Micro development

• Balsahki treatment analysis

• students that are lagging behind are given extra help

• issues with pre vs post

• then difference in difference analysis, know the equation for a diff-in-diff

  • relies on parallel trends, spillovers can bias results

  • good to account for common shocks

• RCT is alternative method of analysis

  • take a large group and randomly assign participants to treatment group or control group

  • important that differences between the groups before treatment should be statistically insignificant

  • pros: simple, credible flexible

  • cons: cost, timing, ethics of giving random people treatment, requires large sample for true randomisation

types of validity, what might cause it to fail

Internal validity: Whether we can conclude that the measured impact is indeed caused by the intervention in the sample

External validity: Whether the impact we measure would carry over to other samples or populations

• environmental dependence

• randomisation bias 1: participants behave differently due to being evaluated

• randomisation bias 2: people need to consent to study so those that choose to consent may have different characteristics to general population

• General eqm effects: results may not hold if everyone does it due to limited resources

RCT applied to study of bed nets

• Cohen and Dupas (2010)

• insecticide treated bed nets effective in malaria infections of pregnant women, with positive externalities

• problem with subsidising it is that unlike something like vaccine effectiveness requires active usage

• positive prices have three important benefits

  • selection effect

  • psychological effect

  • price signals effect

• different levels of subsidy given

• results: demand drops substantially at modest cost sharing, no evidence that usage increases with offer price

• missing long run effects

what are two opposite long run effects of cost sharing

1. Entitlement effect: Subsidies may generate unwillingness to pay more for the product later I

• One-off subsidies may negatively affect long-run adoption

2. Learning effect: Subsidies induce more people to experience the benefits and learning the true value of the product (which they may have been underestimating) I

• One-off subsidies could boost long-run adoption

short run vs long run effects of subsidies on healthcare

• Dupas (2014)

• long lasting insecticide treated bed nets

• phase 1: assign subsidies randomly across households

  • results: usage does not increase with price, steep demand curve

• phase 2: give everyone the opportunity to buy bed nets at medium price a year later

  • results: small but positive effect of high phase 1 subsidy on phase 2 adoption, suggests learning effect dominates entitlement effect

Facts about credit markets in developing economies

1. lending rates are higher than deposit rates, AER 40%-200%

2. Extreme variability in interest rate within the same sub-economy

3. low default rates

4. ex-ante competition in markets, costly to switch so not much ex post

5. rich have larger loans and pay lower interest

information asymmetries in credit market

estimating whether we have hidden action or hidden information

• Karlan and Zinman (2009)

• Randomization occurs along three dimensions:

  • An initial “offer interest rate” featured on a direct mail solicitation

  • A “contract interest rate” that was revealed only after a borrower agreed to the initial offer rate

  • A dynamic repayment incentive that was also a surprise and that extended preferential pricing on future loans to borrowers who remained in good standing

• Hidden information effect can be identified by considering the sample that received the low rc , and comparing the repayment behavior of those who took up at the high ro (cells 2 and 3 in the figure) with those who took up at the low ro

• Similarly, hidden action effect can be identified by considering the sample that took up at the high ro, and comparing the repayment behavior of those who received the high rc (cell 1) with those who received the low rc

• Hidden action effect can also be identified by comparing the repayment behavior of borrowers who both selected in and contracted at identical rates, but faced different dynamic repayment incentives because of randomly assigned future interest rates rf

• find no evidence of hidden information effect but strong hidden action effect through both channels

characteristics of Grameen Bank

• Loans almost exclusively to women

  • equality, less likely to have access to other credit, more likely to use profits on health and education

• weekly repayment schedule

  • for rational individuals less rigid repayment should not increase defaults, studies show no significant effect of repayment schedules on default rate

• group lending with joint liability

  • screening effect: mitigates adverse selection as people only want to go with individuals deemed trustworthy

  • monitoring effect: clients monitor each other for free

  • creates excessive pressure, but banks moving away from it towards individual liability, again no change in default found for individual liability

