Organization Theory

READINGS
1. Garicano, Luis, and Luis Rayo. 2016. "Why Organizations Fail: Models and Cases."
Journal of Eco
2. nomic Literature, ntive54 (1): 137-92.
3. Turco, C. (2018): The Conversational Firm, Rethinking Bureaucracy in the Age of
Social Media.

Topics
Introduction
Methods
Incentives
Vertical Integration
Managerial Practices
Personnel Economics
Knowledge Hierarchies
Reading 1&2
Culture and Leadership
Behavioral

Lesson 1&2 - Methods

Plan of Work:
1. Clinical Papers; 2. Insider Econometrics; 3. Experiments; 4. Introduction to RCT; 5. Field
Experiments with Firms

1. Clinical Papers in Organizational Studies

George P. Baker & Ricard Gil

Vertical integration = two companies in different parts in the value chain merge or one is
acquired. It’s an important phenomenon, which can be seen through many sectors, many
areas…

Example: Samsung now produces the batteries for its phones.

Netflix used to stream only films produced by others, then vertically integrated producers.

In football, teams invest in the junior teams, when players are young.

Story of GM and Fisher Body

In 1919, General Motors and Fisher Body signed a 10 year contractual agreement

for the supply of closed auto bodies. General Motors agreed to buy all its closed

bodies from Fisher. To avoid Fisher taking advantage of General Motors through

establishing a monopoly price, they fixed an ex-ante price. After a while, demand

for automobiles increase and General motor’s considered that the price of Fisher

was too high considering the reduction in production costs. Additionally, General

Motors considered that the production process must become more efficient. This

contractual relationship was intolerable and General Motors began negotiations to

purchase the remaining stock of Fisher Body. Finally, in 1926, they merged.

What is highlighted by this story?

→ Vertical integration decisions

→ Complex and interrelated roles of specific investments

→ Unforeseen contingencies

→ contractual incompleteness

It allows us to understand how firm boundaries work: when two firms integrate, the
boundaries change.

Clinical studies

= A detailed story, often involving a combination of qualitative data, anecdotal evidence and
sometimes quantitative data, about a specific situation or phenomenon
Types

● Teaching case
● In-depth clinical papers
● Empirical large-sample papers

Three purposes

● To tell theorist what to model (what details ignored in current literature)

● To guide on how to model these things (allows identifying which variable must be
included)
● To provide evidence of a certain type for testing theory (to support or weaken
theoretical conclusions)

Areas of concern:

● Representativeness: (will not be representative of any set of real-world settings)

things we learn for one story, doesn’t mean that’s how people behave always
● Bias: firms chosen not at random, they are chosen because they care about
knowledge generation, might be subject to selection bias
● Replicability: (lack of replicability of clinical studies is common) the point of the story
is not to be replicable, but to teach and generate insight

2. Insider Econometrics
Casey Ichniowski & Kathrin L. Shaw
Contents: Definition and characteristics of insider econometrics; Examples; Estimating
treatment effects; Lessons and conclusions

Def. A management practice is a conduct that increases productivity. E.g. giving water
with a coffee

The questions that insider econometrics wants to understand:

1. Why do firms in the same industry adopt different management practices?

Water with coffee ⇒ more people come to the coffee shop and increased profit

2. Does the adoption of a new management practice raise productivity?

 People enjoy the water, free items in general

3. If so, why does the new management practice raise productivity?

Insider ⇒ uses granular micro-level data inside firms and uses insights from insiders
(from managers or employees)

How to conduct an Insider Econometrics Research?

1. ESTIMATE a productivity regression in which productivity is a function of some

management practice: what is the differentiating value, what is the management
practice ⇒ productivity = f(management practice)
2. IDENTIFY why the management practice raised productivity
3. MODEL the adoption of the management practice. Why do you do it? Could be
cultural reasons, …
4. ANALYZE: conduct the statistical test → micro-level data on production units
(employees or teams)
5. FORMULATE testable hypotheses and interpret results. If we increase the water,
people are gonna come back?
Treated Counterfactual (Dashed Line): Represents the hypothetical productivity trajectory
of the treated group if they had not adopted the management practice. Counterfactual= what
we would’ve observed if that the MP wouldn’t have happened.

If I conduct analysis in july and sept: there are less students in the area ⇒
confounding effect = occurs when an external variable (a confounder) influences
both the independent variable (treatment or exposure) and the dependent variable
(outcome), creating a misleading association between the two. This makes it difficult
to establish whether the observed effect is due to the independent variable or the
confounder.

Thus, we get another coffee shop 2 (Yc → controlled) who before and after is not providing
water. There’s also a change for them between July and September: the difference in the
students.

(Type 2 Bias (ΔPᴺ): The non-treated group's productivity is also not a perfect counterfactual,
as their productivity may differ due to reasons unrelated to the treatment.)

⇒ By making the difference between change in Shop 1 and change in Shop 2 = we have
the change due to the water.

Selection bias = selection of the treated company that we bring into the statistical analysis
→ Firms or individuals are not randomly assigned to the "treated" or "non-treated" groups:
the treated could be already more innovative or higher-performing than the non-treated. E.g.
Why did a coffee shop introduce water? Then maybe the manager is thinking about how to
improve, so other measures may have been put into place, which affects what we observe.

Selection bias means something we don’t observe the manager's action → we are
exaggerating the effect of the managerial practice.
Lazear (2000)
Hourly pay to incentive piece-rate pay
If you break your windshield you cannot drive. Then a company will fix it.
At the beginning each employee received a payment per hour, the more hours they worked
the more money they got. Would you work fast or slow? The company noticed that they
would often stop to have coffee, have a pause…
Change incentive scheme: for each car you change a windshield you get paid a bonus on
top of the basic income.
⇒ Better workers are attracted, and worse workers are driven out.
See effect on Productivity.

Worker Types:

● Type 1 (good workers): Workers who were observed both before and after the
introduction of piece-rate pay
● Type 2 (bad workers): Workers hired before piece-rate pay but who left the firm after
the system was introduced → didn’t want to spend the extra effort for a higher salary.
● Type 3 (new good workers): Workers hired after the introduction of piece-rate pay.

Before companies choose the treatment.

Now there are some workers that choose the treatment: Incentive pay impacts not only
productivity but also workforce composition. ⇒ 2nd type of bias:
Adverse Selection Bias: Lower-performing workers may leave after piece-rate pay is
introduced.
+ Higher-performing workers may be attracted afterward

Bandiera, Barankay, and Rasul (2005)

Relative to Absolute Payment
There were two systems for picking strawberries: absolute and relative
systems.
- Relative system: divided in groups, everyone picks up as much
strawberries, and the one who picks up the most gets more, and so
on
- Absolute system: if you pick up 10 pounds you get 10 pounds
Socio network (if you’re friends or not with a person in your team) this might affect
productivity.
Those who work with their friends, when the system goes from relative to absolute, are more
productive. ⇒ social relationships among co-workers do not significantly affect productivity.
In other words, in relative incentive systems, productivity tends to drop when workers are
grouped with friends. This is because workers may internalize the negative externality their
higher productivity imposes on their friends, leading them to reduce their effort.

Lessons and conclusions

- Be careful when choosing the firm to study, the questions to ask, hypotheses to
test, the type of data and the variables to collect, and then the specific econometric
methods to test the hypotheses.
- Which methodology to apply depends on the nature of the insider data sets.
- Previously insider econometrics research tell us that the effects of management
practices on productivity are significant and the productivity of individuals workers is
determined by their environments

3. Experiments in Organizational Studies

- Lab experiments
- Field experiments

Many questions can be approached using experimentation

What do Experimental Studies do?

Ask human subjects (not animals) to respond to incentives in a

controlled laboratory environment which represents the essential economic features of a
theoretical or naturally occurring pattern.
As a model, researchers are trying to avoid any distortions that the environment can
provide and change the behavior.

Experimental economics are not new…

Chamberlain (1948):
- First market experiment in the history of economics (and management)
- Induce a market with a demand and a cost structure in a classroom
- Result: competitive equilibrium performs poorly in explaining the outcomes of
real markets
There are some predictions of the model who could replicate the market well.

Becoming more and more common…

Row player ⇒ player 1 (can choose one of the two rows)
Column player ⇒ player 2

Nash equilibrium ⇒ result of the non-cooperative game

Concept of equilibrium is close to what Adam Smith was saying ⇒ maximize the profit of the
WHOLE economy, I don’t care if the other gets more (2,0 / 2,1 = the same!)

Selection 2/3 test ⇒ Refers to the phenomenon where, over multiple iterations, the
proportion of players choosing certain strategies changes. The average payoff converges
toward a predictable outcome (in this case, 0 for the Nash equilibrium). ⇒ Overtime, the
number of winners is increasing + the avg number is decreasing towards 0.
This game theory predicts how people behave with repeated observations. If we play this for
10 rounds you will intuitively arrive to say 0. This mathematical model we can predict how
people will behave in an experiment.

When can Experimental Economics be useful?

1. To test theory
Under precisely controlled and/or measured conditions that are typically unavailable
in field data.
Institutions: voting rules, communication

2. To evaluate assumptions
- Theory has assumptions
- A theory may fail for a certain set of parameters. Will it do better with others?

3. Other cases
- To identify stylized facts: experiments are often used to identify patterns in
behavior which may or may not be consistent with theory
 Imagine I don’t have a theory, I see a pattern through an experiment and data,
and then tell the theory what we see in the model.
- To compare institutional designs: alternative institutions, or policies, can be
implemented in laboratory and the outcomes compared on the basis of efficiency
There are alternative institutions to democracy e.g Italy changed its
institutional voting design to only males vote.
China conducts institutional changes in some municipalities and if they work
well they make them available in all the provinces in China.

Incentivizing people is key to paying attention to the experiment ⇒ money?

Ethical approval ⇒ before WWII there were none. After that, there are Ethical Committees.

Important to run a PILOT

Other practical matters
1. Make sure you limit the things that can react with your treatment
a. You want to be able to replicate the environment.
b. Provide a precise description of the experimental protocol, so as to make sure
instructions are easy to understand.
2. Eliminate distractions and enable privacy
a. No talking (participants, experimenters) and no distractions
b. If a subject needs help, then answer questions privately and quietly.
c. In smaller groups, reduce the risk that one participant says something out
loud and affects the whole group.
d. Use appropriate screens if privacy is important.
3. Deception is not allowed
a. With deception comes a fear that your subjects will no longer believe what
you are telling them (loss of control) and may even start to outguess you
(changes behavior)
e.g. “I offer you champagne if you do x”, but then I don’t ⇒ deception. Could change in the
future the behavior of the participant
b. But, it is “allowed” to not reveal the whole truth.

4. Preserve anonymity
a. Single blind: no one other than the researchers will be able to ever identify
subjects’ actions or subjects’ payoffs.
b. Double blind: no one can link subjects’ decisions to their identity, even the
researchers. e.g for sensitive subjects involving sex, condoms…
c. Abandoning anonymity: if abandoning anonymity is important for the
experiment, then subjects should know and have the option to withdraw if
they want
 1. Voluntary effort and Reciprocity in organizations
Fields experiments
Kube, Marechal and Puppe (2006)

Setup: Take uni students. Student goes to the library, catalog books on excel and I will pay
her. To some people I pay 15 euros and to some 10 euros.
Question: If a student would expect 15 euros, but actually receives 20, would he work
harder? What if he gets paid 10?

KPM and GL refer to two different studies or experimental setups within the framework of
gift-exchange experiments

Different payment levels

were provided:
- "Kind" condition:
Students received
higher-than-expected pay
(€20 instead of €15).
- "Neutral" condition:
Students were paid as
expected (€15).
- "Unkind" condition:
Students received less
than expected (€12
instead of €15).
- Triangle ⇒ they gave a GIFT!!

Observe:
- People who received what they expected were the most productive
- People who received more than expected: the effect of this perceived kindness
diminishes over time, with effort levels gradually returning closer to baseline.
- Gift: People initially put a bit more effort, then gradually less overtime. If they didn’t
receive a gift their motivation decreased over time.
This study highlights how perceived fairness and reciprocity play critical roles in
determining worker effort

Coordination

minimum effort: Aligning employees towards a common goal with the least friction by
simplifying processes.
managed growth ⇒ Introducing changes gradually rather than overwhelming employees all
at once. e.g. players from Juventus and Inter can coordinate better in the Italian national
team (?)
organizational culture ⇒ people from China work differently from people in Italy.
Systematically 2 way communication improves the productivity of the whole.

4. Introduction to Randomized Control Trials (RCT)

ex. with COVID vax, some people were given a placebo and some the actual vaccine.
When classes are not too large, people learn more. What is the optimal size? There have
been some experiments in which we change the class size.
Observe how people have behaved in the absence of the program (counterfactual).

When we conduct RCT, assigning randomly, we kill the selection bias problem ⇒
whatever is the effect this will be the same.
Treated = counterfactual of the control
Control = counterfactual of the treated
 ⇒ what the RCT is going to do

What is a RCT ?
An experiment designed to isolate the influence that a certain intervention or variable
has on an outcome or event.
If I take a medicine, it means it has been approved by the gov (passed an RCT test)

We need to address the problem

of distinguishing between the
impact of a program and
pre-existing differences in
individuals or groups.
Selection bias→ those who
receive the treatment are
systematically different from those
who didn’t.
Solution: ensure that participants
are assigned to either the
treatment group or the control
group (no treatment) purely by
chance
The effect of the intervention is twice as much as the control.

Sample size, design, and the power of experiments

- Power of the design: probability of correctly rejecting the null hypothesis when in
fact is false.
- Sample size plays a key role → The more data points we have, the more accurate
the results.
- There must be:

50 obs 100 obs 50 obs

Practical design and implementation issues:

Level of randomization
- Should one randomize over individuals (I throw a coin for each person) or some
larger group (I throw a coin for every group)?
- Early randomization of deworming medicines were carried out at the individual level
within schools (Dickson et al., 2000), while Miguel and Kremer (2004) look at similar
programs by randomly phasing-in the program at the school level
- Not always a choice can arise…

E.g. in some impoverished areas in Africa. To half of the kids we’re gonna give the medicine
to kill the worm, to the other half no (randomize at individual level). Some months later, the
people that received the medicine were dying at the same rate.
BUT the medicine worked, so they investigated and found out that the kids go to school
together ⇒ The untreated have a spillover on the treated, who get sick again.
What should we do? Randomize not at the individual level, but at the school level. We take
one school and all the kids have the deworming medicine. Then we see the effect: massive
effects in the treated schools.

⇒ When there are spillovers to the controlled to the treated you need to conduct
research not at the individual but at the group level.

The choice of the level at which to randomize is very context specific, depending on the
nature of the intervention as well as the nature of the interactions between the individuals to
be treated.
Some factors could be taken into account:
1. Spillovers from treatment to comparison (control) groups can bias the estimation of
treatment effects. In such case, the randomization should occur at a level that
captures these effects
2. Randomization at the group level may sometimes be much easier from the
implementation point of view, even if it requires larger sample sizes

Cross cutting designs = methodology to test multiple treatments

or interventions simultaneously while maintaining rigorous
randomization (Participants are randomly assigned to different
combinations of treatments – e.g., receiving one, both, or none)
- To test various interventions and combinations of
interventions relative to comparison groups and relative to
each other, treatments interaction effects, multiple
hypotheses rather than one: The goal is to understand the
individual effects of each treatment, their combined
effects, and any interactions between them.
- Significantly reduce costs!!
- Example Progresa program in Mexico aimed at improving education, health, and
nutrition for low-income families ⇒ interventions Tested: Cash Transfers, Nutritional
Supplements, Educational Workshops

Data collection
What is the value of a baseline survey (= before the experiment happens)?
If we have control variables (by collecting information on gender, nationality etc.), then we
can control better and you require less people participating in the experiment.
You want to know the heterogeneity in the training of the project: what type of person
benefits the most from this training program?
If a survey is conducted well, the fraction of people who are Female should be the same,
since gender doesn’t affect the effectiveness of training.

