IMF subsidies, cancellations, and resumptions: new empirical evidence.

• Author: Urbaczka, Adrian
• Position: Report

For a long time, the International Monetary Fund has been criticized for subsidizing its credits. According to Walter Bagehot (1873), a lender of last resort ought to "lend freely but at a penalty.'" Otherwise moral hazard results (see Dreher and Vaubel 2004). Bakker and Schrijvers (2000) and the Saxton Report (9,002) have presented estimates of the subsidy element in IMF lending. In this article, we present an improved and updated calculation.

We also present evidence on another criticism of IMF policy: that it fails to enforce compliance with policy conditions. The IMF claims that it cancels its programs if debtor governments do not honor their policy commitments. We show that cancellations due to noncompliance tend to be followed by new programs very soon.

Measuring the Subsidy in IMF Credits

Bakker and Schrijvers (2000) tried to quantify the subsidies implicit in the IMF credits that the seven main borrowers received during the Asian financial crisis of 1997-99. They compared the IMFs adjusted rate of charge with the Emerging Market Bond Index (EMBI) published by J.P. Morgan. Bakker and Schrijvers reported yield spreads between 5.7 percentage points in 1997 and 9.5 percentage points in 1999. Multiplying those yield spreads with the amounts of the loans, they obtained a subsidy of $8.5 billion. However, their methodology was suboptimal. First, instead of using the national subindices of the EMBI, they used the overall EMBI index that includes many other emerging economies. Second, they did not look at the actual timing of the payments of interest and principal but assumed disbursements for the beginning of each year. Third and most important, they seem to have compared dollar interest rates and SDR interest rates without allowing for expected or actual exchange rate changes.

The Saxton Report (2002), commissioned by the Joint Economic Committee of U.S. Congress under its chairman, Rep. Jim Saxton, compared the EMBI Global with the yield of comparable U.S. Treasury securities. For the period 1995-2001, it found ex ante interest subsidies ranging from 1 percentage point, as in the ease of Thailand, to almost 30 percentage points for Russia. The approach of the Saxton Report is unsatisfactory in two respects. First, it does not use the national sub-indices of the EMBI even though these are available. Second, it assumes, but does not test for, covered interest parity between dollar and SDR interest rates.

Our analysis improves on the Saxton Report in two ways. First, our standard of comparison is the EMBI sub-index for the country in question. Second, we base our computations on the actual cash flow streams under the IMF programs. Thus, we do not estimate an expected or ex ante subsidy. We compute the actual or ex post subsidy.

The Cost of Borrowing from the IMF

Country-specific cash flow data are available back to May 1, 1984. We exclude the Fund's concessional lending facilities because most of the recipients do not have access to the world capital market. We confine our analysis to the Fund's Standby Arrangements (SBA) and the Extended Fund Facility (EFF), which represent the bulk of the IMF's regular nonconcessional lending activity. We include all countries whose outstanding SBA and EFF debt to the IMF has exceeded SDR 1 billion at some point in time between May 1, 1984, and February 28, 2011. Moreover, to provide complete information on current loans, we include all countries that have been indebted to the IMF under SBAs or in the EFF at the end of our period of observation. This yields a sample of 88 arrangements. The cost of funding is measured by the internal rate of return (IRR), which is also referred to as the yield-to-maturity. For this purpose, it is necessary to identify each borrowing arrangement's stream of cash flows 'along with the respective dates of payments. We have calculated each arrangement's internal rate of return, taking into account payments of principal, interest, and charges.

Some countries had more than one arrangement outstanding at a time. In those cases principal repayments were assigned to the oldest arrangement requiting redemption. Nonconcessional borrowing comes along with a coupon whose rate is linked to the SDR interest rate. Moreover, by taking into account any supplementary charges related to borrowing, referred to as "all-in" costs of funding, we determine the all-in internal rate of return (IRR*) for any arrangement in the sample.

If a country has more than one arrangement outstanding at the same time, the Fund aggregates payments of interest and charges over all outstanding arrangements on certain due dates in "joint transactions." Under these circumstances, we assign payments of interest and charges to the respective arrangement on a pro rata basis according to the share of the arrangement in the country's total outstanding SBA and EFF credit volume.

For the sake of simplicity, we consolidate all payments of interest and charges on a monthly basis on the 15th. Thus, the stun of potential deviations from the considered month's original single payments' settlement dates is minimized. Figure 1 shows the yields to maturity of the 88 arrangements in SDR terms.

The Cost of Funding in Capital Markets

The second essential component to measuring the subsidy of IMF credit is the cost of equivalent borrowing in the subsidy-free, open capital market. As already mentioned, J.P. Morgan provides a large set of emerging market bond indices reflecting secondary market conditions for government debt instruments, including country-specific sub-indices of the yield to maturity. However, three data problems remain. First, not every country in the sample comes with an EMB sub-index. Second, the EMB sub-indices for some major emerging market countries do not start before 1993. As a result of these data limitations the number of IMF credit arrangements for which we can quantify the subsidy drops to 23. Third, the maturities do not usually match those of the IMF credits.

To get a picture of the underlying mismatch of maturities, Figure 2 compares the distribution of maturities in the IMF...