Blog

Mind of an Economist (10)

The rights of a state

Friday, 12/16/2011, at 3:38 PM
What gives the right to a nation to tell its citizens how to behave? Is it to promote peace and prosperity? Is it to maintain its control over the masses? States were formed to protect against outside invasion. When outside invasion is less likely, why does the state persist in its rights over the masses? These rights can be individual-oriented. For example, individuals are punished if they attempt to end their life or in some countries, take away the right to abort. In theocratic countries, individuals have to follow a particular religion and the state takes away the right to convert to another. What gives a national government the right to acknowledge birth, marriage and death thereby giving them legitimacy? Possible rationalist explanations –

INERTIA
Inertia resulting from history of protection. This is a behavioural explanation because it implies that individuals are used to respecting the state because it has been in existence for a long time and individuals can be brainwashed into nationalism either by the school curriculum or by their brainwashed parents.

SPILLOVERS
Externalities arising on other population members – come up with an institution that is on average concerned with the average person and will exercise control in order to promote the well-being of the average person. Why will an individual not expect a back-lash and thereby not conform to the right behaviour on his/her own? Why do we need gun laws for instance? 

IDEOLOGY
Bundling of the state with religion. Religion is propagated by the state in order to enforce its legitimacy of rights over the individual. It is easier to question the state but not easy to question religion. Can stateand religion together prevent individuals from taking their own decisions? 

 IGNORANCE
If individuals are ignorant of the right actions to take such that they may harm others unintentionally, then the state may have a rationale for having a monopoly over individual rights. This is a very strong assumption as it means that government is somehow made of special people who are less ignorant compared to the masses. Even if individuals are ignorant, why not allow their relatives and friends to make them enlightened. 

STATISTICS
Legitimacy of government’s monopoly over individual rights may be a bi-product of the necessity of a government to collect national statistics so that it can exercise dominance in other areas – for instance, collection of taxes.

INCENTIVES
It may incentivize the masses to try to achieve power and may drive them towards gaining authority. If the government did not have control of our rights, there would be no incentive to join the government except the low fixed wages.

CO-ORDINATION
It may act as a co-ordination device for everyone to believe that there is a unitary right and there cannot be heterogeneity of rights.

One of the most enthusiastic students in Econ 111, Andrew Edelman, adds the following to this category:"Multicultural co-unity: These rules allow for a mix of cultures to exist together when they would otherwise conflict.  If you look at countries that were formed through European colonialism, there have both weak governments and, in many cases, tribal conflicts.  In countries such as the United States, England, and even Russia, people with many different nationalities live together under a set of rules separate from their own cultural identities which minimizes conflict."

A related point is that a multiplicity of identities (national, regional, religious) dilutes original identities increasing the possibility of inter-ethnic complementarities.

SIGNAL
Government acknowledgements may be a way to signal high-types from low-types. This is dependent on a common belief in the system that the government can exercise credible and effective control over our rights.

EFFECT ON OTHER PUBLIC INSTITUTIONS
If a lack of control over one’s behaviour can affect other state institutions, for example, the quality of state education or health-care, then there may be an argument for the government to monopolize rights.

David Beron, another excellent student of mine states that this could be related to Michel Foucault’s ideas on Humans in terms of numbers: there is a maximum population at which society can efficiently function (considering diminishing marginal product). If the ideology of people fails to control such matters (religion might be an incentive to procreate), the government needs to intervene. Because no invisible hand exists to match people’s non-capital desires with their nation’s “carrying capacity”, the government has to set limits to the "number of slices" in order for "the cake" to grow as a whole. China can be used as an example: people's behavior is restricted by the one child policy and, perhaps a coincidence, the country is currently going through an economic boom.
 
The above explanations are not mutually exclusive. These reasons are likely to exist in some combination and it would be difficult to differentiate the relative importance of each mechanism.


Thursday, 11/17/2011, at 4:38 PM

Notes from Yale (12-13 November, 2011)


Yale has a campus of impressive buildings with Victorian and Gothic architecture and at this time of the year the flaming fires of yellow and red in the leaves made the rustic buildings look even more beautiful. There was a crowd of Development Economists gathered at the NEUDC conference and I was presenting in the first session titled: Health – Incentives. There were 4 papers being presented in each session and many sessions were happening in parallel.

Sylvia et al. (all papers that I mention here are 2011): they provide incentives for anemia reduction and in the abstract mention that it is the first study to see the performance pay impact in health – perhaps overstating their case! They have 3 treatments and a control group – in the first, they provide information to the principal of schools, the second involves information and an unconditional subsidy and third, information, subsidy and incentive. However, because they are combining these together, it is impossible to evaluate the marginal impact of only the subsidy or the incentive. This can also be described as Oriana’s Critique as it was Oriana who had first pointed this out to me when we were deciding on our 3 treatments. My treatments were information (to the demand-side – here it’s to the supply-side), incentives and info+incentives. This allows us to deconstruct the individual impacts as well as their interactive effect. Interesting paper nonetheless, finds that subsidies may have undermined the combined effect of the incentive and information. This is pretty close to what I find! I should perhaps cite this paper once it becomes publicly available. One quote that I liked in this paper: “He who cuts off two heads, gets two degrees in promotion.” - Chinese General, Shang Yang (390-338 BCE)

Heather Sarsons (Harvard) presented a paper on whether rainfall can be used as a valid instrument for income shocks when looking at civil conflict a la Miguel, Satyanath and Sergenti’s important paper in the JPE. She finds that it depends on whether the rainfall happens in the districts upstream or downstream to dams. She combines the Rainfall data from the climatic center [Center for Climatic Research, University of Delaware], Varshy-Wilkinson’s popular data on Hindu-Muslim riots in India and Dams data by Duflo and Pande.

Oliver (LSE) who was my classmate and is on the market this year, presented his Job Market Paper on Naxalite violence and how it is related to both rainfall as well as shocks in prices of natural resources. He finds that attacks are more targeted towards security forces during good times in more resource rich regions so that the Naxalites can increase their exogenous tax base. If times are bad, civilians are bound to get drawn regardless of the attacks – so Naxalites attack both civilians and the government. He has a model to justify the effects. There are issues with identification of the model as well as not to model the participation decision for locals to join the Naxalite forces (like Weinstein’s models) is a shortcoming – but a very well done paper overall.