- At first glance, randomization renders baseline surveys unnecessary, since it ensures

the treatment and comparison groups are similar in expectation.
- However, baseline surveys:
1. Generate control variables that will reduce the variability in final outcomes
and therefore reduces sample size requirements, reducing costs e.g. more
experienced employees might generally have higher productivity regardless of
the training.
2. Make it possible to examine interactions between initial conditions and the
impact of the program
3. Make it possible to verify that the randomization was conducted appropriately.
5. Field Experiments with Firms
Oriana Bandiera, Iwan Barankay, and Imran Rasul

difficulty in determining causal relationships in the behavior of firms when relying solely on
observational data.
Complexity = taxes, … There is a cost of relationship, if there is a change in policy.

For example, if a company increases wages and sees productivity improve, it's difficult to
determine from observational data alone whether the wage increase caused the productivity
boost or if other factors (e.g., changes in management or external market conditions) played
a role.

Field experiments, in contrast, involve controlled interventions, which help in isolating

causality more effectively.

What do they do?

Illustrate how field experiments, guided by economic theory, can address these challenges
and provide new answers to long-standing questions about firms:
- Do firm choices maximize profits subject to constraints?
- If so, which constraints bind and inform decision-making in firms? If not, why are
firms operating inside the frontier?
If firms all maximize profit, then why is there heterogeneity?

How do they do?

Focusing on experiments designed to shed light on firms' behavior

Experiments within firms Experiments between firms

- Workers are the units of - The firm is the unit of observation
 observations - Discuss how field experiments
- Discuss how to solve agency exogenously varied input availability
problems, from incentive pay to to shed light on constraints firms
social pressure and nonmonetary face.
rewards
Agency problem: principal wants you to work
really hard, and you (the agent) want to maximize
your income

→ Field Experiments within Firms

how can firms solve agency problems and motivate their employees

- Monitoring
- Employees decide their level of effort considering marginal costs and
marginal benefits of shirking and firms choose compensation and monitoring
policies.
- Thus, ↓ monitoring → shirking ↑ . This leads to three concerns: shrinking
behavior is hard to detect, endogeneity (the ability of the econometrician depend
on employers’ management practices, (like compensation or monitoring policies)
which may be put into place because of historical employees' shirking behavior,
which is also what the researcher is trying to study), unobserved factors (changes
in hiring policies)
e.g. in a call center, the principal realized that workers were saying that they made a sale,
when they actually didn’t. So the manager said that they would check a random sample of
their calls, if they didn’t actually make a sale, then she will take out money from their wage.

- Nagin, Rebitzer, Sanders, and Taylor (2002). Field experiments at a call center of
a telephone company. Employees must report the numbers of successful
solicitations and employers corroborated that information by calling back a fraction of
calls reported ‘successful’. If there were inconsistencies, a deduction from each
individual's weekly incentive pay would be made. Optimal monitoring scheme is a
balance between reducing shirking behavior of some workers and monitoring costs
⇒ it is optimal to NOT monitor too much. It reduces shrinking but it’s too costly!!! + If I check
too much of your work you’ll become scared.

- Paying for performance / Monetary Incentive

Monitoring is not always feasible.
so Monetary Incentives seek to align the employees' interests with the principal’s

- Shearer (2004): Workers' productivity increased by 20% moving from fixed wages to
piece rates
- Lazear (2000): selection effects explain half of the 44% increase in worker
productivity that followed the introduction of piece rates (More productive workers
are attracted to firms offering piece rates, as they stand to earn more by working
harder and producing more)
Monetary Incentives and the Social Organization of the Workplace
- Social organization of the workplace is the social relation that exist between a group
of coworkers or between workers and managers
- Bandiera, Barankay, and Rasul (2005, 2007, 2009, 2011) study the interaction among
incentives and social organization of the workplace in a U.K. producer of soft fruit
(strawberries).

- Key findings (2005):

- Productivity is 50% higher under piece rates (when you pay people for each
kg, they work harder)
- (2007): Managerial bonuses increase both the mean and the dispersion of
workers' productivity (dispersion – managers targeted their effort towards
more-able workers ) ⇒ also managers can be incentivized, the avg of the fruit
produced was higher
- (2009): When managers had fixed wages, the average worker was 9% more
productive on days when they were managed by someone who had
established a social connection with them ⇒ the workers that are more
productive are those that are managed by a manager who is a friend or have
a social relationship → manager pays more attention and worker is more
motivated by reciprocity.

When the wages can’t be changed:...

- Increasing motivation through better social relations or status

rewards
New Topics in Within-Firm Field Experiments: Other types of incentives
- More status or social recognition rewards, such as "employee of the month" job titles
- Ashraf, Bandiera, and Jack (2011): Case of 800 agents hired to sell condoms
in urban compounds. Those who received non monetary incentives sell twice
as many condoms as agents who received financial margin on each pack
sold.
The non monetary incentive increased productivity MORE than the monetary one.

- Barankay (2010): provision of individual performance feedback about relative

performance reduces the productivity of workers ⇒ discourages people in the future

Changes in the availability

of inputs change
productivity
Physical Capital and Access to Finance
- Mel, McKenzie, and Woodruff (2008) credit market imperfections and liquidity
constraints may affect firm growth in Sri Lanka
If I have a small business / shop in Italy, I can expand my business through loans. In
developing countries access to capital is harder to obtain (banks: you’re not gonna pay me
back).
These researches gave some money to some and to some they don’t (RCT). The marginal
return on the additional capital is 5-6% (high! - bank return is 1%)

- key finding:
- Return to the additional capital is around 5–6 % per month, giving a real
annual return well above the market lending rate
- McKenzie and Woodruff (2008) had similar results when applying the field
experiment to Mexico

Labor
- Study discriminatory practices
- Scheme: one audit who study two identical applicants with differences in the
characteristics to study( i.e. race, gender, age)
- Bertrand and Mullainathan (2004) study the racial gap in the labor market. They sent
CVs randomly assigned white- or black-sounding names. Its findings show white
names receive 50 % more callbacks for an interview than black names, especially for
high levels of qualification
Experiment with job advertisement: Black and White name?
Researchers say: taking all the jobs, two jobs that are very similar e.g. computer
programming in different banks (same requirements, pay, job). Creating two identical
fake CV with all the information asked to send with different names:
- typical white person name
- typical black person name
Is there discrimination?
For the same CV, the white name CV get 50% more callbacks
Lecture 3 – Incentives
Here we are going to give a first step by thinking about the problems/solutions of these
incentives. People respond to incentives. Companies know this, and try their workers’
productivity by changing the incentives.

Examples
Google: If I die, my partner gets 50% of my salary for the next 10 years.
UltimateSoftware: every two years the company pays your vacation
NetApp: There are bottom workers, middle managers and top managers. Middle managers
try to detect the people that are doing the right thing at the right time, who are called to the
office and the President thanks them (non-monetary incentive). People know that the
company is engaged in what they are doing and appreciate it.
Mayo Clinic: doctors are very good and conduct research. Each employee that is feeling
stressed can go to the “No Stress Zone”, where they can receive massages etc. People
perform better if they’re not stressed. Each employee has a 50 dollar voucher and can give it
to who thinks are the best employees in the area.

Sometimes incentives don’t work in the way you want.

Examples of failed incentives

1. Colombian Military Army in 1810 Colombia: The President of Colombia said that
for each guerrilla fighter that you bring in the more money will be given. Many
guerrilla fighters were being brought in, but they were actually peasants
2. Kindergarten in Tel Aviv (Israel): it’s costly for the school to have parents coming in
late, because they have to pay the teachers more. For each hour that you come late,
I’m going to charge you a fee. From 10% → 20% arriving late. They would rather pay
extra than pay a nanny, it’s more convenient.
3. Schools in Chicago, with a high percentage of black people, performed very poorly.
If your school performs well in the national exams, I’m gonna give you a large bonus.
They perform really well because the profs gave them the answers, to increase their
wages.
4. Colonial India snakes: The British kept dying because of poisonous snake bites.
Let’s pay for each snake that was brought dead → The Indians were raising snakes
in snake markets (after they stopped this incentive, they set all the snakes free)

Here we are going to give a first step by thinking about the problems/solutions of these
incentives.

Firm: Nexus of Contracts

how to deal with such a large heterogeneity
in interests and stakeholders?

Diverging Goals
In these relationships there might be
conflict:
- The firm wants the worker to work,
but the worker wants to be lazy.
Your behavior will be different before than
when entering in a contractual relationship
 - Shareholders want short-term returns, but the firm wants long term returns.

- The firm wants suppliers to supply high-quality inputs, but the suppliers want to save
money and provide low-quality inputs.

Tension
In these relationships there might be conflict:
- It is in everyone’s interests to come together and form organizations that support
specialization and trade.
- But, once they have established these organizations, there are incentives to
misbehave (before and after coming to Bocconi)
- So, parties must safeguard themselves against the bad behavior of others before
they enter into relationships with them.

To solve the problem of misbehavior of people we write CONTRACTS!

Contracts
- A contract is a voluntary agreement to organize a transaction.
- An explicit contract specifies verbally or in writing what each party’s obligations are.
- An implicit contract is a set of expectations that each party has about others’
behavior. (e.g. verbal agreement between me and my bf not to cheat; implicit
consensus, transaction in the supermarket)
- In these slides we focus on explicit contracts!

Enforcement
- An implicit assumption in the study of contracts is that they are enforced.
- In reality, this is not so clear:
1. Some countries’ legal systems are inefficient/corrupt.
2. Going to court is very costly.
- If contracts are not easily enforceable, then they will not provide effective
safeguards.

Contracts and Information

- We will characterize contracting situations according to information problems.
- If everybody can observe everything about everybody else at all times, then
achieving efficient outcomes is easy.
- In reality, people contract without full information about each other, or about the
economic environment.

Asymmetric Information
- The core of this topic deals with environments in which one party knows something
that the other does not.
- In particular we focus on two situations:
1. Before writing the contract, one party does not know something about the
other’s characteristics.
2. After writing the contract, one party does not know what action the other
took.

Timeline
1. Parties acquire characteristics (high/low IQ, healthy etc) ⇒ Ability, preferences,
income
2. Parties design contract.
3. Parties take actions (you use good inputs) ⇒ work, materials, investments.
4. Results of the actions are realized.
 5. Payments are made

Ex-ante Asymmetric Information

- Bocconi: choosing students if they are smart
- Car driving: when you take a taxi you don’t know the ability of the driver
- Health: for a health insurance company it’s important to know before if you’re healthy
or not (thus very costly)
Ex-post Asymmetric Information
- Building Investments: quality of inputs and work → construction project is completed,
but the quality of materials used (e.g., cement, wood) or the effort and expertise of
the workers are not fully observable
- Study hours: you can’t know in group field projects how much effort each person put
in
- Effort in sales: you don’t know the effort of the seller you hired

Compensation Scheme
- In the previous examples, notice that everything depends on the compensation
system.
- If I give you 31/31 in your exam you will not put effort.
- If I do not compensate you for the quality of the construction of the buildings in the
long run, you will not use good materials.
What is it that people care about? Income, recognition?
- If I pay you a fixed wage, you will not put effort into sales.
- An important lesson from these examples is that even if there is a moral hazard, we
can as managers control it by arranging the right compensation scheme!
The correct incentive scheme can avoid moral hazard.

Pre Contractual Problems

Adverse Selection
- In many situations there are bad types with whom people do not want to trade.
- Unfortunately, these types may be exactly the ones that are attracted to trade.
- This may prevent trade with good types.
- Classic example: health insurance market:
- To whom do companies want to sell insurance? healthy people
- Who wants to buy insurance? sick (unhealthy) people
Example: Insurance Market
A = 90 (% probability that they have an accident)
B = 70
C = 50
The price that the insurance company is going to set is gonna be the avg of all the
participants (70%) = avg expected losses.
If my exp loss is 90 (I have an accident and I pay 90 in the hospital) ⇒ you take it!
If my exp loss is 70 (I have an accident and I pay 70 in the hospital) ⇒ you take it!
If my exp loss is 50 (I have an accident and I pay 50 in the hospital) ⇒ i don’t take it, leave
the market

So avg ↑ to 80, but then B wouldn’t take it, it’s too expensive.
Here, the Insurance cannot differentiate between A, B, C so the market disappears and
we have ADVERSE SELECTION (giving the same price to everyone they select the worst
types, the one with the largest expected loss).

Example of Adverse Selection: lying on the CV, on Tinder…

 Communication
- In all our examples of pre-contractual asymmetric information, there is somebody
that loses out on trade.
- This person would like to identify themselves to the seller, and the seller would like to
know who this person is too.
- There should be some role for communication:
1. In some cases, you may be able to prove your type (medical records in
insurance).
2. But certainly not always.

SIGNALING is part of communication

The party with the information takes an action to show to the other party that they are the
“good type”
- A signal is an action that somebody can take to communicate their type to others.
- This action might have nothing at all to do with the nature of the transaction.
- A signal is effective when only some types want to send it, and others not.
- Otherwise, if all types send the signal, nothing is learned, and we are back where we
started.

Conditions for an Effective Signal

- Different types must have:
1. a different benefit from sending the signal or
2. a different cost of sending the signal
Graduating in the best school is a good signal, because there’s a cost in sending the signal:
if you’re smart you get admitted and pass the hard classes in Bocconi.
- If these conditions are met, good types are able to send the signal, and not be copied
by bad types.

A solution? There’s one solution to solve the adverse selection problem. E.g. NHS
(National Health Service) ⇒ the cost is that there’s a large use of the system. The unhealthy
get insured for free, the people that are healthy have less incentive to be healthy.

There are two ways to solve ex ante asymmetry of info:

Screening
- (if I’m the one who has the information) So, signaling is when the person with the
private information takes costly actions to signal the other party that they are the
good types.
 - (If I’m the one without the information) Screening is when a person without private
information design a contract to receive private information. e.g. entry exam in
Bocconi

In general, in order to do screening, the uniformed party must make the transaction
unattractive to those with whom he does not want to trade.
The cost of doing this is to destroy some of the value of trading with good types.
Examples:
1. 7am classes.
2. Low wages in some organizations such as charities. to find the really motivated
people

Solomon Screening Story: the two women each have a child. One day they wake up and
one is dead. They both claim the alive kid. Who has the information? The mothers. The king
(doesn’t have the info) needs to take an action to elicit private information: he asks that the
kid is cut in half. One mother agrees and thinks its fair, the other prefers that the kid is alive
with the other mother. ⇒ the second woman is the mother.

Post Contractual Problems

Principal-Agent Model and Moral Hazard

- The principal engages the agent to perform on the principal's behalf.
- But the agent might take opportunistic actions that harm the principal: moral hazard.
Moral Hazard problem: after the contracts has been signer, you behave differently than
expected
- Moral Hazard: Car Full insurance, NHS (overuse)
- Adverse Selection and Moral Hazard:
1. If you do well in the exam?
2. Health system?

