8. Persistence of Institutional Development

ECON6015_Lecture 8.pdf

Institutions & Economic Performance

Interest in understanding whether the history through its effect on the pattern of institutional development and evolution has a persistent and long term effect on economic performance.

Rafael La Porta et al. (1998,1999, 2000)

Argue when you look at history of colonisation, being a British colony was in some ways different to being a Portuguese, Dutch, or Spanish colony. As the legal system in British colonies was set up in a way that was thought more suitable.

Emphasis on shaping the legal system, as a strong legal system would then evolve as a result of this area being a British colony.

Supported by Douglas North

Different cultures bring in their characteristics to the colonies, which become embedded in the constitution which then persist over time.

Push back by Sokoloff and Engerman (JEP 2000)

Highlight the role of initial institutions and initial factor endowments in shaping these growth trajectories in the New World (North and South America).

Argue that if this was true then you wouldn't see so much of a variance across British colonies.

Can be seen by drawing a contrast between Brazil in the 1700's and now, with the US and Canada. Brazil has the initial factor endowments which were suitable for growing crops requiring large plantations, sugar cane. Large plantations needed a lot of labour and the indigenous population were being wiped out, or unwilling, made it cheaper to import labour (international migration) centuries back (slave labour). As there are large plantations which were supplying slave labour so, in terms of the economic and social political hierarchy they were the bottom, whilst the owners were situated at the top leading to high levels of inequality and the presence of this inequality has shapened the instititutions in that way which preserves the power of the elite. Whereas, the US and Canada were not as favourable for these plantations and cash crops.
Voltaire famously argued, he could not understand why Britain and France were locked in battle for seven acres of snow, referring to Canada. The kind of labour which came in was migration from Britain and other European countries (Germany), and the people kind of moved and replicated this back in the home country, a far more equal country, shaping the constitution and the framework. This explains why social inequality is so much lower, and why growth and investment is so much higher in the US and Canada.
Contrast between Brazil and US/Canada is a very good example which stands as a good substitute for the main argument by Sokoloff and Engerman. We have seen how these patterns have been reflected in terms of the spread in public education, literacy, and extending the franchise to 'men'.

So, they argue that it does not depend whether you are a British colony or not.

Acemoglu, Johnson, Robinson (2001)

Won Nobel Prize in 2024

Mortality rates of first European settlers in the early 17/18th centuries is a very strong indicator as to whether these countries ended up with good institutions.

If Mortality rates are very high, settlers found it difficult to settle so set up more extractive institutions.
Idea of colony was to take resources and bring it back to their home economy to spur development there.
If they were safe, they would decide to stay there and not return to a country going through Industrial Revolution covered in smog. Living conditions were quite low in these now developed nations at these times.
Therefore, settlers set up institutions which mimicked their home country. The good institutions then persisted over time as they appear to be 'sticky', not easily changed.

History Dependence on Economic Conditions: The Role of Institutions

Banerjee & Iyer (2005): History, Institutions & Economic Performance: The Legacy of Colonial Land Tenure Systems in India

History casts long shadows (similar to Sokoloff & Engerman and Acemoglu et al.)

British rent extraction in India

India was a British colony for 2 centuries, nearly became a French colony. Could of been a result of a single battle, sort of by chance.
Highlights the role that history plays in shaping economic outcomes today.
Focuses on one particular aspect/institution and the way that land revenue was collected in a specific country.
Whenever establishing a particular theory in economics, creating a casual link through empirical analysis, these studies previously have used a cross-country analysis. The discerning econometrician would note that comparing across things that are very, very different between countries and comparing growth trajectorites and attributing it to intial factor endowments, colonisation, inequality, etc. They would suggest comparing like-for-like. So, this study washes away a lot of those differences.
However, India remains a very homogenous country, you're not investigating a large set of a particular kind of people, but remains an improvement from an econometrics point of view.
They're also focusing on one particular aspect of the institution and that is land revenue collection in that country.
India had a large ruling diversity, multiple leaders across the regions.
The Mughals were in charge in the 1600's (from Uzbekistan), when the East Indian Company arrived. Initially, the EIC were an economic trading entity, but realised that this country was being ruled by semi-autonomous peoples. The EIC then offered protection between these kingdoms through their money, a private army, and their presence grew.
British took over modern day Province of Bengal in 1757 (modern day West Bengal), most prosperous region in the Indian sub-continent. EIC were now in charge and no longer just being payed, became a pocket ruler.

They began to collect revenues, through agriculture, trade, and slowly expanded the East Indian Company (Not the Crown). There was a revolt in 1857, when Queen Victoria became ruler and first official time Indian became colonised.

EIC began experimenting with different revenue collection systems

Landlord-based rent collection system: districts in the 1990's which historically had landlord-based rent collection systems underperformed. In the mid twentieth century the British left and the whole administration was overhauled. There became a unified method of 'tax' collection.

