The financial architecture for poor countries
Benu Schneider (ODI).
Jonathan Beynon (DFID).
Daniel Cohen (Ecole Normale Supérieure).
1. Jonathan Beynon focused on the impact of aid on growth, conditionality and selectivity. He highlighted two opposing views in the current debate:
(i) The first and more predominant view is that aid only works when government policies are good and a more selective allocation of aid to 'good policy-high poverty' countries will lead to larger reductions in poverty.
(ii) The second view argues that aid effectiveness is not conditional on policy and the implications of the former on more selective allocation is a matter of great concern.
2. The World Bank (WB) position, discussed in their Assessing Aid research report of 1998, largely concurs with the first view. The report concludes that aid is only effective in accelerating growth when the quality of economic management is good. But "Assessing Aid" also found out that the pattern of the actual aid allocations - particularly bilateral aid was highly inefficient, being only weakly targeted at poor countries and even less at well managed countries. Therefore more people could be lifted out of poverty through a more efficient allocation of aid to 'good policy-high poverty' countries.
3. The World Bank perspective seems to have been reinforced by recent empirical work from Collier/Dollar (CD) poverty efficient allocation models which conclude that aid only works in a good policy environment. CD replaced the narrow policy index of the earlier World Bank analysis with the broader 'Country Policy and Institutional Assessment' (CPIA) measure, then estimated the impact of growth on poverty reduction and then optimised aid allocations between countries so as to maximise the number of people lifted out of poverty. The results showed that a more efficient targeting of aid towards countries with high rates of poverty pursuing good policies could double the number of people lifted out of poverty, as much as could be achieved by tripling of present aid budgets under current allocations. A comparison of actual aid allocations with CD efficient poverty allocations suggests that South and central Asia appear to be under-funded, the share going to sub-Saharan Africa would also rise in a poverty efficient world and East Asia is the only other region that would receive aid of any significance.
4. This analysis and policy recommendations, however, has been challenged on methodological and econometric basis, notably by Hansen and Tarp (HT), who argue that the original Burnside/Dollar results are very sensitive to model re-specification and that the impact of aid on growth is positive irrespective of policy environment. Furthermore, aid assists countries to adjust to external shocks and this may explain why some studies show aid to have no significant impact on growth. It is worth noting that an extensive literature review generally supports the view that the impact of aid has in fact been reasonably good and is getting better.
5. More challenges on the emphasis on policy-led aid allocations are that:
(i) CD's own model confirms that the impact of reallocating aid on the basis of a poverty criteria is bigger than reallocating aid on the basis of policy.
(ii) The World Bank's evidence that aid is fungible and that ex-ante conditionality is ineffective can also be questioned.
(iii) Growth is not the only route to poverty reduction nor is growth the only benefit of aid - other benefits include, inter alia, health, educational, distributional effects, environmental development, empowerment and security;
(iv) Adopting the poverty reduction strategy at the level of each country as opposed to a single global target could significantly alter the pattern of poverty efficient aid allocations.
(v) Jonathan pointed out that these concerns plus CD's own acknowledgement that their recommendations are at best a guide have led to some watering down of the World Bank's emphasis on policy as the basis for allocating aid.
6. In conclusion, Jonathan said that the evidence shows that there is still a strong case for aid. However the following policy implications are important for the future viability of aid as a strategic instrument in financing for development:
(i) Aid allocation procedures should be reformed, more rigorously - in an objective and transparent manner taking into account measures of aid need and aid effectiveness to better guide and justify the allocation of aid budgets.
(ii) There is need for a significant shift in aid from middle income towards low income countries such as South Asia and Sub Saharan Africa. Some assistance will have to be targeted towards less poor but populous countries which still have large numbers of poor people (which tend to get left out in the CD models) but primarily for pilot projects and technical assistance than large scale financial transfers.
(iii) Some greater emphasis on 'good policy' countries is warranted but, in assessing policy, donors need to make allowance for the impact of external shocks, while there remains a powerful case for continued engagement in poorly performing countries. Generally, this will mean shifting aid from Eastern Europe, Central Asia, Latin America and the Middle East to South Asia (especially. India) and to Sub-Saharan Africa.
(iv) There is need for a greater degree of ex-post selectivity and flexibility in aid allocations, coupled with shifts to more flexible forms of aid delivery (budget/program support), harmonising donor procedures and improving donor co-ordination.
(v) Evidence that the impact of aid has in fact been reasonably good and is getting better provides powerful support for the case for more aid. This case would be reinforced by, and perhaps dependent on donors being seen to be seeking to improve the allocation of their existing budgets.
