Micro for Development Practice questions

Question 1

Poverty Measurement

What household expenditure do you consider to be absolutely minimal, in that you could not make ends meet with any less?

• Explain how the information collected in this household survey could be used to produce a subjective poverty line (SPL)
• Suppose that over time Langoustine enjoyed a number of economic windfalls (natural resource discoveries, rising world prices of Langoustine’s export products, etc.) and that consequently household consumption levels, estimated from subsequent household surveys, were observed to have risen. Explain why poverty measured on the basis of an SPL, re-calculated in the subsequent survey rounds, might not fall, even though consumption levels were rising for all income groups.
• The concern that not all economic growth leads to poverty reduction has prompted alternative definitions of pro-poor growth. What are the two definitions of pro-poor growth, and illustrate with growth incidence curves how they can be distinguished.

Answer 1

• Draw a graph plotting houseohld per-capita cnsumption against the reported answers to the minimum standard of loving question. Typical plot will reveal that households with particularly low per-capita consumption levels will generally report a minimum consumption requirement above their actual reported consumption. Reverse for rich households. Thus at some percapita consumption level, households will generally report a level that roughly equals their actual observed consumption level. Consumption level at which the equality obtains can be designated as the subjective poverty line: level at which society judges to be sufficient to meet min requirement levels
• As average income rises in societz one might expect that expectaitions, aspirations, preferences etc. of households also change. Thus, households| answers to the question of what expenditure they consider to be absolutely min might also rise with rising average living standards. This results in higher SPL than before, so povert measured on basis of this new, hgher SPL may not be lower than what was found prior to rise in avg incomes
• two definitions
  
  • 1: pro-poor growth occurs if consumption growth amongst poor exceeds taht of population overall
  • 2. occurs if poor benefit from growth in consumption, i.e. if growth is any positive value (any value above x-axis)

Question 2

Growth and Income Distribution

Consider a dualistic economy comprising a low mean rural sector and a high mean urban sector. Assume that within the respective rural and urban sectors all incomes are equal. Suppose growth occurs through the migration of workers from the former to the latter, and that initially everyone is in the rural sector.

• Describe the path of overall income inequality following this growth process.
• Would overall inequality necessarily follow the same path if the urban sector had high inequality? Why?
• Between 1988 and 2010 global income growth has been significant. Global income inequality is very high, but has been fairly stable over this time period, with even a suggestion of a decline. Illustrate by means of a global growth incidence curve who have been the winners and losers from economic growth over this time period.
• What can we conclude from the observation that the Lorenz Curves for the global distributions of income in 1988 and 2010 intersect?

Answer 2

• Initially all workers in tural secotre and overall inequality is 0. Then migration starts and some workers move to urban sector. SInce urban incomes are higher than rural incomes this introduces a source of income ineqaulity between workers; overall inequality rises. As more workers migrate to urban sector, overall inequality continues to rise, but at some point, when ther are no more workers in rural sector inequality must fall back to 0. Thus must be a turning point somewhere and overall inequality traces out an inverted U-curve path
• if urban sector has inequality then it’s not clear that overall inequality must delcine with migration. Could continue to rise uniformly if inequality within the urban areas dominates gap in average incomes between sectors.
• Graph
• Intersecting Lorenz curves imply that amongst the set of inequality masures satisfying the axioms of anonzmitz, scale neutrality, and principle of transfers, there will be some suggesting that global inequality fell between 1988-2010 and others suggesting that inequality rose. Only definitely more qual if curve lies above the other curve.

Question 3

Firms under missing markets

Answer 3

Question 4

Community-based development

• a) Explain why Community-Based Development may involve a trade-off between efficiency and equity. (2 points)

The provision of aid in the context of community-based development can be described as an ‘ultimatum game’ between three players, namely an aid agency, a local leader, and the grassroots. The aid agency provides money to the local leaders with a certain maximum amount, who in turn choose how much of this money to deliver to the grassroots. The grassroots can ‘accept’ or ‘reject’ the money offered by the local leader. The aid agency will only provide money to the local leader if the grassroots accept the offer from the local leader.