Typical Principal Agent Model

Politicians =agents, citizens = principal ⇒ they promise something during the campaign
Performance Pay as solution = we can pay more if they behave better
- In the insurance example, we can force the driver to pay out some of the cost of
accidents. if you go to the doctor you always have to pay a little bit, even if you have
full insurance.
- More generally, we want to pay the agent more when results indicate he acted in the
principal's interests, and less when he didn`t
E.g. Pay for politicians is re-election.

PROS AND CONS for Smart-Working

Elon Musk says it’s “unethical”
During COVID, companies took random screenshots of the employees (w/o them knowing)
⇒ one way of monitoring. → privacy concerns

Modern technology allows for virtual "war room" setups, where people can stay connected in
calls for extended periods. This not only simulates the feeling of being in the same physical
space but also allows for spontaneous interactions, immediate problem-solving, and ongoing
collaboration without the need for constant meeting scheduling.
Exam question!!!

Vertical Integration
1st research paper: The Nature of the Firm - Coase (1973)
What does he want to do?
Obtain a definition of the term ‘firm’ realistic (so that people outside academia understand)
and tractable (well-fitted to other things that people have said about firms) that allows
identifying why, under what conditions firms emerge and what factors lead to its growth.

⇒ Firms exist to minimize transaction costs i.e. all the costs associated to making
transactions in the market. If I want coffee, I need to buy coffee beans (i.e. make a contract)
and have transaction costs.

More on the Economic planning

Transaction costs: Beyond the mechanism price
- Price mechanism (firms are price takers) is admitted as a coordinating instrument but
also the coordinating function of the entrepreneur → informs firms what to produce
and what inputs to buy – The price shows the demand relative to the supply.
- There are costs of using the price mechanism
- Cost of organizing production
- Costs of negotiating and concluding a separate contract for each exchange
transaction
Every time you enter into a transaction, you sign a contract, which is expensive (e.g. a
lawyer on both sides) → Vertical integration is a solution to the problem of paying
transaction costs
- By forming an organization and allowing some authority (an entrepreneur) to direct
the resources, certain costs are saved.

If, inside the firm you save on transaction cost, why don’t we have super big firms where
everything is done? We don’t have cognitive capacity, but can AI? (exam q?)
- Then, why are there any market transactions at all in the real world ? Why is not all
production carried on by one big firm ? Because:

Diminishing returns to management

- As a firm gets larger, there may be decreasing returns to the entrepreneur
function → the entrepreneur's ability to oversee and optimize all parts of the
organization diminishes.
- As the transactions (which are organized internally) increase, the
entrepreneur fails to make the best use of the factors of production
→Different combinations of labor, capital, and resources are better suited to
different products or services.
Coase's theory of the firm, which argues that firms grow until the cost of organizing an
additional transaction internally equals the cost of conducting it through the market.
What defines the growth of a firm?
- A firm will tend to be larger:

The growth of the firm depends on:

- The larger the size, the more likely is to make mistakes (e.g. I need 2kg of coffee for
this day, I actually need 5kg) → how quickly the mistakes grow (firm needs to be
“bounded”)
- How quickly I can decrease the cost of inputs

If a manufacturer merges with a commodity producer is it a vertical integration? YES

Another way to see it: dot is a firm in industry C. Dot in C makes input that sells to industry
B

Conclusions
- Economic definition of a firm is based on:
- Marketing costs (costs of using the price mechanism) ⇒ relying on the market
to handle all transactions incurs certain costs, known as transaction costs
(costs of finding buyers or suppliers, negotiating contracts, and
ensuring compliance)
⇒ A firm exists to reduce these costs by internalizing certain transactions.

- Costs of organizing of different entrepreneurs → if some have to pay different

costs due to inefficiencies they might leave the market
- Cost curve apparatus → how fast mistakes increase, and how the costs
attached to it when I change the size of the firm
- This is a realistic (fits with the legal definition in the real world) definition.

2nd research paper: Four formal(izable) theories of the firm? Gibbons (2005)
What does he do?
- Define and compare elemental versions of four theories of the firm

- With emphasis on why vertical integration is important

Rent seeking
I ask you for a taco-guitar → The price you can sell the taco-guitar for is 350, the price you
asked me for is 1000 ⇒ the difference is called quasi-rents (quasi because it is fictitious that
you’re gonna sell it to someone else). So we bargain for a price between 350 and 1000 ⇒
appropriable quasi-rents.
1000-350 = 650 ⇒ quasi rents
[1000, 350] ⇒ appropriable quasi rents

If instead I ask for a normal guitar → you can sell it to someone else at exactly the same
price I asked you.

Quasi-rents arise when a specific asset, investment, or resource generates a higher return in its
current use than it would in its next best alternative use. This difference in returns represents the
quasi-rent. E.g. a worker with specialized training that makes them highly productive in a
particular industry. The difference between their current wage and what they could earn in a
different job using their general skills is a quasi-rent.
Quasi-rents become appropriable when they are tied to investments or assets that are highly
specialized to a particular transaction or relationship, like a taco-guitar.

- Studied by Williamson, 1971, 1979, 1985; Klein et al., 1978

- Rent-seeking is the individually optimal (but socially destructive) haggling over
appropriable quasi-rents (difference I said at the beginning - price someone else is
willing to pay)
individually optimal ⇒ optimal for me who want to buy a taco guitar
socially destructive ⇒ if someone else asks you the guitar maker a specific guitar, you won’t
provide anymore, stop transacting

- Integration can stop socially destructive haggling over “appropriable quasi−rents”

(AQRs)
→ Vertical integration can stop the rent seeking problems
- Thus larger AQRs make integration more likely, presumably because larger AQRs
make socially destructive haggling either more likely or more costly or both
If the difference is larger, then I’m more likely to do “haggling”.

The difference between the two products is related to how specific the investment is.
Ex. if MC =500, appropriable quasi rents are between 350 and 1000, then minimum 500.

Property rights (not in the exam!)

- Studied by Grossman and Hart, 1986; Hart and Moore, 1990; Hart, 1995
- Inverse of the rent-seeking theory
- Where the rent-seeking theory envisions socially destructive haggling ex
post, the property-rights theory assumes efficient bargaining
- where the rent-seeking theory is consistent with contractible specific
investments ex ante*, the property rights theory assumes there may be
inefficiency
*Even though the terms of the investment are clear, Firm B (specialized manufacturer for Firm A)
might engage in rent-seeking behavior by Threatening to delay delivery unless Firm A agrees to
a higher price per part or Arguing that unforeseen production challenges necessitate additional
compensation.
 - Owning more assets can guarantee a bigger surplus share and so creates a stronger
investment incentive:
- if it is important to maximize one party’s investment, then that party should
own all the assets,
- if the parties’ investment incentives are both important, then dividing the
assets between the parties is efficient.

Incentive system
- Studied by Holmstrom and Milgrom, 1991, 1994; Holmstrom and Tirole, 1991;
Holmstrom, 1999
- Rent- seeking and property-rights theories focus on the make-or-buy problem and
include ‘control’ while the incentive theory focuses on principal and an agent
problem and consider agent’s action space is not affected.
- Asset ownership is merely one of the instruments of incentive system theory.
- So, the agency problem can be structured in two ways: the agent is an employee or
an independent contractor

- If the agent does not own the asset (is an “employee”): all her incentives come
from being paid on measured performance. If the agent does own the asset (is an
“independent contractor”): she receives not only a payment based on measured
performance but also the asset’s value after production occurs, so she has two
sources of incentives.
If you are a sweater maker and you integrate with buying sheep for their wool, you might be
less able to feed them well etc. so they produce less.
if you integrate you earn more per unit, but you sell less units. if you don't integrate
(outsource) you earn less per unit but you sell more units.
Do you prefer a large slice of a small pizza (integration) or a small slice of a large pizza
(outsourcing)?

Adaptation
- Studied by Simon, 1951; Williamson, 1971, 1973, 1975, 1991; Klein and Murphy,
1988, 1997; Klein, 1996, 2000a
- This theory asks whether integration or non-integration better facilitates “adaptive,
sequential decision-making” in environments where uncertainty is resolved over time
- (first best solution → no uncertainty) The second-best solution may be to concentrate
authority in the hands of a “boss” who then makes decisions after uncertainty is
resolved → vertical integration — bringing different stages of production within a single
firm — can be a way to enhance control and facilitate more efficient adaptation, instead
of relying on potentially inflexible contracts
- Similar to rent seeking theory, control is the central issue: emphasis on the boss’s
authority. However, adaptation theory provides evidence that there can be a coherent
elemental theory of the firm without specific investments.
- Suppose two parties that choose between
(a) negotiating a decision before uncertainty is resolved
(b) an employment contract (but contracts are incomplete!)
- Under certain conditions (2), it is optimal for the parties to choose the employment
contract.
- In sum, what this theory claims is that the integration decision is chosen to facilitate
the parties’ relationship.

(2) when there’s high uncertainty

Do Prices Determine Vertical Integration?
Alfaro, L., Conconi, P., Fadinger, H., & Newman, A. F. (2016).

Vertical integration can in theory both cause to increase or decrease prices for third parties
⇒ Changes in vertical integration change prices:
- FV: if a firm vertically integrates with the supplier then it can raise the price to the firm
not integrated ⇒ c < c’ P↑
- EV: if a firm vertically integrates with the supplier then it can increase efficiency and
sell to the competitor at a lower price ⇒ c > c’ P↓
ONLY A THEORY ⇒ NOT TRUE IN REALITY

Introduction
- The relationship between vertical integration and product prices has long been
controversial. (can decrease or increase)
- Firm-level vertical integration indices were constructed for a large set of countries
and industries. Different countries, different firms, different uses.
- The variation in applied most-favored-nation (MFN) tariffs to examine the impact of
tariffs on firm boundaries was exploited.
- The empirical results provide strong support for the view that higher output
prices generate more vertical integration

When prices are low → the cost of VI > benefit ⇒ NO VI

When prices are high → the cost of VI < benefit ⇒ YES VI
↑P ⇒ VI↓

- Now, consider a particular market or merger case in which the efficiency effect is the
more likely to dominate
- A price-taking firm will choose to integrate only if the benefits in terms of increased
profitability outweigh the cost of integrating
- At low prices, the productivity gains resulting from integration are not very valuable,
too small to justify the cost.
- At high enough prices, integration becomes worthwhile.

- Thus, if integration affects productivity, there is a force running in the opposite

direction, from prices to vertical integration

Main Result
Dependent variable
Vertical Integration Index Log of one plus (vertical integration index of firm f, with primary
sector k, located in country c)

Independent variables
Tariff Log of one plus the Most-Favored-Nation (MFN) tariff applied to output in sector k by
country c
0.020 ⇒ POSITIVE COEFFICIENT WITH *** (SIGNIFICANT)
If the gov increases prices, do firms VI more or less?
when there’s an increase in tariff, vertical integration increases
There is a strong effect for firms that not exporting (domestic)

- Domestic_f (1) , (3) ⇒ (0.009) (0.005)

So, we thought that VI has an effect on prices, but actually prices have an effect on VI
(1) When increase tariffs on goods imported (suppliers) and thus VI increases
Splitting up between domestic (non exporters) and tariff (exporters)
(2) This doesn’t happen for non domestic firms but only for domestic
0.003 → for non domestic
0.021 → for domestic

Does vertical integration affect firm performance?

Evidence from the airline industry Forbes, S. J., & Lederman, M. (2010)

- There are major airlines and regional airlines

- These regional airlines sometimes are integrated with major airlines, sometimes not.
- American Airlines (a major airline) arrived in NYC, as well as other airlines… They all
want to go to Ithaca ⇒ if it’s not integrated, it might not be so convenient for
customers, because the regional airline is taking some people from each airline
Introduction
- This paper investigates the effects of vertical integration on operational
performance in the U.S. airline industry
- Large U.S. airlines use regional partners to operate some of their flights. Regionals
may be owned or governed through contracts.
- It is estimated whether an airline’s use of an owned, rather than independent,
regional airline at an airport affects delays and cancellations (how the performance is
evaluated) on the airline’s own flights out of that airport.
- The results indicate that, operationally, there is a performance advantage to vertical
integration.

Main Result → Impact of Ownership on Delays

Dependent variable – Departure Delay
Measure the time between the scheduled departure and the actual departure of an aircraft
from the gate.
Independent variables → Fraction Owned by major: Measure the extent of a major’s
vertical integration with its regionals at an airport.
Owned Regional Flights: Measure the number rather than fraction of all regional flights that
a major has departing from an airport on a day that are operated by a regional that is owned.

The performance advantage increase on days with adverse weather and when airports are
more congested.
When there’s rain or a lot of congestion, the regionals have a lot of delays and there are
more benefits from owning regionals. When there’s rain, with Vertical integration, lower
delay, higher performance.

⇒ Adaptation theory in practice

If there’s more uncertainty, it is better to vertically integrate ⇒ they can adapt better

Conclusion
Integrated airlines perform systematically better than nonintegrated airlines at the same
airport on the same day.
The performance advantage increases on days with adverse weather and when airports are
more congested.
Vertical integration may facilitate real time adaptation decisions

Make Versus Buy in Trucking: Asset Ownership, Job Design,

and Information
Baker, G. P., & Hubbard, T. N. (2003)
⇒ Close to incentive system theory
I need to bring luggage from Milano to Torino. I’m the owner of a luggage
or I outsource to someone who does that service?

Introduction
- A model of asset ownership in trucking was developed, which is tested by examining
how the adoption of different classes of on-board computers (OBCs) between 1987
and 1997 influenced whether shippers use their own trucks for hauls or contract with
for-hire carriers.
- Two types of OBCs began to diffuse in the trucking industry in the late 1980’s: trip
recorders and electronic vehicle management systems (EVMS).
- The results strongly suggest causal links between informational and
organizational changes in the trucking industry

OBC and EVMS

OBC (Own Born Computer)-> Trip recorder adoption -> Better monitoring how drivers
allocate time and effort across tasks
Thanks to a GPS I can reconstruct the trip of the truck → to monitor
We’ll see: If I can monitor better then I will vertically integrate more.
EMVS-> It helps to communicate and localize trucks (we are in 90s no cellphones) ⇒ we’ll
see: more outsourcing

Main Result
OBC Adoption and Asset Ownership
Dependent variable → For-hire carriage share in cohort i at time t ⇒ outsourcing
Independent variables → OBC, the share of trucks with either class of OBC installed, and
EVMS, the share of trucks with EVMS installed.
↑For hire carriage ↓VI
This doesn’t have * but they’re statistically significant.
More OBC more VI
More EVMS less VI

new communication technology that improves coordination ⇒ EVMS

Coase’s theory of the firm Reading

– look at extra (review?) slides

Management Practices
What are we going to do?
Study the role of management and managers as key drivers of productivity at both the firm
and the country/industry level

GDP depends on the Labour, the Kapital, many other factors and the TFP = Total factor
Productivity.

Note: as GDP per worker increases, the TFP increases.

