-Term”-Cause Mortality Across Quantiles: Coefficients

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Contemporaneous Effects
Democrat Governor (1 = Yes) -0.0044 -0.0038 -0.0046 -0.0041 -0.0049* -0.0045* -0.0051 -0.0048 -0.0053 -0.0051
(0.0054) (0.0054) (0.0036) (0.0036) (0.0026) (0.0026) (0.0033) (0.0033) (0.0047) (0.0047)
Democrat Governor & President (1 = Yes) -0.0073 -0.0082 -0.0066 -0.0073 -0.0059 -0.0063* -0.0052 -0.0054 -0.0046 -0.0046
(0.0075) (0.0075) (0.0050) (0.0050) (0.0037) (0.0037) (0.0047) (0.0047) (0.0065) (0.0066)
Democrat Majority, State Legislature (1 = Yes) -0.0142** -0.0142** -0.0105** -0.0104** -0.0065* -0.0064* -0.0028 -0.0027 0.0004 0.0005
(0.0068) (0.0068) (0.0046) (0.0046) (0.0033) (0.0033) (0.0042) (0.0042) (0.0059) (0.0059)
Democrat Majority, Federal 0.0008 0.0002 -0.0006 -0.0012 -0.0020 -0.0026 -0.0033 -0.0039 -0.0045 -0.0050
& State Legislature (1 = Yes) (0.0086) (0.0086) (0.0058) (0.0058) (0.0043) (0.0043) (0.0054) (0.0054) (0.0076) (0.0076)
Democrat Governor & President & Federal 0.0016 0.0027 0.0013 0.0022 0.0010 0.0016 0.0007 0.0011 0.0004 0.0006
& State Legislature Majority (1 = Yes) (0.0095) (0.0095) (0.0064) (0.0064) (0.0047) (0.0047) (0.0059) (0.0059) (0.0083) (0.0083)
Lagged Effects
Democrat Governor (1 = Yes) 0.0036** 0.0036** 0.0028*** 0.0028*** 0.0019** 0.0019** 0.0011 0.0011 0.0004 0.0005
(0.0015) (0.0015) (0.0010) (0.0010) (0.0008) (0.0008) (0.0010) (0.0010) (0.0014) (0.0014)
Democrat Governor & President (1 = Yes) 0.0013 0.0011 0.0023 0.0021 0.0034*** 0.0033** 0.0045*** 0.0043*** 0.0054** 0.0052**
(0.0026) (0.0026) (0.0018) (0.0018) (0.0013) (0.0013) (0.0016) (0.0016) (0.0023) (0.0023)
Democrat Majority, State Legislature (1 = Yes) 0.0059*** 0.0060*** 0.0052*** 0.0053*** 0.0044*** 0.0044*** 0.0037*** 0.0037*** 0.0032* 0.0030*
(0.0020) (0.0020) (0.0014) (0.0014) (0.0010) (0.0010) (0.0013) (0.0013) (0.0018) (0.0018)
Democrat Majority, Federal -0.0059** -0.0055** -0.0054*** -0.0050*** -0.0048*** -0.0045*** -0.0043*** -0.0040** -0.0038* -0.0036
& State Legislature (1 = Yes) (0.0026) (0.0026) (0.0018) (0.0018) (0.0013) (0.0013) (0.0016) (0.0016) (0.0023) (0.0023)
Democrat Governor & President & Federal -0.0063 -0.0059 -0.0063** -0.0060** -0.0062*** -0.0060*** -0.0061** -0.0061** -0.0061* -0.0061*
& State Legislature Majority (1 = Yes) (0.0039) (0.0039) (0.0026) (0.0026) (0.0019) (0.0019) (0.0024) (0.0024) (0.0034) (0.0034)
Lead Effects
Democrat Governor (1 = Yes) 0.0018 0.0013 0.0018 0.0015 0.0018* 0.0016 0.0018 0.0018 0.0018 0.0019