• b) Suppose an aid agency considers providing funds to local leaders for grassroots development for a fixed number of periods (finitely repeated game). Explain why no funds will be disbursed even if the grassroots can observe the total amount of funds but there is no social norm. (2 points)
• c) Suppose there is a social norm such that grassroots expect at least 75% of the aid given by the aid agency. The grassroots do not observe how much aid is given but there is a fraud detection mechanism such that aid is discontinued if any fraud is detected in an earlier period. The probability of detection is increasing in the amount of aid not disbursed to the grassroots. How
• much aid will be disbursed by the agency if the game is played a fixed number of times?
• d) Suppose now that after fraud, aid is discontinued and the local leaders will have to pay a fine. Will this affect the answer under c, and if so, how)? Why (not)? (2 points)
• e) Explain why aid agencies may face a prisoners’ dilemma when reporting fraud in the competition for funds from outsiders. (2 points)

Answer 4

• a) Better information of local conditions and constraints, higher social capital, improved effort through intrinsic motivation against the risk of possible local elite capture
• b) This is an ultimatum game and it is rational for the local government to transfer ε to the grassroots, with ε arbitrarily small, because the grassroots will accept. But then the Aid Agency will not transfer any funds. This holds for the last period and all the earlier periods because of backward induction.
• c) In the last period no aid will be given because the local leader will capture almost everything given that there is no punishment afterwards. But then, by backward induction, no aid will be given in any period.
• d) Now the local leader faces a credible threat of punishment in the last period because he/she may get a fine if less than 100% is disbursed to the grassroots. The local leader will choose to give less than 100% if the cost of doing so (given by the probability of being detected by the fraud detection mechanism times the fine) is lower than the benefit (given by the amount not disbursed). Hence, the higher the probability of being detected and the higher the fine, the more will be disbursed to the grassroots.
• e) It is best if others report fraud and you don’t, and also better not to report fraud if other also don’t.

Question 5

Credit markets

If the project fails, the return is zero. Assume that all potential entrepreneurs in the economy are protected by limited liability only when they have used up all the returns from the project and the collateral to pay back their loan. The bank faces a cost of capital of 135 for each 100 lend.

• a. Is the type of project that is proposed socially efficient? (2 points)

Assume that the bank can ask for proof of collateral in the application process, and thus distinguish between poor and rich borrowers.

• b. Is credit extended to the poor? If so, which interest rate would the bank charge to them? What is the expected profit of the bank from a poor borrower?
• c. Is credit extended to the rich? If so, which interest rate would the bank charge to them? What is the expected profit of the bank from a rich borrower?

Government is concerned that the collateral proof that banks ask for results in terms of lending that are unfavorable for the poor. A new rule is implemented that forbids banks to ask for collateral at the application process of the loan. As a result, the bank cannot distinguish between poor and rich borrowers during the application process. The limited liability clause remains as it was. When it is time to repay, the banks can still claim the assets from rich borrowers which would otherwise have been used as a collateral.

• d. Do banks still grant credit? If so, which interest rate would the bank charge?
• e. Does the policy have the desired effect? Explain why.

Answer 5

• a. Answer: Yes the project is socially efficient : 0.75*200=150>135
• b. The poor will only participate if the expected returns are nonnegative. As they have no collateral, they will participate if the interest rate remains no higher than 100% (ER=0.75(200-r)+0.25*0=0–>r=200–>interest rate =100). (that is the interest rate charged, or a fraction below that) In that case, the bank will make a profit of 0.75*(200-135)-0.25*135=15 per poor borrower. (or a fraction below). Important here: remember limited liability! means tht if revenue=0, borrower does not pay anyhting of loan (if there is no collateral)
• c. The rich will participate if their expected profits are nonnegative. They take into account their collateral, which they may lose. The rich will only participate if 0.75*(200-R)-0.25*90≥0 Rmax=170–>interest rate=70. The profits they derive from that is0.75(170-135)-0.25(135-90)=15.25
• d. In order to keep both borrowers in the market, the bank would have to charge 170. But in that case they would make a loss on the poor borrowers. 170*0.75-135=-7.5. A better strategy is to charge 200. Rich borrowers will not apply for credit, and thus will be credit constrained.
  
