Author Archives: majorgressingham



R. Fishman & S-J. Wei

Journal of Political Economy, Vol. 112, No.2 (2004) pp. 471-96

Principal Research Question and Key Result Does tax evasion increase with the tax rate, and how responsive is this relationship?
Theory Increasing tax rates on imports may reduce collections by reducing imports. This can happen in two ways:

  1. Increasing tax rates reduce true imports
  2. Increasing tax rates reduces the true fraction of imports reported to the Chinese authorities.
Motivation Markets alone may lead to an underprovision of goods such as education, infrastructure and health services. States may step in to correct those market failures with public good provision. This will generally need to be financed out of taxes. However, it may be the case that unless taxes can be efficiently collected, simply increasing tax rates will not result in greater revenue collection due to evasion. In such a situation it would be important to know the responsiveness of evasion to tax rates such that holding enforcement mechanisms constant, tax rates could be set at or below the point where marginal revenues collected is equal to marginal revenues lost through evasion. It would not be desirable to set rates above this level as increasing the tax rate would actually decrease the amount of revenue collected. Past theoretical work has produced inconclusive results, as they are heavily dependent upon the modeling assumptions, whereas empirical work has typically suffered from an inability to precisely measure evasion and hence identify causal mechanisms.


The design of the study allows them to examine three types of evasion: a) an underreporting of unit value b) underreporting of taxable quantities c) mislabeling of higher-taxed products as lower taxed products.


Data Data are trade flow data from WITS and COMTRADE at 6 HS digit level 2,043 products in 1998.

Tax evasion is measured for China as log(export_value) – log(import_value). Export values are for goods exported from Hong Kong to China as reported by Hong Kong. Import values is the value of goods reported as being shipped from Hong Kong to China as reported to Chinese customs.  A similar technique is used to measure the quantity gap as opposed to the value gap.


Strategy They compare the (HK) reported exports from HK to China, with the (China) reported imports from HK to China, and postulate that any difference in the reported numbers is due to evasion. They then run the following regression to examine how evasion changes with tax rates:


log(exportk) – log(importk) = a + Btaxk + Ek



There is a likely problem of measurement error in the dependent variable. This is due to the fact that Hong Kong reports both direct and indirect exports to China (indirect, being from another source nation, going via HK). China reports only what it considers to be direct imports from HK, but it cannot always determine which imports are direct, and which are indirect which leads then to sometimes report indirect trades, as direct trades. This will tend to understate the evasion gap. Whilst this will not bias the estimates (if the error is not correlated with the regressor), it will decrease the precision of the estimates. In order to counter the problem, they exclude product categories for which there is generally a high proportion of indirect trade.


In order to examine the possibility that products are misclassified in order to avoid higher rates, then include a variable called average(tax) which is the average tax rate for products in the same 4 digit category (it is relatively easy to misclassify within 4 digits as the descriptions are quite similar). If misclassification is prevalent there should be a negative coefficient on the variable, as increasing the average tax rate of a category makes evasion less attractive (holding constant the tax rate of the product concerned).

  • Results of the baseline specification are 2.93, and this is significant, although with an R^2 of only 0.02 the model only explains 2% of the observed variation (possibly due to noise from the misclassified indirect imports), although the r^2 is improved somewhat when the data are aggregated into 42 tax rates and averaged.
  • The average(tax) coefficient is negative and significant and also its inclusion drastically increases the coefficient on the TAX variable.
  • For the quantities specification the results indicate that misreporting quantities is not as prevalent or significant as misreporting classification and value.


  • Robust to the exclusion of large outliers.
  • They try to remedy the measurement error (see below), and vary the amount of excluded observation pursuant to this strategy.
  • They interact a dummy for a product being tax exempt with the TAX variable and find it is negative and not significant (i.e. there is no incentive to cheat when the product is exempt from tax).
  • They do a difference estimation using data from 1997 and 1998 in order to control for fixed product characteristics omitted from the main specification that may still be driving results. The results are similar, although reduced in magnitude and significance as much of the variation has been differenced away.
  • They allow the functional form to be varied by estimating the main specification for 4 tax rate quartiles. They find low effects are low rates, and this increases, and then tapers out as rates get particular high.
Problems It is not totally clear that the reported import/export numbers should match, as HK may impose export tax in which case there would be incentives for exporters to underreport. It is also not clear that the Chinese would then rely on HK values, rather than making their own assessments.


If some products are more likely to be misclassified than others then we have an omitted variable problem or a problem of non-random measurement error (as error is now correlated with the regressor). If the proper classification of a shipment as direct or indirect is some function of the amount of effort put in by Chinese customs officials, and EFFORT is a variable that reflects this, it seems plausible that effort may be increasing with tax rates. This could be because a more valuable potential cargo in terms of tax revenue may attract more attention in customs. Likewise EFFORT will be correlated with the gap measure, as EFFORT determines to a certain degree the recorded value of imports. Thus EFFORT is correlated both with the independent variable and the error term, and will therefore bias the results. The direction of the bias is known: at low tariff levels low amounts of effort are put in so there is over reporting of imports due to classifying some indirect imports as direct. This will tend to mask any evasion. At higher levels of taxation more effort is taken, so misclassification is reduced exposing the full extent of evasion. Thus evasion will tend to appear greater at higher levels of taxation, and this will bias the estimates of the TAX variable upwards.


In order to identify evasion by misclassification, they include a variable of average tax across similar products defined as those with the same 4 digit HS codes. What is not clear is the extent to which there is significant variation in tariff levels for products within the same 4 digit code. Whilst the frequency distribution graph 2b does show variation in tariff levels within 4 digit categories, most of this variation is at the low end, between 1-4%. Thus it is not clear that a product imported to china would have tariff significantly different from the average tax rate for that 4 digit category. This means the tax variable and the average tax variable could be quite strongly collinear. This makes it difficult to separately identify evasion by misclassification.


To identify evasion by misreporting of quantity they use a different dependent variable that measures quantities rather than values. This measure should be subject to all the same problems associated with the value measure. It is curious that they find no effect of evasion by quantity manipulation, especially since it seems obvious that in the presence of evasion by misclassification recorded export import quantities should not match as some imports have been fraudulently labeled as different products.


Implications The estimates suggest that the average Chinese tax rate (36%) is already on the wrong side of the Laffer curve. In other words increasing tax rates will actually decrease tax receipts.


Developing countries frequently rely on border measures as a means of raising revenues. This is because with incomes low, and consumption taxes unpopular, they are often left with few options to effectively raise revenue. The evidence supplied by this paper indicate that to the extent that such measures are relied upon, care needs to be taken to set rates such that the maximum revenues are collected and not lost to evasion (assuming that enforcement measures cannot be readily improved due to corruption/lack of capacity).





C.T. Hsieh & E. Moretti

The Quarterly Journal of Economics, Vol. 121, No.4 (2006) pp. 1211-48

Principal Research Question and Key Result Did Iraq cheat the UN oil for food program by setting prices for its oil lower than market value in order that extracted rents could be shared between oil purchasers paying bribes, and the Iraqi government/leadership? The key result is that yeas they did. Around $4.3bn in rents were extracted, and of this they estimate that $1.3bn accrued directly to the Iraqi government. However, this was only 2% of the total value of the programme, so is not actually that bad by international standards.


Theory This is not a theoretical paper as such.

In general the mechanism at work is as follows. After the sanctions against Iraq were enforced, the oil for food programme allowed Iraq to sell tis oil provided that the proceeds were used for humanitarian purposes. Iraq could freely choose the buyers of Iraqi oil. They also had some discretion over the selling price of the oil. This incentivized Iraq to underprice its oil relative to the world price, and then share the generated rents with purchasers who would pay bribes to the administration, and receive oil at a lower than market price.

