Global Economy Journal: Institutional Quality and Economic Growth in Latin America
Global Economy Journal: Institutional Quality and Economic Growth in Latin America
∗
Texas Christian University, [email protected]
Abstract
The purpose of this paper is to relate total factor productivity to the problem of economic
growth in Latin America. Slow economic growth has been the most important problem in the
modern history of the region. The paper extensively reviews the literature on the determinants
of economic growth in Latin America and shows that the slow growth of total factor productivity
(TFP) seems to be the primary problem. Further, this problem is linked to the quality of institutions
in the region. Unfortunately, the concept of institutional quality is very difficult to define in a
meaningful way. This affects public policy towards economic growth in the region. Finally, it is
shown that the problem that has been identified may be crucial for Latin America but the region is
hardly unique in this regard.
∗
The author would like to thank the editor and participants at the annual meetings of the European
Economics and Finance Society, the International Trade and Finance Association, and the South-
ern Economic Association for useful comments on earlier drafts of this paper. The usual caveat
applies.
I. Introduction
Over the last 50 years, the economies of Latin America have been plagued by
three interrelated problems. The most pressing problem was that the region was a
virtual definition of macroeconomic instability. Inflation, high unemployment,
and erratic GDP growth were almost normal economic conditions. Second, Latin
America has had the highest level of income inequality among the world’s regions
for at least as long as reliable data has existed.1 The final, and most serious,
problem has been the relatively slow rate of economic growth. By international
standards, GDP per capita growth in Latin America has been lower than that of
the more successful developing countries. Over the last 20 years, there has been a
moderate amount of research on this problem. To quickly summarize, economic
growth in Latin America has lagged due to low growth of total factor productivity
(TFP). Interestingly, the literature on economic growth in general has become
more concerned with the determinants of TFP during the same time period. The
general conclusion of this research is that the quality of institutions in a country
affects TFP which in turn affects the rate of growth of GDP.
The purpose of this paper is to tie this general insight to the problem of
economic growth in Latin America. The next section outlines just how important
slow economic growth has been in the modern history of the region. The
literature on the determinants of economic growth in Latin America is reviewed to
show that TFP seems to be the primary problem and that it is linked to the quality
of institutions in the region. The next section shows that the concept of
institutional quality is very difficult to define in a meaningful way. A final
section outlines how this affects public policy towards economic growth in the
region. Finally, we show that the problem that has been identified may be more
crucial for Latin America but the region is hardly unique in this regard.
The central economic problem of Latin America is that GDP per capita in the
region is only about 20 percent of that in the U.S. This has not always been the
case. The GDP per capita of Latin America was roughly equivalent to that of
North America as recently as 1700. The traditional historical view was that Latin
America fell behind in the period from 1820 to 1870. The wars of independence
and the extended turmoil that followed led to almost half a century of slow
economic growth. The delayed independence of the region coupled with political
instability may account for a substantial portion of the income gap between North
1
The traditional view is that Latin America also has had the worst distribution of income for the
last 500 years. However, that view has been recently challenged in Williamson (2009).
and South America.2 The subsequent “Golden Age” or belle époque was an
extended period of rapid economic growth lasting from 1870 to the Great
Depression. The period running from 1940 to 1980 is a more complicated story.
GDP per capita in the region grew, but at a relatively slow pace. This result is
related to the percentage difference mentioned above. During this period, the gap
between GDP per capita in the region and in the developed countries barely
moved. Latin America fell further behind during the “Lost Decade” of the 1980s
and much of the last two decades has consisted of making up for that loss.3
By global standards, GDP per capita in Latin America is seemingly not
that low. In 2008, GDP per capita in the region was $7,785. This is more than
double the GDP per capita for an average middle-income country. However, this
needs to be compared to GDP per capita in high-income countries which is
$40,525. Since economic growth in the region has been slow since the middle of
the 20th century, an interesting question is what GDP per capita could have been if
growth had been higher. A frequent comparison is the performance of Latin
America versus the miracle economies of East Asia.4 This comparison may be
less relevant than it seems. The average country in Latin America is not Hong
Kong and is unlikely to ever be so. A similar situation occurs when comparing
Latin America to high-income countries in the 21st century. However, even
allowing for these differences, the growth performance of Latin America has been
poor. In 1955, the average GDP per capita for Latin America was 43 percent of
GDP per capita in high-income countries. By 2000, the regional average was less
than 25 percent.5 A similar story emerges if one examines economic growth in
the best performing economy of the region, Chile. Economic growth in Chile has
been substantially above the average for the region since the 1970s. The result is
that GDP per capita in Chile is over $10,000. A recent study compares economic
growth in Latin America to a more relevant country, Indonesia. Indonesia is an
Asian country that has some of the same characteristics as much of Latin
America. However, economic growth, even in Chile, has been substantially
below growth in Indonesia. The comparison with Indonesia is more relevant as
the country is both a major commodity export and has experienced Latin
American style problems of political instability.6
Explaining the causes of this relatively slow economic growth is an
important question. In answering this question, the obvious solution is to run
regressions on economic growth in the region and examine the results. Over the
2
See Przeworski (2008) for details. However, the extent to which this gap is attributable to these
factors is not universally accepted.