It shows that the smarter I am (TFP), the more work I put in (↑GDP).
This is empirical evidence, but this notion was already in the Wealth of Nations by Adam
Smith.
But we see a lot of variation also within a country! (see later)

TFP: there was a great growth in TFP in 1950s ⇒ years of lots of changes in managerial
practices – Marshall Plan (US came to hep European economies to reconstruct and
changed managerial practices)

Large productivity variation also within countries

 - An average US 4 digit industry plant at 90 percentile
has ~4x labor productivity of 10th (Syverson, 2004)
percentile. In other words, in the U.S., the most
productive plants (90th percentile) are roughly 4
times as productive as the least productive (10th
percentile) in the same industry.
- Controlling for other inputs, TFP difference is about
2:1 → When adjusting for inputs like labor and capital,
TFP accounts for a large portion of these disparities:
in U.S. plants, 90th percentile plants are about 2
times more efficient than 10th percentile plants.
- In India this gap is about 5:1 (Hsieh and Klenow, 2009
QJE). In India, this gap is larger—around 5 times.

FOCUS ON MANUFACTURING
In US: above than 10% of other companies (10th percentile)
→ better managerial practices so it has a lower gap than in
India.

“Frontier” firms ⇒ largest productivity

The gap between the frontier and the other increases over time, both in manufacturing and
the service sector.
The inequality in productivity across firms has increased overtime.

Management and performance

On the role of management and managers
“It is on account of the wide range [of managerial ability] among the employers of labor
that we have the phenomenon in every community and in every trade some employers
realizing no profits at all, while others are making fair profits; others, again, large profits;
others, still, colossal profits.”: wide range of managerial ability ⇒ wide range of profitability
Francis Walker (Quarterly Journal of Economics, 1887)
Why didn't they test if this is true?
- Porter: there is no difference in the managerial ability (if I implement something
successfully, you are going to copy me) ⇒ no point in testing it
- Syverson: very hard to test it

What has been done?

- The World Management Survey

2004, 2006, 2009, 2014

Medium sized manufacturing firms (50-5,000 workers)

⇒ not small firms, because they don’t follow managerial
practices

Extended to Hospitals, Retail, Schools

Asking questions to managers

Testing management and performance relation

log → diminishing returns

Before this management survey, everyone ran this survey without the management.
These are conditional correlations: conditional because we compare two firms that have the
same capital and labor. (exam q!) Conditional on other factors, better management is
correlated with higher firm performance.
Causation = change in management ⇒ change in performance
Correlation: no such thing. We could have reverse causality, a change in productivity could
have a change in management (e.g. productivity down I hire someone better and new for the
management team)
The table reports the impact of management (dependent variable) on several independent
variables: ln(sales), ln(employment)...
The management score is positive and significant on the variables: ex. on sales (β=0.150,
p<0.01), except for the exit binary variable – whether a firm exits the market: better
management reduces the probability of a firm to exit.
⇒ Better management scores increase productivity.

Positive relation: when we increase the management score, we have firms that do better in
all of these.
We’re still talking only about correlation: which is unsatisfactory (reverse causality!)

An empirical study: Bloom, Eifert, Mahajan, McKenzie and Roberts (2013, QJE).
- Randomize management practices delivered by Accenture to 20 plants in large (300
person) textile firms in Mumbai, India
- Control firms get one month of diagnosis.
Treatment firms get one month of diagnosis, four months of intervention.
Why one month of diagnostic: Accenture needs to receive something in exchange →
diagnostic on what they are doing and what they should be doing.
- Collect weekly data for all plants from 2008 to 2010
Control plants ⇒ just receiving feedback on what they should do, not being “treated”,
Treatment plants ⇒ given to plants, not firms.
I’m firm, I have Plant 1 and 2. Plant 1 gets treated, but there are other plant (green line) from
the same firm, that also increase the # management practices.

Treatment plants are much better in productivity.

An empirical study: Giorcelli (2019)
- Evaluation of long term effects of the Technical Assistance Program in Italy after
WW2 using micro firm level data
- Id strategy based on comparison between firms that applied for the program, but had
differential treatment due to exogenous budget cuts
- Sustained effects in the long run in:
- Survival, productivity
- Investments, human capital, markups, exporting

Natural experiment: something in the environment naturally creates the survey.

A+B: With a change in management ⇒ change in 0.4 in TFP; while changes in capital just
0.1

C: The management and the capital are complementary. If you add both you observe
that you get 0.6: you obtain more than the sum of both.

Conclusions
- Measuring management is possible
- Correlational and causal effect that management matters
- 1st change in management → 10% productivity (very large!)
- 30% of productivity gaps across countries → purely because of
management practices: if you give better management practices to
developing countries, they will become MUCH more productive
- Contrary to standard assumption that firms live on the productivity frontier
- Implementation matters and can make a difference

Different managerial practices across firms

Differences across and within countries

Recall:
Porter, 1996, Harvard:
“Few companies have competed successfully on the basis of operational
effectiveness…. Competitors can quickly imitate management techniques… The
most generic solutions diffuse the fastest”
But actually, there is a lot of heterogeneity
⇒ WHY THERE IS HETEROGENEITY IN MP?
Four concerns

Awareness
Lack of awareness

“Excluding yourself, how well managed would you say your firm is on a scale of 1 to
10, where 1 is worst practice, 5 is average and 10 is best practice”- Asked at the end
of WMS survey
No link between self-scores and performance
⇒ People think they are doing better than what they actually are doing. Firms often
rate their own management practices highly (scores leaning toward 6–10 on the
x-axis), yet their actual performance (ROCE) doesn’t always align with this
self-assessment.

How to overcome the lack of awareness?

Experiment ⇒ Nguyen and KIM (2018)
Run an RCT, to explore whether providing randomized information about the benefits
of Quality Management (QM) can help managers better understand and adopt
effective practices.
Key findings:
• Large variance in estimated (perceived) benefits of QM across managers
• Smaller mistakes among educated managers about QM benefits
• Sizeable effects of information debriefing with government officials: when
government officials debriefed firms, it amplified the effectiveness of the intervention.
• Treated companies are more likely to adjust their beliefs on benefits of QM, adopt
QM, and improve performance
Skills
Education for Non-Managers and Managers Appear Linked to Better
Management (in manufacturing and retail)

Firms that have a larger fraction of educated workers or managers have higher
management scores
Related research on 10,000 small (entrepreneurial) firms in LDCs finds a
similar importance for education
illustrates how
business
practices vary
within firms in
these
countries. The
x-axis
represents the
business
practices
score, and the
y-axis
represents the
density or
frequency of
firms with that
score
Digispan recall & Raven test score are Cognitive tests to assess memory.
Owner is male means higher business practices because when the owner is female,
she will have more difficulties in bargaining with government for the adoption of
manager practices (LDC societal problem)
How to overcome the local skills gaps?
- Supply of universities within regions linked to improved management and
higher skill premia (Feng and Valero, 2019)
- Through formal management training courses (Bruhn et al, 2018), but
effectiveness of the training depends on the delivery methods (Maloney and
Iacovone, 2019)
- Acquisitions as direct transmission mechanisms:
- Better managed firms more likely to M&A and improve management of
target firm (Bai et al 2018)
- A way through which MNEs (Multinational Enterprises) may improve
productivity across countries
Incentives → competition
Institutions: Positive effect of competition on Total Factor Productivity (TFP)
There are some firms that don’t want to increase productivity because they don’t
have the incentives to do so.

Positive relation between competition and management

Some experiments on institutions and management

- China accession to the WTO: Growth of Chinese exports by industry using
accession → it’s a natural experiment on significant rises in management
(Bloom, Draca and Van Reenen, 2015)
- Political marginals: Hospital competition in UK under mid-2000s reforms
(Bloom, Propper, Seiler & Van Reenen, 2015), find positive impact of
management
- Labor Regulation: effect of right to work law implemented in Indiana and
Michigan in 2012. They compared them with Kentucky, Ohio, Wisconsin and
Illinois and found incentives practices were statistically lower in the last
What is the impact of a family / founder CEO on the adoption of management
practices?

These firms may centralize control with the founder or family members rather than
delegating authority to middle managers + middle managers may not be hired or
developed with the same emphasis on high-quality skills, especially if their role is
more about executing orders
Their incentive is to maintain personal authority – so no incentive to adopt
management practices.

Organizational frictions → lack of trust

“Many management practices cannot be reduced to well-defined actions that can
be specified ex-ante and verified ex-post.
Instead, the implementation of these practices is crucially dependent on context.
[…] The state-dependent actions necessary to meet these expectations cannot be
the subject of a formal contract. […]
Adoption is based on relational contracts, roughly, understandings that the parties
share about their roles in and rewards from cooperating together, which cannot be
enforced by a court” - Robert Gibbons and Rebecca Henderson, 2012
The idea of a relational contract is that an agent participates thinking that the value of such
a contract will be advantageous in the future. A key part of the relational contract is TRUST.
An organizational friction that doesn’t allow me to implement managerial practices is
indeed the lack of trust.

Example of relational contract - The Andon cord at Toyota

Ex. in Japanese Toyota people trust each other a lot, so their managerial practices do not
transpose to other non-japanese car manufacturing firms.
What does “they trust each other”? If they notice in the assembly line something that is
weird, they pull a cord (called “Andon cord”), which alerts everyone that there is a problem,
the production stops and the manager comes.
Notice how this requires not only the managerial practice of the cord but also a great degree
of trust between workers and supervisors ⇒ there’s a relational contract!
The supervisors believe that:
- the employees won’t pull the cord if they are tired and want to stop working.
- the worker will offer suggestions to improve (even those in the worst case scenario
which will make their job redundant) ⇒ they have work ethic.
The workers believe that:
- the supervisors will also recognize the problem: when the cord is pulled, they will
come and try to figure out how to solve the problem
- the supervisor will implement the improvement without necessarily cutting jobs

Taking it to a different environment where there is no reciprocal trust ⇒ doesn’t work

Conclusions
- Management practices vary across firms and countries
- Awareness, Skills, Incentives and Organizational frictions can explain these
differences
- Diffusion of practices is not frictionless and goes from basic to complex
Role of management in the public sector

These 4 concepts need to be clear:

- Multi-tasking problem ⇒ Issue: as a worker, only the output of task A might

be observable, while B’s is not. If you’re remunerated only for Task A, you’re
going to put all your effort into that
E.g. if a professor at Bocconi is tasked with research and teaching, but they
can only observe the teaching. The multi-tasking model says that the
professor will put all its effort into teaching.

- Performance and promotion is evaluated by the subjective evaluation of the

manager, so workers are going to spend a lot of time trying to influence the
views of the manager, instead of being productive

- Multiple principals deriving from multiple goals, so a worker can only reach a
goal which is maybe not so efficient for one principal

- Preferences and intrinsic motivation: Risk-adversity limits high-powered

incentives (see infra). Some people (risk-averse) might prefer not to be
incentivized.
Additionally, a worker might gravitate towards a public sector job because
they have an intrinsic motivation to help do good – when they start getting
paid to do good, they become demotivated.

PUBLIC SECTOR, we focus on:

Data on public sector

Empirical evidence
1. McCormack, Propper & Smith (2013)
- Performance is measured through perceptions on teaching, research and
good university guides in 112 Universities. Performance basically depends on
reputation, which means they have better management practices
- Results: Older research intensive universities score better– the older
research unis have better management practices – than newer teacher
focused universities (huge heterogeneity within universities: e.g. department
of biology is very different from department of chemistry).
2. Fryer (2012)
- In Asia the public school system is not very good, they implemented the
Charter Schools system: ask very good private schools to manage public
schools
- An RCT on introducing Charter teaching & management practices in Houston
public schools: 100 public schools, half received the management of private
schools, the other didn’t receive anything
- Result: In those who received the management practices, there was an
increased math score (productivity) by 0.28 SD & reading by 0.08 SD in
national test exams
In this graph we observe that these quintiles (distribution cut in 20%) ⇒ as management
improves, student performance improves.

Rasul & Rogger (2013) – EXCEPTION!

- Case of Nigerian Civil Service.
- They adapted and gave the World Management Survey (WMS) for 4,721 projects
across 63 organizations, through which they measure the management practices and
the performance
- Results: increases in management score reduced performance (successful
project completion) and increases in autonomy increased performance.
WHY?
- Multi-tasking
- Intrinsic Motivation
⇒ For each civil servant they have two elements:
- management practices, which followed some protocols
- how much autonomy / decision power
Since there is multi-tasking and intrinsic motivation, these people came motivated to
the public sector to make decisions. When there’s less autonomy and more
management practices, they lose motivation and perform worse.
This is the only example we have like this: here we have other elements like
autonomy which are against following MP (set of rules, checklist).

Another aspect of public sector: HOSPITALS and healthcare

Different Countries Hospitals

The dependent variable is mortality, which can be decreased by better management
practices.
1st row → Bottom line: better management practices decrease mortality.
2nd → how big it is
3rd/4th → for profit or not
other controls…

Managerial actions and performance on US hospitals

 1. Significant variation in managerial pay across hospitals (→ large heterogeneity),
uncorrelated with hospital performance (Joynt et al, 2016)
⇒ no correlation between payment and actual performance of the hospital
2. Considerable variation in pricing across and within hospitals, which is correlated
with higher prices in less competitive areas (Craig et al 2019) → they cannot pay
lower prices when there’s more competition because there are varying management
practices
3. Heterogeneity in adoption of a revenue generating practice (Sacarny, 2018) e.g.
accounting procedure adopted

Similar to the previous graph ⇒ A general pattern is that those countries that perform better
in manufacturing, perform better in other sectors as well, like hospitals..

y axis ⇒ fraction of
firms that belong to
this category
x axis ⇒
management score
India is more towards
the left than the US ⇒
also in the
manufacturing sector.
Between 1-9, we
compare the
distribution of US with
India (which is very
skewed)
The TWO ways that we can solve the management in the health sector:

- Training of the middle management: what do they do?

- Change the CEO / top management

Doctors are peculiar: they don’t want their bosses to not be doctors, as they think they
wouldn’t understand the relevant medical procedures ⇒ interesting and hard for them to
have availability of time to spend studying management.
In an organization doctors want to be covered by a medical doctor, but since this requires a
lot of intensity in training, it’s difficult that they have space in training in management →
putting people who know about management are disregarded
⇒ an option is to give short specialization courses for doctors to train in management

#1 The role of managerial training

- Selection to managerial roles often based on seniority, not skills (and typically
dreaded by clinicians)
→ No systematic managerial training available once promoted
- Clinical leaders with management training associated with significantly higher
management scores and better clinical outcomes (WMS)

#2 The role of CEOs

- CEOs with clinical training are more effective in academic hospitals and CEOs with
private sector background (higher competition) are more effective in highly
competitive environments
- Although, there is no evidence of systematic and persistent differences in hospital
outcomes across CEOs
~ Ongoing work on this area. There is already evidence on the influence of the
CEOs’ management style in the private sector and in the public sector for the case of
schools

Many outstanding research questions more broadly

1. Why are so many firms badly managed – what matters most?
Open question, let’s think about it in GFP: are they good? How is that related to other firms
in the industry?
 2. How does management change over the firm’s life cycle start-ups versus mature
firms?
GFP shows how the management practices evolve overtime.
3. Are different management practices complementary, or are their impacts more or less
additive?
First is correct
4. What is the role of leaders and culture?
5. How can organizational frictions be reduced in the public sector?
Personnel economics
Lazear and Oyer (2007)

What are personnel economics?