If prices are set endogenously (by Iraq), then they would ideally set the official price at 0, which would allow them to collect 100% of the revenue generated by selling oil at the unofficial price. As it was, they were constrained from doing so because the UN had to agree on the official price, and Iraq would be punished if the UN was reasonably certain that they were setting the official price too low. The probability of detection is assumed to be increasing with the distance between the official prices i.e. f(Pmarket – Pofficial/SD). The price differential is normalized by the standard deviation of prices to capture the idea that it was more difficult for the UN to determine whether the official price was too low in periods of high oil market volatility. Given this constraint, Iraq would set the official price such that the marginal gain from lowering the official price was equal to the marginal increase in detection from doing so. This yields two hypotheses:

  1. Increased volatility in oil prices decreases the official price paid for Iraqi Oil
  2. Cost of detection was probably higher for respected multinational companies than it was for obscure individual trades (who are more willing to pay bribes), thus as the scope for rent extraction increased (with volatility), the make-up of purchasers would shift toward individual traders (although it is still possible that supply was such that only multinational companies were able to absorb Iraq oil output).


Motivation The oil for food programme was in terms of $$$ one of the largest humanitarian efforts undertaken in recent history. It is a rough example of conditional assistance (in that the proceeds from oil sales could only be used for humanitarian projects, and prices had to be agreed by the UN). This paper looks at how the designs of such initiatives can create incentives or opportunities for cheating/corruption. In general then, lessons learned from this episode could be used when thinking about how to design interventions. In particular, that there are striking differences between outcomes when the later policy of retroactive pricing was introduced shows how small details of programme design can have large effects on results.
Strategy Two strategies:

  1. Compare the price official selling price of Iraqi oil to its nearest substitute (Arabian light, and Urals [the Iraq equivalents being Basrah light and Kirkuk respectively]). The show evidence that the price gap between the Iraqi oil and its substitutes averaged 0 in the year pre-programme, and again after retroactive pricing is introduced. The following is estimated relative to the years before the programme (1980-1995) which is the excluded category captured by the constant alpha:

ΔPt = α + β1Program1t +  β3Program2t  + β3Program3t + εt

Where the program variables are periods within the years of the program 1997-99 2000-01, 2002.

  1. 2. Compare the official selling price of Iraqi oil to the eventual spot price (net of transportation costs). The data for this estimation are not as of high a quality due to different sources used, but the estimates are consistent with the story.


  • Results of the first model indicate an average difference in price of $2.44 for the whole period, and $2.07 for the 1997-99, $3.91 for 2000-01, and 0.68 for the period after retroactive pricing was introduced (and the coefficient is insignificant). This generated rents between $2.28 and $4.12 bn.
  • Results are similar for the spot rate comparison, although generally less strong and there seems to be a significant effect for Basra Light even after retroactive pricing was introduced which is rather confusing.
  • They examine the relationship between volatility and underpricing in the years before and after retroactive pricing, and find that price differentials are higher in weeks of high volatility in the pre retroactive pricing period, but insignificant afterwards, which adds weight to the story about volatility allowing greater underpricing.
  • There is some suggestive evidence as scope for rent extraction increased there were more individual trader purchases. The correlation is 0.48
  • They cannot say for sure how the rents were shared as between Iraq and the traders, but they estimate that $1.3bn was had by Iraq (for details of the estimation see the paper).


Robustness Looking at documentation from the Volcker review of a specific set of 5 trades made, they estimate the gains from rent extraction using their model for the same limited time period and get very similar numbers. The Volcker review was based upon review of the small amount of documentary evidence that exists.

They explore a variety of alternative explanations:

  1. There was a stigma associated with buying Iraqi oil and this lead to lower prices: this is not really feasible, as there is a persistent gap between the official price and the oil’s own spot price in the market (for exactly the same oil). Additionally, there is not likely to have been any major shift in moral perception of Iraq oil in 2001 when the price differential all but disappears.
  2. Decline in Shipping facilities: There is no reason to think that facilities improved drastically post 2001. Additionally, they collected data on waiting times for ships at oil terminals in Iraq and find no significant relationship between the official price and the waiting times.
  3. Decline in quality – again, they are comparing markets for the same oil, so this can’t explain differences in official/spot prices. Also, oil quality can’t have improved dramatically in 2001.
  4. Increased supply: There is no relationship between output and the price differentials, so it is unlikely that lower prices were caused by shifts in output.
  5. Market volatility: underpricing estimates are not changed significantly when controls for volatility are included.
Problems This only looks at one possible source of rent extraction. Iraq could also have been overbilling for humanitarian supplies.

External validity is pretty low as this was a remarkable series of events.


Implications Programme design is of great importance. The paper does not show that aid is not effective, indeed the programme was thought to be a success in terms of its humanitarian goals, and as the extracted rents represent only around 2% of the total value of the programme, and this is quite low by other development programme standards. However, there is no guarantee that humanitarian investment would continue after the programme was to end.





I. Kuziemko & E. Werker

Journal of Political Economy, Vol. 114, No. 5 (2006) pp.905-30

Principal Research Question and Key Result Does having a temporary seat on the UN Security council increase aid receipts due to vote buying type behaviour? In other words, is aid allocated pursuant to geo-strategic positions? The results indicate that on average non-permanent seat holders receive a 59& increase in total development aid from the US, and 8% increase from the UN itself.
Theory There is good reason to think that aid will not increase to non-permanent seat holders. Firstly, candidates may be seeking the non-financial benefits of the seat (increased geopolitical influence, access to information etc.). Secondly sticks could be used instead of carrots (e.g. Yemen had its aid cut for not supporting Iraq War II). Thirdly nonpermanent members have very little power as the permanent 5 have vetoes, and so it may not be worth actually buying anyone off.


However, there are other reasons to think that aid might increase. There are three theorized reasons as to why:

  1. Vote selling
  2. The public eye on the seat holder allows them to raise the profile of their internal needs, and as the west becomes more aware of their internal problems, more aid is given.
  3. A country becoming more integrated into the world economy may improve their chances of being on the council, and the West’s willingness to supply aid and this could be driving the correlation.


Motivation If aid is allocated strategically then this may be a partial explanation as to why it is so difficult to quantitatively uncover benefits from aid spending across countries.


Data The council had 10 non-permanent members, and every year two new ones join and two leave. There is competition and a beauty parade to get on the council so it is by no means exogenous as to which countries accede to the council seat.

The data they use is limited to developing countries. Aid measures are taken from the USAID figures. ODA data for grants from the UN come from the OECD.


Strategy They run a fixed effects model with log(Aid) in country i at time t as the dependent variable, and a dummy for being a security council member on the right, with  controls for whether the country was at war, and the Polity variable of ideology. Additionally they interact the membership dummy with dummies that indicate whether it was an “important” year to be on the security council as measured by the number of mentions the council gets in the NYT (e.g. lead up to Iraq, Falklands, etc. were important). If the ability of a country to get aid, and a seat on the security council is driven by some omitted variable (such as increased economic integration), then there should be no differential effect between being on the council in an important year, as opposed to a non-important year. If the interaction is significant, then it is plausible that the security council effect on aid is causal.


Additionally they have time dummies indicating the year before election, the election year, the two years of service, and the two years after service on the council. If aid increased in the year before election, this would undermine the causal story. Similarly if aid remains high after the years of service, then this would indicate that the country in question had permanently raised the awareness of its needs, and this would detract from the vote buying story.


  • Overall the coefficient on the membership dummy is 0.47, which translates to a 59% increase in aid from the US. The interaction terms indicate that when it was an unimportant year, the seat holder received essentially no extra aid, but when the council was most newsworthy the interaction becomes significant, and translates into an extra 170% of aid. Adding the political controls does not change this result. This is consistent only with the vote buying story.
  • The year prior to election does not see any increases in aid, but there are significant increases in the year of election, reaching a highly significant figure in the second year of service, that drops off pretty much immediately after service is terminated, indicating that aid increases are intimately tied to council membership. This is consistent with the vote buying story.
  • The results are similar for the UN funding although the magnitudes are smaller (presumably as more people have to approve it).
  • When the split the ODA by UN agency they find that the biggest increases are from UNICEF and the UNDP which are generally thought to be controlled by the US.
Robustness n/a
Problems The conclusions of the paper may be slightly overblown in terms of implication for aid in general. Whilst it may be the case that aid is used strategically in the UN, there is no evidence supplied that this is the case for the entire US aid budget. Additionally, if on average there is an increase of only 8% for ODA supplied by the UN, then there must be a large amount of ODA that is not applied for strategic purposes, and we might then wonder why we can see no cross country effects from this type of assistance.


