Aid Lecture Slides 2

The Economics of Foreign AidLecture 2 “Political Economy of Aid” : 

The Economics of Foreign Aid Lecture 2 “Political Economy of Aid” Justin Sandefur Weeks 7 & 8, HT 2006 “Economics of Developing Countries” Option

Last week: “Poverty Efficient Allocation of Aid”: 

Last week: “Poverty Efficient Allocation of Aid” Building on Burnside & Dollar (2001): If aid only works in a good policy environment, how should we allocate aid? Collier & Dollar (2002) compute a “poverty efficient aid allocation” on the basis of the Burnside & Dollar results Selectivity: allocate aid to poor countries (as opposed to allies, etc.) with good policies already in place (as opposed to funding reform efforts).

“Poverty-efficient aid allocation”: 

“Poverty-efficient aid allocation”

“Poverty-efficient aid allocation”: 

“Poverty-efficient aid allocation”

Key assumptions underlying Collier & Dollar allocation: 

Key assumptions underlying Collier & Dollar allocation Impact of aid on growth depends on the policy environment Aid is fungible Conditionality doesn’t work Growth is good for the poor (and policies which are good for growth are good for poverty reduction) Reasonable assumptions?

Outline for Today:: 

Outline for Today: Political Economy of Aid: Fungibility Conditionality & Contract Theory Moral Hazard Time Inconsistency Evidence on Conditionality Dependency: Is Aid Oil? Conclusions

Fungibility: 

Fungibility “When a thing which is the subject of an obligation…must be delivered in specie, the thing is not fungible, i.e. that very thing, and not another thing of the same or another class in lieu of it must be delivered.” (OED) If a donor gives aid for education, will it actually improve education in the recipient country? If a donor gives aid for education, will it increase gov. spending at all? NB: This NOT (necessarily) corruption!

Conditionality: 

Conditionality “Underlying the litany of Africa’s development problems is a crisis of governance.” WB 1989 1980s: ‘Washington consensus’ focus on getting prices/incentives right, rather than pouring in resources => trend toward conditionality. 1990s/2000s: Disillusionment “It seems clear that the lending cum conditionality process works well only when local polities have decided, largely on their own, possibly with outside technical help, to address their reform needs, effect certain policy changes sequentially, and approach the international community for financial help in getting there.” (Ranis, 1995)

Analytical Framework to Analyze Conditionality: Aid as a Contract: 

Analytical Framework to Analyze Conditionality: Aid as a Contract The Principal-Agent Problem Principal: (donor) designs contract to achieve some end (poverty reduction) Contract is generally and aid/policy combination {a,p}. Agent: (recipient gov) accepts or rejects contract and chooses actions in light of contract incentives Moral hazard, adverse selection, and time inconsistency all addressed in this setting.

Moral Hazard: 

Moral Hazard Hidden action: agent’s action (poverty reduction efforts) imperfectly observed by principal. Azam & Laffont (2003) Donor utility Recipient utility Theta<1 indicates recipient cares less about consumption of the poor

Moral Hazard: 

Moral Hazard Autarky (no aid) Some benchmark level of consumption by rich cRA and poor cPA Full observability. Donor offers a ‘conditional’ contract, {a, cP > cPA}. Same is possible with imperfect observability, so long as recipient is risk neutral. Favoritism Interesting case: recipient government has a preference for one sub group of the poor. Will allocate cP1 > cP2 if unconstrained. Even with full observability, optimal contract will only reduce but not eliminate favoritism. Risk aversion? Standard result is moral hazard literature is that if there is a noisy signal of effort and agents are risk averse, the optimal contract will not achieve the 1st best. Not clear how this applies here. Main point: Conditional aid well-suited to address moral hazard. Caveat: Strong assumptions about commitment (time consistency).

Time Inconsistency: 

Time Inconsistency “Over the past few years Kenya has performed a curious mating ritual with its aid donors. The steps are: one, Kenya wins its yearly pledges of foreign aid. Two, the government begins to misbehave, backtracking on economic reform and behaving in an authoritarian manner. Three, a new meeting of donor countries looms with exasperate foreign governments preparing their sharp rebukes. Four, Kenya pulls a placatory rabbit out of the hat. Five, the donors are mollified and the aid is pledged. The whole dance starts again.” (The Economist, 1995)

Time Inconsistency: 

Time Inconsistency Occurs when it is not in the best interest of a player (donor) to carry out a threat or promise that was initially designed to influence another player’s actions (recipient gov). If economic policymakers lack the ability to commit in advance to a specific decision rule, they will often not implement the most desirable policy later on. Familiar example: Kydland & Prescott, monetary policy and inflation expectations. “Samaritan’s Dilemma”: special case where charitable urge undermines beneficiary’s incentives to be cautious

“When is foreign aid policy credible?” Svensson (2000): 

“When is foreign aid policy credible?” Svensson (2000) Similar setup to Azam & Laffont: both donor and recipients have preferences for own consumption & poverty reduction. 2 recipients ‘competing’ for aid. Donor chooses an aid policy to encourage policy reform leading to poverty reduction. Nature creates noise between reform effort and poverty reduction outcome. Two key considerations: Can the donor observe recipients’ reform efforts? Can the donor commit ex ante to a (conditional) aid allocation in the following period?