Another paper (Walker et al) studied the impact of access to AIDS treatment on employment outcomes in South Africa where HIV rates are 17% and blacks make up 80% of the infected population. The unemployment rate is 25-30%. ARV (anti-retro viral) treatment is very common for HIV/AIDS and has been shown to increase the life expectancy to close to 70 years in developed countries – in South Africa it was expected to increase life expectancy by 7 years. Of course the paper finds that employment went up due to access to this treatment as at a particular moment, some had access and others did not. However, the paper fails to consider the general equilibrium effects of the increase in access to ARV – in particular the impact on wages of those who did not have access or those who were not HIV. This matters for welfare implications.

Ozier (World Bank) finds evidence for long-term effects of externalities of deworming in Busia, Kenya (done by Kremer and Miguel) on Cognitive effects. This paper was really interesting as it used several methods to get at cognitive effects of the spillover population. In particular, it used Raven’s matrices (measures fluidity of intelligence and is culture-neutral), verbal and math vocabulary, verbal fluency through pictures of foods and animals, Peabody tests, memory digit span forwards and backwards. It finds an interesting effect – there are no long-term health benefits of externalities from the deworming treatment but there are long term effects on cognition on those who were very young when their neighbors got dewormed. Could this be pointing towards compensatory [crowding out] behavior by parents in health but not in education? This paper also fits into the classic economics mould: Senator Ron Paul often cites Economic in One Lesson by Henry Hazlitt–which argues that economics should care about the long term and not the short term. It should also recover the impact on all population members not just those who are directly affected. This paper is very clean as it takes the advantage of a randomized experiment. It also deals with the early child health intervention literature that has long-term effects on neighbors and on an outcome that was not directly the target of the intervention. Beautiful. My favorite paper at the conference. I also had a talk with him later and he explained to me how the cognitive tests worked. He recommended to me work by Sophie (?) on pre-school program intervention in Mozambique with Save the Children NGO and they use a Bayley index for cognitive tests for 2-4 years. Playing games, pictures, etc. are the usual way of testing kids on cognition.

Günther Fink presented his paper on health and agricultural productivity in Zambia through an RCT design looking at ITNs (insectiside-treated bed nets) and the removal of credit constraints. One net improves harvest value by 10% - then why is the adoption so low? Could it lack of information – no according to the author as they knew about how mosquitoes can cause malaria and it’s ill effect on productivity. Could it just be credit constraints? (Udry and Anagol, 2008)? Could it be hyperbolic discounting (Duflo/Kremer)? Risk aversion story is possible but not probable. Bed nets may be a peculiar technology anyway!

Another paper looked at the impact of mother’s education on infant health using Duflo’s idea and setting – Indonesia. However, closure of schools did not lead to a fall in infant health. Not the most original or interesting paper but the methodology is diff-in-diff. The paper is also subject to the Angrist and Krueger’s quarter of birth critique (1992).

Gato, Aida, Aoyagi and Sawada (Univ. of Tokyo) examine individual piece rates (IPR) versus group piece rates (GPR) in a field experiment (and lab experiments within field experiment). They find positive incentive effects in IPR, free-riding in GPR and moral hazard in FW as expected. Introducing monetary incentives seems to crowd out intrinsic motivation in their experiment. They are able to estimate parameters on altruism and guilt aversion, which I thought was neat to isolate the behavioral channels.

Gil Shapira’s (UPenn) work on AIDS in Malawi showed that the actual likelihood of getting HIV from infected partner is much lower than what people think. Achyuta Adhvaryu (Yale) tested how misdiagnosis leads to a change in the speed of learning and found that in Tanzania, misdiagnosis for malaria can reduce learning and adoption.


One of the more interesting papers was by Nicholas Wilson (Williams), who studied if there was an overcompensation effect of circumcision in men in a RCT in Zambia. Peltzman discovered the overcompensation effect of seat-belt rule (makes people drive faster increasing the risk of accident). The hypothesis was that circumcision (which reduces the HIV/AIDS risk by 60%) may increase risky sexual behavior inhibiting the impact of circumcision. What he found was the opposite. Circumcision reduced the likelihood of engaging in risky sexual behavior. There may be two channels for this effect: first, belief based and second, non-belief based. This appears to be going through a salience effect (a non-belief based) channel as circumcision increases the salience of tradeoff. One issue here may be a justification effect (real or artificial) that kicks in to justify the original action of getting treated.

Subha Mani (Vassar College) is studying the impact of providing vocational education on earnings in Delhi. This can be replicated in Punjab and she showed interest in collaboration. She also knows Olga Shemyakina who works on conflict and has written a paper on gender, schooling and conflict in Tajikistan. She is also working on a paper with Tarun Jain (Indian School of Business) who has a paper on the impact of languages on economic growth that exploits remarking of borders along linguistic lines. He invited me for a talk at ISB, Hyderabad when I was in India.

John Strauss, a famous health economist at USC, gave a talk about his work on height shrinkage in the old population and how if this shrinkage is not taken into account, can lead to an attenuation bias of the impact of height on earnings. On average, there is shrinkage of 3.4 cm in females over 70 years and men shrink 1.9 cm once they are old. He used the China Longitudinal Healthy Longevity Study and China Health and Retirement Study. Also literates and rich shrink less. Pre-shrinkage height matters more for men.

Cas et al. find long-term effects of early life public health intervention of a suffent increase in the availability of village mid-wives (like Anganwadi workers/Ashas) on adolescent cognition using the best survey data out there – Indonesian longitudinal study. Can address concerns with mother-fixed effects. Uses Duflo-type Diff-in-diff for treated and untreated cohorts.