  • There is support for the US power hypothesis. This indicates that aid is allocated strategically.
  • These results may explain why it is hard to uncover the positive effects of aid in macro studies; because it is being applied not based on either need or effectiveness, but on strategy. In a way, this is a sanguine finding, because it indicates that if aid is actually directed to developmental ends it may have positive effects.





W. Easterly, R. Levine & D. Roodman

The American Economic Review, Vol. 94, No. 3 (2004) pp. 774-80

In a Nutshell

They reconstruct the Dollar and Burnside data set, but add additional countries for which data are newly available, and extend the analysis to 1997. They otherwise maintain the exact same methodology. The DB results do not hold – the interaction term coefficient changes sign and is insignificant, and this is so for the 2SLS and OLS and the restricted (low income) sample. The results chop and change sign depending upon the specification but are rarely ever significant, showing how fragile the results are on aid effectiveness (perhaps because the aid figures involved are so small relative to other sources of finance).

This paper indicates that we should treat the DB results cautiously. Often the literature around this question is not informed by theory, and there can more plausible specifications than there are data points in the sample.




C. Burnside & D. Dollar

The American Economic Review, Vol. 90, No.4 (2000) pp. 847-68

Principal Research Question and Key Result  Is aid only effective when it is applied in a good policy environment? They cannot detect a significant effect of aid on growth except when aid is interacted with a good policy variable. Therefore aid is effective conditional on a good policy environment.
Theory Aid is an income transfer, and this transfer may or may not produce growth depending upon whether it is invested or consumed. To the extent that it is invested it will be effective. In turn, the extent to which is invested is dependent upon a good policy environment being in place in the recipient country. 


Motivation Aid can be unrestricted, or conditional. In the case of unrestricted aid, governments receive lump sum transfers to use as they please. This is attractive as it preserves the sovereignty of the state, and it is possible that the home government knows best how to direct aid. On the other hand, it may then be used to finance personal consumption by the politicians, and there is likely to be misalignment between the preferences of the donor and the recipient.

Thus conditional transfers may be preferred, where a type of contract is agreed upon between donor and recipient about the use of funds. This has the benefit of aligning preferences and restricting rent seeking (to a degree), but there are enforcement issues as well as issues relating to “ownership” of the aid programme. Additionally, as money is fungible, aid for one project, may then reduce the amount the government has to spend in that area from its own budget, and those freed up funds can then be directed to arms, or other projects that may be unrelated to development.

If it can be shown that aid is only effective in good policy environments, then this suggests that aid should be directed only to those states that have good policies. However, the states that have those policies may not be the neediest, in which case an argument could be made for directing aid to where it is most needed, but attaching conditions that create some incentive to create a good policy environment.


Data Panel of 56 countries and 6 four year time periods.

They have data on GDP growth; they include the Sachs/Warner openness dummy (trade policy proxy), inflation (proxy for monetary policy), budget surplus and consumption over GDP (measure of government consumption policy).

The aid data is a novel and welcome improvement to past studies as it includes not only grants, but the grant component of concessionary loans.


Strategy  Basic growth equation with time/country fixed effects. The key coefficient is on the (Policy_Index*Aid) interaction.

Initially they interacted each policy variable with aid, but they could not get precisely estimated coefficients (i.e. they were insignificant). So they created a single policy index which measured overall policy. They did so by regressing growth on the policy variables with controls and used the separate coefficients to construct weights for the policies that would be used in the index (the justification being that the coefficients show how important each policy is for growth).

The OLS estimates could be biased due to correlation of aid with the error term. This could be negative if donors respond to negative growth shocks by providing more aid, or positive if donors supply aid strategically, so as countries grow in income (and hence influence) more of an effort to court them is made by the international powers. Hence the also do an IV.

The instrument is based on the notion of “political influence” and is constructed using measures of population, arms imports, Egypt dummy, franc zone dummy and central America dummy. It is quite a weak instrument, although it does just pass the F-test rule of thumb test.


Results In the regressions without the interaction term aid never enters significantly in either the OLS or the 2SLS estimations.

The interaction term enters significantly in the OLS regression, but not in the IV regression. They include an interaction between (aid^2*Policy_Index) which enters negatively implying that the impact of aid is a positive function of policy, but a negative function of the amount of aid (diminishing returns).

They run regression with aid a dependent variable in order to see how aid is allocated. They find that smaller and poorer countries get more aid. Egypt gets 2 % of its GDP in extra aid, and the policy index has a positive coefficient, although the magnitude is small and it is not significant.

They also use government consumption as the dependent variable and find that bilateral aid is more closely associated with increased consumption than multilateral aid.


Robustness They drop middle income countries (Brazil etc.) and find that both the OLS and 2SLS coefficients are significant on the interaction term of interest. This implies that policy is more important for aid effectiveness in low income countries


Problems The good policy definition is pretty restrictive, and may not really capture what is important about the domestic policy environment. For example, free press/separation of powers could be equally important if sections of society/government are able to hold the executive to account for his spending decisions. Additionally, that the individual policy interactions were not reported, and then dropped as they could not be precisely estimated makes me suspicious. That they then composed a policy index using a method that whilst intuitive, was not backed up by any theory, and this was found to get the results they wanted, makes me even more suspicious – it starts to look like data manipulation.

The exclusion restriction almost certainly does not hold. For it to do so, the measure of population and arms (to take but two elements) need have effects on growth only through their effect on aid. This is clearly absurd, as the size of the population, and population growth is a key component of the neoclassical growth model irrespective of aid.

Throughout they treat policy as exogenous, which it very likely is not.

The significance of the results is pretty variable and fragile to the specification. The 2SLQ estimates are only significant for the lower income countries. Whilst this is not necessarily a problem, it indicates that the evidence is not particularly firm – as will be confirmed by Easterly (see later summary).


Implications That bilateral which is presumably most closely tied to donor interests, increases government consumption, may be evidence of why aid is typically not found to have desired growth effects in recipient countries.

Given the problems associated with the estimation, it is hard to have a huge amount of faith in the results. In a sense, they are intuitive – if a government has policies conducive to growth and they invest aid pursuant to those policies, then by definition growth will increase. However, the question remains as to how aid should be allocated, need or effectiveness? There may be political constraints that prevent governments from allocating to where aid may be most effective, as those countries may also be wealthier. However allocating on need may be equally difficult if there are no benefits from doing so. Conditions could be used, but these have their own problems as documented elsewhere.





There are generally thought to be two rationales for the giving of aid:

  1. Fix market failures – such as building institutions, legal systems etc.
  2. Get resources to needy individuals – this may look more like humanitarian relief.

However, the actual resources devoted to aid are tiny when compared with the total amount of resources devoted to development (i.e. government expenditure, personal expenditure etc.). However, it is a very visible political issue and hence there is generally a lot of media/academic attention devoted to whether aid achieves its goals. However, given the small amount of resources that are actually devoted to aid (only 4 countries maintain the UN goal of donating 0.7% GDP), perhaps it is not that surprising that quantitative studies find it difficult to uncover any beneficial impact of aid on growth etc.

Often aid is seen as a political instrument, e.g. rewards for assisting in war on terror, and historically Egypt received lots of funds due to the presence of the US military in that country.  Thus, if aid is a political tool that is directed at strategic interests rather than based on need/effectiveness, then we should not be surprised if it is difficult to uncover meaningful benefits of aid programmes.

The debate is quite high profile. Sachs and his crew argue strongly in favour of aid, whereas Easterly tends to assert that we need to search for mechanisms of improving outcomes rather than just throwing money at the problem. Moyo argues that aid actively harms. It is hard to model a situation where aid is bad as opposed to simply a waste (although it may cause conflict, crowd out local markets, encourage rent seeking and a dependency culture).