7. The strength of each of these policy recommendations will vary between donors, depending on individual mandates and political priorities, areas of comparative advantage and existing patterns of and processes of aid allocation.
For a more detailed examination of the issues raised here, please see Jonathan Beynon's recent paper Policy Implications for Aid Allocations of Recent Research on Aid Effectiveness and Selectivity.
8. The discussion focused on the following:
(i). What should be regarded as a good policy environment - so far mainly based on public sector management, macroeconomic, structural and social policies, some form CPIA weighting.
(ii) The need for more flexible transparent aid delivery systems such as budget support, more harmonised donor procedures and decentralised decision making within aid agencies.
(iii) The need for instruments and procedures to prevent aid from crowding out domestic aid and vice versa.
(iv)It was noted that it is important to highlight that good policies generally enhance the positive impact of aid.
(v) The matter of the various types of aid and respective impact still has to be streamlined. What should be included in the calculation of true aid, how should technical aid, humanitarian aid be treated? There may be need to dis-aggregate aid and take each element separately. Jonathan noted that this matter is unresolved among donors up to the World Bank level.
(vi) It was highlighted that policies and recommendations seem to have remained the same for the past three decades. There is therefore need to encourage more collaborative work and consensus at domestic and international levels.
9. The main message of Daniel Cohen's presentation was that expectations of the potential impact of HIPC are overstated.
10. His presentation was based on his recent paper The HIPC Initiative: True and False Promises (OECD Development Centre, Technical Papers No. 166, October 2000). Cohen contended that, if properly managed, HIPC could potentially have a similar positive impact on very poor countries to that which the Brady deal had on middle income such as Mexico in the late 1980s . However, Cohen argued that current analyses of HIPC lack perspective on the 'market value' of debt to be written down, thus ignoring the crux of the argument that emerged in the literature surrounding the Brady deals. In the case of HIPC, the net present value of debt is considered, leading calculations to re-assess the discount factor and correct for the conditionality of the debt but not take into account the risk of non-payments. Consequently, expectations of the potential effects of HIPC overstate its actual reach. More importantly, by overvaluing the amount of debt write-down, donor countries may be induced to lower the additional of aid granted to these countries.
11. If one considers a write-down of debt owed by a HIPC country from an initial debt to exports ration (D/X) of 300% to 150%, representing the new target as set out in the Cologne Summit (1999), the impression is that the country is effectively receiving a huge debt reduction, and thus a significant resource transfer, equivalent to 150% of its exports. However, it is clear that these numbers confuse face values with real values; paper money with true resources.
12. In comparison, in the case of the Brady deal, which was largely concerned with commercial debt, one sees that it was dealt on the secondary market, where investors' expectations on the outcome of the claims translated into prices over the same claims. Obviously, debt prices in the secondary market were particularly low during this period after the 1980s crisis since investors expected only a certain percentage (in the case of some countries as low as 10%) of debt to be recouped.
13. Daniel made a fundamental distinction: the average price embodied in the market price and the marginal price corresponds to the effect at the margin of reducing the debt. If D is the outstanding debt of a HIPC country and X is the amount of resources creditors expect that country to produce, the market price of debt equals X/D. Since the creditors as a whole will never receive more than X, getting an extra dollar at the margin will not increase their total profit. Therefore, a write-down of debt will not bring about any change in the overall market value of debt, as long as the outstanding debt is still smaller than X.
14. If a HIPC country continues to have to pay the same amount of X after a deal, the net benefit to that country from a debt reduction may be nil. This is expressed by the marginal price dV/dD, which indicates at the margin how much debt reduction can bring in expected payment to a country. As long as D>X, the marginal price will stay at zero and, consequently, there will be no change in the overall market value of debt (dV=0). An example of this can be found in Bolivia after the 1980s debt crisis: sponsored by donors, the government repurchased half the face value of its debt on the market, with the latter responding by doubling the market value of the remaining debt (from 5 cents to 10 cents to the dollar). Bolivia had, therefore, paid for nothing.
The above points clearly demonstrate the need to distinguish a strict nominal accounting of debt write off from a 'market based' account, which analyses the extent to which expected forthcoming payments are reduced.
15. Cohen argued that, in contrast to the middle income countries involved in the Brady Plan, HIPCs debt is mainly owed to public donors and there is no secondary market where the price of debt can be assessed. Therefore, the market value of debt has to be estimated. In response to this, Cohen has developed an econometric exercise that attempts to estimate secondary market prices of middle income debtors in the late 1980s.