A Yale PhD candidate, Yaniv Stopnitzky, analyzed the recent advertising campaign in rural Haryana on “Marry only if the boy has a latrine” at home. He does a Diff-in-Diff-Diff (Triple Diff) to get at the effect comparing families with marriageable boys after and before the campaign in Haryana relative to other parts of India. He uses the NFHS rounds in 2004 and 2008 and finds a significant increase in the latrine usage among families with marriageable boys! Fascinating. He also looks at the differential impact of this latrine adoption among families located in districts with a more skewed sex ratio in Haryana. He finds that wherever the demand for girls exceeds the supply the most, there the families are much more likely to adopt latrines. I asked him to also study the effects of different demographics (one boy, one girl versus two boys) as well as girls of marriageable age. Also he could try out different qualities of latrines (from pit to flush).

The plenary session at Yale was completely packed. Mark Rosenzweig (Yale) introduced the speakers. First up was Alain de Janvry (UC Berkeley) who spoke about how the discipline of development economics had evolved since Hirschman and Lewis who were the post-world war 2 pioneers. The second phase of development economics focused on growth, Washington consensus and millennium development goals and finally, we are beginning to see the dichotomies between small game and big game, formal versus informal, cottage and small versus large enterprises, self employment versus wage labor, microfinance versus commercial, transfers versus autonomous incomes, behavioral constraints versus profitability and risk. Next up was Angus Deaton (Princeton) who has written some of the best textbooks in health economics and household surveys. He said that we have a better description of the world than what we had in the past – better material and health maps as well as better micro data from several developing countries in the world. DHS has 88 surveys from 27 countries freely available online. Gallup wants to now survey 1000 individuals randomly from across the world every day! In macro, the process started with the Heston-Summers paper, but we are moving closer to linking micro and macro, yet there exists a discrepancy. 30-40% of the GDP is imputed and not actually what it is – there is a considerable shortfall of measurement knowledge now. Economists do not know or care about PPP numbers and how they should be reformed. Happiness economics needs to be integrated into Development economics. In the earlier era, Chris Udry’s field studies in Northern Nigeria were shocking (as he wasn’t even tenured). Now, field experiments are being undertaken by graduate students. In 1980, there was no STATA! Deaton had to write down a program to calculate the coefficients (probably an overstatement). Two aspects of development are emerging to be very important. One is the relevance of History and second is Behavioral Economics. However, Deaton was pessimistic about economists’ role in contributing towards economic development thus far, and argued that this needed to change as well.

The third speaker was Abhijeet Banerjee (MIT) who talked about the earlier paradigms in development: workers do not save (Kaldor), farmers do not migrate (Lewis), poor people are unproductive (Leibenstein, Mazumdar), there were too many babies and theory revolved around co-ordination failures (Rosenstein-Rodam, Nurkse, Hieschman). Starting in the 1960s, Becker, Stiglitz, Schultz, Behrman, Bardhan, Srinivasan, Rosenzweig, brought development into mainstream economics by assuming that people maximize utility and firms maximize profits. Moreover, markets tend to clear even though there may be market failures. According to Banerjee, 1980s was the worst decade for development, mostly dealing with macro stability and exchange rate “fix-all” treatments. In the 1990s there were dynamic models built around market failure (Banerjee) as well as detailed micro-level analysis (Udry, Townsend and Deaton). This evolved into two fundamental approaches: one dealing with firms and politicians and the other with consumers and votes. Nick Bloom now has questioned the profit-maximizing assumption of firms. But the biggest revolution has been in the way consumers are prone to making irrational decisions (as written in his book with Duflo). Banerjee left the tone on a pessimistic note asking how theory should be constructed? If there ever could be an overarching theory or if we will simply cut ourselves to behavioral rules for every technology, place and time.

Finally, Daron Acemoglu (MIT) spoke about the role of economic vs. politics in economic development and how institutions are formed and how they evolve. However, his view was more optimistic and he stressed that there had been a convergence in development economics. Almost everyone now recognizes that institutions, policies, endowments, social norms, “culture”, geographic constraints all matter a great deal for development.

 Two papers that I want to write more critically about though I like them both are one by Locatelli et al. (UCL) and the other by Solis (UC Berkeley). The first paper studies the crowding out effect of a randomized intervention of using IRS (Indoor Residual Spraying) for removing mosquitoes and preventing malaria in Eritrea. This is a companion paper to the one looking at the impact of treatment. The authors find no crowding out effect on buying insecticide treated bed-nets. If it all, the effect was the opposite on beliefs (salience effect). The control group was not pure. Everyone in the sample had access to ITN’s and LHS – both are anti-malarial household treatments. However, crowding out may occur with diminishing returns to increasing the number of interventions. This may be one reason why there is no impact. The main problem was that it was not a diff-in-diff making it very difficult to analyze the causal impact. The trends in beliefs may be different in treatment and control regions leading to a spurious significant difference in beliefs between the treated and the control post-intervention. The authors also did not specify empirically whether the salience effect was due to a technological complementarity between ITN and IRS or due to a behavioral channel, which they assume to be the case. Finally, crowding-out is not a specific phenomenon to health – it has been observed in several educational studies too (Bonesronning, etc.), and was perhaps missed by the authors. There also appeared to be a selection bias as 25% of the households were not provided with IRS in the treated villages. This selection bias plays into the pre-treatment beliefs being different in the treated and control villages.

The paper by Alex Solis was on the impact of removal of credit constraints for higher education on higher educational inequality between rich and poor. He looked at Chile with a Regression Discontinuity Design where loans were given out to those who asked with a minimum score threshold of 750. This formed the basis of the regression discontinuity design. He uses the Imbens and Lemieux (2008) linear spline regression and finds significant decline in educational inequality due to this policy between those who managed to score just more than 750 relative to those who were just below. This raises some concerns: first, if unobservables drive both the performance over 750 as well as the ability or likelihood of borrowing money (for instance, if they are more driven or are less likely to procrastinate), then there may be a bias. Second, the role of expectations: if people were not expecting this policy, then this could be compared to an exogenous intervention. Expectations may also be interacting with unobservables here. There was a conflicting finding – higher GPA scores seem to have access to more funding and poorer students scored less than richer students. He later told me that this was because greater scores opened up a higher likelihood of being admitted to the course that matched the demand of the student and this increased the probability of getting a loan. Third, half the loans were not repaid – but this was not factored into calculating the benefit-cost ratio of the program – also the cost of capital was not factored.  Fourth, there may have been supply-side changes that took place simultaneously – increase in school size, quality of teachers, their wages, etc. so it is impossible to rule out these interaction effects with the threshold rule in affecting the selection bias. Finally, to assess welfare implications, where could the money have been better used? This question is commonly evaded in a partial equilibrium framework but is very important.