What do the data say? Well macro data is hard to analyze as aid is not applied in the same way in different regions, and this heterogeneity leads to difficulties in interpretation. Additionally the debate is rarely theoretically justified. If poverty traps are real, then aid will be useless until it is applied such that capital accumulation can occur at such a level that the trap is broken. Thus is aid is constantly applied to poverty trap countries, the results will always be negative growth until the right level of aid is disbursed. As ever, we do not observe the counterfactual, so even if growth is negative in the presence of aid, it may have been even more negative without it. Detecting results may also be difficult for the reasons noted in the opening paragraphs.

Potentially, micro studies could be of use, but without taking into the general equilibrium effects of aid, results will likely be of limited use due to poor external validity.





S.Subramanian & A. Deaton

Journal of Political Economy Vol. 104, No.1 (Feb., 1996) pp.133-162


Principal Research Question and Key Result How elastic is caloric intake with respect to expenditure? The key result places the elasticity at 0.55.
Theory The development literature posits two links between income and caloric intake, with causality operating in different directions for each theory. Firstly, some argue that productivity depends upon nutrition. In extreme cases this can lead to a poverty trap as those who do not get enough to eat are insufficiently productive to make it profitable for employers to hire them above the wage threshold that would give them sufficient income to purchase enough calories. This implies unemployment and underemployment and an inability of businesses to attract sufficient workers. This inability to work is thought to act as a break on economic growth.

The other line of enquiry holds that nutrition is conditioned by income, and in this regard we try to estimate the Engel curve, which is how nutrition changes with respect to income. The demand for calories should rise with income, perhaps not 1:1 as people substitute quality for quantity, but with an elasticity greater than 0.

However, some academics have argued that elasticity is 0, in particular Bouis . This implies that economic growth will not result in improvements in nutrient intakes. If this is true then there is a real challenge for economists who tend to measure welfare in terms of income. If income does not guarantee a good standard of living (perhaps because people do not really know what is good for them), then the whole approach to increasing incomes needs to be rethought. In a further example, economists believe that the ability to substitute across goods is welfare enhancing as it allows consumers to protect themselves from price shocks on one particularly product. However, a nutritionist is only concerned with individuals having a good diet, and if in fact as incomes rise people substitute toward food that is bad for them, then this decreases welfare irrespective of the newfound ability to substitute.

Motivation See above
  • National Sample Survey for rural households in Maharashtra, western India.
  • 5,630 households, 10 in each of 563 villages
  • Report expenditure on over 300 items including 149 food items (and then tables are used to convert these into calories)
  • No income data collected, so total household expenditure is used as the welfare measure.


Method Summary Stats etc.

They regress total available calories on the number of meals given to guests, employees and those taken at home to find out how many calories are contained in each type of meal. They then subtract/add those meals given away/received to the total calories available, to create an adjusted figure. If they did not do this the elasticity for richer households would be grossly overstated as they have a large number of available calories as they give away many more meals to employees/guests. When the data are tabulated it becomes clear that the poor spend a lot less money per 1000 calories than the rich. This is because coarse cereals provide a much larger share of caloric intake of the poor. As people get richer it appears they substitute between food groups away from cereals toward dairy and meat products etc. which have many fewer calories per unit of expenditure. Thus although the total food elasticity is 0.772, the price of calories elasticity of 0.32 drives a wedge between the food and calories elasticites. This is due to substitution.

Empirical Method

There are two strategies. The first models the log per capita calorie expenditure against the log of total household expenditure using smooth local regression techniques, similar to kernel density analysis, with weights assigned such that observations closer to the x in question have greater weight. The bandwidth is set such as to balance the variance-bias tradeoff. Rather than doing a regression for each value of x they use an evenly spaced 100 grid points ini the distribution of log per capita expenditure. This method is useful for modelling non-linearities between bivariate variables, but becomes too complex when controlling for other variables.

Thus there is also an OLS specification that can take into account the effect of omitted variables not included in the above method. Such important covariates include household demographics including size of household and age, as well as individual village effects.

Results The results of the kernel analysis do not reproduce the findings of elasticities close to 0. The predicted line is close to linear and negative such that lower expenditure households have a greater elasticity (0.65) that higher expenditure  households (0.4), which is to be expected. The elasticity estimates do not contain 0 even when applying 2 standard deviation confidence bands.

Similar analysis investigating price per calorie and expenditure indicate that price per calorie increased with expenditure, this is found to be the product of quality upgrading within food groups and between food groups.

In the OLS results including household size decreases the coefficient from 0.4 to around 0.35, and this is then robust to the inclusion of other variables including caste, labour type, religion, and other demographic variables. The elasticity is food as regards to income is 0.75, and this is pretty evenly split between increase in calories available (0.37) and price per calorie (0.38). Thus the elasticity of price per calorie with regards to expenditure, drives a wedge between the amount of food purchased and the amount of calories that are actually available.

Robustness Various controls added. They check for non-linearity using the specification noted above.


Interpretation The data are not easily reconcilable with a poverty nutrition trap. Food calories are cheap in the study and as households become richer they substitute toward lower calorie foods, which is not consistent with a picture of individuals unable to work due to lack of food, as if that were the case elasticities would be much greater. Additionally, food calories are found to be cheap in the region, and as such if there is poverty trap it is one that is easy to escape.


Problems The analysis is based upon available calories, and thus if wastage etc. is a salient issue, then results may be biased.

Expenditure is a good proxy for income but it is not perfect as it may involve spending money from relatives, remittances, and government programmes, and spending using those types of fund may exhibit different patterns than more traditional labour income.

Endogeneity: if hunger caused poverty as well as poverty causing hunger there would be reverse causality problems. The argument is that lack of hunger reduces productivity and thus wage earning capability which prevents calories from  being purchased. Hunger thus creates a “poverty trap”. They cannot rule this out by using e.g. IV regression, but they claim that as the 600 extra calories that are needed to sustain physical work, could be purchased for 4% of the daily wage, that the barriers to sufficient nutrition are not high enough to create a poverty trap.  




D. Acemoglu & S. Johnson

Journal of Political Economy, Vol. 115 No. 6 (2007) pp. 925-85

Principal Research Question and Key Result Have the 21st century’s increases in life expectancy translated into increased economic growth? No statistically significant effect on growth is found and in fact per capita indicators show significant negative association with life expectancy, indicating that any aggregate growth gains are more than offset by increases in population and birth rates.


Theory In neoclassical growth theory increased life expectancy raises that population which initially reduced capital to labour ratios thus depressing income per capita. This may be compensated for in the medium/long run by higher output as more people enter the labour market and more capital is accumulated. This compensation may eventually exceed the initial per capita measures is there are significant productivity benefits from longer life expectancy. This may not occur however if some factors of production are supplied inelastically (not sure I get this last bit). None of this should be taken as an indication that welfare will not be greatly increased from increased life expectancy, only that there is no discernible relationship with GDP.


Motivation There is a growing consensus that improving health outcomes can have indirect payoffs through accelerating economic growth. Whilst Marco studies of these effects are plagued by problems of comparability and also that countries characterized by ill health are also disadvantaged in other ways, micro studies (which show positive effects of health outcomes on growth) cannot account for general equilibrium effects. The most important of these would be that as there are diminishing returns to effective units of labour, micro studies that cannot account for the effect of population growth pursuant to improvements in health, will tend to overstate the economic returns from doing so. This paper takes a novel approach by instrumenting fairly successfully for life expectancy by exploiting what they term the international epidemiological transition that occurred in the 1940s which was essentially the development and diffusion of curative and preventative medicine.


Strategy There is a long differences strategy (primarily to look at effects of life expectancy on other demographic variables), and an IV strategy (to look at effects of life expectancy on wealth variables).

The long differences strategy compares variables in 1940 (prior to the transition) with 1980 (being prior to HIV). By taking differences they are excluding all the country fixed components of the growth model ( technology, initial human capital etc.), with an additional vector of controls. The dependent variables of interest are GDP, population, births, and age composition of the country. The primary explanatory variable is life expectancy.

The IV strategy uses potential mortality to instrument for life expectancy. They collect data comparable data for 15 of the most important infectious diseases, and create time dummies for when interventions in that disease started to occur at the medicinal frontier. The instrument is constructed as follows:

Mit = ∑[(1-Idt)Mdi40 + IdtMdFt

Mdit denotes mortality in country I from disease d at time t. Idt is a dummy for intervention for disease d at time t (and thus equla to 1 for all dates after the intervention). Mdi40 is the pre-intervention mortality from disease d in country I, and MdFt is the mortality form disease d at the technological frontier after the intervention.