• Regular meetings

  • social capital, ineraction can lead to support and reputation, defaults found to be lower if members live close by and share culture

• Small initial loans

  • same idea as building up credit score

Effectiveness of microfinancing in alleviating poverty

• Banerjee et al. (2015)

• RCT of slums in India where some were given microcredit immediately and some after 2 years

• microcredit follows grameen characteristics

• results,

  • only a quarter of eligible people use it,

  • no difference in consumption,

  • only helps existing businesses, not more entrepreneurs

  • No HDI change

• microfinancing not transformational

two main categories of risk coping strategies

• Ex-post strategies (after the realization of the shock)

  • They are aimed at stabilizing consumption in the face of realized income fluctuation

• Ex-ante strategies (before the realization of the shock)

  • Risk-averse households use them to reduce the variance of their incomes

Ex-post strategies

• Credit

• transfer from government

  • benefits are rare in developing countries, govt only intervenes with large shocks like drought

• transfer from social network

  • most beneficial when income streams across members are negatively correlated

• Selling Assets

  • Paxson (1992), use rain shocks and finds that savings are used to smooth consumption, but thailand has well established financial systema

  • livestock as insurance finds little evidence of selling animals to smooth consumption

  • everyone selling at the same time creates low price

  • reduce investment in education during hard times

• Labour Supply

  • may increase labour supply to weather shock, however shock lowers labour demand so wages even lower

Ex-ante strategies

Consumption side

• Accumulate savings

  • allows for ex-post dissaving

• Buy insurance

Other

• Investments that are shock resistant

  • requires large initial sums

• Diversification

  • returns to different income streams should not be positively correlated

  • extent to which this is possible is limited

• Sharecropping

  • rent out land to pass risk on to tenant, but only receive portion of profits

• Avoidance of risk

study on weather risk and portfolio riskiness

Rosenzweig and Binswanger (1993)

• capture the risk that farmers face through the coefficient of variation of monsoon onset date in India

• do individuals that face higher risk invest in less risky assets

  • yes

• does this relationship change across wealth levels

  • yes higher wealth means more risky but more profitable assets

• estimate relationship between weather risk and farm profits

  • one standard deviatino decrease in weather risk raises average profits by 35% for poorest farmers

• poorest farmers are most inefficient because they are poor, choose to be inefficient

issue with formal crop insurance and proposed solution

adverse selection: farmers inherently more risky will select in

moral hazard: once insured farmers excert less effort

very high transaction/ verification costs

• Weather index insurance

  • pay-out determined by level of rainfall

  • premium and payout can hence be determined based on historical trends and is unaffected by household characteristics and behaviour

  • dont need monitoring

study of impact of insurance on magnitude and riskiness of agricultural investment

Karlan and Udry, (2014)

• find that constraint for farmers is uninsured risk not availability of credit

• look at investment response to cash and insurance grants

• set up model to show

  • complete credit and insurance: input choices indpeendent of wealth

  • imperfect capital market only: cash grant leads to increase in I insurance leads to lower investment

  • imperfect insurance market only: incurance and cash grant lead to increase in investment

  • both imperfect: in this model capital constraint dominantes so same results as in imperfect capital market

• RCT results

  • investment increases with insurance grant but small effects with cash grant

  • liquidtiy constraint not as binding as thought

why has weather index insurance had a low uptake

• Price

  • some are actuarily unfair but even if they are low uptake

• Financial Literacy

  • apparently farmers are unfamiliar with millimetres of rainfall

• Behavioural explanations

  • when considered in isolation insurance product may be perceived as unattractive risky asset, mental bracketing

• Credit Constraint

  • need to buy insurance to start off with, wealthy individuals can cope with risk better

• Basis Risk

  • likelihood that harvest fails but you also dont get paid since rainfall isnt low enough