Application of economics and math to traditional topics of human resource management
= everything companies do to consider human resources, in particular how they impact
productivity.
Questions to answer: how to hire people? When should I fire people? How to incentivize
people? How to promote people?

(-) A normative side of this area (what it should be?) e.g. there should a grading system, a
certain way to incentivize people
(+) A positive side (what is there to decide?)

Personnel economics handle the alternatives (what is better for the organization?):
1. form of compensation: pay per hour or per day?
2. individually or teams?
when coordination costs are high, better to do individual
when the complementarities are large, better to do in teams
3. recruiting talent: where to place job advertisement
4. benefits and stock grants: we can pay with bonuses and incentives schemes, but
also stock options ⇒ workers are gonna work harder, increasing the value of the firm
and thus receiving higher returns (incentives of the company ~ incentives of the firm)

COMPENSATION SCHEME
Determine the wage of the worker (compensation scheme) to maximize profits, subject to
two constraints: incentive compatibility and individual rationality.
I make sure that for doing what I want them to do they:
- receive more than doing something different – INCENTIVE COMPATIBILITY
 - receive more than what they would receive in the post possible other job (outside the
organization) – INDIVIDUAL RATIONALITY (=participation)

max pi maximize profits you have to set

appropriate wages

s.t. subject to two constraints:

- IC (incentive compatibility The utility for the worker for executing

constraint) ⇒ the action that I want the worker to do x*
u(x*)>= u(x) needs to be larger than the utility of
doing any other action x.

- I.R. (individual rationality) ⇒ The utility for the worker for executing
u(x*)>=0.0 the action that I want the worker to do x*
why 0 should be larger than the outside option,
which is exactly the utility she would get
by being in an alternative organization

UTILITY

This Utility function is concave in income ⇒ u( )

→ when I have 0 income, and you give me a 100 euro
I increase my utility by a lot
At the beginning it increases a lot, overtime the
increase is smaller and shorter.
Utility function is concave ⇒ person is risk-averse i.e.
needs extra money to participate in lotteries.
If workers are risk averse, they would not like to take a
contract in which the income might change a lot from
one period to another.
Ex. I’m an umbrella seller and people buy them only if
it rains. Lottery: rain / not rain. Differences in income
means that I’m unhappy, if I’m risk averse.

⇒ By putting together the compensation scheme subject to constraints and the risk of
adversity, there’s an interesting trade-off between insurance and incentives(!!!!):
Employers face a trade-off between:

● Providing insurance: Offering stable or guaranteed income to employees to

address their risk aversion.
● Providing incentives: Linking compensation to performance to encourage effort.

Workers don’t want to work (Moral Hazard problem), but workers’ behavior can be changed
through incentive scheme
⇒ However, since some are risk averse, if there’s an incentive scheme, they
might not want to participate in the contract and leave the company → you can
only incentivize those workers that are less risk-averse

Relative Performance Evaluation

- Use the performance of a peer group of competitors to filter out shocks that are
common to the whole peer group.
Examples:
- the exam was too difficult, it was raining, the whole class was sick ⇒ we look at
relative performance
The downside of relative performance is that there can be collusion among workers.
- if I pay my three workers depending on how much relative effort they put, they can
just relax all day and I have to pay them regardless of the job not getting done
efficiently

The standard economic model says that: the wage should

be equal to the marginal product of labor i.e.
w = mpl
mpl = if I put an extra hour of labor, how much extra value I
produce
w = wage I’m paid

Issues with this:

- Increases in wage happen i.e. promotions. BUT
Decreases in wages don’t really happen ⇒ strong
culture against it, they would rather leave the
company.
→ Implication: if next year I’m less productive than
this year, my wage should decrease in this model. In
practice, this doesn’t happen.
- Superstar effect: see Chris Martin and singer in the
metro: are they the same? they have the same mpl
(they are both good), but they don’t earn the same wage? See red line: for some
jobs, everyone for a certain level has a low wage, but at a certain point you reach a
level in which you earn exponentially more (everyone wants to pay and hire them)

→ Tournament theory: if you reach a certain level, then you earn a lot. But if you’re slightly
lower you won’t. It’s a tournament because a lot of people are competing for the highest
prize.
- Prizes are fixed in advance and depend on relative rather than absolute performance
(if others do bad, you do good)
- Larger spreads in wages at different levels motivate those at lower levels to effort
more
⇒ we observe a lot of inequality, which makes people work hard to get to the “prize”.
E.g. CEO earns 2 million, VP earns 400.000 → why would allow such an inequality?
Because that incentivizes people (at the bottom) to work hard to get promoted.

Matching Firms and Workers

Two ways to think about it:
1. There are very good workers, who will be the best in any environment (in Apple,
Amazon, Bocconi, coffeeshop) OR
2. The matching that the worker has with the company has an effect on productivity!

⇒ Matching:
- Labor is not like any other input… it has a huge heterogeneity!
- The issue is matching the right firms to the right workers. There is not one worker
which is the best for everyone.
- Study the selection process from two approaches: game theoretic models of
asymmetric information and models of efficient matching with symmetric learning
about worker productivity.
 Learning Models
1. Firms and workers equally well informed

Theoretical Model of efficient matching with symmetric learning about worker productivity.
The production of a worker i has on firm j in period t, depends on three things:
- alpha_i = ability of the worker (innate ability)
- mu_ij = how much this firm increases my productivity, i.e. makes my ability come out
⇒ depends on the culture and the mission of the company, where the company is
located, what my boss values…
- epsilon_ijt = error, since there are random productivity shock e.g. leak inside my
office so I can’t work

2. Asymmetric Information Models → how to solve the problem?

How employees match to firms when one party is better informed:

Common solutions …But they do not always bring efficient results!

- Costly signals to infer the ability of - A firm could take advantage of highly
applicants. (Spence 1973) productive individuals, avoiding their mobility
cfr. Incentive slides on Signaling (Akerlof 1970) → Talent Hoarding ⇒ if you
- Self selection. (Salop and Salop, 1976): are valuable in a team (you work hard to get
workers decide to work in some places, not in promoted), the boss is not going to promote
others you to keep you in her team, generating
- Incentive schemes. Lazear (1986a): changing frustration in the employee
the incentive scheme can result in some - Discrimination. (Milgrom and Oster 1987) – see
people being attracted or not to the infra paper on women musicians
organization

Firing and Displacement

Standard Economic Model: decrease mpl, lower wages. In practice, not true:
- There are cultural norms against wage reduction
 - Wage compression: irl wages are not in decimal points (10.1→10.2) but 100 →150.
So with increasing mpl ⇒ they are in the same “step” of the function. e.g. if mpl1=60,
mpl2=65, they have the same wage, the wage is compressed.
Theoretically, if an organization wants to change the workforce (e.g. new IT system), a way
to do it is by firing people. In practice:
- There are a lot of legal restrictions to fire a large number of people (Unions)

How to determine the level of compensation?

- In a perfect scenario, enough to keep employees from leaving the firm and equal to
marginal product of the employee’s labor (w=mpl)
- In an imperfect (real) scenario, there is a learning process.
Farber and Gibbons (1996)
- Expected productivity depends on the innate ability and observable
characteristics (education) ⇒ can have an IQ test and pay that, but pay more
for having a good education…
- They found a strong correlation between the last and the current wage →
when the company sets a wage, it’s trying to infer the previous wage.
- Also that employer’s learning plays a key role in the development of wages:
how the wages evolve inside the organization, because the employers learn
the ability of workers.

Organization of Work

Three questions:
1. Job design: if I have 10 tasks, how do I decide the allocation among workers? → The
answer is comparative advantage (e.g. if you’re very good at the computer, I will give
you this task)
2. How to form teams → enhance collaboration
3. Teams and managers hierarchies are substitutes of incentives → hierarchy enforces
discipline and compliance through oversight + incentive to be promoted
For example: if I have a strawberry picking business, is it important that my employee is
educated or not? How much variation does it make?
If I’m looking for a waiter, how important is it that they speak italian?

Conclusion
- Personnel economics is an essential tool to learn about the underlying economics
of human resource policies
- Trade-off between insurance and incentives: THEORETICALLY, when you change
the incentive scheme → the fraction of income that comes from the incentive bonus
scheme compared to the fixed gain increases → We predict that, since there are
risk-averse people, some are going to leave.
However PRACTICALLY, studies in real companies show that, when schemes
change, they don’t leave
- Studies of single firms are useful for understanding the effects of various human
resource policies such as incentives, team-based production, and the role of jobs
Questions that are worth further pursuit: How important is it that firms find the right workers
for their particular context? Why do firms send employees to executive education training?
Or What explains the lack of evidence to support the risk/incentive trade-off predicted by
moral hazard models?

TWO EMPIRICAL PAPERS:

1. The Effects of Human Resource Management Practices on Productivity: A Study of
Steel Finishing Lines
Ichniowski, Shaw and Prennushi (1997)
Look at the variation of steel finishing lines companies.

What do they do?

- Provide new empirical evidence on the productivity effects of
alternative employment practices using data on 60 steel finishing
lines

How do they do?

1. Minimize heterogeneity by collecting a unique data set on a sample of steelmaking
operations
2. Conduct field interviews for one to three days at each site (to each of the 60 steel
finishing lines) collecting information on:
a. HRM practices: what effect to they have,
b. the performance of the finishing lines,
c. the capital equipment used in the production processes → note: the more a
machine is on, the more it produces
d. other inputs into the production process, and wage data
3. Estimate productivity (capital equipment, employment practices, and other
line-specific determinants of productivity)
Linear regression

The coefficient γ′ captures the effect of HRM practices on productivity.

Xit​: Line Characteristics →Includes technological features of the production process,

such as capital equipment quality, automation level, or input use. The coefficient β′ measures
the impact of these line-specific characteristics on productivity.

error: any unobserved factors or randomness that affect uptime productivity but are not
included in the model

4. Consider fixed effect

Why subtract the Mean? By differencing, this transformation eliminates unobserved, time-invariant factors that
are specific to each line, such as management quality or baseline productivity levels. So you only have the
effect of introducing the HRM practice
Four systems of HRM practices

Each HRM system represents a combination of practices. For instance:

● HRM System 1 includes all practices with a proportion of 1.00, indicating that every
production line in System 1 has adopted these practices.
● HRM System 4 has no adoption of these HR practices (all values are 0.00)

How are they combined?

Consider 7 practices:
1. Incentive pay: are we promoting, incentivizing workers?
2. Recruiting and selection ⇒ how much screening is put in to interview the workers
3. Teamwork ⇒ is there high participation; are there multiple teams and formal team
practice: do teams meet and try to find solutions together
4. Employment security: if they get fired, are they going to have an income for a while
5. Flexible job assignment → Job rotation: people learn about different dimensions
of the job. Are people going to go around the different branches (rotate) to learn
about the company?
6. Skills training: to low or high skill workers?
7. Communication ⇒ how do workers communicate and how do they meet (officials
channels or not), how do they receive feedback on which channels
RESULTS – Table 4 provides estimates of productivity (measured as percent uptime)
across the four HRM systems, using different statistical models

% uptime = variable which measures how much time the production line is running as
opposed to down ⇒ greater productivity
FOCUS ON Column 1:
1. HRM System 4 is the baseline, everything is read against it: how much are the
system 1/2/3 more productive than system 4 (not having any practices)?
HRM System 1 is 9.7% more productive than System 4 ⇒ since they are all
significant, the data indicates that having all practices (System 1) or at least some
(System 2 / 3) is better than having none.

2. SIZE OF THE COEFFICIENTS: S1 has a larger coefficient than S2 and S3, so it’s
better to have all the practices.

Conclusions
- (!) Innovative HRM practices raise worker productivity and tend to be complements:
when I have two HRM practices I have a LARGER productivity than what I would get
if I had each of them independently and then sum.
- Systems of innovative HRM practices have large effects on production workers'
performance, while changes in individual employment practices have little or no
effect ⇒ how you treat a specific individual

2. Orchestrating Impartiality: The Impact of “Blind” Auditions on Female Musicians

Goldin and Rouse(2000)

Introduction:
- A strong presumption exists that discrimination has limited the employment of female
musicians, especially by the great symphony orchestras.
- Among the Big five, no more than 12% were women
- Discrimination of music directors against women. Zubin Mehta-Los Angeles
Symphony (1964-1978): “I just don’t think women should be in an orchestra”.

What do they do?

They study if there is discrimination against female musicians during the various rounds
(preliminary, semifinal, and final) of a hiring process.
How do they do?
- Audition records of eight major symphonies orchestras from 1950 to 1995
- First name of the musicians to determine sex = 96% determined
- 7,065 individuals and 588 audition rounds

- Differences-in-differences strategy (cfr Methods class)

β,γ,δ: Coefficients measuring the effects

- They don’t show significance here – but the results are all significant!!
- 8 columns: the even (2, 4, 6, 8) columns have all the controls ⇒ more coefficients.
- FOCUS on coefficient “Female x Blind”: positive and significant ⇒ in a blind audition
when I have a female candidate she is more likely to pass in the next round than the
male candidate.
- FOCUS on:
- column 1 ⇒ there are effects in preliminary rounds
- column 5 ⇒ no effect without preliminary rounds in semifinals
- column 7 ⇒ blind audition however helps in the finals

Result: Positive effect of the screen at preliminary round (without semi-final) and the
finals
- Blind audition ↑ the likelihood that a woman will be selected by about 11 pp.
(at the preliminary round)
 - For female musicians who made it to the final round, column 8 shows that
screen has a positive effect of 33 percent
⇒ very strong evidence of discrimination against women in classical music
orchestras

Conclusions

Why care about discrimination? There are ethical reasons, but also a great motivation: talent
is everywhere and in an organization, as managers, we want the best people, no matter their
gender, age, nationality…
Knowledge Hierarchies
Hierarchies and the Organization of Knowledge in Production
Garicano (2000)

Introduction
- Hayek (1945): each individual can acquire knowledge about a narrow range of
problems. ⇒ how organizations organize human assets to manage the knowledge
- Knowledge-based hierarchy: the organization decides who learns what and whom
each worker should ask to deal with an unknown problem. So, people that have more
knowledge are at the top.
- Communication: workers acquire only the most relevant knowledge and, when
confronted with a problem they cannot solve, ask someone else.

A Model of Communication and Knowledge Acquisition in Production

- Production
- A worker operating a machine confronts a range of problems that must be
solved to produce
- Assumption: there is a set of possible problems and a set of knowledge to
solve some of these problems. Also, workers face a cost of learning
- The problem a worker confronts is to choose the length of the interval of
knowledge acquired to maximize expected output.
Each worker is going to choose how much knowledge to acquire.
Z = knowledge of a worker
if the knowledge required > Z ⇒ they need to ask questions.

- Communication and Organization

- Cost (h): wasted time → cost of asking questions, independently if the person
asked knows the answer or not.
- Assumption on cost (simplification): aggregate all the communication losses
in the worker who is being asked and consider communication cost even
when the worker asked does not know the answer.
- Assumption on organization: the size of the organization is large enough
and everyone in the list of a particular worker may be eventually asked if
necessary

Characteristics of the organization

- Proposition 1: Specialization
There are two types of workers:
- PS ⇒ Problem Solvers → if they don’t know, they ask the others
- PH ⇒ Problem Helpers → do not operate machines, only specialize in helping the
others
For any given allocation of knowledge, workers in each class specialize either in production
or in the transmission of knowledge about solutions. Moreover, only one class specializes in
production, and all other classes are formed by problem solvers who support workers in that
class.
- Proposition 2. Non Overlapping knowledge.
No solution is known by two different classes.
What the PH know is something that the PS doesn’t know.
E.g. The janitor knows how to operate the cleaning products, while the rector of Bocconi
doesn’t.
- Proposition 3. Organization by frequency
Production workers learn to solve the most common problems; problem solvers learn the
exceptions. Moreover, the higher up in the list of production workers a problem solver is, the
more unusual the problems she is able to solve.
The most frequent problems are going to be solved by PS, when we shift to the right there
are less frequent problems, which the PH are going to solve.