(1-Idt)Mdi40this part of the expression will equal the mortality rate in the country before there is an intervention as I only equals 1 when there has been an intervention, and so when there is no intervention 1-I = 1 and 1*Mdi40 equals the mortality rate pre intervention. Then when there is an intervention I turns on to equal 1, and this part of the expression will equal zero.

IdtMdFt – this part of the expression will equal zero in the pre-intervention period, and the mortality rate at the frontier after the intervention turns the dummy to 1.

Critically the value of the predicted mortality derived is in no way dependent upon how successful a country is at implementing the intervention. The dummy switches to one for all countries at the same time, which strengthens the exclusion restriction, as variations in predicted mortality are in no way correlated with the country specific error. Additionally, the Mdft is set to zero such that predicted mortality from a disease equals zero after the intervention, and so is uncorrelated with the error.

They show that there is a strong first stage relationship between the predicted mortality and life expectancy, and this is the case even excluding the richer countries from the sample

Results Long difference: the 1940-80 estimate of the effect of life expectancy on population is 1.6 which suggests that a 1% increase in life expectancy causes a 1.6% increase in the population. The coefficient on births is 2.35 and the on the % of population under 20 is 0.94. These are robust to using longer time windows (1940-2000). When the GDP measures are the dependent variable, there is a 0.85% increase in GDP associated with 1% increase in life expectancy, but this is insufficient to compensate for the population effects, so GDP per capita and GDP per working age population are both significant and negative. These results are only tentative, as there are endogeneity problems i.e. richer countries may invest more in health.

The first stage yields a coefficient implies that improving predicted mortality will improve life expectancy by 21% and this is significant at the 1% level. This is the case then shorter time windows are used also. The run the 2SLS IV estimate with different dependent variables:

  • Population – countries with a larger decline in predicted mortality experienced larger increases in population. The coefficient on the baseline sample is 1.65 and is significant at the 1% level.
  • Births – coefficients vary between 2.15 and 2.9 which are quite high
  • % Population under 20 – the coefficient is large and significant, but using the longer time window becomes insignificant indicating that the new drugs saved the lives of young people initially, but this effect was averaged out over time.
  • GDP – the coefficient on life expectancy is now only 0.32 and is not significant, although given the size of the standard errors, economically significant effects are impossible to rule out. This effect is smaller when only looking at the subsample of poorer countries.
  • GDP per capita – any increase in GDP was more than offset by the population growth as the coefficient is now negative and significant, and this seems to have affected working age population as much as any.


  • They use some alternative instruments which I will not go into, results are similar.
  • In the IV estimates they control for institutions, initial GDP, continent dummies etc. and there are no significant changes to the estimates.
  • They do a falsification test by showing that lead predicted mortality had no effect on life expectancy, and they further see if predicted mortality in 1940 is a good predictor of life expectancy in the period of 1900-1940 which it is not indicating that preexisting trends are not driving the validity of the instrument.
  • The analysis only takes into account mortality. If the epidimiological transition has also greatly reduced morbidity and this has had an effect on individual productivity, then there may have been economic effects not captured by life expectancy measures.
  • If the disease environment in 1940 in some way was a reflection , or cause of the growth trajectory/path of a country, and that trajectory is persistent over time, then using the 1940-80 window may cause endogeneity issues, as outcomes then, and in 1980 will be correlated with this unobserved trajectory variable and this could be biasing results.
  • The results may not be applicable to the world today as the transition was a very singular event, and the disease environment today is very different particularly taking into account HIV which tends to affect adults in the prime of their lives, rather than the diseases analyses here whose burden fell predominantly on young people outside the workforce.
Implications Increasing life expectancy increases the population, and birth rates did not decline sufficiently for there to be any meaningful compensation in terms of increased productivity. Overall this has led to a decrease in income per capita. This would seem to suggest that for economic (as opposed to simply welfare) benefits to be felt from improved health outcomes, there needs to be concurrent efforts to improve productivity such that capital can be accumulated at a faster rate than the population growth diminished capital per worker. Naturally none of this means that promoting health outcomes does not have its own intrinsic value.





J.L. Gallup & J.D. Sachs

Journal of Tropical Medicine and Hygiene, Vol. 64 (2001) pp. 85-96

A Summary

What effect does Malaria have on GDP growth in Malaria endemic regions? They estimate that countries with severe malaria incidence have on average 1.3% lower growth rates per year.

It is quite clear that poor countries predominate in the same regions as malaria, and that economic growth in those regions is much lower than elsewhere. What is not clear is whether the observed correlation between malaria and low growth are causal or merely coincidental. Specifically, malaria could merely be proxying for other determinants of low growth such as poor quality tropical soils, inaccessibility to world markets, different patterns of colonization etc. However, in regressions that include these types of geographic variables, malaria is still found to have a significant effect. This is also the case when Africa is excluded from the sample, which shows that it is not merely Africa that is driving the results.

What is not clear however is whether malaria is a cause of poor growth, or the product of poor growth? If for example, malaria eradication/prevention becomes possible only when countries reach a certain level of development, then the true effect runs from growth to malaria, not the other way around. The authors claim that this interpretation does not tally with the facts. They claim that Malaria eradication is determined by ecology and climate, and not by personal behviour, general development, or the extent of urbanization (although they may be important, they are of second order importance). This can be shown when it is considered that the regions with the worst malaria in 1965 had the least reduction in malaria in the next three decades. They claim that in endemic areas of Africa with up to 300 infectious bites per night, control simply is not possible and so it cannot be argued that control is the effect of growth.

Whilst seductive, this line of reasoning is not really sufficient to prove the case. As they attest to there have been successful eradications of malaria. For example in Greece where up to a quarter of the population was infected in the malaria season, it was successfully eradicated. This also occurred in Italy, and other parts of Southern Europe. Whilst the virulence and prevalence of the disease in those areas may indeed have been less severe than in Africa, it is not at all clear that they were not able to eradicate the burden due to comparatively higher levels of development.  The sort of largely anecdotal evidence presented in the paper is not really sufficient to substantiate the claim that causality runs from malaria to growth and not vice-versa.

Additionally, they show growth patterns for Taiwan that successfully eradicated malaria, and compare their pre and post malaria growth rates with that of the rest of East Asia, and the simple difference in difference is only +0.9% which is hardly compelling, and is not estimated using regression techniques so no statistical significance can be evaluated. Additionally Mauritius did not see growth after eradicating malaria, although this may have been due to the closed nature of their economy – this indicates either that the effect does not run from malaria to growth, or that malaria interacts with other features of the economy such that simple eradication does not guarantee a growth episode.

They run a cross country growth regression using a new malaria index which is the fraction of population living in areas with high malaria risk in 1965 times that fraction of malaria cases in 1990 that are due to the most severe malaria vector. On the right hand side are initial income levels, human capital, institutional variables, geographic and economic. The results indicate that both initial levels of the index and subsequent changes are significantly associated with GDP growth. One might argue that the malaria index is only capturing the worst type of malaria, and there seems to be little sense in leaving out other types, as restricting to the worst variety essentially confines analysis to Africa, and Haiti. Indeed when Africa is excluded the results persist, however this could be being driven exclusively by Haiti, which is extremely poor, and burdened with very high malaria levels. Nevertheless, the results show that for a 10% reduction in the malaria index on average would result in 0.3% growth increase.

In order to combat remaining problems of endogeneity they instrument for the malaria index using the prevalence of mosquito vectors in each country in 1952. No first stage is reported and thus it is not possible to evaluate the strength of the first stage. Additionally it is not clear that the instrument and the index are substantially different, they both seem basically to be measuring the same thing. However, the results do not substantially change the OLS estimates.

When they include other tropical diseases in the regressions they do not find any significant effects. Whilst this indicates that malaria is not proxying for other diseases, it is not theoretically clear why malaria should have significant effects, but yellow fever etc. should not. This leads me to question he results.

