What’s the marginal product of capital in developing countries?

Financial officials are meeting in Washington, D.C., for the Spring meetings of the World Bank and the International Monetary Fund (IMF) between April 13 and 18, 2026. There will be a lot of commentary about the short-, medium-, and long-term impact of the war with Iran, as well as concern about the outflow of portfolio capital—some $70 billion exited from emerging markets and developing economies (EMDEs) in March alone.

This raises the obvious question: Do private capital markets know something that official finance lending institutions do not? Why are multilateral development banks lending when others are taking their money out as fast as they can?

The answer lies in the very different business models and risk profiles of official and private lenders. Official lenders bear very limited risk. Both the International Bank for Reconstruction and Development (IBRD) and the International Development Association (IDA) have a “no write-off” policy. IBRD has had a 0.7% annual default rate and an economic loss (no interest is charged during nonaccrual) estimated at 9.6% for each default, implying a total loss of less than 0.1% of its lending. IDA, similarly, has had very low losses, although it has written off credits when compensated for this by bilateral official creditors. Only two countries, Eritrea and Zimbabwe, are currently in arrears to IDA, putting the share of loans in nonaccrual status at approximately 0.4% as of end-2025.

For these multilateral lenders, then, the key question they should ask is whether their loans and credits are helping their clients.

The economic answer to this is to ask whether the marginal product (or rate of return) of capital (MPK) is higher or lower than the lending rate. Somewhat surprisingly, given the central importance of this figure, there are no published official data for this variable.

Inspired by a recent paper in the Quarterly Journal of Economics (QJE) by Pellegrino et al., we derive the MPK for over 100 developing countries. The methodology is a simple one of calibration.¹ We start by estimating the share of output attributable to capital as one minus the shares of income accruing to labor and natural resource rents. The MPK is then calculated as the product of the resulting capital share and the output-side GDP-capital stock ratio. The capital stock and GDP data are in 2021 PPP terms to facilitate comparison between countries. We took 2019 as the reference year, as this is the last year that precedes the COVID crisis with comprehensive data coverage. (MPKs do not dramatically change from year to year, although they may well change substantially over longer periods).

The results in Figure 1, rough though they be, are quite striking. Despite well-known issues of poor governance, weak institutions, and the limited access to complementary inputs identified by Robert Lucas (human capital and total factor productivity), the MPK for poor countries is well above the real lending rate for IBRD (and by definition above the IDA threshold as well). For comparison, the MPK for a developed country like the United States averages around 13% using this methodology.

The implication is that there is considerable welfare benefit to clients from multilateral lending, even given current policies. The strategy of rapidly expanding lending to developing countries, as laid out in various G20 documents such as the “Triple Agenda,” seems sound. Of course, if policy reform and sound project preparation can augment the return to capital, the argument is further strengthened. As one of us has reported on before, the actual returns to projects funded by multilateral development agencies tend to be even higher than these numbers would suggest.

However, the math only works if the dollars lent actually translate into dollars invested. Here, corruption, relative costs of project preparation (especially for small projects in developing countries), time delays in implementation, and the like can have a serious impact in driving a wedge between the MPK and the financial return to a private firm. Efforts to mitigate these losses in turn have their own “costs of doing business in country X.”

These considerations weigh heavily on private financiers. The World Bank and other multilateral development banks (MDBs) do not have the scale to finance the needed investments in developing countries and increasingly see their role as mobilizing and catalyzing private capital. But unlike the MDBs, the providers of private capital do have to worry about repayment risk.

A simple metric of this is given by the 10-year bond yield of developing countries on sovereign loans. The higher the risk, the higher the yield that countries have to pay.

Figure 2 subtracts the 10-year bond yield from the previous estimates of MPK for each country where there is data (the sample is smaller because not all countries, especially the poorest, have issued bonds).

Most developing countries, but not all, have relatively high spreads between the MPK and the real 10-year bond yield. Some countries have a negative spread between MPK and the perceived risks (e.g., Turkey and Brazil), but this is due to high levels of perceived macroeconomic risks rather than low levels of MPKs.

What are the implications? First, for most countries, the issue does not seem to be the relative risk-return of investing in EMDEs. Instruments such as guarantees, which lower risk, can of course affect the calculations, but it may be the case that other factors, such as the frictions associated with individual transactions, are even more important.

This is why a considerable part of the conversation will revolve around country platforms (getting projects approved rapidly in a government-approved framework, with complementary public investments and policies to raise returns). Public-private partnerships are another example of joint project design, often covering multiple years. Many of these activities, however, are complex to coordinate and oriented toward sectors such as infrastructure and digitalization. They therefore work better in countries with better regulatory regimes that tend to be middle-income.

The greatest potential, identified by Cull et al. from the World Bank, is for syndicated loans, where MDBs and their private partners can lend for shorter maturity, faster return projects, and where the private sector can simply piggyback on the project preparation work done by an MDB to get to scale and reduce transaction costs.

All MDBs concerned with growth and development worry about the additionality of their lending and the line of sight between lending and investment. However, additionality cannot be understood in the context of a single project that a government might invite an MDB to finance. It must concern itself with the MPK in the country. MDBs should put greater effort into understanding (and transparently publishing) time series data on MPKs.

At the same time, MDBs must unpack the root causes of limited private capital mobilization. In some countries, as we have shown, it is undoubtedly about the fact that risk, proxied by the 10-year bond yield a country faces in global capital markets, is simply too great compared to the expected return. But in other cases, it seems more likely that the frictional costs of project investments dominate. These can range from project preparation and time to market to policy uncertainty and other factors. Understanding these factors and tailoring country-level private capital mobilization (PCM) to the root cause is needed to avoid simplistic solutions. If general guarantee facilities do not pay special attention to these factors, they will simply add to private-sector returns without altering PCM volume flows in a material way.