So why in the 1990's did those districts who had the landlord-based rent collection system underperform?

They underperform in terms of agricultural yields, investment in agriculture, and public investment in education and health.
Wheat yields are 23% higher and infant mortality 43% lower in historically non-landlord-based rent collection systems.
Therefore, how you used to historically collect land revenues seems to have an effect even though they have been removed.

Two Channels to Explain this Disparity

The Landlord System:

Landlords were the landed gentry. They were very rich, autonomous in their own areas, similar to what the British had going on.
They were prosperous and the rest of the people were not.
These landlord-based districts created inequalities that persist to this day.
Perhaps they created social antagonism that has limited collective action to redistribution.
As there was less trust between the kingdoms, they could not go the central government for investment. Historic distrust between elite and non-elite.

Inequalities

Maybe the landlord-based areas just like in Brazil (Sokoloff & Engerman) had created a lot of inequality which had persisted even through to today which is what makes all these areas worse just like how Brazil is now worse than the US and Canada.
So, economic inequality can be detrimental for growth. However, nobody claims that perfect economic equality is perfect for growth, too much is bad but maybe a little is okay (economic consensus).

Banerjee and Iyer ignore the latter channel, as in the 1960s-80s India held massive land-reform programs and the large inequalities were brought down. Greater effect in landlord regions, as landlords had held a lot of land:

Land reforms were actively pushed, so to argue that inequality back then has survived in not correct. So, this simply cannot explain the disparities.
The gap between these two districts actually widened in the 1960s and onwards. So, clearly this could not have been due to the inequalities as by this time the inequalities had come down. The fact that the gap was widening, when Indian was investing in rural areas, there was agricultural development programs meant that these landlord explanation couldn't capture the gains compared to the others.
So, history matters as it shapes the social and political environment.

Revenue Collection

Started in 1765, starting in Bengal and Bihar, then moved out

Kolkata was set up by British India, and became the capital. Then in 1911 they moved back to Delhi.

Conquests of:

Orissa (1803)
Assam (1824-26)
Madras Presidency (1765, 1792-1801)
Gujarat (1803)
Bombay Presidency (1817-18)
Central Provinces (up to 1860)
Oudh (1856)

They sent the same leader who succeeded in Bengal to the rest of the country. So, at one point or another a large part of India was under British control, leading to the revolt in 1857, last attempt to get rid of them but never worked.

As these conquests were happening, different revenue systems were being installed.

Manufacturing was not large, so the primary sector was large and so land tax was 60% of British Government revenue in 1841, then fell thereafter.

In many cases the British tried to keep any interaction minimal, and give the right incentives to the landlord or the people in charge so they opted for a fixed rate, which was then adjusted to a certain degree but not changed year after year or season either.

Methods of Revenue Collection

Zamindari (landlord) and Zamin (land), Ryot (individual cultivator), mahal (village/house).

Zamindari: landlords pays fixed rent to British, collecting freely from peasants.

Some of these fixed rents were subject to the Permanent Revenue Settlement Act of 1793.
No regard to output, or number of labourers, the British wanted a fixed amount.
Hope is to incentivise landlords to behave like entrepreneurs and maximise profits, but it did not work this way. Many were rich and noble, and many lived in mansions in Kolkata and were not entrepreneurs. They paid the rent then moved on.
Bengal, Bihar, Orissa, Central Provinces (MP), some parts of the Madras Presidency (Tamil Nadu + Andhra Pradesh)

Ryotwari: individual cultivators pay directly

Most areas of Madras and Bombay Presidency

Mahalwari: village-based revenue collection with a village head in charge

North-West Provinces, Punjab

The Data

These are the current states in India (not including Bangladesh, Pakistan, or Myanmar), not the old areas in British India like how big Bengal back then. Some were excluded because of lack of data.

There is a lot of variation in terms of which states were and which were not landlord.

West Bengal for example has 0.00 non-landlord, making them fully a landlord state.

Maharashtra containing Bombay & Mumbai, with the latter containing Bollywood, is one of the most prospering states in India. This state was 78% non-landlord.

Punjab, village-based, used to be a once prosperous state, but unfortunately not so much now.

In the Map:

Landlord = plus signs
Individual = dots
Village-based = dashes

Comparison is between the plus and the rest. There is of course variation, but these parts are mostly not plusses, not landlord, and these states are much more prosperous.

These are newer boundaries:

Run regressions like landlord vs non-landlord, and outcomes today which is yields in public health and education.

Questions is, is being historically landlord or non-landlord random? No

But it is systematically related to outcomes today? Could there be a feedback? No, the feedback is ruled out by the separation in time. Systems back then could not be influenced by systems today. But there could be other omitted variables, something about the place which determines the kind of revenue collection system it was assigned.

