How Concessional is Aid Lending?

Abstract

The method used by Development Assistance Committee countries for measuring the concessionality of aid loans has remained unchanged for nearly 20 years. It was designed to measure the net cost of aid to donors not the net benefit to recipients. The discount rate used takes no account of changes in the value of the currency of the loan or of changes in prices for goods traded by recipient countries. Furthermore, it does not consider the implications of tying of aid or of policy conditionality. This paper suggests an alternative measure that shows the real net benefit of aid finance to recipients. It argues that the discount rate used by the Development Assistance Committee is too high and that changes in the value of the currency in which a loan is taken out can be important. Nevertheless, real rates of interest for developing countries remain surprisingly high despite low nominal rates due to falling prices of traded goods. This finding has implications for the future real cost of debt service to recipients.

Acknowledgements

The authors are grateful to John Weiss, Mike Tribe, Sam Cameron and two anonymous reviewers for helpful comments on an earlier draft of this paper.

Notes

1. The issue of immediate disbursement is not considered here. Delayed disbursement usually worsens the benefits to the borrower.

2. Estimation of the grant element from the point of view of donor cost is legitimate if we are trying to measure donor generosity but, if the stated purpose of aid is to provide assistance to the recipients then it is the benefit to the recipient country that is of more importance in evaluating aid. In this respect, the purpose for which the measure is used is a relevant issue.

3. For example neither Uganda nor Ethiopia allow public sector projects to use external funding that is not deemed to be concessional and in defining the meaning of concessional the grant element calculation is used.

4. Interestingly, the International Development Association (IDA) of the World Bank uses a similar approach to determining the discount rate in estimating the grant element of loans (that is, the opportunity cost of capital to the lender) but derives a different value. For the IDA, the discount rate is determined by ‘IDA's long term return on liquid assets, which is currently 6 per cent for five years ended in FY 2003’ (World Bank, 2004: 14).

5. While there has been extensive criticism of the practice of using high discount rates in relation to environmental issues (see Cline, 1993; Price, 1993; Potts, 1994; Livingstone and Tribe, 1995; Kula, 1997); the literature on aid concessionality has tended to focus on issues such as tying of aid and conditionality (see Morrissey and White, 1996; Killick, 1997; White and Morrissey, 1997; Morrissey, 1998; Maxwell and Riddell, 1998; Leandro and Schafer, 1999; Morrissey, 2002; White, 2002) and has generally not challenged the underlying assumption that the discount rate to be used in calculating concessionality should be 10 per cent. A notable exception to this is Raffer and Singer (1996: 10) who point out that a 10 per cent discount rate at recent nominal interest rates potentially allows donors to issue government bonds at a typical rate of 6 per cent and lend these sums on as ODA at no cost to themselves.

6. It should be noted that reference to commercial rates assumes that recipient countries are sufficiently credit worthy to have access to commercial funds. This is not always the case. Where access to the international capital market is not available, an alternative estimate would be the domestic opportunity cost of capital, that is the IRR on the marginal project excluded from the public sector investment programme. In such cases it could be argued that the opportunity cost of capital should be higher because of shortage of funds. On the other hand, it could also be argued that a country that was actually able to achieve a consistently high rate of return on public sector projects would probably be able to borrow commercially.

7. A one year LIBOR rate is also available but not for the whole period under consideration. A similar approach to the one used here is described in Potts (1994).

8. It should be noted that the fall in the prices of traded goods affects both the real cost of concessional borrowing and the real cost of commercial borrowing. It is, therefore, neutral in its direct effect on concessionality but increases the overall burden of servicing loans. However, the fall in the level of the nominal rate of interest in recent years is related to the low level of inflation and so it could be argued that these circumstances have contributed to the reduction in the difference between commercial rates and concessional rates and, therefore, to reduced concessionality.

9. It is common for a risk premium to be added to the interest rate paid on loans taken by organisations from countries where credit worthiness is regarded as an issue by lenders. Strictly speaking, the premium would be added to the nominal rate so the 1 per cent premium used in this example should actually be 1.01 per cent if the rate of inflation applied in estimation of the real rate of interest is −1.0 per cent. Clearly this is not significant in determining a discount rate to the nearest whole percentage point.

10. Our argument is that commercial loans are a relevant alternative to aid lending and the relative importance of such loans is increasing. However, aid loans are subject to government guarantees, so comparison with loans without such a guarantee is not appropriate. Griffith-Jones and Fuzzo de Lima (2004: 7) cite evidence based on World Bank data on the difference between the interest rate spread (over the US Treasury bill rate) on guaranteed loans and non-guaranteed loans. The US Treasury bill rate was on average (1978–2005) about 1.2 per cent below the $US LIBOR rate for one year (International Monetary Fund, 2006). The simple average of the spread for nine developing countries for guaranteed infrastructure loans was 2.2 per cent above the Treasury bill rate or about 1 per cent above the LIBOR rate. The simple average rate for unguaranteed loans was 3.9 or 2.7 per cent above the LIBOR rate but these loans are not strictly comparable. A premium of about 1 per cent is therefore justified as a reference point, if guaranteed commercial borrowing is seen as the next best alternative to a concessional aid loan. Guaranteed loans still carry some risk because of the possibility of government default and the transaction costs of invoking the guarantee.

11. Extending the end date for the period under consideration from 2003 to 2005 resulted in a decrease in the average rate of depreciation of the Korean Won from 3.9 to 2.8 per cent suggesting that the results can be quite sensitive to the choice of time period and reinforcing the point that what is required for operational use is a forecast rather than a simple extrapolation from the past. On the other hand, the effect on the value for the appreciation of the Japanese Yen was relatively small, increasing from 1.7 to 1.8 per cent.

12. The Won depreciated drastically following the Asian crisis of 1997 but then appreciated from 1999–2000 and has been relatively stable since then. The value used for the rate of depreciation is dependent on the beginning and end years and is, therefore, not distorted by the unusual circumstances of 1997 and its immediate aftermath, however the change brought about by extending the period by two years suggests that the recovery of the Korean exchange rate may not yet be complete.