• Alternative Risk coping strategies

  • informal insurance can crows out formal insurance

  • social networks, informal insurance better equipped to handle moral hazard

• Trust

  • of insurance provider or product

• Catastrophe type insurance

  • if insurance only pays out rarely have higher perceived basis risk

  • less likely to learn from it and trust it more

ROSCAs

• rotating savings and credit associations

• pass a pot of money around a group

• very popular, 50% of savings in Cameroon

• random ROSCAs or Bidding ROSCAs

• although negative interest and risk of loosing it still very popular

• because more likely to afford indivisible good earlier than through individual savings

• mostly women, have stronger preference for indivisible good like education

• discipline and commitment device

Internal vs External constraints to saving

External: limited availability of saving at formal institutions, demands from social network

Internal: lack of self control

using a commitment device to counter hyperbolic discountign study

Ashraf et al. (2006)

• test whether individuals open savings account with commitment feature that limits access

  • quarter of people that were offered such an account would accept

• are individuals with hyperbolic preferences more likely to open such an account

  • women yes

• does this lead to higher savings

  • yes by 81%

• RCT

• ROSCA is an actual commitment device

how does a lack of self control help alleviate principal agent problem, how does timing of self control problem change

• since agents are risk averse, tying pay to performance usually requires risk premium. However if if workers are aware of self control problem they will value sharper incentives as a way to motivate future selves and this increases probability of reaching high state

• closer to payday reduces lack of self control since reward and cost are closer together

study of self control problem at work

Kaur et al. (2015)

• test for

  • demand of dominated contracts

    • offered to only get half the pay per unit if quota not met

    • third opted in and it increased their production and earnings

  • presence of payday effect

    • all paid weekly but randomise on which day of week

    • output per worker is 8% higher on payday

  • heterogeneity in extent of payday and contract effects

    • those with higher pay day effects were more likely to opt into dominated contracts

    • if option to choose dominated contract further away from payday then the treatment effects are greater since the self control problem is greater

study on reason for low fertiliser use and how to increase in

Duflo et al. (2011)

• claim that procrastination can explain it

• need to purchase fertiliser in advance which they put off until it is too late

• offer commitment device in the form of discount for early purchase of fertiliser

• prediction

  • small cost reduction in period 1 should increase usage more than same cost reduction in period 2

  • if they want to use it as a commitment device given the option they should prefer to have the discount in period 1

• yes offering cost reduction increases usage by up to 70% and this effect greater for period 1

• suggests policy should offer time limited subsidies

What are the three reasons that make social learning difficult

• Difficult to define set of farmers an individual learns from

• Difficult to observe what an individual learns from

• Difficult to distinguish learning from other reasons that may cause similar observed outcomes

study on effects of social learning from fertiliser usage

Conley and Udry (2010)

• collect detailed data on who the farmers talk to to establish information links

• dates of planting provides timing at which new information is revealed

• Prediction 1: farmers adjust inputs towards suprisingly succesful input levels

• Prediction 2: adjust away from unsucseful

• Predicitoin 3: observed profit above expectation induces switch to that input level

• Predicition 4: probability of changing input levels in response to new info is decreasing in farmers experience

• idnetifying assumption is that information is uncorrelated iwth unobserved determinants of changing in growing conditions, maybe mistake changes in conditinos with changes to fertiliser

• control for change in common growing conditions by taking difference of fertiliser input between average plots close to i and the i

• know the regression in lecture 10

• all predicitions hold

what are the benefits of learning from other and what might it depend on

• can be a slow process

• if learning from other is possible depends on whether agricultural technology is sensitive to local cricumstances

what are two opposite effects that learning from neighbour might have on adoption

• Spillover: On the one hand, it can fosters adoption because it increases expected profits

• Free-Riding: On the other hand, it may delay adoption because the farmer may refrain from experimenting himself and instead wait for enough useful information to be accumulated by his neighbours

impact of technology on markets, fishing

Jensen (2007)

• price at market just determined by amount of fish caught nearby since fishers cant check prices and can only go to one market