  • OR at which r is bank profit (min r they accept): 0.5*0.75(r-135) + 0.5*0.75(r-135)-0.5*0.25*135-0.5*0.25*(135-90)–>r=165. With that, bank profit on poor would be negative (0.75*30-0.25*135=-11.25), so bank would not lend to poor; thus better for bank to charge 200, which would drive richt out of market
• No, it leads not equally unfavorable terms for the poor, while the rich borrowers are excluded from credit.

Question 6

Weather risk

Rosenzweig and Binswanger, in their article “Wealth, Weather Risk and the Composition and Profitability of Agricultural Investments”, investigate the effect of the weather variation in a particular location on the riskiness of the investment portfolio of farmers.

• a. Which investment portfolio do the authors refer to?

The authors investigate the relationship between the coefficient of variation in the onset of the rains, and the coefficient of variation in farm profits. They find a relationship as shown in the figure below.

• b. Indicate which line (straight line, dashed line, or dotted line) shows the relation for the wealthiest farmers, and which one shows the relationship for the poorest farmers. Provide an economic explanation for the answer.

The authors distinguish between poor and rich farmers on the basis of their total wealth. As a robustness check, they also do the analysis where they use inherited wealth rather than actual wealth.

• c. What problem does using inherited wealth instead of actual wealth solve?

The authors conclude the paper with “The results suggest that improvements in the abilities of farmers to smooth consumption, perhaps via increased consumption credit, would increase the overall profitability of agricultural investments”.

• d. Which finding in their paper makes them come to this conclusion?

Answer 6

• a. They refer to how farmers use their capital, investing in different types of land, cash, animals and crops.
• b. The straight line is for the rich, they do not care about riskiness and invest in more risky portfolios. They are thus more affected by variations in rainfall.
• c. That there are omitted variables which could influence both investment decisions and wealth. This would bias the estimate of wealth
• d. The finding that average profits are also negatively affected by rainfall variation, and that this is stronger for the poor. Apparently the poor adopted low return, low variation strategies to protect themselves against weather risk. Rainfall insurance could result in them adopting strategies that have higher overall profits

Question 8

Returns to Skills in Developing Countries: Ability, Credentialism and Skills

• a) Which of the above coefficients can be interpreted in terms of ‘credentialism’? Why?
• b) Suppose a primary school-leaver in Tanzania earns 500 Shillings (KSh). Using the reported estimates from column (1), how much would she earn as a secondary school leaver? Note: S is a dummy variable equal to one if a worker is a secondary school leaver.
• c) Suppose the ability of a primary school leaver in Kenya increases by one point. How much more will she earn keeping S, H, E, F, G, B, L

Boissiere et al. (1985) emphasize that apart from the direct impact of ability (keeping S and H constant), there is in principle also an indirect impact, which turns out to be larger in their empirical analysis. In order to estimate the indirect impact, they the full recursive system of equations (1)-(3).

• d) What is the total indirect effect of ability on earnings within their recursive model? Write your answer in terms of the coefficients of the recursive model.
• e) Someone notes that the estimated coefficients of H are larger than those for R in Kenya and Tanzania, and concludes that cognitive achievement explains a larger part of the earnings difference between primary- and secondary-school leavers than ability. Give two reasons why this reasoning is faulty.