67
(0.0021) (0.0021) (0.0014) (0.0014) (0.0010) (0.0010) (0.0013) (0.0013) (0.0018) (0.0018)
Democrat Governor & President (1 = Yes) 0.0019 0.0022 0.0026 0.0028 0.0034** 0.0034** 0.0041** 0.0039** 0.0047* 0.0044
(0.0031) (0.0031) (0.0021) (0.0021) (0.0015) (0.0015) (0.0019) (0.0019) (0.0027) (0.0027)
Democrat Majority, State Legislature (1 = Yes) 0.0021 0.0025 0.0010 0.0012 -0.0002 -0.0001 -0.0014 -0.0013 -0.0023 -0.0024
(0.0025) (0.0025) (0.0017) (0.0017) (0.0012) (0.0012) (0.0015) (0.0015) (0.0022) (0.0022)
Democrat Majority, Federal -0.0031 -0.0028 -0.0029 -0.0027 -0.0027 -0.0025 -0.0025 -0.0023 -0.0024 -0.0022
& State Legislature (1 = Yes) (0.0035) (0.0035) (0.0024) (0.0024) (0.0017) (0.0017) (0.0022) (0.0022) (0.0031) (0.0031)
Democrat Governor & President & Federal -0.0130*** -0.0124*** -0.0115*** -0.0110*** -0.0098*** -0.0094*** -0.0082*** -0.0079*** -0.0069* -0.0067
& State Legislature Majority (1 = Yes) (0.0047) (0.0047) (0.0032) (0.0032) (0.0023) (0.0023) (0.0029) (0.0029) (0.0041) (0.0041)

Observations 105542 105542 105542 105542 105542 105542 105542 105542 105542 105542
Location Health Controls Y Y Y Y Y Y Y Y Y Y
Population Health Controls Y Y Y Y Y Y Y Y Y Y
Income Inequality Controls N Y N Y N Y N Y N Y
County FEs Y Y Y Y Y Y Y Y Y Y
Year FEs Y Y Y Y Y Y Y Y Y Y
State Time Trends (Linear) Y Y Y Y Y Y Y Y Y Y
State Time Trends (Quadratic)
County Time Trends (Linear)
Quantile 0.10 0.10 0.25 0.25 0.50 0.50 0.75 0.75 0.90 0.90

Standard errors are clustered at the county level. Dependent variable in logs. Location health controls include population shares across 18 different age categories. Pop-
ulation health controls include population shares by education (HS degree, some college, and at least a 4-year college degree) and population shares by gender and race
(white/male, white/female, black/male, black/female, and other/male). Income inequality controls include state income shares for the top 10% and top 1% and each in-
teracted with county urban status. Lagged values correspond to the effects over the prior four years (constrained to be equal). Lead values correspond to the effects over
the future three years (constrained to be equal). * p <.10, ** p< .05, *** p<.01.
Appendix G Possible Mechanisms
As stated in Section 1, a thorough investigation of the possible mechanisms through which liberalism
and, to a lesser extent, political party impact mortality is beyond the scope of this paper. Nonetheless,
we can offer a few comments. First, our analysis has shown that controlling for income inequality has no
effect on the association between politics and mortality. Second, anecdotal evidence suggests a possibly
important role of the social safety net. Aside from differences in the generosity of programs for which states
have some discretion, states and localities can affect the efficacy of these programs and, in particular, the
ease of access. A 2023 article in the Washington Post argues:

“State lawmakers gained autonomy over how to spend federal safety net dollars following Repub-
lican President Ronald Reagan’s push to empower the states in the 1980s. Those investments
began to diverge sharply along red and blue lines, with conservative lawmakers often balking
at public health initiatives they said cost too much or overstepped.”26

The result is that “States’ politics — and their resulting policies — are shaving years off American lives.”
To investigate this, we examine two prominent safety net programs: Unemployment Insurance (UI)
and the Food Stamp Program (FSP; now called the Supplemental Nutrition Assistance Program (SNAP)).
Access to both programs by eligible households came under scrutiny during the Great Recession and Covid-
19 pandemic. For instance, early during the pandemic, it became clear that “Florida’s unemployment
system was the slowest in the country to process claims” and “residents described nightmarish experiences
as they tried to get benefits.” Governor DeSantis pointed to the prior administration during which the
system was created, arguing that while “he wasn’t sure if it was Scott’s intention to discourage people
from applying for benefits ... the system was certainly designed to produce that outcome.”27 A similar
situation has recently materialized in Texas, where 600,000 individuals have lost access to Medicaid due to
“procedural errors,” leading to “reverberations” causing individuals to also lose access to SNAP benefits.28
Figure G1 shows the distribution of state UI recipiency rate averaged over 1987-1996 and State FSP
participation rate in 1997 by state policy liberalism.29 The recipiency rate is the ratio of individuals
collected UI to the number of unemployed. The participation rate is the share of FSP-eligible households
that collect FSP benefits. We examine data from the 20th century as this is the time period where we find
evidence of partisan mortality cycles. In both cases, but for UI in particular, more conservative states have
significantly lower rates.