16. In order to begin empirical analysis, one needs to establish a minimum benchmark above which debt should be interpreted as being too large. Three rations are usually taken into account: debt-to-export ration; debt-to-GDP ratio; and debt-t-tax receipts. HIPC mainly relies on the first ratio, with the benchmark reduced from 200% to 150% at the Cologne Summit. There are several methods to identify the critical benchmarks for the different ratios. One can look at the average debt ratios at the time of the first rescheduling; at the share of debt countries were servicing at the climax of the debt crisis; at the primary surpluses accumulated by large debtors to stabilise their debt-to-export ratios; or by an estimation of profit model that considers among its regressors several of these variables, together with other variables such as the degree of openness of a country and the level of liquidity in its economy. The outcome of the estimation (i.e. the ratios for which the risk of a debt crisis exceeds 60%) are shown in the following table. Above these thresholds, the debt is expected to become valueless.
17. It is also possible to estimate the secondary market price of debt and Cohen carried out such estimation at the time of the Brady-deals. The estimated model relates the price of the debt to several variables, among which are the D/X ratio and other crisis-indicators, such as the amount of arrears and rescheduling a country has had. The results are shown in the table below and clearly shows that the price diminishes as the debt-to-export ratio rises and/or the country experiences a crisis.
Cohen posed the question: how can it be that with a D/X of 200% a country is considered as solvent and the price is 100%, while after the crisis the price is less than 30%? In response, Cohen believes there are basically two lines of explanation: one is based on the concepts of 'multiple equilibria' and 'self-fulfilling expectations' and the another is based on the concept of 'myopic markets'.
18. Finally, one can compute the market values of the debt reduction based on the estimates from Brady-debt-data. The following table shows the initial debt-to-export ratio (D/X)o and the market value reduction in percentage of export (dV/X), both in Cologne (D/X=150% target) and Pre-Cologne (D/X=200% target) terms.
nitially, HIPC was established to write down the outstanding debt to 200%. Thus, Cohen demonstrated that if a HIPC country has reduced its debt-to-export ratio from 300% to 200%, in terms of export, the value of the write-down would not be 100%, as largely believed, but less than 3%!
In Cologne terms, the initiative is more significant. Bringing down a country's D/X ratio from 300% to 150% implies a resource transfer amounting to about 11% of its export, which, although still a considerable relief, is still far below the 150% as the face value would suggest.
19. Cohen finished his presentation by outlining some of the policy issues emerging from his analysis:
(i) As the correct prices of debt is important, if donor countries base their book keeping on face values instead of market values, other aid might easily get wiped out, which would have a detrimental effect on countries in terms of net resource transfers.
(ii) HIPC makes countries prone to crisis by bringing D/X down to 150%, only 50% below the critical threshold (with the debt becoming value-less). Total cancellation of debt would be a better way of proceeding because the average price could be applied immediately (since it shows how much the debt is worth). Countries could, perhaps, be given the possibility of purchasing (at a reasonable discount) more debt than that reduced by the target of 150%.
(iii) The Meltzer Commission argued that subsidised loans to developing countries basically represents a grant and, there, the World Bank should just provide grants in the future, accounting them as such in the budget, instead of entering them in the books in terms of loans. Cohen believes that this position is too extreme, particularly as many developing countries have managed to repay their debts (e.g. Korea during the 50s-60s). Rather, Cohen argues that the World Bank should separate grants from loans at commercial terms (instead of giving concessional loans) and perhaps let the World Bank market these loans after some time.
(iv) Collective action among creditors is a must. Since no individual creditor intends to bear the burden of a loss, policies easing collective action have to be implemented.
20. The discussion focused on the following:
(i) Concern was raised at the suggestion of a new approach to aid in the form of grants instead of loans e.g. by World Bank especially since it is almost certain that HIPCs would not be able to repay.
(ii) The use of stock data is a better benchmark which reflects how close a country is to insolvency compared to indicators such as GDP, imports, exports etc which are flows reflecting the level of liquidity.
(iii) It was suggested that the current approach to HIPC could be improved by the setting up of sovereign debt agency that would administer cancellations to ensure objectivity, transparency and compliance. It would also deal with the problem of moral hazard in the case of using instruments like debt buy-backs.
(iv) Freezing repayments to allow debtors to recover may not work as it creates a vicious circle of rescheduling, and borrowing to repay accumulating debt.
21. The main conclusion of this seminar was a persuasive case for more aid and that measures were needed to improve the public's perception of aid.
This event investigated two key questions: is there a case for aid today?; and the financial architecture for poor countries, does HIPC pave the way.