In addition, there is evidence that access to markets is also important as well for economic growth. If a region is cut off from larger markets either because of its natural geography or because of man-made trade barriers, then the incentives for entrepreneurial activity and investment are again reduced. In 1999, Jeffrey Frankel and David Romer cautiously concluded that the converse holds as well: better trading opportunities do lead to faster growth. There is still some debate among economists about the relative importance of institutions and trade, but it seems likely that both are important and that in fact they complement each other. Several years ago, Kenneth Sokoloff found that rates of invention were extremely responsive to the expansion of markets during the early industrialization of the United States by examining how patenting activity varied over time and with the extension of navigable waterways. For example, as the construction of the Erie Canal progressed westward across the state of New York, patenting per capita rose sharply county-by-county. The United States had a good system of protecting these intellectual properties, and the development of transport links to broader markets stimulated individuals and firms to invest more in developing new technologies.

Indeed, looking back over the past century, locations with access to markets and good property rights have generally prospered, while locations disconnected from markets and with poor property rights have remained poor. Many of the features that we associate with underdevelopment are therefore results of these underlying weaknesses in institutions and policies. In such environments, there is little incentive to invest in equipment or education and develop modern industry.

But these symptoms have often been mistaken by aid donors as causes of underdevelopment. If low levels of investment are a problem, then give poor countries foreign aid to invest in capital. If a lack of education is a problem, finance broad expansion of schools. If modern industry is absent, erect infant-industry protection to allow firms to develop behind a protected wall. All of these approaches have been pushed by aid donors. In poor countries with weak underlying institutions, however, the results have not been impressive.

Over several decades, Zambia received an amount of foreign aid that would have made every Zambian rich had it achieved the kind of return that is normal in developed economies. If lack of capital was the key problem in Zambia, then that was certainly addressed by massive amounts of aid; but the result was virtually no increase in the country's per capita income. Similarly, large amounts of aid targeted at expanding education in Africa yielded little measurable improvement in achievement or skills. Donors financed power plants, steel mills, and even shoe factories behind high levels of protection, but again there was virtually no return on these investments.

The recent thinking in economic history and development suggests that these efforts failed because they were aimed at symptoms rather than at underlying causes. If a government is very corrupt or dominated by powerful special interests, then giving it money, or schools, or shoe factories will not promote lasting growth and development. These findings suggest that much of the frustration about foreign aid comes from the many failed efforts to develop social infrastructure in weak institutional environments where governments and communities cannot make effective use of these resources—not from the intrinsic inability of aid itself to generate positive results.

There are a number of important caveats about these findings on aid effectiveness. First, humanitarian or food aid is a different story. When there is a famine or humanitarian crisis, international donors have shown that they can bring in short-term relief effectively. Second, there are some health interventions that can be delivered in a weak institutional environment. In much of Africa, donors have collaborated to eradicate river blindness, a disease that can be controlled by taking a single pill each year. That intervention—and certain types of vaccinations—can be carried out in almost any environment. But many other social services require an effective institutional delivery system; other health projects in countries with weak institutions have tended to fail without producing any benefits.

Persistent Institutions

A second important finding from recent work in economic history is that institutions are persistent. Last year, Stanley Engerman and Sokoloff showed how differences in the natural endowments of South and North American colonies centuries ago led to the development of different institutions in the two environments. Furthermore, many of these institutional differences have persisted to this day. If institutions are important and if they typically change slowly over time, then it is easy to understand the pattern of rising global inequality over the past century. Locations with better institutions have consistently grown faster than ones with poor institutions, widening inequalities. Because it is relatively rare for a country to switch from poor institutions and policies to good ones, countries that began at a disadvantage only fell further behind in the years that followed.

The importance of good institutions and policies for development in general and for aid effectiveness in particular is something that donors have gradually realized through experience and research. International donors' first instincts were to make improved institutions and policies a condition of their assistance. In the 1980s in particular, donors loaded assistance packages with large numbers of conditions concerning specific institutional and policy reforms. Some World Bank loans, for example, had more than 100 specific reform conditions. However, the persistence of institutions and policies hints at the difficulty of changing them. There are always powerful interests who benefit from bad policies, and donor conditionality has proved largely ineffective at overcoming these interest groups. A 2000 study that I co-authored with Jakob Svensson examined a large sample of World Bank structural adjustment programs to find that the success or failure of reform can largely be predicted by underlying institutional features of the country, including whether or not the government is democratically elected and how long the executive has been in power. Governments are often willing to sign aid agreements with large amounts of conditionality, but in many low-income countries the government is either uninterested in implementing reform or politically blocked from doing so. "Aid and Reform in Africa," a set of case studies written by African scholars on 10 African states, reaches similar conclusions: institutional and policy reform is driven primarily by domestic movements and not by outside agents.