Overall, I came back to Amherst feeling enriched both intellectually as well as socially. It has also spurred a spirit in me to write more (and better).

[Correction: Prof. Walter Nicholson adds that regression analysis was common in 1980. Bob Hall (his classmate) was one of the inventors of TSP which later became E-Views. SPSS had existed way back in 1971 and economists such as Walter, Griliches, Jan Tinbergen, and Yale's Cowles Foundation members regularly used regression analysis in their papers. He also cites a Cowles Foundation monograph from 1953 which deals with causality in regressions: http://cowles.econ.yale.edu/P/cm/m14/index.htm]

Institutional successes of the modern world

Tuesday, 9/27/2011, at 10:05 PM

What are the great institutional successes of the modern world? Democracy, rule of law, fiscal policy and monetary policy. How do we know these have been successes? Rules or constraints at the level of the state have been followed by individuals to co-ordinate on an equilibrium such that there is little uncertainty and more opportunities for growth. There is greater power for the individual in a constrained political framework where the executive has constraints on the power and repression. Where the individuals can compete to become leaders and have information about the political parties which compete in an open and transparent manner. These measures are hallmarks of a state that does not have to tinker with exogenous rules to maintain growth. The changes to rules come from within. However, these are examples of political institutions. There can be institutions that work to promote law and order, to collect taxes and to affect the efficiency and equity of the banking system. What I would like to talk more about are the institutions governing the expenditure of taxes. These can in the extreme case go to the leader’s pocket if there are no restrictions – either constitutional (in terms of getting re-elected) or de facto power that the leader exerts by virtue of the weakness of other institutions – such as that of independence of judiciary or media. This is the case in many African countries (like Zimbabwe). How do these institutions change to bring about better targeting of expenditures. To reduce poverty, to improve the quality of life and to bring peace and prosperity into the lives of the national population. Does change come from external sources like conditional aid? Does it come from within – through revolutions and coups? Does it come with time naturally?
Now take a developing country with fairly strong democratic traditions (like India). Here, how should expenditures be targeted. Of course, health and education would take precedence. Within health, how should health workers be compensated? How much to spend on new equipment? How much should be spent on building more hospitals and improving access? These are institutions that are again linked to other political and economic institutions (for example, democracy and how effective the tax system is or if law and order is a problem). There can be failures at this level of institutions if the targeting of expenditures does not have a desired impact. These failures can be sustained if there is a lack of knowledge on the demand side (for example, the population) or even if there is no lack of knowledge – there could be problems of collective action. On the other hand, there could be a supply-side problem too. Here, the workers may not be motivated enough by what the government is paying them and they end up demotivated and reduce their quantity and quality of effort. Successful fixing of such institutions can then mean making workers motivated and improving knowledge among the beneficiaries or by solving collective action through community-based interventions. An example of solving both demand-side and supply-side problems is the microfinance model a la Grameen Bank. Here supply-side problems of adverse selection and moral hazard that lead to high non-performing assets are solved by group credit (social capital is the collateral) and also promising more credit if the loan is repaid. The demand-side problem of inadequate information (about saving, financial instruments, investment opportunities) as well as self-control issues are addressed through regular meetings with stakeholders. Another example could be paying performance pay to child health workers and at the same time educating mothers about health and nutrition. If only performance pay is offered to workers and they improve their inputs, there could be an endogenous feedback such that mothers may reduce their feeding to the children. Similarly, workers may reduce their efforts if mothers increase the inputs at home. Incentivizing workers and mothers to improve their inputs would fix both types of market failures simultaneously and help sustain an institutional equilibrium that improves lives of both health workers and beneficiaries without any negative spillover effects on political and economic institutions.

Link between Agri-growth and malnutrition?

Tuesday, 9/27/2011, at 9:58 PM

In their Economic Times article, ‘Agri-growth and malnutrition’, Gulati, Kumar and Shreedhar put forth an interesting hypothesis: agricultural growth is necessary for a decline in child malnutrition in India. The authors proxy agricultural growth (which is quite volatile) with land productivity and estimate a strong negative correlation between agri-growth and malnutrition. They argue that this implies agri-growth is a necessary condition for combating malnutrition. However, their reasoning is notoriously flawed. First, it is highly contentious if land productivity can be taken as a good proxy for agri-growth. Agri-growth may depend on a state’s institutions – for instance, trading opportunities, law and order, the banking system, quality of governance, access to technological innovation and so on. A state’s institutions are very likely to be correlated with land productivity. Richer countries also tend to have more favourable climates and more fertile lands. This may be driven historically, by higher population growth in fertile areas which fuels more innovation or by the government having better fiscal capacity. In any case, better geographic and soil conditions do seem to lead to a more “developed” state on average. This means that some states are born “lucky” and others are not. If this is the case, it is no surprise we find a negative correlation between land productivity and malnutrition. The real result may be that some states that have better soil conditions tend to have better institutions and also lower malnutrition. It is quite another question, then, to ask what makes better institutions apart from luck.
It is not easy to find determinants of malnutrition. However, the only way to try to unbundle the causes is not to take a state-level approach but a micro-level study. At the level of a city or a village, it may be possible to see what is truly correlated with malnutrition, given the same access to technology, trading and credit opportunities, etc. I study malnutrition among five thousand children (aged 3-6) enrolled in Anganwadis in slums and find that diet, family income, mother’s expenditure on food, mother’s nutritional knowledge, education and other fixed local environmental factors (like prices of products, local water quality, quality of Anganwadi, etc.) all together contribute in explaining only 8% of the variation in child malnutrition this year. This means that child malnutrition among 3-6 year olds may be difficult to predict even given detailed information about their local environment let alone land productivities at the state-level.
There are at least two possibilities: it could be driven through genetic factors – underweight mothers on average have underweight babies and so on. It could also be the knowledge of the mother at the time of child-birth. Several medical studies point to moderate and severe malnutrition being linked to malnutrition within the first two years of birth. It could be dietary habits or the history of immunization and disease during the formative years. If this is the case, it would be futile to improve agricultural growth as it is likely to fail to lead to better outcomes for health. In order to make policy implications, one must make statements that are micro-founded at the level of the family and not the state. It is a common blunder to adhere to state-correlations without digging deeper. Child malnutrition is a complex and difficult problem. It demands a more rigorous solution.        