Answer 8

• a) c1. Credentialism implies that schooling impacts earnings even after controlling for ability and cognitive achievement.
• b) exp(0.28)-1=32.3%. Hence 500 becomes 662 KSh.
• c) The coefficient for ability is zero (column 3) and therefore no change in earnings.
• d) The indirect effect is the sum of the following effects: (i) c₃*b₁ (impact ability through human capital, relation D in figure 1); (ii) c_{<span>1</span>}*a₁ + c₃(b₁+a₁*b₂) (impact ability through education, relation E in figure 1)
• e) 1) One should also look at the significance of the coefficients (large standard errors), 2) there is an indirect effect from R on H, 3) scaling of variables also matters, and 4) variation of variables matters for explaining variation in earnings.

Question 9

corruption

A common definition of corruption is the misuse of public office for private gain.

• a) Give two reasons why a corruption payment (bribe) is different from a tax or a fee.
• b) A common policy response to reduce bribe-taking by public officials is to increase their wages. Argue why this might actually increase the amounts requested in bribes.
• c) Reinikka and Svensson (2004) analyze the extent of local capture in a Central Government Transfer Program in Uganda. They show that local capture is higher for communities with lower income. Give two possible reasons for this correlation.
• d) Discuss how such a correlation between local capture and income (as in question c) affects benefit incidence analysis of public expenditures.
• e) Olken (2007) estimates the experimental impact of auditing on corruption, where corruption is measured by the percent of funding for a road project unaccounted for. His estimate of corruption most likely suffers from serious measurement error. Does this imply that his estimated treatment effect of auditing is biased towards zero (‘attenuation bias’)?

Answer 9

• a) 1) no transfer to government budget, and (2) bribes involve higher transaction costs than taxes because of uncertainty and secrecy (bribe-taker may renege on agreement and ask for additional bribe).
• b) If the wage increases, then a public official may ask for a higher bribe because (1) he/she is less liquidity constrained, and (2) the risk is higher because he/she might risk losing a better paid job. (3) now higher risk aversion by receiver since does not need to take as muchrisk as before anymore as now due to wages has higher income
• c) (1) communities lack the income to file a complaint, (2) poorer communities may be less informed about their rights (3) community with lower income might have less political power / leverage to stand up against leader (4) more in need of aid, so happy with even very small amounts
• d) If local capture (as a percent of total public transfer) falls with income/expenditures, then the benefit incidence of actually received transfers will be less pro-poor than the initially provided public transfers.
• e) If the measurement error is uncorrelated with the treatment variable, then measurement error in the dependent variable will not bias the estimated treatment effect, but it will reduce the precision of the estimates (standard errors will be larger). The measurement error is likely to be uncorrelated with treatment, because the latter is random (although it is not impossible that the engineering teams were somehow affected by the treatment)

Question 10

Poverty

• a) Suppose that it is claimed that poverty has fallen in a given country. Analysis reveals that second-, but not first-, order stochastic dominance is observed over the range of poverty lines from 0 to Z_max during this time period. What does this imply for poverty measured on the basis of the Foster-Greer-Thorbecke class of poverty measures?
• b) Suppose the growth incidence curve for a given country over a given time period lies everywhere above zero, and is monotonically upwards sloping. Explain why there might be disagreement as to whether this indicates that growth is pro-poor.
• c) In conventional analyses of poverty household consumption is typically divided through by household size to obtain a measure of per capita consumption. It is assumed that there is no intra-household inequality and so family members receive an equal share of household consumption. Poverty is then measured by comparing each individual’s per capita consumption against a poverty line expressed in per capita terms. Even if one were to accept the assumption of equal sharing within the household, what are two other objections one might have to the convention of dividing household consumption by household size? Provide a brief description of the reasoning behind one of the objections.