• staggered adoption of cell towers along the coast

• identifying assumption: without introduction of cell service no differential in outcomes across these regions

  • should test for parallel trends assumption for all three before and for the two that dont have adoption after 1 adopts

• use dummy regression

• significantly reduces variance of prices and hence markets are more efficient due to price competition

two channels through which poor property rights might influence investment

• fear of expropriation

• access to credit might be limited if land cannot be used as collateral

study on property rights and agricultural investment in Africa

Goldstein and Udry (2008)

• land in western Africa is under the control of a chief

• land is allocated to individuals on perceived need and political influence

• do individuals with better political connections have more secure tenure rights and does this lead to higher investment

• fallowing for which land is left empty to increase fertility, is long term investment strategy that increases probability that land is lost

• identification strategy: plots are owned individually so can control for household characteristics, only look at variation between individuals within the same household

• empirical strat: effect of political standing on fallowing choice conditional on plot characteristics and household characteristics

• results

  • start with profitability by gender and find women make less even after controlling for characterstics of land and household

  • concluded that this difference in gender is because women do not have the same political standing as the men, do the husbands not care about their wives?

  • gender coefficient becomes insignificant after controlling for fallowing decisions, household, spacial and year fixed effects

  • can also control for if household comes from inherited office or not to account for reverse causality

• might still have endogeneity if productivity is correlated with political status

  • can control for if land comes from chief allocation or commercial for given individual, commercially purchased plots have property rights regardless of political standing, results robust

study on alternative channel though which property rights can increase economic welfare

Field (2007)

• looks at efficiency gains from transferring role of property protection to the state

• opportunity cost higher in urban settings

• Peru gave formal property rights to more than a million people at staggered intervals

• diff in diff comparing labour supply of benefeciearies across neighbourhoods that received compared to did not receive

  • as always this requires parallel trends assumption

  • need treatment to be random

• results show that households work more hours, and work further away and not at home, around 13 hours worked more per household per week

two hypotheses for the creation of permanent labour contracts

• Monitoring hypothesis: can hold individuals accountable, thus permanent contracts more likely for activities that are harder to monitor

• Insurance hypothesis: risk averse workers value consumption smoothing so it is cheaper to hire workers on a fixed basis for risk neutral employer

impact of productivity shocks to poor study

Jayachandran (2006)

• productivity shocks lead to larger change in wage if labour supply is inelastic which occurs if workers are close to subsistence, more credit-constraint, less able to migrate, which this paper tests for

• income effect dominates substitution effect

• rainfall used as exogenous productivity change

• use model given on ipad to find theoretical results of relationship between poverty, ability to smooth consumption and migration

• Proposition 1: wage elasticity increasing in poverty, poverty given as ratio of subsistence level to average productivity

• Proposition 2: wage elasticity incresing in banking cost

• Proposition 3: wage elasticity is increasing in migration cost

  • intensive margin: effect of local labour supplied

  • extensive margin: effect on migration

• Use of instrumental variable

  • crop yields depend on productivity but also on other inputs and amount of labour

  • use rainfall as IV for crop yield

    • relevance and exclusion restriction ( rainfall can must only effect wages through its effect on productivity

• presence of banks and roads could also be correlated with omitted variables

• Results

  • wage responds to rainfall shock since lot of casual employment

  • baking presence using different measures leads to lower wage elasticity

  • same with cost of migration measured by things like road density, railway closeness to city

  • poverty proposition is not significant, but higher the proportion of landless people the more wage is inelastic

three views on the role of microenterprises in development

La Porta and Shleifer (2008)

• Romantic view

  • entrepreneurs are constraint by environment, financial and regulatory

• Parasite view

  • small firms can only exist because they exist in the informal sector and avoid taxation and regulation so can afford to be unproductive

• Dual economy view

  • small and large firms use different human capital. high quality goes to large firms whilst small firms have decreasing returns to scale and are only focused on covering subsistence

study on if small and informal firms are credit constraint and this is hindering them

de Mel (2008)