Do CCTV's deter terrorists?

Tuesday, 9/27/2011, at 9:27 PM
There has been a surge in the amount of low-intensity warfare as a tool to propagate terror. The obvious question to ask is: why has there been a spate of low-intensity bombings in local densely populated areas as opposed to say, the high-intensity bombing of a strategic location?
First, the cost of planning a low-intensity operation is much lower. Second, the cost of executing it by placing ammonium nitrate inside a tiffin box atop a cycle is also very small. Third, the benefit gained by the terrorists through media coverage of the destruction and accompanying terror is relatively significant and contributes to creating distrust between communities. Fourth, the extremely low risk of getting caught makes engaging in terrorism more lucrative for a few local people and organisations funding such activities.
This means that there are broadly only three ways in which we can reduce bombings in low-intensity attacks and not compromise the security of strategic locations:
  • To raise the costs for terrorists of planning and executing low-intensity attacks.
  • To reduce the “benefit” the terrorists obtain from such attacks.
  • To increase their risk of getting caught.
Figure 1 illustrates the equilibrium number of bombings terrorists execute. The two black diagonal lines represent marginal cost and marginal benefit curves, i.e., the planned cost or benefit of bombing one more location in a single terrorist attack. Equilibrium is reached where these two curves intersect. Here, implementing (a) would mean a parallel upward shift of the marginal cost curve and implementing (b) would lead to a shift of the marginal benefit curve leftward. These have not been shown on the diagram. Acting on (a) and (b) by interfering with efficient market mechanisms like putting controls on use of fertilizers or on free media is not advisable as it can have negative spillovers and inefficiency losses.
The only convincing way of dealing with terrorism of this nature is to increase a terrorist’s risk of getting caught on camera in at least one of the locations. This not only raises the costs of planning an attack (making the curve move upward), but also makes the marginal cost curve steeper than before as chances of getting caught increase like an ad valorem tax. The marginal cost curve now becomes the red line and the equilibrium number of bombings carried out in one attack comes down dramatically. By exponentially raising the costs of carrying out one extra bombing, two effects may take place. First, there would be a decline in the number of bombings per terrorist incident (as shown in the graph). Second, there would be a greater incentive to go in for one high-intensity blast at a strategic location. This would work like a substitution effect, whereby a worker tends to reduce the amount of work if marginal tax rate increases. However, it would be a preferable situation for India if we consider this a zero-sum game, whereby raising the costs for terrorists works in the same manner as increasing benefits for India.
Installing Closed Circuit Television (CCTV) cameras in susceptible places like markets, shrines, hospitals and colleges would help garner information on terrorists, should they decide to attack. These cameras may be visible or hidden. These can also be used to detect traffic violations, violence, drug abuse, shoplifting and property theft. Britain had started using CCTV cameras to combat IRA terrorism during the 1970’s and today has the highest amount of camera surveillance in the world. In London, there is a good chance that one will be captured by over 300 cameras in a single day. The only dilemma facing the authorities should be whether to install fixed cameras or Wi-fi enabled cameras. The latter are more expensive, but give the option of higher mobility and thus may indeed be the better option since terrorism seems to be a long-run issue. There are also costs of maintenance and staff to monitor the pictures that must be taken into account in any cost-benefit analysis.
Although studies show a decline in vandalism in London buses and reduced robberies in London Underground stations when they are equipped with CCTV cameras, there is very little quality evidence to suggest the extent of their achievement. This is mainly because of the problem of lack of randomisation when the cameras are installed.
For instance, in order to quantify the degree of success of a drug, a pharmaceutical company will undertake randomised trials, in which a random sample of people is given the drug and others are not. Everything else remains the same for these two groups. This is helpful in finding the true extent of potency of a drug. Similarly, to find out if smaller class groups increase learning ability, a randomised solution has been carried out by economists in Israel. In another experiment, a bank in South Africa lent money at different rates to randomly selected individuals, in order to find its effect on non-performing assets.
Therefore, in order to judge the efficacy of cameras, the police would need to install these cameras at pre-planned locations in randomly selected cities (“treated” cities). The difference in terrorism and crime between the treated and the control cities would be monitored. If the difference changes significantly after the introduction of cameras, it would imply that surveillance is having an effect and, more importantly, the quantitative effect can be found. Furthermore, dummy cameras (which are readily available online) can be used in half the cities and real cameras in others to distinguish the effect of deterrence from just the perception of being under-watch and deterrence from actual catching and punishment.
The problem lies in implementing such an experiment on a large enough scale. In order to give policy recommendations, it is of great importance to delineate causal impacts of policies. Carrying out randomised social experiments will give a fillip to finding out “clean” effects of surveillance on terrorist activities.
Generally the police would like to set these cameras in cities where crime is expected to be high, but this would rule out a way to find the causal impact of CCTV on crime as there would be a selection bias. Moreover, it would be a multi-pronged approach (increased policing, hiring more personnel, etc.) to tackling crime and terrorism, and thus to isolate the impact of CCTV on crime would be impossible.
CCTV is also likely to change the behaviour of people who are not potential offenders. In a survey carried out in Germany to assess perceptions of people to CCTV, it was found that a majority of people like to have cameras installed in malls, banks, railway platforms, and along motorways. However, there is displeasure at placing them at the entrance of residential buildings, in public toilets and in changing or dressing rooms. There is a general concern that being constantly under inspection will adversely affect civil liberties.
Between 1996 and 1998, more than three quarters of the UK government spending on crime prevention went towards CCTV installation and monitoring. It has been shown that effects of CCTV on deterrence are short-lived and not persistent. For instance, the underlying patterns of car thefts in British cities re-emerged after three quarters of low-crime.
The Indian police are waking up to the yawning gap in their surveillance infrastructure. In a recent news article, it was predicted that by 2010, all NCR towns of Haryana would come under hi-tech CCTV surveillance through Wi-fi, hi-band and fiber based surveillance systems covering all major sensitive locations such as the expressway, malls and MNC offices, bus stands and railway stations. Even though there are no randomised studies on the benefits of installing these and several concerns on their use remain, there seems to be a consensus emerging on advantages of having “someone watching over us”.