Answer 10

• a) second order stochastic dominance indicates that all poverty measures which are strictly decreasing and weakly convex in consumption of the poor will point to declining poverty irrespective of the precise location of the poverty line within the range[0:Z_max]. This corresponds to any poverty measure within the FGT class with a parameter value of 1 or greater. However the FGT0 measure – the headcount rate – will not point to an unambiguous decline in poverty. There will be at least some poverty line within the range [0:Z_max] for which the headcount will have risen over time.
• b) simple answer: in relative terms not pro-poor, in absolute terms yes; An upwards sloping growth incidence curve suggests that income growth amongst the richer percentiles of the income distribution is more rapid than amongst the poorer percentiles. However, because the growth incidence curve is everywhere positive, even the poorest segments of the income distribution will have seen positive income growth. This means that for those who define pro-poor growth as occurring simply when growth benefits the poor, there will be pro-poor growth. However, those who require that pro-poor growth is accompanied by pro-poor redistribution will disagree, given that the rich are benefiting more than the poor.
• c) dividing household consumption by household size to obtain a measure of per capita consumption ignores first, the possibility of different family members having different needs, and second, the possibility that there may be economies of scale in household consumption.
  
  • The idea with respect to the former is that family members of different ages, gender and activity levels may have different consumption requirements in order to achieve a particular welfare level. Households of different compositions should first be rendered comparable by converting them into, say, equivalent adult males. Household consumption should then be divided by equivalent adult males to yield adult equivalent consumption, and this should then be attributed to each family member. Suppose that an equivalence scale exists showing that children “need” only half the consumption of adults to achieve the same welfare level. Suppose there are two households, one comprised of four adults, and one comprised of two adults and two children. Suppose that the two households enjoy the same total consumption. Now, instead of assessing poverty on the basis of per capita consumption (which would be the same for the two households) it should be assessed on the basis of “adult-equivalent”consumption which for the second of the two households in the example would require that total consumption be divided by 3 rather than 4.
    
    • With respect to economies of scale the argument is that the cost of achieving a particular welfare level may decline with size of the household. Suppose part of household consumption is devoted to the consumption of services from public goods within the household (e.g. a television, a water pump, etc.). Consuming the services of a television by one family member does not preclude the consumption of the same services by another family member. Yet, for a one person household, the per capita cost of = consuming these television services is the full cost of the television. For the two person family, the per capita cost of consuming these services can be divided by the two family members, and is thus half the cost of the television. Thus, the per capita cost of achieving a given welfare level might be different for households of different sizes. The extent to which this would likely apply will depend on the extent to which there is consumption of such “public goods”. Food items more closely resemble private goods, but certain non-food items – notably durables – may well embody such public-good characteristics.

Question 11

Inequality

• a) It is often argued that inequality measures should satisfy the following axioms: anonymity, scale independence, population replication independence and the principle of transfers. Describe briefly what each of these axioms mean.
• b) In which way does checking for Lorenz Dominance allow one to probe the robustness of inequality comparisons?
• c) Explain, with reference to inequality decomposition, why it is possible for income inequality in all provinces of a country to be rising, while at the country level inequality is falling.

Answer 11

• a)
  
  • Anonymity: shifting the identity of members of the population around will not change measured inequality.
  • Scale independence: a proportional change in everyone’s income will not affect measured inequality.
  • Population replication independence: simply replicating the original population will not affect measured inequality
  • Principle of transfers: transferring income from the poor to the rich must increase inequality.
• b) Checking for Lorenz Dominance involves comparing Lorenz curves for two (or more) distributions. If Lorenz curves intersect then there will be at least one inequality measure within the class of measures satisfying the conventional axioms of inequality measurement that will fail to agree on the inequality comparison. However, if the curves do not intersect all measures of inequality within this class will point to the same conclusion. Note that the robustness of conclusions is confined to ordinal, not cardinal, comparisons of inequality
• c) Inequality decompositions break overall inequality into a term capturing the effect of average income differences between groups (provinces, in this case) and that of differences within groups. In our example, it is claimed that within-province income inequality is rising in all provinces, but that overall inequality is falling. This is possible if average incomes between provinces are converging so rapidly that the between-group inequality decline offsets the within-group inequality increase.