26
See https://www.washingtonpost.com/health/interactive/2023/republican-politics-south-midwest-life-expe
ctancy/.
27
See https://www.npr.org/sections/coronavirus-live-updates/2020/08/06/899893368/gov-says-floridas-unemp
loyment-system-was-designed-to-create-pointless-roadblock.
28
See https://www.texastribune.org/2023/09/28/texas-hhsc-medicaid/.
29
UI data: https://stats.bls.gov/mlr/1997/07/art2full.pdf. FSP data: https://www.fns.usda.gov/snap/reachi
ng-those-need-food-stamp-participation-rates-states.

68
 Figure G1: State Participation Rates in Safety Net Programs & State Policy Liberalism

(A) UI (B) FSP

Notes: Higher values of all ideology scores correspond to greater liberalism. UI = unemployment
insurance. FSP = Food Stamp Program.

Another policy that has been shown to have large health impacts and have a strong partisan division
is the excise tax rate on cigarettes. Figure G2 shows that the distribution of tax rates in 1993 is much
higher in liberal states, consistent with Golden et al. (2014).30

Figure G2: State Cigarette Excise Tax Rates & State Policy Liberalism

Notes: Higher values of all ideology scores correspond to greater liberalism.

In sum, we suspect that the ability of governments at the federal, state, and local levels to influence
the efficacy of the social safety net and policies that impact health behaviors is the source of partisan
30
See https://www.ncbi.nlm.nih.gov/books/NBK236771/.

69
political mortality cycles, at least until 2000. Moreover, even though these are short-term changes in
policy, they may influence all-cause and short-term mortality by exacerbating the deleterious effects of
chronic conditions. For example, Borgschulte and Vogler (2020) found that areas covered by Medicaid
expansion in 2014 reduced all-cause mortality by more than 11 deaths per 100,000 individuals within two
years. Obviously, more research is required.
Finally, the absence of partisan mortality cycles in the 21st century, a time period characterized by
stagnant and even declining life expectancy in the US (Mehta et al., 2020), is noteworthy and important.
This holds even if we pool the demographic groups and estimate the baseline specification for each time pe-
riod.31 One possible explanation is that government gridlock has increased over time with the polarization
of the US (Binder, 1999, 2014); see also Appendix F. Thus, political regimes face greater challenges in con-
verting policy preferences to actual policy. Binder (2014, p. 17) argues that “levels of legislative deadlock
have steadily risen over the past half-century” and “stalemate at times now reaches across three-quarters of
the salient issues on Washington’s agenda.” Figure G3 provides evidence supporting this hypothesis. The
figure plots two measures of the legislative productivity of US Congress from 1789 to 2003. The measures,
derived in Grant and Kelly (2008), reveal an overwhelmingly downward trend since each index peaked in
1965.

Figure G3: Historical Legislative Productivity of US Congress

Source: Grant and Kelly (2008); Kelly (2018).

Another possible explanation is the opioid epidemic. Drug overdoses tripled between 1999 and 2020,
mainly due to opioids (Currie and Schwandt, 2021). According to Baker et al. (2021, p. 36), opioids
have been an “equal opportunity scourge” affecting all communities and demographic groups, although
mortality rates are higher for low income, low education individuals. However, there is little difference
in the overdose rate among individuals above and below the poverty line (Currie and Schwandt, 2021).
Regardless of cause, the results are not encouraging in terms of the political system generating solutions
to geographic and socioeconomic inequality in mortality moving forward.
31
Results available upon request.

70