Framing of a title

Tuesday, 9/27/2011, at 9:21 PM
Think of answers to the following questions:

1. Would you rather have:
a) 99% pure drinking water or
b) Drinking water with 1% impurities

2. Would you rather have?
a) A bird-flu vaccine that is able to save 90% of the lives or
b) 10% of the population will die despite the bird-flu vaccine

3. Would you rather have:
a) Buy 1 + Get 1 free or
b) Buy 2 + Get 50 % off

In all these cases, most people would choose a) even though both options are equivalent. This psychological phenomenon is called “Framing Effect” whereby altering the structures of sentences can lead us to prefer one to another. Neuro-scientists think that there are sections in the brain responsible that help emotions overrule rational decisions leading to framing effects. This has lead many researchers to believe that some people may be more susceptible to framing effects than others. MRI scans on the brain reveal that these may be caused by a section of almond-shaped neurons called amygdalae.

Framing biases affect investing, lending, borrowing decisions and are used by advertisements to capture the attention of a potential consumer. They are used by politicians and economists to mask a snail-slow economy in seemingly robust terms. They are used by students to show off their achievements on the curriculum vitae. By Sainsburys to sell Orange juice and the 'Orange juice' to (mis-)communicate the amount of calories or sugar it has. You may call it a play on the numbers or the words, but it is a combination of amygdalae doing their work and firms maximizing profits.

Prospect theory in economics explains the framing effect by saying that people weigh losses more heavily than equivalent gains and this leads to a bias in favour of the 'positively-framed' statements. In our case, this would correspond to choices a on all the three questions. This revelation puts framing effects on the forefront of Behavioral Economics.

So next time you see John F. Kennedy's famous quote: “Ask not what your country can do for you, but ask, what you can do for your country,” think of framing effects!

For you I have the following question: On a sunny Sunday afternoon, you want to play cricket. So you go to Lillywhites and ask them for a bat and a ball. The salesman says, “The bat costs £1 more than the ball and they together cost £1.10.” Can you guess the price of the ball?

Role of London School of Economics in Indian History

Tuesday, 9/27/2011, at 9:18 PM
The London School of Economics is not merely a congested cluster of buildings in the heart of London. It has also produced leaders who have gone on to shape the lives of millions of people.


When India became independent, the LSE-educated Bhimrao Ambedkar wrote India’s constitution and later went on to champion the rights of untouchables. His bronze bust stands at the entrance of Clement House at LSE. Another figure from LSE who made a deep impact on post-partitioned India was Sardar Tarlok Singh. He was the Director General of Rehabilitation in the Punjab responsible for the resettlement of over ten million refugees. Making two key innovations, he made the biggest refugee resettlement operation a resounding success.

India’s first Prime Minister, Jawaharlal Nehru chose Krishna Menon (from LSE) as a key advisor and India’s defence minister. Menon, who had earlier been India’s first Ambassador to the UK, helped shape the mindset of India that was suspicious of Pakistan but a supporter of socialism. In 1962, China’s victory in a war against an ill-prepared India forced Menon to resign.

Stepping into her father’s shoes in 1966, Indira Gandhi chose P.N. Haksar (also from the LSE) as her Chief Advisor. Haksar had earlier interned with Menon in London.

In an obituary to Menon, P.N. Haksar described how he met Krishna Menon:

I can recall in all its detail the day I met him. I recall the year. It was 1937. The month was November. It was lunch time. I was stepping out of LondonSchool of Economics. Next door to it, in the
Houghton Street
where the School was situated, was a small cafe run by a cheerful Italian. As I was turning towards it. I met Feroze Gandhi (also at the LSE who went on to marry Indira Gandhi). Feroze and I walked into Aldwych past a group of buildings called the Bush House, then on to the Strand, up a flight of dark stairs, we entered a room. There for the first time I saw Krishna Menon.

Under Haksar's strategy, Indira moved sharply to the left. She became the icon of hundreds of millions of India's poor, by adopting socialist economic policies. She eliminated the allocations to India's ex-royalty and nationalized the banking system. Haksar also sought to boost India’s military forces, which helped India win the war against Pakistan in 1971. Her landslide victory soon after was accredited to Haksar’s political and military strategy.

In recent years, India has had KR Narayanan as President in the 1990’s, who happened to study to LSE. IG Patel, ex-Governor of India’s Central Bank served as the Director of LSE from 1984-90. The IG Patel Chair was established in his honour at the LSE last year. In 1998, Amartya Sen was awarded the Nobel Prize in Economics for his work in Development economics, much of which was done as a teacher at the LSE. The influence does not stop here. Many of the world's leading economists working on India are here at LSE, notably, Maitreesh Ghatak, Tim Besley, Robin Burgess, Lords Stern and Desai

and several graduate students.

If a country’s history is decided by the institutions its policy makers went to, LSE can rightfully claim to have had the biggest hand in shaping modern India

Why did Punjab breed terror? A rationalist perspective

Tuesday, 9/27/2011, at 9:13 PM
In many ways the story of Punjab is anomalous to the stylized facts of civil war literature. Empirical literature on civil wars points to a negative correlation between income and likelihood of conflict. There is also cross-country evidence to suggest that rough terrain contributes to greater possibility of violence. Both these facts did not hold in the case of Punjab. However, recent research on terrorism supports the view that terrorism is not correlated with poverty and lower education levels. This is consistent with terrorism in Punjab, where terrorists on average were moderately educated and came from families owning small and medium sized farms.