3
For a more extensive discussion of the issues covered in this section see Edwards (2009).
4
For more detail, see Edwards (2010).
5
For more details, see Grier (2007).
6
See Reyes and Sawyer, (2011).
DOI: 10.2202/1524-5861.1710 2
last 10 years a small literature has developed on this subject. A relatively new
and small literature has both advantages and disadvantages. The advantage is that
even the oldest papers include many of the insights on economic growth
developed during the 1970s and 1980s. The disadvantage is that this literature is
not nearly large enough to fully answer the questions that are critical in terms of
enhancing economic growth in the region. At this point, the literature clearly
shows that the primary problem of growth in the region is the rate of growth (or
lack thereof) of total factor productivity (TFP). Further, the literature is new
enough to incorporate the insight that TFP can be heavily influenced by
institutional factors in an economy.
So far, the literature on this issue has shown that a limited number of
measures of institutional quality seem to have an impact on TFP in the region. At
this point, the results perhaps are suggestive rather than definitive. This may be
related to the lack of knowledge about how to define institutional quality. In turn,
this creates two problems. First, the choice of a variable representing institutional
quality may be susceptible to either using “what works” or simply using what has
been used in the past. The second problem is that the measures of institutional
quality that are commonly used are rather broad. Using such broad measures for
this variable may be limiting our understanding of the contribution of institutional
quality to growth and/or making it more difficult to make specific policy
recommendations that might enhance growth. The next section of this paper
reviews the major empirical studies on economic growth in Latin America.
7
For the seminal works in this area see North (1981, 1990).
that many of the variables now commonly used to measure institutional aspects of
economies are rather new. The results of this literature are briefly summarized
below.
Chong and Zanforlin (2004) performed the first attempt at putting
institutional variables into a standard growth model for Latin America. While
they did not measure the effects of these variables on TFP, their paper is
important as the first in a series of attempts to quantify the effects of these factors
on growth. They measure institutional quality from two sources, but only the data
from the International Country Risk Guide (ICGR) has been used in subsequent
studies. In this study, institutional quality was proxied by an average of five
variables from ICGR: risk of expropriation; repudiation of contracts by
governments; law and order tradition; corruption in government; and quality of
bureaucracy. The results of this paper are important as they show that higher
institutional quality is associated with higher economic growth in the region. The
limitation of the study is that the specification does not allow one to quantify the
effects of institutional quality on TFP. This is a critical point that is still not
resolved. Does one simply add an institutional variable(s) to a standard growth
model or does one attempt to measure the effects of institutional quality on TFP
more directly? As we will see, there is still no consensus on this. de Gregorio
and Lee (2004) produce results similar to the paper by Chong and Zanflorin
(2004). Institutional variables are included in the regressions, but the effects on
TPF are not explicitly modeled. In this case the institutional factors were proxied
by Government consumption; the Rule of Law; the Inflation rate; Democracy; and
Openness. All of these variables seem to have the expected effects on economic
growth but their effects on TFP are not measured in this study. This paper
illustrates another trade off. Many of the measures of the rule of law and other
institutional variables are of relatively recent vintage. Some of the other measures
such as inflation and openness obviously can be obtained for a much longer time
series. This can create a tradeoff between data availability and the quality of
measurement. How to deal with this problem isn’t clear from the papers in the
literature.
The empirical work of Fernandez-Arias, et al (2005) was the first to
explicitly consider how TFP and institutional quality interact. As is now well
known, they found that the source of lagging growth in the region is TFP. The
difference here is that they go on to attempt to determine the factors that are
influencing TFP growth. Among the factors that were tested for their influence
on TFP are education; Life expectancy; Openness; Imports of machinery and
equipment; Credit to the private sector; Government consumption; Inflation; the
Black Market Premium; and the ICRG variable as defined in the previous study.
