Examining the economic returns from infrastructure investments

It's considered an article of faith that infrastructure investments, especially in developing countries, pay for themselves and have large multipliers. This belief has generated both demand (the argument that developing countries have large infrastructure financing requirements) and the supply (the very vocal movement on private participation and cross-border capital flows into infrastructure). Reports like this and this fuel and sustain these beliefs.  

In this backdrop Peter Blair Henry and Camille Gardner of NYU urge a very important cautionary note. 

The paper's objective in the words of the authors is important,

Specifically, the paper presents a simple framework for evaluating the productive potential of investing in infrastructure that explicitly and simultaneously compares the social rate of return on infrastructure in emerging economies with the rate of return on alternative investment opportunities in: (a) the local environment, and (b) advanced economies. This dual- hurdle rate framework yields a unique classification that sorts countries into quadrants according to their potential for efficient investment in infrastructure. The quadrant-based classification flows naturally from a simple but powerful result: adding to a given country’s stock of infrastructure capital is both economically efficient and potentially financeable through a market- based allocation of foreign savings only when the return on doing so exceeds the return on all capital (infrastructure and non-infrastructure) in both the local economy and the developed world.

Their finding,

There are a total of 53 developing countries, 26 of which have data for the return on paved roads and 49 for the return on electricity generating capacity (EGC), yielding a total of 75 country-infrastructure-return observations. Of these 75 observations, 71 exceed the mean return on the corresponding type of infrastructure in advanced economies, and the returns differ, on average, by a factor of 21 in paved roads and a factor of 6 in EGC. In spite of this reality, however, opportunities for efficient investment in the infrastructure of EMDEs are not nearly as widespread as the large prospective return differentials would otherwise seem to suggest.

Just 39 of the 75 observations (distributed across 32 developing nations) clear both the cross-country and within-country hurdle rates for efficient investment. This means that 21 of the 53 countries, or roughly 40 percent, do not clear the dual hurdle for either type of infrastructure and are therefore not obvious candidates for additional investments in paved roads or electricity generating capacity. Furthermore, of the 32 countries with projects that clear the dual hurdle, only 7 are places where it is efficient to engage in additional investments in both kinds of infrastructure. The reality that the data from less than one-seventh of the countries present a clear case for investment in both roads and electricity raises questions about the wisdom of big-push approaches to infrastructure in the developing world.

Their explanation for these findings,

First, developed countries are over-invested in infrastructure relative to their other forms of capital. This has driven down their return on infrastructure to the point where the yield on another dollar invested in the infrastructure of advanced economies is less than their overall return on capital. Accordingly, the binding constraint for market-driven cross-border investment from developed countries into emerging-economy infrastructure is actually not the return on infrastructure in developed markets but rather the return on all capital in developed markets, and there are fewer developing countries whose returns on infrastructure clear this more stringent hurdle rate. Second, in order for additional investment in infrastructure in a given developing country to be efficient, its rate of return on infrastructure must exceed its own return on all capital. The imposition of this second hurdle rate renders still fewer countries as candidates for efficient investment in infrastructure.

They propose a twin-test for foreign investment in infrastructure 

A country whose return on all capital is less than that of advanced economies can nonetheless be an efficient destination for advanced-economy savings to finance infrastructure if: (a) the return on investing in that country’s infrastructure exceeds the return on all capital in advanced nations, and (b) the country’s return on infrastructure exceeds its own return on all capital.

They point to countries where this holds, 

Ten of the 32 countries whose social rates of return on infrastructure justify greater investment in paved roads or electricity generating capacity have rates of return on all capital that fall below the rate of return on all capital in the developed countries... For example, of the 39 country-infrastructure-return observations that suggest opportunities for efficient investment, 21 are in paved roads where the average (median) social rate of return is 10.2 (5.99) times larger than the return on capital in advanced economies. The average (median) prospective return for the 18 efficient investment opportunities in electricity generating capacity are more modest: 2.2 (1.87) times as large as the return on capital in advanced economies. In addition to the opportunities for efficient investment in EGC being substantially smaller than in the case of paved roads, another notable difference is that the potential for efficient investment in EGC manifests primarily in low-income countries— particularly those in Africa—whereas the opportunity in paved roads falls mainly to middle- income developing nations.

It is to be noted that, based on 1985 data, investment only in electricity generation fails to clear both hurdle rates and in paved roads clears only the cross-country hurdle rate. 

In other words, the purely economic case for domestic investments in infrastructure does not exist for a country like India. 

A problem with the study is that it uses the 1985 data to calculate hurdle rates. While it makes no difference to the framework, its outputs may vary based on factors like changes in cost of infrastructure, elasticity of returns on infrastructure, and so on. It would be great if someone could update this study with the latest data on returns to investment and analyse the economic case for investments in infrastructure.