Having a rural base in Punjab was essential to the terrorist movement as this helped them remain hidden from the police forces. This was despite not having forest cover or rugged terrain. The key to inducing the local population not to denounce the rebels is local knowledge or information about who is doing what at the village level. Local knowledge allows the rebels to credibly threaten retribution for informing the police. Evidence from Algeria by Kalyvas (1999) suggests that ethnic insurgents use this informational advantage to a great extent, often threatening and inflicting harsh sanctions on their own people. In Punjab, kidnapping was an efficient technology used by terrorists to extract their rents and terrorism was informationally extremely localized, especially in villages. This suggests that farmers were deeply affected by terrorism and were acutely aware of the attacks due to a high degree of social capital locally.

The year 1984 saw the Indian Army being deployed in the state because of several incidents of terrorism. A separatist group, Khalistan Commando Force proclaimed the independence of Punjab (calling it "Khalistan") and they along with Bhindranwale took refuge in the Golden Temple, the holy shrine for the Sikhs. In April, extremists simultaneously attacked thirty nine local railway stations located in twelve different districts. On 2nd June, the army sealed off Punjab from the rest of India and seven tanks rolled into the Golden Temple. As has been observed elsewhere, specific events suddenly mobilize large numbers through grievance and coordination and this seems to have been the case after Operation Bluestar. A couple of months later, Indira Gandhi was shot dead by two of her Sikh bodyguards.

In 1985, President’s rule was lifted with the Rajiv Gandhi-Longowal Peace Accord, which referred all contentious issues to several commissions. But the accord was never fully implemented by the central government. In May 1987, due to political instability, President’s rule was imposed again. It has been shown through a theoretical argument that a peace pact may increase conflict in later years because the extremist faction of the separatists would want to grab power from the moderates who signed the accord. This appears to have happened in Punjab.

The death toll due to terrorist violence rose from 1,333 in 1987 to 5,265 in 1991 and tapered to 871 in 1993. By 1994, the police under the leadership of K.P.S. Gill declared that terrorism had been defeated and that normalcy had returned. Along with an increase in police personnel, changes in foreign support due to sealing of borders with Pakistan and the curtailing of ISI funding by Benazir Bhutto contributed to usher in peace. This eventually resulted in both internally generated terrorism and external abetment being wiped out.

In the rationalist strand of explanations for civil conflict, the theory of commitment problems seems to fit the scenario of Punjab. Despite a series of negotiations, there was no breakthrough between the Centre and the separatists. This may have been because of two reasons. First, there were commitment problems, i.e. no credible commitment mechanism for the separatists to believe in the centre’s offers. This would mean that once the extremists handed over their weapons, there would be an incentive for the government to renege on its promises. This commitment problem is prevalent in many civil conflicts and usually can be solved through a credible third-party commitment agency or if the movement is crushed completely as was done in Punjab. Second, issue indivisibilities namely, river water sharing and Chandigarh being made capital of Punjab were issues that did not seem to have a bargaining equilibrium as there may have not been enough alternatives for the groups to choose from. This problem was accentuated by the divisions among the militant groups themselves despite trying to forge alliances through committees. Which of these reasons was more important is an open question, but I think that issue indivisibilities could have been resolved using monetary transfers, but trustworthiness of the government would have always been suspect (which is rational for the terrorists).

During 1987 to 1992, the agricultural growth rate plummeted from 6% to 2%. To analyze how terrorism impacted Punjab’s economy, it is critical to understand how violence affected investment decisions of firms and farmers. A likely mechanism through which terrorism affects investment is through firms facing the threat of extortion of their employees (human capital losses) or loss of property (physical capital losses). The threat of extortion may increase especially if the investment is visible to outsiders, for instance if the farmer has a tractor or a firm has a huge plant. Another mechanism could be through the risk of migration. Due to deterioration of such property rights, investment is likely to decrease. Also, as conflict increases, land would be more difficult to collateralize (and interest rates set higher). This would lead to a reduction in land investment. Terrorism may increase the risk of a regime change and the accompanying redistribution of land would make long-term investment unprofitable. An increase in the probability of dying would also make investment less attractive. Finally, an increased risk of migration would negatively impact investment.

Which of these factors was responsible for a decline in agricultural investment and shutting down of several factories can only be found out by ruling out the others. As of now, because of data availability issues, it is difficult to say which channel was the most important. However, regression analysis shows that violence leading to an increase in the probability of kidnapping (proxying for threat of extortion) turns out to be a significant channel. Such a channel has been ignored earlier in the empirical literature on conflict. It reinforces the anthropological stylized fact that individuals were well informed about terrorist activities. And this knowledge helped them better predict future kidnappings and reduce their investment accordingly.

Corruption Economics 101

Saturday, 10/1/2011, at 5:13 PM

In the 1980’s, the prevailing view was that corruption can be good for growth. Huntington said, “In terms of economic growth, the only thing worse than a society with a rigid, over-centralized, dishonest bureaucracy is one with a rigid, over-centralized, honest bureaucracy.”

In other words, it acted as the lubricant between the rusty wheels of bureaucracy.

But, now, several studies later, the consensus is that corruption is actually bad for growth. Many people compare taking bribes to taxes. However, it is worse than taxes, because taxes distort incentives but corruption goes one step further. It distorts incentives but the outcome is uncertain too. The only incentive to provide a good after taking a bribe for it is, if there is reputation cost involved. It may be better for growth if bribes were fixed and the outcomes certain. Then, corruption would simply be an additional tax.

Corruption is the misuse of public office for private gain. During Suharto’s regime in Indonesia, there was rumored to be widespread corruption and nepotism. Nevertheless, Indonesia recorded high growth during his tenure. This may have been because bribes were reportedly fixed at a constant proportion and the outcomes were generally certain. Suharto’s son, meanwhile, was running the biggest media conglomerate in the country. Of course, Suharto said that it was due to his son’s hard work and talent that he had reached where he had. Edward Miguel, a bright young economist, decided to investigate. What was the effect of being Suharto’s son on the firm’s value? He looked at the firm’s stock market price before and after Suharto fell ill (which happened on a number of occasions and was taken to mean that he may not be President) and compared the fall in the price of the son’s firm with a control portfolio of similar looking firms. What he found astonished him. There was a huge impact on the son’s firm but the control portfolio suffered only a blip. The difference between the two portfolio prices before and after Suharto's illnesses gave us a measure of the value of political connection or nepotism.