- Proposition 4. Pyramidal organization

An organization with multiple layers has a pyramidal structure, with each layer a smaller size
than the previous one

These organizations (organized like this) are efficient in the use and allocation of knowledge.

Let's examine An Alternative Interpretation: Organization When Problems Differ in

Their Difficulty (= Knowledge requirements)

- Then the organization can be characterized, analogously to the frequency-based

organization:
- Proposition 1: Workers specialize either in production or in problem solving. Only one
class specializes in production (same)
- Proposition 2: Knowledge acquired by different classes does not overlap (same)
- Proposition 3: Production workers learn to solve the easiest problems and
problem solvers learn to solve the harder ones. Moreover, the higher up in the
list of production workers a problem solver is, the harder the problems she is
able to solve
- Proposition 4: The organization has a pyramidal structure, with each layer a smaller
size than the previous one.

Comparative statics
→ less cost ⇒ production can ask more questions, helpers need to know more, and
production needs to know more (span of control not in the exam)
→ cost of acquiring knowledge up ⇒ less knowledge
→ complexity of environment (frequency of difficulty problems) down ⇒ range of expertise is
ambiguous, can increase or decrease.

The Distinct Effects of Information Technology and Communication Technology.

Bloom, Garicano, Sadun and Van Reenen( 2014)

Introduction Effects on information technology and communication are commonly aggregate

but … authors suggest it is a mistake!

I and CT are different concepts: information and communication technologies.

- IT: no more big expensive archives + if my workers can acquire easier knowledge,
they can have more autonomy, because they don’t require knowledge, they can just
look for it easily
- CT:

What do they do?

- Study the cognitive view of hierarchy by testing for the differential impact of
information and communication.
- Consider
 - Knowledge horizontally vs knowledge hierarchical
- Information acquisition costs vs communication costs
- Theoretical starting point: Garicano (2000).

Theory: The Race Between Communication and Information Technology

Simplified version of Garicano (2000)
Assumptions
1. Production Requires Time and Knowledge
2. Knowledge Acquisition Is Costly
3. Knowledge Can Be Communicated

Cheaper Communication Centralizes;

Cheaper Information Access Decentralizes

- INTRANET: system which allowed workers to communicate between each other →

Communication Technology ⇒ less autonomy to the workers
Negative theoretical indicator

- ERP/CAD/CAM: Information Technology ⇒ more autonomy to the workers

Positive theoretical indicator

How do they do it?

Test the empirical relevance of the model
Using an international micro data set from the United States and European countries

Main results: Plant Manager Autonomy

Line managers
CAD/CAM:
INTRANET: decreases autonomy: negative and significant coefficient.

Main results: Worker Autonomy

What about the other types of workers (bottom worker)?
CAD/CAM: increases the worker autonomy
INTRANET: negative coefficient but not significant ⇒ partial empirical evidence of less
autonomy

Main results: Plant Manager Span of Control

Concluding remarks
- Aggregate information and communication technologies (ICT) is incorrect
- Information technologies (IT) are associated with increased autonomy and span of
control - facilitating more effective employee decision making.
- Communications technologies (CT) are associated with decreased autonomy -
decisions will be passed up to the centre of the firm
- Automation theories → information technologies associated with centralization
- Coordination theories → communication technolgies associated with decentralization

READINGS
Readings review
DISCUSSION ON CONVERSATIONAL FIRM

- Author: Catherine Turco

- Hack Night: spending a night to solve the problem
- Experiment Dinner:
- Mystery Dinner: company pays people to go the restaurant with people of the
organization. cost: working at 8pm, sharing lessons, thoughts, …
- “Sunlight is the best disinfectant” → if we’re transparent, leaders are more
accountable, no mistakes are done and better decisions are taken
- Transparency is good? Can be distracting and overwhelming.
- Opinions of employees: “Transparency for the sake of transparency”but also

Chapter 2. Open Communication

practical and symbolic value of transparency – hack nights and dinners are not exploitative,
they are millennials (speaking up, using social media..), they like “being treated as adults”

- What about our generation, Gen Z?

- Employees have voice rights, consequences:
- some can write more, if someone is very vocal (“opinionated people” – want
to impose) they can overcrowd others.
- wiki trolls: giving a hard time to everyone, doesn’t help to the conversation
- social media enhance power asymmetries: we all have access to the wiki,
- employees use the platform to promote themselves
- inequality between men and women
- hostile bubbles, just like “housing bubble”. Positive bubble: want to be
always positivew
- “ask the executive” ⇒ report information in an anonymous
communication

Chapter 3.
- “Experiment Meetings”
- might distract
- executives are giving too little feedback ⇒ not iterative project
- Interesting: employees want less control, but when executives don’t give too
much feedback, so to not discourage the creativity of the team
- This happens because “UGJ” policy: they can do anything they want as long
as they use good judgment, solutions:
- Provided training to UGJ: lack of detailed instructions but questions
are welcome
- people asking questions
- So that these decision rights work you need relational contracts
- The Engineers didn’t have any problem in UGJ: manager was very clear,
more centralized, and their job is from input generate output,
They all had the same background and experience and education ⇒
homogeneity, common judgment

Chapter 4. Openness control

- Brian Cassidy: wrote a controversial comment, they didn’t fire him, but they made
him make talks publicly
- TechCo allows employees to be online and speak their mind but they need to face
the consequences

Chapter 5: Open Culture

Culture deck version 2.0 → marketing tool to attract better employees usually in
companies they use it in training.
- Employees started to criticize – it wasn’t written by employees but executives, why?
- Even if all the employees knew that it wasn’t [true], it still would be good to have that
brand externally. It reminds me of Range Rover ads where they show people
off-roading. How many Range Rover drivers do you know who actually off-road?
They’re all soccer moms who are driving through manicured suburban streets!
- this is what we want to do

Chapter 6: HR
- initially “no HR” policy
- this became problematic: there’s the need for an HR department to deal with things
like maternity leave

Chapter 7: Conversational Spaces

- They criticize it (noise, smells), but they wouldn’t have it differently
- Teamwork is facilitated
- Rotation: They change places 4 times a year. Pros: different views, see different
colleagues,
- Managers can sit in open space but can sit in offices (because they have one-to-one
meetings)
- Open surveillance: everyone can see if you’re doing something you shouldn’t be
doing, increased accountability
- Less stressful environment, background noises

DISCUSSION ON WHY ORGANIZATIONS FAIL

6 issues:

INCENTIVE PROBLEMS (agents don’t WANT to do something):

1. Short-termism:
MULTI-TASKING: short term tasks VS long term tasks
The problem of the organization is that some employees care about the short-term too
much, rather than the long-term.
Examples:
● financial companies focused a lot on short term goals
● BP refinery explosion in 2005

Solutions:
- clawbacks for short-term bonuses
- contract with limited share of variable pay
- CRO: Chief Risk Officer

2. Decentralization
Benefits:

Examples:
- FBI, during the 9/11 attack they weren’t incentivized to share knowledge . In each US
State there’s own FBI office, there’s no communication among them to detect
patterns.

Solutions:
- Hybrid solution and arrangements like PepsiCo
 - Aid to the issue to the not talking to each other: put a measure performance on
knowledge sharing (like McKinsey did).

3. Communication Failures
- soft information: providing information that is useful ⇒ true
→ cheap talk: transmit information without providing hard evidence⇒ might be true or not
ex. “we will have a guest lecture with Garicano” you provide soft information… this won’t
have consequences.

- hard information

- intrinsic communication: at Pixar, people are left very free because they are
very motivated. They choose people that care about the company and give
truthful feedback.
- extrinsic communication

4. Inability to Adapt to Change Due to Organizational Rigidities

example:
- Kodak
- political economy: people in power don’t want to change, who are the only ones who
can change.
- NYSE:
- “kicking the loser upstairs” ⇒ keeping people in the company even if they weren’t
relevant anymore – estating a culture is rigid

BOUNDED RATIONALITY PROBLEMS (they CAN’T do something):

5. Hierarchy and Allocation of Talent

example:
- family firms

6. Coarse communication
example:
- Black Hawk incident: US air force and US army, so the US army downed an
helicopter from the US airforce. They had different codes ⇒ coarse coding: “they are
not american”.
Solution:
- specific investments in codes like GE and Universal made efforts to translate their
codes: executives held workshops on the business lexicons used within each
organization to foster mutual understanding and cooperation.

KNOWLEDGE HIERARCHIES

HIERARCHIES AND THE ORGANIZATION OF KNOWLEDGE IN PRODUCTION

How can we solve problems within firms? With knowledge.
Each individual can acquire knowledge about a narrow range of problems (only a
certain level of knowledge can be achieved)
How can firms organize human assets to extract the maximum knowledge?
Through knowledge-based hierarchy: the organization decides who learns what
and who each worker should ask, to deal with an unknown problem → People with
little knowledge are at the bottom of the pyramid while people with a lot of knowledge
are at the top of the hierarchy.
There is communication: workers acquire only the most relevant knowledge and,
when confronted with a problem they cannot solve, they ask someone else – if they
know how, they solve the problem, otherwise they ask.

Model of Communication and Knowledge Acquisition in Production

A worker operating a machine confronts a range of problems that must be solved to
produce. There are workers and there are problem solvers. (another example: call
center)
Assumption: there is a set of possible problems and a set of knowledge to solve
some of these problems. The workers are working with machines, sometimes they
can have problems: in case they have no knowledge on how to solve it, they ask.
But, workers face a cost of learning to acquire more knowledge.
The problem a worker confronts is to choose the length of the interval of knowledge
acquired to maximize expected output (expected because clearly there is uncertainty
regarding the problems that will need to be solved), i.e. they need to balance costs
and expected benefits (so, higher output due to reduced problem-solving delays).
Each worker chooses how much knowledge to acquire. ( z = level of knowledge)

There is Communication and Organization.

Cost (h): wasted time (the cost of asking questions), even when no answer is
provided.

Assumption on cost: aggregate all the communication losses in the worker who is
being asked (receiver), i.e. when a worker asks a question, the productivity loss from
communication (time wasted) is attributed to the receiver. Consider communication cost
even when the worker asked does not know the answer (inefficiencies)
Assumption on organization: the size of the organization is large enough (we are
not thinking of start-ups etc.) and everyone in the list of a particular worker may be
eventually asked if necessary.

Then the organization can be characterized like this:

Proposition 1: Specialization. For any given allocation of knowledge, workers in
each class specialize either in:
- production (problem-helpers)
- the transmission of knowledge about solutions (problem-solvers): don’t
operate the machinery but just help on problems).
Moreover, only one class specializes in production, and all other classes are
composed of problem solvers who support workers in that class.

Proposition 2. Non Overlapping knowledge. No solution is known by both different

classes. Problem-helpers know how to solve problems that the problem-solvers are
incapable of solving, and vice-versa

Proposition 3. Organization by frequency

Production workers (problem-helpers) learn to solve the most common, frequent
problems; problem solvers learn the exceptions.
Moreover, the higher up in the list of production workers a problem solver is, the less
frequent the problems she is able to solve.

Proposition 4. Pyramidal organization

An organization with multiple layers has a pyramidal structure, with each layer a
smaller size than the previous one, and this is to maximize the efficiency in the use
of knowledge. (size of the layer of the people = number of people solving the
problem, increasingly less frequent)

A Model of Communication and Knowledge Acquisition in Production:

Organization When Problems Differ in Their Difficulty
(difficulty: there are ≠ knowledge requirements)
Problems differ by frequency and difficulty. Then the organization can be
characterized, analogously to the frequency-based organization:

Proposition 1: Workers specialize either in production or in problem solving.

(SAME)
Proposition 2: Knowledge acquired by different classes does not overlap. (SAME)
Proposition 3: Production workers learn to solve the easiest problems and problem
solvers learn to solve the harder ones.
Moreover, the higher up in the list of production workers a problem solver is, the
harder the problems she is able to solve !!!! DIFFERENT
Proposition 4: The organization has a pyramidal structure, with each layer a smaller
size than the previous one. (SAME)

COMPARATIVE STATICS
What happens if I change one thing, if everything else is constant?
1) When the cost of asking questions (h) ↓, I am asking more questions, so more
preparation is needed from problem solvers. This also implies that the level of
expertise (z of PH) is lower (increased reliance on PS)
2) If the cost of acquiring skills or knowledge decreases (c ↓), PS and PH are more
prone to acquire knowledge, and vice-versa.
3) Lamda is the complexity of the environment, hence the frequency of difficult
problems. Changing complexity, the effect on expertise of both PS and PH is
ambiguous → If the complexity of the environment (λ) decreases, production workers
may invest less in expertise (fewer challenging problems) but also PS might also
reduce their specialization (lighter workload), ⇒ it depends on the balance between
reliance on PS and the need for independent problem-solving by production workers.
(span of control not in the exam)

THE DISTINCT EFFECTS OF INFORMATION TECHNOLOGY AND

COMMUNICATION TECHNOLOGY
15 years ago, everybody was talking about ICT: information and communication
technology. Effects on information technology and communication are often
aggregated but experts suggest it is a mistake: IT and CT have different impacts on
the organization.

There are two key ideas:

IT → 1. If my workers can acquire more knowledge they are more autonomous:
decentralization of the workforce
CT → 2. If communication is easier and they can ask questions quicker, leading to
less autonomy: centralization of the workforce

They studied the cognitive view of hierarchy by testing for the differential impact of
information and communication.

Consider:
- Knowledge horizontally vs knowledge hierarchical
- Information acquisition costs vs communication costs
Theoretical starting point: Garicano (2000).

Theory: The Race Between Communication and Information Technology

Assumptions:
I. Production Requires Time and Knowledge
II. Knowledge Acquisition Is Costly
III. Knowledge Can Be Communicated

COMMENT ON THE TABLE. Three types of technology:

INTRANET: systems allowing workers to communicate easily within the workplace
(e.g., Teams). It is a communication technology decreasing workers’ autonomy
(centralization) ⇒ Cheaper communication centralizes.
ERP/CADCAM: workers have access to much information in storage within the
organization. It is an information technology increasing workers’ autonomy
(decentralization) ⇒ Cheaper information access decentralizes.
Considering the plant manager’s autonomy:
- an information technology such as ERP increases autonomy: positive significant
coefficient.
- a communication technology such as INTRANET decreases autonomy: negative
significant coefficient.
Considering the worker’s autonomy:
- an information technology such as CAD CAM increases autonomy: positive
significant coefficient.
- a communication technology such as INTRANET decreases autonomy: negative
coefficient which is not significant. This implies partial empirical evidence.