The findings were that Openness, Inflation, and the ICRG variable all seem to
have the most significant relationship to TFP. While all of these variables seem
DOI: 10.2202/1524-5861.1710 4
to “work” it is less clear as to why they work or perhaps how they interact. As is
usually the case, these results are suggestive and very interesting but leave many
questions unanswered.
Cole, et al. (2005) reach very similar conclusions concerning relatively
slow growth in Latin America. Their empirical work reaches the strong
conclusion that weak growth in the region is almost wholly a function of low
growth of TFP. They further conclude that the problem is not low growth in
human capital. Their primary focus is explaining problems with TFP. However,
they take a very different approach to explaining the problem than the previous
study. They argue that barriers to competition such as tariffs and barriers to
internal competition such as high entry costs, underdeveloped capital markets, and
labor regulation have hampered innovation. This in turn has reduced the
productivity of the countries of the region. While the intuitive reasoning of this
approach is powerful and the authors supplement it with case-type examples, it is
difficult to implement empirically.
The paper by Chumacero and Fuentes (2006) provides confirmation of the
importance of TFP for explaining the poor growth performance of Latin America.
However, the paper is more focused on determining structural breaks in the
empirical relationships and does not explore the possible causes for low TFP
growth in the region. Grier (2007) provides a useful update to the paper by Cole,
et al. This paper estimates growth equations for Latin America for the period
1955 to 1995. The results confirm low TFP as a cause of low growth. The author
then compares Latin American estimates to the estimates obtained for a sample of
developed countries. The empirical results presented are focused on explaining
this gap as opposed to estimating their effects on TFP directly.8 The paper then
goes on to attempt to explain this gap as a function of: Learning-by-exporting;
Technological Diffusion; Capital Openness; Educational Quality; Macroeconomic
and Political instability; Government spending; Government type; and Diversity
of population. Dummy variables for Landlocked, Climate, Debt crisis, and Oil
were also included. The results indicate that Government spending, Diversity,
and Government type all reduce TFP. Further the dummy variables for
Landlocked, Debt crisis, and Oil were also significant. Further estimates in the
paper indicate that manufactured exports can increase growth.
The reader may have noticed a puzzling problem with this literature. For
those used to thinking of changes in TFP as being related to changes in
technology and human capital the studies cited above seem deficient in this
regard. None of these studies attempt to control for the effects of human capital
accumulation or R&D on TPF before beginning to investigate the effects of
institutional quality.9 While this is an unfortunate omission, it is perhaps
8
In this regard, the paper is quite similar to Fernandez-Arias, et al. (2005).
9
For an example of this type of control before adding other variables see Woo (2009).
Data on institutional quality are controversial – they are based mostly on surveys,
for one thing, so they reflect perceptions and not objective measures - but what
there is does not show Latin America in a favorable light.
Andres Velasco
DOI: 10.2202/1524-5861.1710 6
on institutional quality have begun to appear and are now being used both in
studies of economic growth and in other contexts.10
For the most part, this new data is in the form of surveys of businesses,
academics, and government officials, and multilateral agencies. In the absence of
more concrete data, economists are increasingly using this “softer” data out of a
lack of better alternatives. The general sense seems to be that while this sort of
data isn’t perfect, it is all that is available at this point. Some of the more widely
used data in this literature is the aforementioned ICGR data; Governance Matters;
Doing Business; the Competitiveness Index produced by the World Economic
Forum; or the Economic Freedom Indexes produced by the Heritage Foundation.
More specialized surveys such as the corruption index published by Transparency
International show up in other types of research beyond economic growth. This
sudden rush of data creates two problems. First, the data is relatively new. This
is particularly a problem in studying economic growth. One can either combine
new institutional data with much older data when studying economic growth or
use older data that may or not be an adequate proxy for institutional quality such
as inflation or openness. At this point, this is an uncomfortable choice with no
obvious solution. A second problem is now the sheer wealth of data. Institutional
quality is something that everyone understands. Rigorously defining it is a bit
more of a problem. In the current circumstances, it is not at all clear which of the
many choices for institutional quality is best. For example, is the ICGR data
“better than” the Governance Matters data? Or is the Doing Business data to be
preferred to the Competitiveness Index? Uncomfortably, there are no known
answers to these questions. In such a situation, perhaps it doesn’t matter. All of
these measures should be positively correlated with economic growth and more
importantly for Latin America with TFP. In the next section, a small empirical
exercise indicates some of the potential problems in this regard.
