Week 9 - Institutions II

Question 1

True or false: Institutions are humanely devised?

Answer 1

True - we, as a society, choose the institutions our society ends up with.

Question 2

What determines who has the political de jure and de facto power, as outlined by Acemoglu and Johsnon (2008).

Answer 2

The type of political institutions you have, and how resources are distributed across various groups in a country

Question 3

The people who have this power, what do they determine?

Answer 3

The type of economic and political institutions we end up with, which effect economic performance and distribution or resources, and it continues as a cycle.

Question 4

When modelling institutional differences, what 3 institutional characteristics do we need to consider?

Answer 4

1.Economic institutions
2. Who has the Political power
3. Political institutions

Question 5

Why do economic institutions matter?

Answer 5

Matter for economic growth because they shape the incentives of key economic factors. Also important for the distribution of resources across various groups in society.

Question 6

Where do the economic institutions come from ?

Answer 6

They’re affected by history and chance, but ultimately comes down to the collective choice of societies.

Question 7

Since economic institutions affect the distribution or resources, how is there a conflict on interest?

Answer 7

Because different groups will want a different distribution of resources which only benefit them . The deciding factor then is political power

Question 8

What is de jure and de facto political power?

Answer 8

de jure - derived from political institutions (e.g. democracy vs dictatorship)
de facto - political power which isn’t given by political institutions e.g. gained by starting a revolution, or leverage over government or lobbying.

Question 9

Where does de facto power come from?

Answer 9

Originates from both the ability of groups within society to overcome their collective action problem and from the economic resources available to them. E.g. trying to overthrow a dictatorship, need money for weapons. High cost high reward.

Question 10

So why would institutions hardly change?

Answer 10

If you have de jure/de facto power, you have an incentive to keep the status quo to keep your political power. There is a lot of persistence emphasised in AR.

Question 11

How could you get change?

Answer 11

A big external shock in society which could shift the de facto political power to a new group, which can then push political and economic institutions in a new direction.

Question 12

What is an example of this shock in the UK?

Answer 12

Monarchs used to have de jure political power and they decide on economic institutions which only benefit themselves and their friends. This leads them to also holding all the wealth, strengthening their de facto power. But then there are changes in the land market and the expansion of atlantic trade (new opportunities to become rich). This new class of wealthy land owners, building up de facto power. They overcome the Stuart monarchs in the civil war, changing the political institutions, paving the way for the industrial revolution.

Question 13

Why didn’t these traders just keep all the power for themselves?

Answer 13

They were a very big group of people, so each of them wouldn’t benefit very much simply from extracting wealth from society. More beneficial to set up better economic institutions.

Question 14

Why might structural adjustments of the institutions may not work for reform?

Answer 14

If political institutions and political power remain unchanged, those in power will simply try and achieve their goal via a difference set of instruments e.g. privatisation of gov’t assets organised to distribute assets to political supporters.

Question 15

Why is there still no guarantee that we’ll end up with better institutions if de jure power changes as well?

Answer 15

Since sources of de facto political power might not have changed

Question 16

What is an example of this?

Answer 16

Abolishion of slavery. the same people still held on to power and found other ways to repress black people e.g. literacy tests to vote, threat of violence, underinvestment into education, controlling local politics.

Question 17

Why is there still guarantee to end up with good institutions by changing de jure and de facto power?

Answer 17

Those who now hold power may wield it in the same way as those before. Another reason is that the new elite might have to be as bad as the traditional elite to take power in the first place.

Question 18

What is the iron law of oligarchy by Acemoglu and johnson?

Answer 18

de jure and de facto power shift to new elite, but this new elite now simply uses extractive institutions to hold onto power.

Question 19

What are some radical ideas to improve institutions?

Answer 19

Ibrahim prize
Romer’s charter cities

Question 20

What is the Ibrahim prize

Answer 20

Monetary prize for ‘African leaders who have developed their countries, lifted people out of poverty and paved the way for sustainable and equitable prosperity’. The aim is to increase the pay-off to setting up good institutions over extractive ones.

Question 21

What is Romer’s charter cities?

Answer 21

Find uninhabited plot of land and create a new set of rules from scratch, and allow willing participants to opt in and form a large city.

Question 22

What was Romers rationale?

Answer 22

If its too hard to change rules for a nation as a whole, so instead start from a clean state and give people a choice about whether they want to participate under these rules or not.

Question 23

What is an example of a charter city?

Answer 23

Hong Kong, geographically in china but until 1997 administered by the British and operating under a different, more market oriented, set of rules than China. This resulted in fast growth and a place where people wanted to migrate to. It also offered a model to be copied when rest of china moved towards markets.

Question 24

Why might this not work?

Answer 24

problematic logistics - infrastructure mainly. In honduras, there were protests against loss of sovereignty. People won’t trust the state to set up a city with fair rules and good institutions. Rules guaranteed by a rich country has some colonial overtones.