It is important to distinguish the different types of corrupt activities. There are two types: with theft and without it. With theft, the government official just pockets all the money and provides, say a driver’s license. Without theft, he gives the government the official price paid by the consumer, but takes an additional fee which he hides. In corruption with theft, corruption can propagate as price charged is usually lower than the official price. Interest of buyers and sellers are aligned, making corruption more persistent. An accounting system that limits theft may reduce corruption.

Banerjee argues that corruption is caused by bureaucrats wanting to make money and governments making laws to prevent them from doing so. Economists try to model corruption as a way to get a product with a certain probability of getting caught. It is the marginal benefit and marginal cost (which depends on law enforcement) of doing so that determines the equilibrium level of corruption.

However, sociological evidence suggests that corruption may also be a cultural phenomenon. In order to test if the sociological explanation holds water, Raymond Fisman and Edward Miguel had a look at the data for New York Parking tickets. Due to diplomatic immunity, foreign diplomats were not required to pay their fines until recently (until when the paper was published!). This meant that effectively the fear of punishment was zero for diplomats of all countries. So the only explanation of systematic patterns of unpaid parking tickets could be due to cultural norms. They found that countries that were more likely to be perceived as corrupt (there are many indices that measure this including a World Bank index and a Transparency International index), were also more likely to have more "corrupt" diplomats.

Apart from trying to change the culture of corruption, there are two ways of reducing it. First, a bottom-up approach in which the locals are given the authority to supervise on projects. Second, the top-down approach, where the authority sends central representatives for monitoring. In a randomized experiment by Ben Olken in Indonesia that focused on road infrastructure, the top-down approach was found to be more effective in reducing over-reporting of use of funds. The grass-root level channel was subject to elite capture as has been found elsewhere in studies on aid and decentralization.

The Big G-School Puzzles

Tuesday, 9/27/2011, at 9:04 PM
1. Why do poor students in government schools undertake extra tuitions costing 12 times more than the government school fees?
• Bad teaching: Teaching in government schools is only to the level required for answering questions in the exam, and not catered to applying concepts.
• Peer pressure: Children see other students doing better than them (who take tuitions) and make the causal inference that tuitions = better marks.
• Parental conscience: “If Sharma ji’s son is taking tuitions, we should also send our son. We can compromise on our living expenses but not our son’s education. Even though we are not educated, we cannot deprive our son of a good education.” [Note: the gender preference in the above statement is intentional.]
• Tuition teachers work harder: As they are paid on an hour basis, they have a greater incentive to put in more effort. The market is competitive as there are a large number of teachers who can provide tuitions.
• Job prospect theory: If a child goes for tuitions, he will get better marks, which will get him to a good college, which will get him a good job.

2. Why do government colleges have a better environment than government schools?
• Motivated students: As there is a cutoff for entry even in government colleges, above average students self-select themselves into government colleges.
• Better teachers: The teachers are paid more and thus better skilled teachers are chosen for college teaching.
• Higher fees: Colleges are not as subsidized as primary schools, so there is a greater incentive to provide better quality because it is charging the students.
• Age: As children mature, they become more subdued (and have a better understanding of complex concepts).
• The good trap: If students understand better, teachers are more effective and this increases the students’ level of understanding leading to a self-reinforcing mechanism.


3. Why do we have “bad teaching” in government schools?
• Self-selection: Bad (under-skilled, under-motivated and over-weight) teachers self-select themselves into positions that require less effort per hour and there is less chance of them getting fired.
• No explicit goals: There are no set goals and often teaching is only to cater to the exams rather than making students better at applying concepts or exploring creative answers.
• Tuition disadvantage: Teachers know that students are going to take tuitions, so why bother? This in turn, leads to more tuition.
• No feedback: Students hardly ever give feedback. They are often shy or instead, they fear retribution if the feedback is not kept anonymous. They are required to uphold teachers as gods and listen to every word without ever questioning them. This leads to moral-hazard (hidden action) for teachers.
• No top-down sticks: The principals of the schools do not care about the quality of teaching as they too have no incentive to monitor classes. There are no explicit goals (except in some cases, marks obtained) which can be addressed to by the teachers by awarding marks copiously.

4. Why do religious schools tend to do “better” than government schools?
• Discipline: Religious schools enforce discipline in code of conduct which regulates the students and makes them fulfill their daily commitments.
• Intrinsic motivation: These schools are often run by people who do not have a monetary goal in mind, but are driven by their passion to instill a value-based education in students (the value being correlated with the religion they represent).
• Self-selection: These students are often brought up in a religious environment at home too. Religious environments at home are arguably correlated with less deviance, less domestic violence, more discipline and less junk food and less idiot box watching (which reinforces their faith).
• Similar preferences: Peers are more often than not of the same religion and the same sex. This leads to a greater social capital and trust and more productivity or greater retribution if one is deviant.
• External support: These schools get a great deal of funding from religious organizations that have a stake in the system and are driven by their zeal to instill their values in the youth.

5. Why do students in government schools lack fluency in English or knowledge about current affairs?
• Teachers’ ability: The government school teachers are on average worse in spoken English and general knowledge.
• Family income: As I observed in our interviews with children who had applied for scholarship, family income is a strong predictor of English fluency of the child. On the other hand, parents may have lower earning potential because they are not good at English. This also leads to a “Bad English trap”.
• Exposure to people who speak better: Low exposure to English-speaking students, teachers, and parents’ social network lowers the bar and makes children end up like those whom they have coffee with (in this case mid-day meals).
• No exam: There are no exams on verbal ability and current affairs in government schools and so no immediate incentive for students.
• Infrastructure and technology utilization: Libraries have locked shelves. Librarians are bored individuals who only care about books not being stolen. The perceived marginal benefit from reading an Enid Blyton or a J.K. Rowling is cipher. There is hardly any hands-on experience to prepare students for the ever innovating tech-world (internet, email, google, video chat, Powerpoints, Windows, etc.).