There is little agreement as to what channels malaria works through to lead to lower GDP. Suggested in the paper are:

  • Lower productivity due to morbidity (although potentially mitigated by partial immunity)
  • Lower levels of cognitive development (see for example the Spanish flu paper in EC454)
  • Malaria keeps away tourists and investors.
  • Malaria limits internal movement of people and hence goods/services.

If the results are taken at face value, then a huge amount of emphasis should be put on prevention and eradication. This is in fact what we see in the development community. LLINs are being widely distributed and provide a very cost effective way of reducing the incidence of malaria, particularly since only 50% of a community need to sleep under a net in order for spillovers to be created (the mosquito dies when it lands on the net so is unable to bite anyone else). Naturally these interventions are not made solely with GDP in mind as there are welfare benefits from not being sick that are potentially more of concern than long run GDP growth.




A. Abadie & J. Gardeazabal

The American Economic Review, Vol. 93, No. 1 (2003) pp. 113-32

Principal Research Question and Key Result Did the conflict in the Basque country affect the economy? The results suggest a 10% loss of GDP due to the terrorism.
Theory Terrorism could affect GDP in various ways. The most important is likely to be investment. If earning a return on investment becomes uncertain because either the return may be extorted or the entrepreneur killed then this acts as a random yet significant tax on investment. Under such a circumstance investment is depressed and this will affect output and hence GDP. Additionally foreign investment in the affected region could be reduced if conducting business in that region is thought to be risky, although the mechanism is exactly the same, although it operates on international rather than domestic actors.


Motivation Political instability is often said to have strong effects on economic prosperity. However, studies to date have largely been cross country studies which suffer from comparability issues (as conflicts are rarely similar). This study seeks to explain how the richest region in Spain subsequently dropped to the 6th position in terms of GDP per capita. As it is focused on only one such conflict the heterogeneity issues outlined above are circumvented to a certain extent (although, as ever at the expense of external validity).


Data They have panel data for 1968-1997 which includes variables on deaths and killings, as well as GDP and other variables that can be thought to determine GDP such as investment ration, and human capital measures.


Strategy They exploit the fact that ETA was created in 1959 but did not implement large scale terror operations until the mid-70s. Additionally in 1998 a ceasefire was declared which was subsequently cancelled, and this provides testing ground for looking at how economic outcomes varied during both the scale up of violence (largely killings and kidnappings), and the cease fire.

They essentially do a DID, however they cannot simply compare the Basque country to another region, as there was no comparable region – the Basque country was the richest, most industrialized etc. So they construct a synthetic control group. They do so by identifying a list of variables that are thought be drivers of economic growth (agriculture share etc. table III) and assign weights to the other possible control regions such that when aggregated the weighted averages of the variables resemble the observed variables for the Basque country subject to the constraint that the variable that should best be reproduced is the GDP per capita for the Basque country in the 1960s. When this is done, they end up with a synthetic control group that is comprised of 85% Catalonia and 15% Madrid.

During the ceasefire they look at the cumulative abnormal returns of stocks listed as Basque stocks, relative to other stocks on the Spanish market. Asset prices should reflect all available information, and if instability is important then Basque stocks should have performed better when the ceasefire was announced and became credible, and worse as the cease fire broke down. They categorized the stocks using market professionals.


Results They plot the GDP growth of the synthetic control and the Basque country and they follow each other closely until the mid-70s, when the Basque country falls behind. This suggests a loss of 10% of GDP due to the terror. The gap in the GDPs of these regions seems to spike at the same time as the deaths from terrorism in the Basque country. 

The results of the ceasefire study are that the good news dummy coefficients are significant and positive for Basque stocks and negative for Bad news.

Robustness The do a placebo study, by comparing Catalonia and a synthetic Catalonia (constructed as above but excluding the Basque country as a possibility) and find that there is no significant gap in GDP, although the real Catalonia did outperform the synthetic one by 4% around the time of the Barcelona Olympics.


  • The synthetic control is made up almost exclusively of Catalonia, thus it is not very balanced, or impervious to idiosyncratic shocks in that region. Additionally, it is not clear that selecting weights so that GDP is matched is the best strategy, as similar GDP levels in the 60s does not guarantee that what is salient in terms of future growth has been captured.
  • They do not actually estimate the DID using regression techniques as far as I can see, so we have no idea what the standard errors are, or what the other significant factors were in determining outcomes in the Basque country. This does not allow us to verify how important e.g. industrial decline was in explaining GDP in the Basque. Without such results it cannot be said conclusively that the higher industrial share in the Basque country pre-terror was not driving lower GDP in the face of industrial decline post-terror.
  • The authors state that Catalonia and the Basque country were both highly industrialized regions. If one of the effects of terror was to incentivize entrepreneurs or businesses to move away from the Basque country due to instability, the chances are they would relocate to somewhere that was similar to the Basque country, which surely would be Catalonia. As the synthetic control is made up predominantly of Catalonia, any significant movement of human capital from the Basque to Catalonia could have affected GDP outcomes in Catalonia, and hence this would tend to overstate the results.
  • It is not clear that they have isolated anything to do with property rights as such.


Implications Conflict can harm the economy. This is not a new idea. Not sure what the policy implications are, other than avoid civil conflict if possible.





T. Besley & R. Burgess

The Quarterly Journal of Economics, Vol. 115, No. 2 (2000) pp. 389-430

A Short Summary 

In a Nutshell

Investment in the asset base of the poor has been seen as a central way to fight poverty. In the case of poor agrarian economies this typically means improving the terms upon which the poor have access to land e.g. improving tenurial security, removing intermediaries etc. Using panel data for Indian states (which had control over reforms) they find that there is a robust link between land reform and poverty reduction, although the effects are strongest when the reforms are concerned with improving the terms of contract rather than actual redistribution. The reforms increased agricultural wages consistent with the reduction poverty, and there is some evidence that overall the reduction in poverty came at the expense of lower income per capita.

They classify the reforms into four categories:

  1. Tenancy reforms – i.e. how much rent can be charged etc.
  2. Reforms to abolish intermediaries – these intermediaries were feudal remnants and were allowed to collect a larger share of the surplus than the landlords themselves.
  3. Reforms that put ceilings on landholding such that land could be redistributed to peasants.
  4. Reforms that allowed for consolidation of disparate landholdings.

A measure of poverty is then regressed on time and state fixed effects, the cumulative aggregate (but also disaggregated) land reform variables, lagged as they take time to implement.

The results indicate that (1) and (2) significantly reduce poverty, but that (3) and (4) have little effects. Whilst this may be explainable theoretically, it could also be being driven by the fact that there was no serious attempt to implement (3) and (4). The split up the results by urban and rural sector and find the effects only hold for rural populations which is consistent with the idea it was the reforms and not some wider economic fluctuations that lead to decreased rural poverty. They add various other variables that could be explaining the reduction poverty (e.g. health spending, state taxes for redistribution) and find the results are robust, although the coefficient on state taxes is also significant.

Reforms could be endogenous to poverty: if reform is targeted at poverty reduction then it will be focused on areas where poverty is highest and hence we are underestimating the effect. If it is targeted at areas that will respond the best then we are overestimating. Thus the instrument using lagged political identity of governments as instrument for the reforms. Whilst the F-test is very low (weak instruments) the results are broadly similar which adds weight to the main specification.

With agricultural wage on the left hand side they confirm that the reforms did indeed increase wages. However, with income per capita on the left they find weakly negative effects from the reforms. This hints at an equity-efficiency trade off. The reforms that have an effect on poverty (those that alter production relations rather than distribution of land) decrease poverty, but may also decrease state wide incomes



A. Banerjee, P.J. Gertler & M. Ghatak

Journal of Political Economy, Vol. 110, No. 2 (2002) pp. 239-80

Principal Research Question and Key Result Did operation Barga in West Bengal, which aimed to strengthen laws relating to regulating rents and improving security of tenure for sharecroppers, have an effect on agricultural productivity.
Theory The impact on productivity from the strengthening of land rights can come from two sources:

  1. Bargaining power – As the tenant has increased bargaining power the landlord is forced to offer him a greater share of the crop, and this increases the tenant’s incentives and hence he will produce more
  2. Security of tenure – this has two conflicting effects. When output is low the landlord could threaten eviction. However, now that eviction is not permitted, this option is not available and this could in theory reduce the incentives on the tenant to be efficient. On the other hand it encourages the tenant to invest in the land as he knows he will be in situ and therefore able to enjoy the fruits of said investment.