Time Inconsistency: 1st Best: 

Time Inconsistency: 1st Best If answer to both questions is yes… i.e., observe effort and able to commit First best outcome: High aid level (direct poverty effect) Higher reform effort than withot aid (indirect effect) Donor bears all risk (from nature’s shocks) Requires…

Time Inconsistency: 2nd Best: 

Time Inconsistency: 2nd Best Suppose donor has only a noisy signal of reform effort, but can still commit. What’s the best they can do? Second Best Outcome: Risk is shared between donor & recipient Optimal effort lower than 1st Best, higher than autarky. Recipients better off than 1st Best; Donor (and poor) worse off.

Time Inconsistency: 3rd Best: 

Time Inconsistency: 3rd Best Finally, return to perfect information, but suppose donor can’t commit: Third Best Outcome: 1st Best allocation of aid is achieved GIVEN the level of policy reform But reform level is reduced. Recipient knows donor will give aid in second period regardless of effort.

Potential Solutions to Time Inconsistency in Aid: 

Potential Solutions to Time Inconsistency in Aid Analogy to Rogoff’s Conservative Central Banker: improve welfare by delegating allocation to “an (international) agency with less aversion to poverty.” Tied Aid. Sign contracts with private companies to implement projects. Ties donors hands – can’t reward economic mismanagement.

Evidence on Conditionality: 

Evidence on Conditionality Is conditionality enforced? Svensson (1997) examines data on WB loan tranche releases. Finds that in majority of cases when conditionality is violated, tranche is still released. When does conditionality work? Dollar & Svensson (2000) examine the success and failure of structural adjustment programs (SAPs). Success = WB OED grade; inflation and bdgt deficit. Domestic political variables predict success (or failure) correctly in 75% of cases. (Democracy, gov tenure, ethnic fractionalization, etc.) Does aid reinforce reform? Sachs (1994) finds that while political factors set the stage for reform, aid has helped to reinforce successful reforms.

Dependency: 

Dependency Basic argument: two reasons politicians listen to their constituents: Constituents’ votes reelect them Constituents’ taxes pay their salaries Friedman 1957: by strengthening governments “at the expense of the private sector” aid will “reduce pressure on the government to maintain an environment favorable to private enterprise.”

Variants of Dependency Argument:: 

Variants of Dependency Argument: Aid reduces incentives to adopt good policies. Aid overwhelms the administrative capacity of recipient governments. Aid reduces accountability to citizens. Aid works like ‘welfare dependency’, implicitly a high marginal tax rate. Aid flows are highly volatile and thus a source of volatility rather than sustained growth.

Dependency continued: Aid & Governance: 

Dependency continued: Aid & Governance Brautigam & Knack attempt to test the basic Friedman & Bauer hypothesis: aid undermines good governance. Mechanisms (somewhat vague) Funds political patronage Creates moral hazard (see above) Soft budget constraints => lack of fiscal discipline Econometric test: Data on 32 sub-Saharan African countries, averaged over 1982-1997. Regress change in ICRG Corruption measure (and later, tax revenues as % of GNP) on aid and controls. Find a signficant NEGATIVE effect of aid on governance, particularly after instrumenting aid.

Dependency: “Is Aid Oil?”: 

Dependency: “Is Aid Oil?” Collier (2005) notes that aid and oil are the 2 largest external resource flows to Africa. Large literature on the “natural resource curse”, both in terms of monetary economics, and political economics. Why is aid different? Technical Assistance. Projects & knowledge transfer Conditionality

Review & Conclusions (1 of 2): 

Review & Conclusions (1 of 2) Aid allocation (actual): primary driven by strategic considerations Aid and growth: A weak positive relationship in cross-country data, subject to diminishing returns. Some evidence that this relationship is conditional on good policies – though may be spurious. Aid allocation (ideal): taking the aid-x-policy result seriously, aid should be allocated on the basis of income and policy to achieve maximim poverty reduction. Political economy considerations Moral hazard: conflict between donor and recipient (gov) preferences suggest a productive role for aid conditionality. Time inconsistency renders this result somewhat suspect. Dependency: some weak evidence which should possibly cause concern.

Conclusions (2 of 2): 

Conclusions (2 of 2) Overall message: History of aid effectiveness is dismal. Has not lived up to expectations in terms of growth or poverty reduction. “Past performance is no indication of future returns.” Precisely because… Aid has not been allocated for growth or poverty reduction Aid modalities have not necessarily been target at growth Aid policy has not been credible. Conditionality has not been enforced. Emphasis has been placed on large resource transfers, ignoring domestic political considerations. Inasmuch as we can recognize these failures, we shouldn’t be forced to repeat them.