• RCT of different cash grant amounts or capital grants

• measure change to profits

• find positive impacts of both grants to capital and profit, also hours increased a bit for some grants, but this stops us from measuring the effect of capital purely

• thus need IV, try to net out changes in profits that are due to labour profits

• find that value of return to capital which is around 60% is greater than market real interest rate implying credit constraint

study on the effect of management on business performance

Bloom et al. (2013)

• textile plants in india received five months of management consulting randomly

  • control plants only got one month of diagnostic consulting (say whats wrong but dont help)

• basic management practises were implemented and this led to

  • quality defects fell by 50%

  • inventory fell by quarter

  • output rose by nearly 10%

  • TFP rose by 15%

• so why dont more firms adopt

  • didnt think changes would impact profits for basic changes they were aware of

  • some consulting suggestions firms were unaware of

• reaon why low productivity is sustainable is due to low competiiton

Business training

• one of the most common forms of support given to small firms

• research has shown mixed results of this

• differences in quality and intensity of training as well as starting out conditions of firms play large impact on effectiveness of training

study on how conflict of interest between principal and agents leads to barriers of technology adoption

Atkin et al. (2017)

• more effective cutting of footballs

• three groups

  • tech group were allocated new technology

  • cash group were gien cash equal to new technology

  • control group given nothing

• found that almost no one in any group adopted new technology

• workers were resistant and misinformed owners about productivity benefits

  • this is due to pay structure as paid per unit but new technology reduces quantity in short run and potentially in long run

• second experiment suggested change to pay structure

  • although small sample results show better uptake of new technology if change to pay structure is implemented

importance of long term realtionships and contract enforcement, study

Machiavello and Morjaria (2016)

• when efficient legal remedies and reliable information are missing buyers are less willing to switch to low cost sellers

  • especially in the absence of formal contract enforcement

• Kenya exporters to foreign buyers do not have enforceable contracts

• flowers is better to have contract, sellers can plan production, buyers can combine into bouqets more easily if they know supplu

• thoeretical model

  • no contract enforcement, uncertainty over seller type ( reliable or unreliable)

  • rational contract must be such that neither party has incentive to change

  • two tehoretical results

    • if lack of enforcement atters volume of trade is limited by incentive compatability constraint so decreases with temptation to deviate

    • learn about sellers type means value of relationship increases with age of relationsihp

    • can also include violence which is negative shock, inverted U shape between deliveries with violence and age of relationship, first increases as relationship is valuable so want to keep it so go through extra effort to still deliver, if relationship is sufficiently old sellers ability has been proven so can get away with not delivering in one perio

• backed up by empirics, sellers might have to accept initial losses to acquire good relationship

two aspects of corruption that can effect firm performance

• value of political connections

• incidence and level of graft, which firms must pay bribes and how much do they pay

study on estimating the value of political connections

Fisman (2001)

• indonesia has highly centralised and stable political structure

• Event study triggered by rumors about poor health of president

• measure stock price of firms with varying connections to president

• give rating on the degree to which firms depend on political connections

• if rumors of very poor health then the most politically connected firms should see the largest drop in share price

• empirics support this

study on what exactly political connections help with, more specifically if they help with preferential lending

Khwaja and Mian (2005)

• political connectiveness measured by if board has a politician

• fixed effects

• measure if borrowing from government or pivate bank is different

• resutls

  • bigger loans, worse return to banks

  • results most striking for government bank, results even negative for private banks

• another strategy is exploiting changes in political connectedness within the same firm

  • allows to control for firm fixed effects

  • use variation in terms of electoral success of politicians party

  • measure changes to loan size within same firm

  • results support this idea, both when politician is in power and when politicians party is in power

  • power of politician also matters

what determines size and sector that pays bribes study

Svensson (2003)

• large variation in bribes theorised to be due to control rights and bargaining strength across firms