Question 12

Risk

Answer 12

• a) Dependent variable: Non medical household consumption; independent variable: a dummy which equals 1 if a household member experienced a health shock. Insurance implies that beta is not significantly different from zero.
• b) Calculate the variance in consumption resulting from health shocks by beta²var(H); mu is average consumption. R=2
• c) You would expect the hypothesis gamma=0 not to be rejected while
• beta=0 is rejected.
• d) We expected gamma<0. Risk averse people will be less affected by shock because they do not like fluctuations in consumption So we expect this to mute the effects , thus theta>0
• e) We expected gamma<0. People with less assets holding cannot dis-save to smooth consumption. They will be more affected by the shock, thus theta<0
• f) We do not have any prediction based on economic theory

Question 13

Credit

Consider a competitive bank which could lend 100 to a risk neutral farmer at the beginning of the planting season. If the farmer does not borrow, he cannot farm, and the yield at the harvesting season will be zero. If the farmer receives the loan, he can invest this in his farm, and the harvest will depend on the effort the farmer puts in, and the rains. The farmer is protected by limited liability, that is, the bank can never seize more than the yield. The relevant information is summarized in the table below. There is a 25 percent probability of no rains. If there are no rains, the yield depends on the effort of the farmer. If he puts in effort, the yield is 140, if he does not put in effort, the yield is 140 with a probability 0.5, and zero otherwise. If there are rains, the probability of a good yield without effort increases to 0.75. The cost of effort is 10. The bank faces a cost of capital of 115.

• a) Is taking a loan and investing it in farming socially efficient if the farmer puts in effort? And if the farmer does not put in effort?
• b) If the bank charges an interest rate to cover the cost of capital, will the farmer want to borrow? Will he put in effort?
• c) Will credit be extended?

The bank teams up with a weather insurance company to provide a new joint product instead of the existing product. The new product combines the loan with an actuarially fair weather insurance. The cost of the insurance is 20. The farmer is protected by limited liability, that is, the bank can never seize more than the yield and the insurance payout combined.

• d) If the bank charges an interest rate of 15 percent, will the farmer put in effort if the new product is offered?
• e) Will credit be extended if the new product is offered?

Answer 13

• a) Yes , it will generate 140 for sure, 140-10=130>115.
  
  • No, it will generate an expected return of 0.25*0.5*140+0.75*0.75*140=-18,75 negative, so not socially efficient; important: here look at effor/no-effort across rain/no rain; could also split up in 4 scenarios (rain+no effort, rain+effort etc.) to be more precise
• b) Yes, he wants to borrow. When successful, in all cases, he will make 140. Even with putting in effort that is still more than 115. But shouldn’t we do more complete analysis here, i.e. including rain/no-rain effort/no effort scenarios? then total ER=0.75(140-115-10)+0.75*0.5(140-115)0.25(140-115-10)+0.25*0.5(140-115)+27.5, positive, so yes borrows
  
  • Will he put in effort. Expected return if put in effort is 15, see (a). Expected return if not put in effort 0.25*0.5*(140-115)+0.75*0.75*(140-115)=17.2 that is higher than 15, so farmer will not put in effort
• c) No. Because farmer will not put in effort at low interest rate. Bank needs to mark up interest by 115/(0.25*0.5+0.75*0.75)=167.3 . That is above 140, the participation constraint
• d) actuairlly fair means C=ER of insurance, so look at each scenario how much it pays (e.g. in case of effort, ER=20=profit rain + profit no rain=0.25*x+0.75*0–>x=80) Expected return with effort is 15. Insurance does not change that because it is actuarially fair. Expected return without effort will change =-20+.25*0.5*((140-115+80))+0.75*0.75*(140-115)=7.18 which is less. So farmer will put in effort at the low interest. 80 is only received if the project is successful, otherwise the bank will take it.
• e) Yes, and the farmer will put in effort because the bank knows the farmer will put in effort.