Conclusions
● Aggregate information and communication technologies (ICT) is incorrect.
● Information technologies (IT) are associated with increased autonomy and
span of control - facilitating more effective employee decision making.
● Communications technologies (CT) are associated with decreased autonomy
- decisions will be passed up to the center of the firm
● Automation theories → information technologies associated with
decentralization
● Coordination theories → communication technologies associated with
centralization

We do not care about the span of control, but it is the number of people under a
manager.
 CULTURE and LEADERSHIP

Cultural Conflict and Merger Failure: An experimental approach

The majority of corporate mergers fail. Widespread merger failure is at odds with the
public and media perceptions that mergers are grand things: merges are almost sure
to create enormous business synergies that are good for employees, stockholders,
and consumers.
Differences in cultures and in points of view often lead to mergers’ failures.

Example: Daimler-Chrysler merger

Daimler:
- German
- Formal and structure management style: they wanted to dominate the
decision making
Chrysler:
- USA
- Relaxed, informal and freewheeling management style.

Due to the German unit’s increasing dominance, the differences in cultures and
different points of view led to performance and employee satisfaction at Chrysler to
decrease.

“While cultural conflict often plays a large role in producing merger failure, it is often
neglected when the benefits of a potential merger are examined”
When people think about merging, they overlook the differences in cultures that
might make the merger difficult.

Culture is a general shared social understanding (involves multiple people), resulting

in commonly held assumptions and views of the world among organizational
members. This helps tacit coordination: no need to make everything explicit.
Very often, this organizational culture is the result of a joint experience (catholic
people tend to have the same unified view on the world).

EXPERIMENT
Phase 1 – Participants internalized their firm’s "culture" through storytelling.
Every subject is presented with the same set of 16 pictures depicting office
environments.
The manager describes to others the pictures to the employee, saying what happens
within the office. A story for each photo. (pictures about things that happen in the
office). Subjects alternated roles (manager or employee) in each round.

Phase 2 – Employees from different cultures were merged and had to

collaborate in tasks that required coordination
They do it for two different firms: one subject from the “acquiring” firm was selected
to be the manager for the rest of the experiment, and one subject from the other firm
was selected to be the “acquired” employee.
The two employees (one acquiring and one acquired) play a coordination game and
could finish at different times and receive a different amount in each round for
completing the task. There are coordination games the two employees have to play:
if they coordinate better they produce more and higher quality results. The manager
received the average of the earnings of both employees.

RESULT
1. Tension and hostility between the manager of the merged firm and the new
employee.
2. Subjects still place blame on members of the other premerger firm and
attribute the post-merger difficulty to their lack of competence, not realizing it is
cultural differences. They have no clue it is an experiment of culture and put the
blame exclusively on the other part not being competent enough.
3. There are some successful cases of integration of cultures.

CONCLUSION
- Differences in culture show a decreased performance for both employees
after the merger (not only the acquiring but also the acquired), and subjects
underpredicted the extent of this decrease.
- Creation of a new hybrid language (or explaining the old language) when
the successful merger occurs.

PAPER: Does Culture Affect Economic Outcomes?

Sapienza Guiso Zingales (2006)

Consider culture is a vague notion. Everybody speaks about culture, nevertheless it

has an undefined role and impact for culture in economics.
In the paper they try to define culture and see how it has an impact.
Recognize that better techniques and more data have made it possible to relate
people’s preferences and beliefs to measures of cultural legacy. Summarize a
cultural approach that can be tested.

How do they do?

Step 1. Define culture customary beliefs and values that ethnic, religious, and social
groups transmit fairly unchanged from generation to generation (if your whole family
is catholic, you will be very likely to be catholic as well. You are more likely to feel
sorry for a homeless person because you imagine they had a hard life compared to
the others who blame him for laziness).
Step 2. Show that those beliefs and preferences have an impact on economic
outcomes.
Step 3. Claim a causal link: if I change the preference I change the economic effect.
It is not just correlation.

Having in mind that: “Individuals have less control over their culture than over other
social capital (= capital related to their socio-economic status).”
GRAPH:
X-axis is the index of trust, y-axis we have different religions and the category in
general.
In a statistical estimation, it is statistically relevant and different from zero.
If we group everyone who is religious, they are much likely to trust others with
respect to non-religious people.
The second bar is about being raised religiously, they have more trust in people, but
a much bigger difference compared to the ones who are religious.
Jewish people have a lot of variation in trust (this is why it is not really significant).

GRAPH
Japanese and Scandinavians are more trusting, although for the former the results
are not significant.
People who trust the least are afro-americans.

GRAPH: The impact of religiosity on the probability of becoming an entrepreneur.

We have a control: whether or not the respondent is male
● If male, more likely to become an entrepreneur.
● If white it is more likely to become an entrepreneur.
● The older the person is the more likely to become an entrepreneur, but the
age squared is negative,which implies a complex relationship: if I am either
too young or too old I won’t become an entrepreneur.
● The more education I have, the less likely I am to start a business.
● Even controlled by characteristics like gender, race, age, education, income,
the higher the trust people have in others have a higher statistical probability
of becoming an entrepreneur. ⇒ Trusting others the probability of being
self-employed by 1.5 pp.

What is the channel through which culture affects preferences?

- The world value survey (40 year survey) asked “Here is a list of qualities that
children can be encouraged to learn at home. Which, if any, do you consider
to be especially important?” to a random sample of people across countries.
- Regress this variable on several dummy variables for various religious
affiliations.
- Results: Catholics are 3.8% more likely and Protestants 2.7% more likely than
nonreligious people to view teaching thrift to their children as an important
value.
In countries in which parents teach more thrift, the national savings are larger.
⇒ Culture impacts economic outcomes.

GRAPH:
Culture affects political preferences.
Catholics, Protestants and Jewish have a more negative attitude toward
redistribution than those with no religion.
The URSS were a socialist society where everybody got the same regardless of how
hard they worked. In this context, they were not welcoming some religions due to
their adversity towards redistribution

PAPER: Cultural Biases In Economic Exchange?

They document how much cultural biases affect economic exchange.
Trust is affected by cultural aspects of the match between trusting country and
trusted country.
Consider the history of conflicts, religion, genetic and somatic similarities.

GRAPH:
Level of trust among European managers –where 1 is the best and 5 the worst.
The most trusted country is Germany; the least trusted is Italy: even Italians do not
trust Italians, they prefer Germans. This impacts the level of economic exchange
across countries.

We want to understand how certain historical variables affect the trust between two
countries and, in turn, how this affects the economic exchange (particularly, exports)
These are the determinants of trust:
● RELIGION: Commonality of religion to measure a country’s cultural tradition
● SOMATIC DISTANCE: differences in looks (red haired etc): Indicator of
somatic distance, based on the average frequency of specific traits (hair color,
height, etc.) present in the indigenous population (Biasutti 1954) +
INFORMATION OF INDIGENOUS POPULATION: understand how people in
a certain area were a long time ago (for example, before massive immigration
occurred)
● Country’s history of wars
● Cross-country trade: control by estimates of transportation
GRAPH:
- The more years spent at war, the lower the level of trust
- Religious similarities increase the level of trust
- The larger the somatic distance (the more different we are in how we look),
the lower the level of trust

HOW DOES TRUST AFFECT EXPORTS?

GRAPH: The first coefficient is positive and significant, and it shows that e.g.,
If I trust French more than Polish, I will import more from France. The higher the
level of trust in a country, the higher the imports from that country (here higher by
36%).
The perceptions rooted in culture are important determinants of economic exchange.
Lower bilateral trust leads to less trade between two countries.
Trust among European countries differs in systematic ways, which are correlated to
their different cultural heritages (wars, differences in religions etc.).
These differences in trust seem to have economically important effects on trade.

Work Environment and Individual.

Background: Explaining Regional Shirking (i.e., not working as hard as
expected) Differentials in a Large Italian Firm
There appear to be significant regional differentials (north-south) in the prevalence of
shirking among the employees of a large Italian bank.
In particular, absenteeism and misconduct episodes are considerably more
frequent in the southern branches of the bank.
Examine several potential explanations for this fact:
1. Individual backgrounds: they are more laid-back or "chill"
2. Sorting effects: the parties are better in the south, so I sort to the south
3. Different local attributes: offices in the south do not have ventilation

1. To study individual background:

Use the full sample of workers to evaluate absenteeism and misconduct. Examine
the statistical significance of the incidence of shirking (average number of absence
episodes and the frequency of misconducts) by region of birth and region of work
(whether or not they were born/they work in the South or not)
2. To study sorting effects:
Report statistics for the incidence of absenteeism based on the subsamples of move
between or within regions, or stay in the period 1975–1995.

GRAPH:
People who work in the North, independently of where they were born, are exactly
the same.
People who work in the South tend to be, overall, more absent.
GRAPH:
GRAPH:
PANEL A: If you were born in the South, you are more likely to be absent with
respect to people born in the North.
PANEL B: If you work in the South, you are more likely to be absent with respect to
people who work in the North.
PANEL C: combines the previous effects.
The following three combinations Born = south; Work = north; Born = north; Work =
south; Born = south; Work = south are more likely to be absent compared to Born =
north; Work = north.
These effects are progressively increasing: Born = south; Work = north have fewer
episodes of misconducts and absentees than people Born = north; Work = south and
people Born = south; Work = south. (exam = gives the table and a statement: TRUE
OR FALSE)

Striking regional shirking differentials within a large Italian bank with branches
distributed over the entire country. The branches are exactly the same: same
ventilation, same facilities etc.
In particular, absenteeism and misconduct episodes are substantially more frequent
in the south, but it is not a matter of being born in the south. Also Northen people
who move in the South are affected by these results.

PAPER: VALUE OF DISSONANCE AT WORK (Espinosa)

Looking at the World Management Survey, we infer that there are large disparities in
productivity: some firms have very little productivity and others have very large
productivity.
The most common explanation for these differences in productivity is organizational
culture (organizations differ in values shared by all their members).
Is it possible that culture also explains performance inside the organization?
How can values explain performance heterogeneity within firms?
One view: only alignment between employees and organizational values matter (and
increases productivity). When workers are aligned with the mission and goals of their
managers, their productivity increases.
But the relationship between middle managers and workers is fundamental to
understanding organizational performance!
Managers and workers bring their personal values to the workplace: there are
subcultures. How do these affect employees’ performance?

Research question: does diversity in personal values affect employees’

performance? If so, why?

WMS ⇒ They partnered with one of the biggest banks in the world with 200,000
employees and in more than 50 countries. They made a survey, delivered both to
managers and workers to measure their personal values. Then they constructed
measures of the differences in personal values that you have with your manager and
workers.
FIRST POINT: having different personal values with your managers has a negative
effect on productivity (even controlling for gender). Misalignment with colleagues
doesn’t matter.
SECOND POINT: there are both hard (objective) and soft (subjective) productivity
measures, but misalignments only regard the hard productivity measures.
The main reason why this happens is workers’ voice: if the worker and the manager
have different personal values, the worker will be less likely to have meetings with
the manager and ask questions.

What are firms doing about this? Implementing training programs to improve the
quality of interpersonal interactions, as this can attenuate misalignment costs and
can help in managing people's differences (e.g., psychological safety, cultural
awareness).
Observing effects from:
- Survey measures: Firms often use surveys to measure employees'
perceptions of their ability to speak up, to challenge bad policies…
- Quasi-exogenous variation in communication quality (workers' voice):
refers to variations in communication quality that are not fully controlled or
randomly assigned but occur due to external or semi-random factors. e.g. The
presence of a new manager with different communication skills.

Result: in environments with more workers' voice, VAM (Value Added

Misalignment), i.e. mismatch between manager’s personal values and the ones
of the workers, does not affect productivity

They have taken 11 values and asked to rank which ones are better.
Most people (80%) say responsibility is fundamental to teaching children, 10%
people say faith. Also, distributions are similar across gender.

REMARK 1: They classified personal values in some that don’t directly affect
productivity and some that affect productivity (e.g., people that say hard work is
important are very productive, while people who say faith is important are less
productive on average).
⇒ They classified the personal values in 2:
- factors that matter for the production function,
- factors that don’t.
Results are not driven by performance relevant values → all the effects are driven by
those differences in value that do not directly enter the production function, like hard
work.

REMARK 2: Results are not driven by other sources.

REMARK 3: Differences in personal values with my manager has a twice as large
effect than differences in gender.

Why is this the case? 4 reasons:

1. Managerial biases: you have different personal values from the manager;
therefore she will evaluate you worse ⇒ this is not the case
2. Workers’ motivation: ⇒ don’t have strong evidence
3. Job barriers, managers do not mentor, unfair task allocation ⇒ this is not
the case
4. The only relevant factor is communications flows: workers’ voice.
When you have more misalignment you are less likely to discuss problems of
objectives with your manager, less likely to participate in meetings and speak
up.
Conclusion:
There is random allocation of people sitting in the office, sometimes people sit with
their boss and sometimes not: this is exogenous. In environments with more
workers’ voices (hence, people speak up more) there is no effect of misalignment.
81% workers said that the higher the communication quality, the lower the cost of
having different personal values.

MAIN IDEAS:
Value misalignment with the boss lowers productivity of workers. First evidence of
this effect: companies are unaware of such implications of culture.
The paper wants to underscore the need for a deeper understanding of what
diversity of the workplace means beyond visible traits.

—-------------------------------------------------------------------------------------------------------------
Guest Lecture – Intesa San Paolo
Pierluigi Dialuce (Head of HR)
- from finance and risk management to HR?
- there’s the perception of HR being very bureaucratic, is it true? but from you
presentation it seems that its more about people and transparency than processes, is
an HR today a bureaucrat or has more to do with relationships and the value their
bring?
- What are some good questions to ask in an interview?
be honest and consistent with what i talked about before regarding your values (social
issues) – if the company is uncomfortable to answer, then probably they are not so true to
my values.

Soft skills framework:

- Continuous learning appetite and smart thinking
- social influence and connecting leadership
- courage and entrepreneurship
- This framework inspires the virtuous behavior of people
- crucial to maintain also ethical values
- inspire people to act and think proactively

Gender pay gap: women in the same position as men are paid less. Girls are pushed more!
—------------------------------------------------------------------------------------------------------------------------
 BEHAVIORAL ECONOMICS
Psychologists have conducted extensive research on how people think. While our
models often assume that individuals are perfectly rational, in reality, our perceptions
are influenced by biases. Here, we will explore how behavioral insights can be
applied to decision-making within organizations.

A few examples:
EXAMPLE OF THE TWO TABLES: Which is longer? There actually the same, it is a
matter of perception
EXAMPLE OF THE TWO COLORS: Two colors may look different, but when
isolated, they are revealed to be the same. This illustrates how context and
surrounding elements can distort our color perception.
EXAMPLE OF THE TWO BLACK CIRCLES: One black circle appears larger than
another, yet both are the same size. This illusion is caused by the surrounding
circles, which influence our perception of size.
KITAOKA Images created by Kitaoka give the illusion of movement, despite being
completely static. These designs exploit our visual system in a way that mimics
psychedelic effects.
McGurk effect: What we hear can be altered by conflicting visual information. . For
example, the sound "BA" may be perceived as "FA" if the speaker's mouth
movements suggest it. This effect persists even when we are aware of it. The brain
tries to make sense of it by making us hear “FA”: our senses are misleading us.
The Default Bias: When examining organ donation rates in European countries, we
find a striking bimodal distribution: some countries have very high donation rates,
while others are notably low: Denmark, Netheralnds, Germany NO – France, Austria
YES
Despite their similarities, the key factor influencing this difference lies in how the
default choice is framed on their driver's license applications:
- Low Donor Countries (Opt-In):
These countries use an opt-in system, “Check the box if you want to participate in the
organ donor program”
- High Donor Countries (Opt-Out):
These countries use an opt-out system, “Check the box if you DO NOT want to
participate in the organ donor program”.
Depending on the outside option you get extremely different results.