We measure the rule of law using part of a set of aggregate indicators based on
hundreds of specific and disaggregated individual variables measuring various
dimensions of governance, taken from 33 data sources provided by 30 different
organizations.
Martin Krause
10
For examples see the recent papers by Blanco (2009) and Fullerton, et al (2007).
variable is the gap between TFP in Latin America and East Asia which is a
common comparison in the literature.11 One of the unique features of this data is
that there is data on this gap for every country in the region. The gap is expressed
as TFP for Latin America minus TFP for East Asia so the gap is negative for
every country in the region. The smallest gap is -0.3 for Chile and the largest gap
is -2.0 for Venezuela. This gap is then regressed on three common measures of
institutional quality for each country. The global rank for the countries is taken
from the Doing Business data. The scores for Rule of Law and Regulatory
Quality were collected from the Governance Matters data. The Doing Business
rank should have a negative sign indicating that as the rank increases the
aforementioned gap narrows. For the other two measures, the coefficient should
be positive. This relationship is in no sense designed to be anything other than
suggestive. It is more in the way of indicating that the choice of institutional
quality may be troublesome.
The results above show that all of the measures of institutional quality
have the correct sign. That’s a good start. However, the sizes of the coefficients
vary widely. Further, none of the coefficients are significant. In this simple
exercise, the rule of law seems to work best. It is even marginally significant. In
this one simple case the rule of law “wins”. Would this always be the case? The
levels of significance are not that different and all have the correct sign.
Somewhat different data could have easily yielded a different result. The exercise
was merely given to illustrate a point. Different measures of institutional quality
might easily yield different empirical results in the case of Latin America. The
example above only illustrates three of the more commonly used measures. There
are many others available.
IV. Conclusions
The purpose of this paper was to stimulate thought on the issue of institutional
quality and economic growth in Latin America. The primary problem of
economic growth in the region is that the growth of TFP in the region is low. The
11
This data was gathered from Fernandez-Arias, et al (2005).
DOI: 10.2202/1524-5861.1710 8
relatively recent but small literature on growth in the region almost universally
shows that this is the case. There also seems to be general agreement that poor
institutional quality is the primary driver of low TFP growth. Understandably, the
newer empirical research in the area is focused on this relationship. However,
there are a number of different possible measures of institutional quality. On a
purely logical basis, it is not clear which of these variables should be used for
empirical work. A proposed solution is an index number constructed from the
various indexes.12 While this avoids some uncomfortable choices, an index still
leaves one with the problem of weights.
In the interim, researchers in this area perhaps need to be more clear about
the choice of the measure of institutional quality. It may well be that such choices
are not trivial. Some variables may work and others not for reasons that are not
entirely clear. Just the recognition of plausible choices may be helpful. Using
what works is not necessarily in a case like this when there are choices and no
clear decision criteria. What may need to be avoided at this point is the use of a
particular variable purely because others have used it. The literature seems far too
underdeveloped for that approach.
References
Cole, Harold L., Lee E. Ohanian, Alvaro Riascos, James A. Schmitz, Jr. 2005.
“Latin America in the Rearview Mirror.” Journal of Monetary Economics
52 (January): 69-107.
Comin, Diego, Bart Hobijn, and Emilie Rovito. 2008. “Technology Usage
Gaps.” Journal of Economic Development. 13, pp. 237-56.
12
See Krause (2009).
de Gregorio, Jose and Jong-Wha Lee. 2004. “Growth and Adjustment in East
Asia and Latin America.” Economia 5 (Fall): 69-115.
Edwards, Sebastian. 2010. Left Behind: Latin America and the False Promise of
Populism. Chicago: University of Chicago Press.
Owen, Ann L., Julio Bideras, and Lewis Davis. 2009. “Do All Countries Follow
the Same Growth Process?” Journal of Economic Growth, 14, pp. 265-86.
DOI: 10.2202/1524-5861.1710 10
Przeworski, Adam. 2008. “Does Politics Explain the Economic Gap between the
United States and Latin America?” In Falling Behind: Explaining the
Development Gap between Latin America and the United States, ed.
Francis Fukuyama. New York: Oxford University Press.
World Bank. 2008. Doing Business, 2009. Washington, D.C.: World Bank.
World Bank. 2008. Governance Matters, 2008, Washington, D.C.: World Bank.
11