Transcript
Page 1: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Development Aid and Portfolio Funds: Trends, volatility and

fragmentation

Paris

February 2009

Emmanuel Frot and Javier Santiso

Stockholm Institute of Transition Economics

OECD Development Centre

Page 2: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Research agenda in development finance

• Compare flows along several dimensions– Quantity: are portfolio capital flows more important than ODA?– Volatility: is ODA a more stable source of income than portfolio capital flows?– Efficiency: is the fragmentation of ODA increasing?

• Complementarity between flows– Pro or counter-cyclical role of ODA (with respect to private capital flows)– Can we expect ODA to insure against capital flows shortfall in the future?– In the current crisis, what role is there for ODA if we consider the sudden stop

of private flows?• Towards an aid efficiency index?

Different sources of finance for developing countries

Page 3: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

ODA and capital flows quantities

• Growth of all inflows– FDI and remittances increased

much faster– FDI and remittances are more

important than ODA.– Equity flows almost as high as

ODA recently.• Any fall in capital flows

expected to severely harm developing countries.

• Role for ODA

ODA has been less important than other external income sources in the past 15 years

Source: Authors based World Bank and OECD data.

Page 4: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• ODA is accused of being too volatile• But it is much less than capital

flows.• However:

– Its volatility has increased while those of private capital flows fell.

– In recent years ODA has actually been more volatile than remittances (during the year 2008 trend different).

Volatility

ODA is less volatile than capital flows, hence a potential role for cushioning against shocks

Volatility of flows, coefficient of variation, 1970-2006

Total ODA FDI Bond Equity Remittances Mean 0.67 1.23 2.35 3.06 0.74

Source: Authors based World Bank and OECD data.

Volatility of flows, by decade, 1960-2006

Total ODA

FDI Bond Equity Remittances

1960-1969 0.73 n.a n.a n.a n.a 1970-1979 0.61 0.99 1.79 2.20 0.36 1980-1989 0.37 0.97 1.74 2.19 0.38 1990-1999 0.46 0.90 1.83 1.86 0.50 2000-2006 0.46 0.88 1.72 1.83 0.41

Source: Authors based World Bank and OECD data.

Page 5: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

A shock in capital flows does not trigger an ODA inflow. ODA is not countercyclical.

Correlation between aid and capital flow shocks

Coefficients of correlation

FDI-5-year moving average of FDI

Bond-5-year moving average of Bond

Equity-5-year moving average of Equity

Remittances-5-year moving average of Remittances

ODA-5-year moving average of ODA

0.009 -0.04 -0.03 0.008

FDI-5-year moving average of FDI

0.12 0.09 0.19

Source: Authors based World Bank and OECD data.

Does ODA compensate for capital flow shortfalls?

• In the paper:– ODA and capital flows are

substitutes across countries: redistributive role of ODA.

– They are neither complements nor substitutes at the country level.

• A country experiencing a capital flow shock does not see a change in its ODA.

• No observed risk insurance against capital flow shocks of ODA.

Page 6: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Fragmentation has become more severe: aid efficiency is reduced

1,000,75

0,50

0,25

0,15

0,10

0,00Non-developing country / missing data

1970-1979

1990-1999

1980-1989

2000-2006

Aid fragmentation for recipients: Hirschman-Herfindahl index

Source: Authors based on OECD DAC data.

Page 7: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• DAC uses a different fragmentation measure for recipients (DAC10): number of donors that represent less than 10% of total receipts.

• Normalizing the Hirschman-Herfindahl index to make it comparable to DAC10 actually yields very similar results.

Aid fragmentation for recipients

Average recipient fragmentation 1960-2006

Source: Authors based on OECD DAC data.

Page 8: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• ODA has a role to play in a globalized world of development finance with large quantities of private capital flows.

• The current crisis reinforces the need for countercyclical mechanisms, particularly because of the importance of sudden stops of private flows.

• Aid fragmentation is severe whereas it increases transaction costs and reduces the value of aid.

• There are gains to be made by implementing coordination and adopting the recommendations of the Paris declaration.

• However ODA is a potentially more efficient tool in the development process. It does not currently use opportunities to smooth away variations harmful to developing countries. Even worse, it often adds to them.

Conclusion and policy implications

ODA is not exploited at its potential value

Page 9: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Herding in Development Aid Allocation

Paris

February 2009

Emmanuel Frot and Javier Santiso

Stockholm Institute of Transition Economics

OECD Development Centre

Page 10: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Are aid donors herding?

• Herding has been suspected for years.– Cassen (1986): donors move in herd, suddenly disbursing money into star countries,

and sudden increases are followed by long aid declines.– Riddell (2007): “herd instinct” among donors.

• But no study has ever quantified its size or its causes.• Herding is defined as the tendency for donors to follow the crowd, a

trend, to mimic each others’ decisions.– We look at simultaneous decisions about aid increases and decreases.– Even without herding this is expected because donors react to similar variables, and

we will control for this.– Other reasons are closer to those identified in finance: informational cascades,

strategic behaviors.

Many claims that they are, and that this decreases aid efficiency.

Page 11: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• Herding sometimes creates benefits for recipients (humanitarian aid following an emergency).

• But also costs– It provokes aid swings and so contributes to volatility.– It increases fragmentation if it results in many uncoordinated missions.– It potentially creates aid darlings and orphans:

• Reinhardt (2006): “I can't get IDB money if I drop the ball with the World Bank”.