We expect the net effect to be positive however. This is because incentives were improved, and initial research indicated that threats of eviction were rarely used to punish small outputs. Rather landowners would just cultivate the land themselves or sell.


Motivation These reforms were not a wholesale redistribution of land; rather they strengthened existing rights in order to make relations for tenants vis-à-vis landlords more equitable, and to strengthen tenure. Whereas the reform laws in China were probably only feasible due to the authoritarian nature of the government, these types of reform could more easily be pursued by other developing nations.


Data The reforms enforced long dormant laws. By registering with the government share croppers could enforce their inheritable incumbency rights to the land, and landlords could only demand 25% of the output. They could not be evicted if they paid this 25%. Operation Barga sought to increase the number of registered share croppers which it did by many millions. At the time West Bengal was the state with the highest proportion of registered share croppers, it is land scarce and highly agriculture dependent. In general sharecroppers were forced to pay on average 50% of the output prior to reform.


Strategy Difference in Difference comparing with Bangladesh. They are similar in terms of observables, and Barga happened because of what was going on in India, not specifically because of West Bengal. Thus the parallel trends assumption is likely to hold. Rice yields were approximately the same for both states until Barga at which point West Bengal increased its output (except for the two drought years). The outcome variable is log(rice yield per hectare) in district d in time t. There are district and time fixed effects, a treatment and post dummy, and their interaction. The coefficient on the interaction (as ever) is the DID estimator.

A second strategy is to use the intensity of sharecroppers that registered within West Bengal and to compare outcomes to see if productivity rises faster in areas where registration as more intense. As the opportunity to register was not distributed to everyone at the same time, they use lagged variables of the proportion of tenants that were eligible in a time period. They recognize that the spread of the opportunity was probably related to demand, and also to supply which could be correlated with various other factors that affected outcomes.  Ideally they want an instrument but do not find one.


Results In 1979-83 West Bengal grew slower due to drought. In 1984-88 and 1988-93 rice yields were 5% higher in West Bengal. This is from a simple DID.

In the full specification with additional controls for irrigation and rainfall (to control for the Green Revolution), and the amount of rice under cultivation, results are similar. However, it is only with the full set of controls that that the 84-88 and 88-93 coefficients are significant. With only half the controls only the 88-94 coefficient is significant, and with no controls it is the same. That the results increase in magnitude when the green revolution is controlled for indicates that Bangladesh was more proactive in adopting the e principles of the Green Revolution. Barga explains around 28% of subsequent growth.

The results of the second model are that registration rate was positively associated with yields. All coefficients are positive and highly significant including all the controls indicated below. This adds credibility to the DID results.


Robustness Using data for 1969-78 they regress changes in yields over the period on a West Bengal dummy (including the Bangladesh data), and they cannot reject the hypothesis that the coefficient on the dummy is zero, indicating that yield growth was substantially the same for both states.

In the second specification the omission of other programmes could be important:

  1. Expansion of infrastructure – controlled for using roads and irrigation
  2. New seed varieties  – controlled for using area of such under cultivation
  3. Left front leads to better implementation – dummy for left front
  4. Closer to Calcutta means better implementation – southern dummy to control
  5. Registration targeted to high density sharecropping areas – controlled using density of sharecroppers interacted with time dummies.
  • There could be some sample selection bias. Faced with the new laws some landowners preferred to sell up. This lowered the price of land and allowed the sharecroppers to purchase. This means that some of the pre-reform sharecroppers were taken out of the sample as they now farmed in their own right. If these tenants had previously been low efficiency (hence why the landlord preferred to sell) then this would bias the results upwards.
  • Significance in the full DID model is not hugely strong, and seems to be driven by the inclusion of control variables. Not necessarily an issue, but it is curious.


Implications Tenancy law strengthening can lead to improved crop shares and security of tenure which can in turn affect agricultural productivity. As noted above, as this was a limited transfer of rights rather than full on redistribution and so it could be implementable in other countries. However, what is also clear is that these laws were already in existence and Barga only sought to implement them effectively. Thus, any land reform needs to take into account that it will only be successful if it does not create incentives for landlords to cheat, and if the domestic institutions are willing to enforce the reforms.





 J. Y. Lin

American Economic Review, Vol. 82, No. 1 (1992) pp. 34-51

Principal Research Question and Key Result Which of the package of reforms that were enacted in 1970s China caused agricultural output to rise so precipitously? Was it the de-collectivization (improved individual access to land), higher prices, more flexible quotas, or a change in inputs? What factors were responsible for the subsequent slowdown?  The paper finds that the dominant factor was the conversion to the Household Responsibility System, away from collectivization.


Theory Much of the developing world is dependent upon agriculture, yet there are large inequalities in land ownership. Land reforms such as the ones discussed here can improve both efficiency and equity. Having land title allows those who work the land to command a greater share in the output, and this may improve incentives to invest in new technologies, new seed varieties etc. such that output is increased. This may then have a subsequent effect on agricultural wages. Additionally, land title relaxes the credit constraints faced by the poor, such that when an individual at the bottom of the income distribution is deciding whether to be an entrepreneur or a wage earner, he may be more able to start a business if he has land as collateral. This can improve the allocative efficiency of capital, which in turn will increase aggregate output.


Motivation In 1978 after a period of sluggish agricultural growth, a package of reforms were enacted that included raising the prices of major crops (by up to 50%), a change to the HRS (which initially began in secret without permission, but then the government allowed it to be rolled out, but to the poorest regions first), the increased ability of farmers to sell into the market (after they had produced their quota), as well as a change in availability of inputs such as fertilizer. Quotas were also reduced, and interregional trade restrictions were relaxed. There was subsequently a period of rapid growth in the sector (1978-84), although after 1985 much of this growth came to a sudden halt.

If the HRS was the major source of growth then future reform should be aimed at strengthening the position of household farms, on the other hand if the other reforms were more salient, then it is possible that decollectivization actually harmed agriculture. It is important to know which mechanism was most important for designing future policy and for gleaning lessons pertinent to other developing nations.


Data Province level panel from 1970 to 1987 for 28 of 29 provinces. Agricultural output is for 7 grain crops and 12 cash crops. There is data on inputs such as land (cultivated land), labour (number of workers in cropping sector), fertilizer (gross weight of fertilizer consumed) and capital (tractors, bulls etc. measured in horsepower). There are also measures of the % of HRS, index of prices, and market prices relative to manufacturing prices, % sown acreage, multiple cropping index.


Strategy One way (province) and two way (province and time) fixed effects model. In the model with time fixed effects the price variables have to be omitted as they are national, and hence region invariant. The time dummies already capture the impacts of year to year price changes on productivity. He estimates using OLS and Estimated Generalized Least Squares. Variables are normalized by number of team/farms as the provinces vary greatly in size.


Results HRS had a positive and significant effect on agricultural growth, as did the growth in state and market prices.

The total output growth between 1978-84 was 42%, and 45% of this came from growth in inputs. Fertilizer growth alone accounted for 1/3 of total output growth. Increases in labour and capital had only minor effects on output. HRS contributed 48% of output growth through increased productivity.  The switch from grain to cash crops also appears significant. This leaves around 5% of growth unexplained. The results for HRS are similar when the input variables are excluded which indicates that HRS operated through efficiency gains rather than increased use of inputs.

The drop off in growth was most likely due to the end of the period of expanding HRS, as that reform was complete in 1983, therefore even without any other cause growth would have fallen by around 50%. Additionally, there was a slowdown in growth of fertilizer use, and also labour was drained by the manufacturing sector.


Robustness There are no real robustness checks as such.