Question 25

What are issues with cross-country empirical literature on institutions and development?

Answer 25

1. Measuring institutions
2. Small samples

Question 26

Explain the issue that has to do with how institutions are measured

Answer 26

An often used measure is ‘constraints on the executive’ (1-7). A larger number means higher quality of institutions, more constraints on executives. Even if the ranking of countries is correct, interpreting the effect of such a variable is problematic, because the actual numbers are arbituary e.g, does Australia (7) have 7 times as many constrains as cuba (1)?

Question 27

What is the actual probem?

Answer 27

While ranking may be meaningful, the actual numbers are not (they are ordinal not cardinal). Including this variable linearly in a regression assumes that going from 1 to 2 is the same as going from a 5-6, when it actually might not be.

Question 28

What does vollrath argue?

Answer 28

When you turn this measure into a set of dummies then most of the effect on development comes from comparing the 7’s with the 1’s. Countries with few constrainst might not see much extra development when they increase constraints a little bit.

Question 30

What is the issue with small samples?

Answer 30

There are often 50-70 observations, or even less. This is especially worrying when combined with IV. IV estimator is at best consistent, and never unbiased. This also makes it hard to reliably control for alternative explanations, since we have few degrees of freedom.

Question 31

What is a problem with AJR?

Answer 31

Settler mortality rates affect development via other channels, not just through institutions e.g. poor agricultural conditions, (geographical features). AJR do control for things like temperature and humidity, but their controls don’t fully pick this up.

Question 32

What is AJR’s defence?

Answer 32

the IV results are quite stable across a range of specifications, each with different control variables.

Question 33

in AJR, the IV effect is much larger than the OLS effect. Why is this an issue?

Answer 33

Because we’d expect OLS to be larger due to more positive biases in this. ‘measurement error’ is an easy explanation, but almost impossible to asses whether this is the case. Or it could be that

Question 34

What could explain this?

Answer 34

‘measurement error’ is an easy explanation, but almost impossible to asses whether this is the case. Or it could be that settler mortality is not a valid instrument: correlated with other things that reduce income per capita.

Question 35

What does Albouy (2012) point out about issues with data in AJR?

Answer 35

36/64 countries do not have settler mortality data - instead they are assigned a value from other countries. These are also not very reasonable.
AJR more often use mortality rates of soldiers on campaign , compared to soldiers in barracks. This means that AJR use artificially high mortality rates for bad institutions and low GDP per capita

Question 36

What does Albouy do?

Answer 36

Drops the 36 countries and adds dummy variables when campaign or labourer rates are used. 1st stage and 2nd stage relationships now much less clear. But AJR point out that the campaign and labourer dummies are insignificant. Suggests these mortality rates are not differernt. They are al

Question 37

What does AJR reply?

Answer 37

AJR point out that the campaign and labourer dummies are insignificant. Suggests these mortality rates are not differernt. They also argue dropping 36 countries is too harsh and examine various alternative assignments and data sources for some of these observations. They found similar results to their earlier work.

Question 38

When can you use regression discontinuity design?

Answer 38

When we are interested in estimating the causal effect of some treatment e.g. policy, a gov’t programme

Question 39

When is this mostly useful?

Answer 39

When whether or not you receive the treatment depends on which side of a certain threshold you are on a continuous variable e.g. to get a scholarship, you have to score 60 or more on an exam. Continuous variable = exam score, threshold = 60

Question 40

What 2 things do you need to apply RD?

Answer 40

1. A continuous eligibility index, on which the population of interest can be ranked e.g. test score
2. Clearly defined cutoff score or threshold for treatment eligibility.

Question 41

Going to be using this example:

Answer 41

Want to know the effect of cash transfer programme on the daily food expenditures of poor households. Treatment is cash transfer. Poverty index 0-100. Threshold: 50

Question 42

What does RD do?

Answer 42

Exploits the discontinuity in the outcome variable around the eligibilty threshold to estimate the causal effect of the treatment. e.g. HH with index just above 50 should be very similar to HH just below 50. Any difference between them should now be that those just below 50 receive cash transfer.

Question 43

What do you do if you have lots of observations around the threshold?

Answer 43

e.g. 49-51, you just compare the average food expenditure for HH’s receiving vs those which are not for this smaller sample.
As observations decrease, you widen the sample size

Question 44

What happens when you widen the sample to e.g. 20-80

Answer 44

You can’t say HH’s at 20 are similar to HH’s at 80 because they have much higher food expenditure because they are much richer.

Question 45

How do you solve this?

Answer 45

Run a regression where we control for the continuous eligibilty measure by adding the poverty index as a control variable which allows for a smooth linear effect of this poverty index on food expenditure.

Question 46

We’ve assumed the effect of the control (index) on y is linear. What is the problem with this?

Answer 46

If the effect is normally non-linear then some of this non-linear effect might get picked up by beta1, which then no longer only reflects the causal effect of the treatment

Question 47

not finished