• Additionally, we identify which variables prompt donors to increase aid and so we improve our understanding of donor allocation policies.

Motivation

Why is it important to learn about herding mechanisms?

Page 12: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Different types of herding

• Natural disasters cause donors to react similarly

– Though beneficial, it may cause an overflow.

• Debt relief is often granted in a coordinated fashion.

• Current crisis: sudden stops in capital flows should trigger simultaneous decisions by donors.

Beneficial herding: exogenous shocks trigger simultaneous reactions

Source: Authors based and OECD data.

Proportion of donors increasing aid to countries hit by the 2004 Asian tsunami

Page 13: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• A fall in capital flows does not induce more donors to increase aid.

Beneficial herding following capital flow sudden stopsNo evidence of beneficial herding shows no counter-cyclical decisions from donors

-100

01

002

00P

ropo

rtio

n o

f do

nors

incr

eas

ing

aid

-1000 -500 0 500 1000Capital Flows

Percentage change to a 3-year moving average, 1970-2006

-100

-50

05

01

00P

ropo

rtio

n o

f do

nors

incr

eas

ing

aid

-1000 -500 0 500 1000Capital Flows

Percentage change to a 3-year moving average, 2000-2006

Source: Authors based and OECD and WDI data.

Page 14: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Herding is detected by exploiting deviations from an average behaviour.

How to measure herding in aid allocation

• Carefully define aid (gross ODA net of debt relief, humanitarian and food aid) and carefully select a group of donors and recipients.

• Adopt a time horizon suitable for aid:– Look at 3 and 5-year periods to avoid capturing small changes.

• Idea behind the measure:– If “many” donors simultaneously increase or decrease aid to a recipient there is

herding.– How to define “many”? We need a benchmark: Global proportion of aid changes in a

period that are increases– If in a recipient-year the proportion of increases is far from the benchmark then it is

interpreted as herding.

Page 15: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Finance provides some guidance about herding measurement

• Use two measures from the finance literature that use this intuition.– LSV developed by Lakonishok, Shleifer and Vishny (1992).– h proposed by Frey, Herbst and Walter (2007) to correct for the downward bias

present in LSV.

• Results– All the measures adopted in the paper indicate that herding is present.– Using 3-year periods: LSV has a value around 3% (similar to financial markets), h has a

value close to 10%.– It means that if 50% of all allocation changes are increases then 60% (or 53%) of

donors take the same decisions for each recipient.

Two herding measures

Source: Authors based on OECD DAC data.

Page 16: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

Herding determinants

When do donors simultaneously decide to increase aid to a recipient?

• Estimate the effect of various shocks on aid allocation decision:– Economic growth– Political transitions– Natural disasters– Armed conflicts

• Results:– No effect of growth (decisions are neither pro nor counter-cyclical)– Positive effect of “new polity”– No effect of democratic transitions– Negative effect of authoritarian transitions– Positive effect of natural disasters– No significant effect of armed conflicts

Asymmetry in political transitions

Page 17: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• Netting out the effect of these determinants leaves the herding measures quite unaffected.

• The “corrected” levels are due to herding caused by strategic behaviours, informational cascades, etc.

• Such behaviours are more than anecdotal: Marysse et al. (2006) argue that political considerations and donor coordination problems have created aid darlings and orphans in the region of the Great Lakes in Africa.

How much do these determinants account for herding?

“Rational”, observables causes explain very little of herding

Original measure

New polity

Foreign intervention

Democratic transition

Authoritarian transition

Natural disasters

Conflicts

LSV 2.981 2.950 2.947 2.936 2.887 2.844 2.836

h 9.83 9.76 9.77 9.76 9.63 9.53 9.49

Source: Authors based and OECD data.

Page 18: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• Beneficial herding occurs for natural disasters but not for other expected reasons: capital flow stops, democratic transitions, recessions.

– Beneficial herding may also be harmful: humanitarian aid overflow is no panacea.• Once again aid allocation decisions are not found to be pro or counter-cyclical for many

variables: growth, democratic transitions, wars, capital flow stops.– Nancy Birdsall argued this week that because of capital flight, credit drying up, and declining

remittances, most developing countries will experience big shortfalls in revenue this year, and called for $1 trillion to be urgently unlocked. Given past experiences, this is unlikely to happen.

• Overall it seems herding does not occur for observable reasons, but more because of some unobserved motives: there is still a lot to understand about donor allocation policies.

• The aggregation of donors’ individual behaviours potentially leads to large aid swings and donor darlings/orphans.

• Coordination in the donor community would help to prevent such outcomes , or to create swings when they are most needed.

Policy Implications

Herding was expected to be caused by external shocks, but it is only weakly so.

Page 19: Development Aid and portfolio Funds: Trends, Volatility and Fragmentation

• Volatility and fragmentation have so far been measured at the country level: extend the analysis at the sector level.

– Is fragmentation more pronounced in some sectors? If yes, why?– Is volatility influenced by some sector specificities? What can be done about it?

• Similarly, herding is likely to be stronger in sectors– Fads and fashions

• These intermediate characteristics of donor allocation will ultimately be combined to build an aid efficiency index.

Future research

These results call for further investigation


Top Related