  • It is not totally clear what it was about the HRS reforms that increased output although results indicate that it was through efficiency gains. However, would these gains have occurred had they not been able to then sell on the open market for a better profit than was offered by the state quota? In other words, it could have beent he whole package of reforms that allowed fro gains from HRS to occur. If HRS was pursued in isolation the results may not have been the same.
  • There are no real robustness checks in the paper. For example a placebo test could have been done using a different time period to check that the results are not driven by spurious correlations.
  • This paper measures only the short term effects of the HRS system. If it improves investment incentives, there should be effects that linger way beyond the time period of study.
  • Limited external validity as the Chinese government is so centrally strong. Such reforms problably would not be possible in another setting.


Implications Whilst small holding are often thought to be a barrier to efficiency in the developing world, this paper suggest that strengthening the land rights of small holders can leads to large gains in output as they make more efficient use of available inputs, and may be more inclined to invest in their farms.





D. Stromberg

The Quarterly Journal of Economics Vol. 119, No. 1 (2004) pp. 189-221

Principal Research Question and Key Result Does penetration of mass media such as radio create better informed voters that consequently receive more favourable policies? In the context of early radio expansion, this paper finds that an increase in the share of households with a radio by 1% increases spending in that area by 0.54%. 
Theory Mass media creates a distribution of informed an uninformed citizens. Informed citizens may be able to achieve better policy results. For this to occur they must vote, and they must know whether their representative has done something for them, and information from the media helps them. Thus is it more costly for politicians to neglect voters with access to political information via the media. This indicates that government spending s should be higher on groups that have access to the media, higher on groups where more people vote and voter turnout should be higher where people have access to media.The model indicates that if:

xi(uc)(zc) – βi η > ui

then the incumbent will be reelected. X is 1 if the population knows that something has been done for them. U is the utility they receive from the amount of spending Z. Beta is the ideological preference for the challenger, and Eta is the general popularity of the other candidate.

The governor knows that the voter will vote with some probability t and that the voter knows of his responsibility for the relief programme with some probability α that is an increasing function of r (radio coverage)

This generates the following propositions:

  1. If the voters cannot know if money is spent in their county or not (x = 0) then the politician has no incentive to spend there, as he will get no political credit for doing so.
  2. If Beta is distributed such that the ideological preference for the challenger is such that the incumbent cannot win, then he will not spend in that county as he will not be reelected in any case.
  3. He will allocate more funds where there are more gains to be had on the margin i.e. where turnout is higher, and there are more radios, there are more swing voters and the need for relief spending is high (where Uc is high).


Data 2500 US counties in panel from 1920-1940. This was in the middle of radio expansion, and also during the FERA programme which distributed funds to those whose income was inadequate to meet their needs. It was locally administered and local officials decided who would and would not receive the assistance. Governors were the main arbiters.

Ln(zc) = αln(rc) + βln(tc) + δ1xc1 +εic

State specific fixed effects are also included and standard errors are clustered within state. The main hypothesis is that α>0


Results Factors indicating low socio-economic status are positively correlated with spending indicating that income assistance was directed to places where utility was likely to be highest (i.e. where they needed it most).The elasticity of spending with regard to radio ownership imply that increasing radio coverage by 1% would raise spending by 0.54%, and increasing turnout by 1% increases spending by 0.57%. The most important explanatory variable is unemployment which indicates that this was not just pork barrel politics, but that spending was directed where it was needed.

If radio use increases turnout, and turnout increases spending, then this is another mechanism through which radio is working. A fixed effects PANEL regression is estimated with turnout as dependent variable, with a host of controls. The coefficient on radio coverage is 0.117 and significant at 5% levels. Thus increasing radio coverage by 10% would increase turnout by 1.2%. Since every increase of 1% of turnout increases spending by 0.57% then the effect of radio on spending through turnout is 0.12*0.57 = 0.07%.


Robustness It is recognized that there could be bias in the estimates. Specifically, if richer counties (not otherwise captured by controls) have lots of radios, but no need of assistance then results would be downward biased. But it more people seek out radio ownership and are also better at getting their preferred policies, then this would create upward bias. In recognition of this, he implements an IV strategy, which uses geological features ground conductivity and woodland cover as instruments for radio ownership (as these variables both affect the quality of the received signal). The F-stats in the first stage are all strong. Exogeneity might be questioned, as geological features especially wood cover could be correlated with poverty or exclusion and hence relief spending which would downward bias the IV estimates. However, despite these concerns the IV results are actually more positive than the OLS results. As the author therefore takes the OLS results as his baseline, the IV just indicates the direction of the bias (i.e. people seek out radios who are better at getting what they want), and as such the main results of the paper are conservative, and this lends credence to the story.Property values, employment stats, income, wages, bank deposits etc. are all controlled for as well as share of votes in last election, voter density etc.

If the model is correct there should be more spending where elections were more closely fought. This is tested by excluding noncompetitive states, and the coefficients are nearly twice as large.

The effects should be larger in rural areas, as urban dwellers had better access to other types of media. When the specification is tested on a rural subsample the coefficient increases nearly 50%

If radio use is simply proxying for some other variable relating to the use of consumer durables then we should see similar results for other durables e.g. car ownership. Indeed gasoline sales are shown to have correlations with wages, employment etc. (just as radio does), but gasoline sales per capita are not related to spending in regressions.


  • This is a cross section, with data being pooled cumulatively. Panel data would have been ideal as we could see how outcomes changed with increased radio penetration, and particularly if funds are limited, then as radio coverage becomes near universal the limited pot of FERA funds may not be significant enough for use for political capital in all counties covered by radio. This would be akin to a general equilibrium effect. Panel data would have allowed.
  • Sadly no interaction terms are used. For example an interaction term between unemployment and radio coverage could have given an estimate of the differential effect that radio coverage has in the presence of a given level of unemployment, or need. This would have been interesting to see, as the levels estimates are not as readily intelligible.


Implications Mass media can carry politically relevant information to voters who can then use this to update their voting positions. This can make politicians more accountable as people are more likely to vote.Simply extending the franchise to the poor is not enough as this paper makes clear. What is important is how informed people are, for if certain sections are not informed as to the spending policies of the government, then such spending may be cut without fear of losing votes, and redirected to areas that may have less objective need for the spending.

As the inclusion of welfare indicators made the estimates stronger it seems clear that spending was not just directed at those who were rich enough to own radios.

The bottom line is that radio improved the relative ability of rural America to attract government transfers.





R. Chattopadhyay & E. Duflo

Econometrica, Vol. 72, No. 5 (2004) pp. 1409 – 1443

A Very Short Summary 

In a Nutshell

As women are universally underrepresented in politics, at times political reservations are made for women. There is some evidence that men and women have different policy preferences. Whether these will translate into different policy outcomes is not at all obvious, as one of the implications of the median voter theorem is that the identity of the politician is not as important and the preferences of the median voter.

Using the fact that India decided that 30% of seats and chairs had to be for women in local councils (and the councils subject to the law were made so at random), the paper looks to see if there are effects of reserving seats for women, by looking at the provision of public goods within the regions that had a female head of council. The results indicate that gender does influence policy decisions and hence the identity of a decision maker does influence policy. In particular, in West Bengal, female populations complain more about drinking water and roads, and spending on those goods is differentially higher in those regions that have a female head of council. In Rajasthan, women complain more about water, and less about roads, and this too is reflected in public good spending. Thus it seems that policy outcomes are closer to what women want than what men want. Women prefer programmes that increase their opportunities (better water means less time on housework etc.).

Other results indicate that a female head increases female participation in the council meetings, and makes women more likely to complain i.e. it increases the role of women in the political process. The reason that the results are more in line with what women want is because their preferences are more closely aligned with the personal preferences of the leader.

The paper concludes that leaders under the reservation policy invest more closely with the interests of general female concern, and these results are robust to inclusion of the identity of the leader and observable community variables. This implies, that whilst democracy may be an important way to make government accountable to the citizens, the identity of the leaders is of importance. In particularly if there are cultural or other norms that tend to prevent women from being politicians, it is possible that the interests of women will be substantially overlooked, and this is not fair given that they represent 50% of the population. Thus, along with free media as advocated by Besley and Burgess, encouraging participation by women in politics may be a means of increasing the accountability of government which in turn may affect socio-economic outcomes (such as infant mortality as in the Africa paper summarized above).