The Identification Problem

What the determined the rental system? Banerjee and Iyer emphasise that the allocation of rental system was very chancy, it was not inherently systematic:

Individual influence: Mundro in Madras wanted one system, Elphinstone in Bombay wanted another. This made it a bit random.
Political Events: lead to reversal, like Oudh. Surrounded by non-landlord when itself is landlord. This decision was made after revolt. Originally, British thought that they would keep it village based like the rest, but the village said after the 1857 mutiny that they would keep it landlord. Again this makes it random
Oudh is famous for Biryani.
Date of Conquest: the later the date of conquest, the greater the non-landlord it seemed to be. The more familliar you are with the terrain, you can interact with the people and figure out what works better. More ryotwari later. Direct dealings with cultivators made easier once the administration systems had been explained. Maybe the high inequality landlord were conquered easier, and it was easier to set against each other by the British.

Why was this reversal?

These are some worrisome issues, but a good paper has to go out on a limb. The allocation of the system was in part by chancing varying local factors, nothing systematic, so you can treat it as random.

Specification

There is a dummy for every year, main variable is NLi (measure of how much of that district is non-landlord).

Dominantly non-landlord call is NLi = 1

Endogeneity Concerns

Neighbouring districts

To address these, they did a limited regression by comparing pairs, being neighbouring districts. Therefore, one of them landlord, one of them non-landlord. As they are close in proximity they would have similar geographical attributes. This works

Instrumental Variables: Conquest between 1820-1856

They then used an instrumental variable. They used this conquest between 1820-1856, the latter part as an instrument as to whether it is landlord or non-landlord. There is a correlation, if they were captured later you would be more likely to be non-landlord. The idea is that the time of your capture has nothing to do with your agricultural productivity in general.

Normally, what you would have is that each column is a whole regression but this is not the case here. These are actually the y-variables, and the reported coefficients are the coefficients on that main variable non-landlord, which is measured in two ways:

As a dummy variable (0 or 1)
Or as a proportion

When you look at non-landlord as a proportion, that has a positive effect on the amount of crop area irrigated now. Similarly for fertiliser usage, all of these coefficients are positive and significant. The first five are measures of agricultural investment, now measures of agricultural productivity.

Non-landlord areas have higher amounts of both agricultural investments and agricultural productivity.

In the second column, they exclude Bihar and Bengal because if you think that they are the main drivers, the effects are the same. So, it is not one particular area driving the results.

Main takeaway is that if you had historically non-landlord state revenue systems, then today your agricultural investments (irrigation, fertiliser use, usage of high-yield variety seeds) or in terms of actual output of crops is high.

Non-landlord state is GOOD.

Same effects when. they run the regression with neighnours, agricultural effects are higher when there is a non-landlord district. Effects disappear for one of two variables, but you are trading off as it is a smaller sample.

Controlling for irrigation, adoption of HYV (high yield variety seed) and fertiliser use, NL has no further impact on yields.

Why is it that in areas which were historically non-landlord do you see good effects today?

Banerjee and Iyer argue that the reason you are getting high yields in terms of agriculture is because you get more investment in agriculture, there is better use of fertiliser, there is more selection on the type of seeds you plant.

The difference between investments of landlord (bottom) and non-landlord (top) begins to diverge (widen) after 1965. Maybe it suggest that the landlord districts didn't manage to get as much investment in agriculture as the non-landlord districts.

In 1965 many farmers were given loans, subsidies as a nation wide initiative from the central government as part of the Rural Development Programs such as:

high yield variety in rice and wheat
public infrastructure (fertiliser delivery) for agriculture

Banerjee & Iyer argue that former landlord districts were worse at collective action to get public investment:

"One way to characterise the difference in the nature of public action is to say that landlord-dominated states were busy carrying out land reform exactly when the non-landlord states started focusing on development."
They could never solve the collective action problem, there were elites, and large group of non-elites who never thought elites cared about them.
Whereas, in non-landlord districts these distinctions were fuzzier. No clear big land owner. Closer in economic prospertity so could ague you have much more in common.

Conclusion

The landlord dominated states were really pushing for land reform to wipe out inequality. Whereas, the non-landlord did not have to do as much land-reform and focused more on investment.

The main driver for these results is the amount of investment made by the state into agriculture. Those area which were more landlord, the investment by those states were lower:

they were doing land-reforms and were stretched
it was more difficult to implement any sort of investment as people could not agree.

Again, non landlord were positive and significant results for:

A = agricultural investments
B = agricultural productivity
C = education and health investments
D = education and health investments

If you can do it whilst accounting for state expenditure per capita for development purposes, then in some of the cases the coefficients become weaker.

Finally, if you directly include a state dummy, the coefficients almost completely disappear. So really, it is the state investment which makes a difference, and the reason why the state investment could never really take off could never really take off was due to similar to the Brazil vs USA story. Back in the day there was so much inequality that the groups could not relate to each other, and if they are so divided how can you expect them to come forward and declare their needs.

Whereas, a more equal district are much better organised and can come together.

This aspect is reinforced by history.

Page updated

Google Sites

Report abuse