Doctors
Even experts like doctors are not immune to decision-making biases. Consider an
experiment conducted in 1995 involving decisions about hip replacement surgery:
- Scenario 1 (Single Omitted Medicine):
A patient is scheduled for a hip replacement, but the doctor realizes they forgot to try a
simpler option — prescribing one medicine. When presented with the choice to cancel
the surgery and administer the medicine, most doctors agree to cancel and try the
medication first.
- Scenario 2 (Two Omitted Medicines):
In this scenario, the doctor realizes they forgot to try two medicines before proceeding
with surgery. Surprisingly, very few doctors choose to cancel the operation in this case.
Forgetting to prescribe one medicine feels like a manageable oversight, but
forgetting two medicines appears to demonstrate greater incompetence. To avoid
acknowledging or signaling such a mistake, doctors are less likely to alter their
course of action, even if trying the two medications might be in the patient's best
interest.

“The Economists” Ad
The Economist advertising experiment highlights how irrelevant options can
influence our choices:
Initially they are offered these options:
- 59: online subscription 16%
- 125: print-subscription 0%
- 125: print-subscription + online 48%

Then the options change to this:

- 59: online subscription 68%
- 125: print-subscription + online 32%
An option that seems "irrelevant" can heavily influence decision-making by framing
other options more attractively in comparison. (ChatGPT dice che questo si chiama
decoy effect: where introducing a less desirable choice alters our perception of the
remaining options, leading to a different outcome).

TOM AND JERRY

Tom-Ugly Jerry-Jerry Preferred Tom
Tom-Ugly Tom-Jerry Preferred Jerry
By including an unattractive version of one option (e.g., Ugly Tom), the original option
(Tom) becomes less appealing by comparison, leading people to prefer the
alternative (Jerry).

Three waves of Behavioral Economics

1. Identify ‘‘anomalies’’ of economic theory and some alternatives conceptualizations:
how do people take decisions and whether it is similar to the decision making
predicted.by models
2. Once behavioral patterns are understood, you formalize these alternatives in
models and identify empirical validation
3. Fully integrate into economic analysis by embedding old and new assumptions as
special cases of general models and formulate new theoretical results, empirical
tests, and applications (which is our focus). Develop comprehensive models that
incorporate both traditional economic assumptions and newer behavioral insights.

Two premises of third-wave behavioral economics.

Premise 1: Adding untraditional assumptions does not at all mean abandoning
traditional methods. We can keep using same methods.
Premise 2: Adding untraditional assumptions does not mean abandoning traditional
assumptions. We can continue the other assumptions instead we have strong
evidence it doesn’t hold within the model.

In fact, with greater psychological realism we can improve economic analysis by:
- Explaining behavior studied by economists that traditional analysis has had
difficulties explaining
- Explaining behavior that seems economically important enough that one would
have thought economists would have been studying—but haven’t been. Miguel’s
Alumnus increased sales of his trucks by placing them next to an ugly truck.
- Beyond explaining behavior, better understand normative/hedonic effects of
observed behavior. [positive analysis = how it is; normative = how it should be]
Once we know these insights we want to make the model more complicated by
making and less tractable but more realistic in a trade off on the scale of what
economists do all the time (when invoking more familiar assumptions)
MORE REALISTIC PREFERENCES:
1. REFERENCE DEPENDENCE:
SIMPSON EXAMPLE. Bothered in the face in the groin back to the face and now it
feels nicer
His preferences are relative to the option he has.

LOSS AVERSION (Poll 1): Would you accept 50/50 lose $600, gain $700 bet?
MAJORITY OF PEOPLE TURNED DOWN THE PROPOSAL. The traditional model
would say accept as expected value > 0. NOT BECAUSE OF DIMINISHING
MARGINAL UTILITY, BUT BY AVERSION TO MODEST SCALE RISK. Loss
Aversion: people quantify more the losses and the gains. You are sadder in
magnitude to lose 100 than to win 100.

WHICH ‘‘FEELS’’ LIKE A BIGGER DIFFERENCE? (Poll 2):

Gratification 87 days from now vs 88 days VS Gratification 0 days vs. 1 day.
The ratio of people preferring today should be the same as the ratio of people
preferring 87 instead of 88 as the time difference is always one day, but from the
survey the ratio is different. People look at the relative difference compared to when
is it in time.
SOCIAL PREFERENCES: We like people who like us and vice versa. Reciprocity
plays a key role.

PRESENT-BIASED PREFERENCES
It is the most active and fruitful area of 3rd-wave behavioral economics.
We tend to over pursue immediate gratification vs long-term preferences.
Present bias:
- Today matters more than the future
- Today we care roughly equally about well-being on any two future dates: I see
20thDecember 2025 and 21st December 2025 as the same, although one is more future
than the other, but when the future arrives and the first of these dates becomes ‘‘today’’,
then we care more about that first date than about the second date
We have used economics and management to discount the future with the
exponential model, which clashes with real preferences of people.
PRESENT BIAS PREFERENCES: we discount too much the future. We don’t care
enough about the future.

QUASI-UTILITY MAXIMIZATION:
Behavior reflects maximization of coherent utility function at each moment in time.
Life is an infinite series of (potential) choice sets, When facing X and Y, should:

We are going to maximize our utility that depends on two variables

But to what extend are we considering the whole set of options? The models state
we are always considering all possible options, but it is not actually true.
The concept of focusing effects suggests that we only consciously make choices in
an infinitesimal percentage of the infinite number of choices sets we actually have
available. When choosing what to eat, you don’t consider each and every kind of
cuisine existing. While utility maximization theory says we choose the best option
across all possibilities, the focusing effect highlights that we tend to concentrate only
on a small subset of those choices.
Another related concept is narrow bracketing: e don’t fully integrate our decisions
with other decisions even when clear our utility would be higher if we did. You take a
decision without considering the effect on other alternatives. We think of the options
are independent objects.

There are four mispredictions implications

Addiction and Habit Formation: When making decisions, such as going to the gym
or drinking, we often underestimate the likelihood of becoming addicted to these
behaviors over time.
Irreversible Decisions (HOT): In moments of intense emotions, like anger, we may
make decisions (such as sending an angry email) without fully predicting their
long-term consequences. Our judgment tends to be biased when we're in a "hot"
emotional state.
3. Benefits of Cooling Off: Some companies implement internal policies that
require a waiting period, such as one week, before finalizing a deal. This allows
individuals to reassess the situation objectively and avoid making decisions based
on initial impressions or external pressures.
4. Excess wealth- seeking/consumption: We often underestimate how little we will
actually enjoy wealth. For instance, going from $75,000 to $100,000 may feel
exciting initially, but over time, the higher amount becomes the new norm, and the
extra wealth provides diminishing returns in terms of happiness.

PROSPECT THEORY: more detail

Expected Utility Theory asserts that preferences must be:
- Complete: We always have a clear preference between any two options.
- Transitive: If A is preferred to B and B to C, then A must be preferred to C.
- Continuous: If A is preferred to B, small changes to A won't make B more
preferred.
- Independent: Adding a new option doesn't affect our preferences between the
existing choices.
This implies:
- Linear expectation of a utility function
- Asset integration–gambles are evaluated by integrating gains or losses with current
wealth

ALLAIS PARDOX (Poll 3)

The Allais Paradox imply that people prefer B to A and C to D, but this is A
VIOLATION OF THE EXPECTED UTILITY THEORY
If B> A:
0.89 u * (1M) + 0.11u(1M) > 0.89 (1M) + 0.01u(0M) + 0.1u(5M)
0.11u(1M) > 0.01u(0M) + 0.1u(5M)
If C> D
0.1 (5M) + 0.89(0M) + 0.01(0M) > 0.89 u* (0M) + 0.11u(1M)
0.1 (5M) + 0.01u(0M) > 0.11u(1M)
The two inequalities are exactly the opposite to what we expect from utility
maximization.
This is visible from the Value Function (Nobel Prize).
What matters are changes in wealth or welfare, rather than the utility function for final
wealth levels as in EU.
Gains and losses are defined relative to a reference point. This reference point is
normalized to the individual’s current level of wealth. For example, an additional 10
euros has far greater value to a homeless person than to someone like Elon Musk,
even though traditional economic theory would treat them as equal.
Perceptions are characterized by diminishing marginal sensitivity.
Loss aversion plays a significant role: we are much more sensitive to losses than
to equivalent gains.

This also explains Poll 1!!

In the Allais Paradox, when choosing between options A and B, you focus on the 1
million option in both, and you choose B because it offers a higher probability,
making it look more appealing.
When comparing options C and D, you prefer C because, even though the
probability differences are small, the actual monetary values in C are much larger
than in D.
Prospect Theory can explain the Allais Paradox, unlike the traditional utility
maximization model.
Loss aversion plays a key role: even small losses (like a 1% chance) are weighted
much more heavily than equivalent gains. This helps explain why people’s choices in
the Allais Paradox deviate from what traditional utility theory would predict.

LIMITATIONS: One weakness of prospect theory is the somewhat arbitrariness of

reference point adjustment (you can always pre-define it) it adjusts to tell whatever
story it needs to.

Application: Endowment Effect and the Coase theorem

We tend to value things we own much more than things we do not own. This
phenomenon is known as the Endowment Effect.
Consider the following experiment: two groups are created, one with a mug and the
other without.
Before receiving the mug, the willingness to pay (WTP) for the mug is roughly equal
to the willingness to buy. However, after receiving the mug, the group with the mugs
is asked to sell them to the other group. Interestingly, no trade occurs because the
willingness to sell (WTS) is much higher than the initial willingness to pay. This
happens because people value the mug more once they possess it. The act of
ownership increases the perceived value of the object compared to before they had
it. Despite there being no change in information (everyone knows the value of a
mug), those who own the mug are now asking for more money than they would have
before owning it.
Loss aversion (prospect theory): The contribution of Behavioral Economics
The individual evaluated trades and transactions relative to some reference point.
For calculating WTA, the individual requires enough $ to offset moving from {mug} to
{no mug}. Loss aversion towards the object implies that this will be higher than WTP.
The loss aversion explains the endowment effect: I value more the loss than the gain
and that is why I increase the price.

Impact on the loss affect on the effects of the Coase theorem

((We have seen this economist before in Vertical Integration))
**COASE THEOREM: In the standard case, final allocation of resources should not
depend on the initial assignment of property rights and an efficient solution could be
reached.

Imagine a factory and a farmer. The farmer’s land is located next to a river, and the
Coase Theorem suggests the following scenarios:
- If the government clearly grants the factory the right to pollute, and the farmer
values a clean river, they can negotiate a solution where the farmer pays the factory for
the right to reduce pollution. This allows both parties to reach an efficient level of
pollution that makes them both happy.
- On the other hand, if the government grants the farmer the right to a clean river,
the factory would need to compensate the farmer for the pollution.

The Endowment Effect comes into play once rights are assigned. When the factory
holds the right to pollute, they may value that right more highly than before, making
the price for pollution higher than the farmer is willing to pay. Conversely, when the
farmer holds the right to a clean river, they value that right more (due to the
Endowment Effect) and every little particle of dirt will be taxed more.
Loss aversion, which helps explain the endowment effect, influences how we
interpret the Coase Theorem, particularly the idea that it leads to an efficient
solution.

Prospect theory is a suitable replacement (can explain more than the utility theory)
for Expected Utility Theory (EU) because it can explain the economics anomalies
mentioned as well as the regular phenomena EU explains.
In certain contexts, Prospect Theory acts as a complement to EU.

Intertemporal Choice and Time-Inconsistency

Poll 4: Suppose it’s January 1st. Which would you prefer?
A: 7 hours of boring work on April 1st.
B: 10 hours of boring work on April 15th.
Now suppose it’s 4 months later (April 1st). Which would you prefer?
C: 7 hours of boring work today
D: 10 hours of boring work in 2 weeks
The difference in days between A and B is equivalent to the difference in days
between C and D. According to the Management Standard Theory, we should make
the same decision for A and C, as well as for B and D. However, inconsistencies
arise in practice. Some people alter their preferences on a given day, showing how
the passage of time can influence decision-making.

Poll 5:
Which would you prefer?
A: $2000 right now
B: $2400 in a year from now
Which would you prefer?
C: $2000 in 10 years
D: $2400 in 11 years
The difference in days between A and B is equivalent to the difference in days
between C and D.
Time Inconsistency is when the optimal decision at one point in time is no longer the
optimal choice at another point in time
"Preference reversals" when both outcomes were in the future, you preferred 7 hours
on April 1st, or the more delayed $2400 and when the choices were more
immediate, you flipped your preferences and preferred 10 hours of work in 2 weeks,
or the immediate $2000 !!!
The change is due solely to the passage of time.

What is the explanation and the alternative proposed by Behavioral

Economics?
Preferences are dynamically inconsistent because we do not discount all time
periods uniformly(exponential discounting) Discount factor differs among periods.
== Hyperbolic discounting
We overweight time periods that are closer to the present relative to time periods
that are further in the future. For instance, the difference between now and next year
feels larger than the difference between 2030 and 2031. For example, we exhibit a
higher discount rate between now and 1 year from now than over 7 years from now
and 8 years from now.

The traditional model of time discounting is the exponential model, which assumes
that people discount future rewards at a constant rate. However, Behavioral
Economics suggests an alternative called hyperbolic discounting.
Hyperbolic discounting places a stronger emphasis on near-term delays than the
exponential model does. Hyperbolics discounts STRONGER nearer delays than the
exponential model predicts. Further delays are less aversive

Poll 6:
Which would you prefer?
A: You receive $200 right now
B: You receive $300 in 4 years
Which would you prefer?
A: You pay $200 right now
B: You pay $300 in 4 years
Gains are discounted more heavily than losses. People tend to be more patient
when dealing with losses (willing to wait longer for something they are losing)
compared to gains. Patience also varies depending on whether the amount is
positive or negative.

Poll 7:
Which would you prefer?
A: $10 right now
B: $20 in 3 years
Which would you prefer?
A: $1,000,000 right now
B: $2,000,000 in 3 years
You should remain consistent, so if in the first you choose A, you should choose A
also in the other one. While the majority first chooses A and then B this happens due
to the magnitude effect or the amount effect – people tend to be more patient as the
amounts at stake rise.

Experiment: Wertenbroch
1/2 of students were allowed to set their own binding deadlines for when to turn in 3
class papers. The other 1/2 was simply given exogenous, evenly spaced deadlines
Th The “rational choice” would be to set all deadlines for the last day, and then have
the flexibility of setting your own deadlines. The majority of deadlines were set before
the last day possible. The performance was higher for those students which were
given given exogenous deadlines. Those given more freedom actually performed
worse.
Variation Of The Experiment
2nd experiment: Students proofread papers (created with a text generator) that
contained errors. 3 conditions: Exogenous evenly spaced deadlines, exogenous all
at the end deadline, self-imposed deadlines
Again, people did indeed demand commitment, choosing to self- impose costly
deadlines. The number of errors detected (and thus total earnings) was highest in
the exogenous evenly spaced deadlines, followed by the self-imposed deadlines
condition, and worst in the end deadline condition