Economist, European Department, IMF
The last decade and a half has witnessed a general deregulation of international capital markets in the industrial world, resulting in the free movement of capital between countries. How does this freedom to borrow and lend between different nations affect key variables such as investment and saving? In a world characterized by unfettered capital flows, countries with a high level of investment need not rely on an equally high domestic saving. In such an environment, the gap between domestic investment and saving can always be financed by foreign saving via a current account deficit. Hence, it might be thought that there would be little correlation between saving and investment across countries. Yet, recent research shows that this is not the case within industrial countries. This article explains some of the factors responsible for this unexpected result.
To illustrate the issues at stake, first imagine a world in which countries are unable to borrow from, or lend to, each other. All investment within the country must, therefore, be financed out of its own saving, hence saving and investment rates between countries will be perfectly correlated. As the country neither borrows nor lends abroad, the current account will be in balance, because it reflects the net amount of foreign saving flowing into a country. In the less extreme case of low capital mobility, where there are high costs to international borrowing and lending, domestic saving and investment would still be highly correlated.
For a detailed treatment, see “Saving-Investment Correlations: Immobile Capital, Government Policy or Endogenous Behavior?” by Tamim Bayoumi and “Saving, Investment, Financial Integration, and the Balance of Payment,” by Michael Artis and Tamim Bayoumi, IMF Working Papers WP/89/66 and WPI89I102, respectively, available from the author.
Now consider a world in which capital markets of different countries are fully integrated. For example, think of countries as if they were individual states in the United States. How would saving and investment be related in different states? Since there are no barriers to people in, say, New York borrowing from people in California, the level of saving in any state need not be related to its level of investment. Rather, the level of saving will differ from state to state according to the saving preference of people within each state. Similarly, the level of investment would depend upon investment opportunities in each state. Assuming that the factors causing people to save and invest are different, there will be no connection between the two. On the basis of this reasoning, Martin Feldstein and Charles Horioka of Harvard University suggested looking at the correlation of saving and investment across countries as a test of the degree of international capital mobility.
The results of this test, however, were far different than expected. Saving and investment rates between countries turned out to be highly correlated both during the 1960s, when capital controls between industrial countries were extensive, and during the 1970s and 1980s, when these barriers to capital mobility were relaxed. Charts 1 and 2 show investment and saving as a percentage of output for several members of the Organization for Economic Cooperation and Development, with data averaged over 1966–70 and 1981–85, respectively. The 45-degree line indicates a current account balance of zero. While there is some reduction in the saving-investment linkage over time, the 1980s still show a remarkably close relationship between the two.
How can we explain this result, known as the Feldstein-Horioka puzzle? Various explanations have been proposed. Feldstein and Horioka argued that international capital mobility is not as high as is generally believed, perhaps because of structural factors such as the lack of information, risk aversion, and differences in legal systems.
Others have suggested that the observed correlations between saving and investment may not reflect the level of capital mobility, but rather the behavior of the private sector or government. For example, if private saving and private investment both responded to the same factors, such as changes in the rate of growth of population or productivity, then saving and investment between countries could be correlated even if capital flows freely. Alternatively, if governments target their current account through, say, changes in fiscal policy or interest rates, the same result would apply. In both of these cases, even in a world of high capital mobility, there would be a close link between saving and investment.
One way to test these hypotheses is to look at how highly correlated saving and investment are in different sectors of the economy, or under different policy regimes. For example, if the correlations are primarily due to private sector behavior, then private saving and investment should be at least as highly correlated as the corresponding values for the total economy. The data show, however, that this is not the case; there is little relationship between private saving and investment in the 1980s. To examine the validity of the other two explanations, that the result is caused by a genuine lack of capital mobility or is a result of government policy, is considerably more difficult. It is not enough simply to look at data for the government sector, since under both hypotheses the government and private sector saving-investment balances will offset each other, although for different reasons.
We could look for a situation in which the government did not interfere much in the economy, and capital controls were not stringent. If saving and investment are not closely linked under this regime it could indicate that government policy is the primary cause of the behavior in Charts 1 and 2. If, however, there were a high correlation between saving and investment, we could conclude that capital is actually not very mobile across countries.
The gold standard and now
The classical gold standard, stretching from 1880 to 1913, represents exactly the type of situation suggested here: capital controls between countries were low or nonexistent and governments were noninterventionist. Chart 3 shows the relationship between saving and investment over this period. It is clear that saving and investment were much less correlated over 1880–1913 than in the 1980s, implying large movements of capital between countries. A counterpart to this were persistent current account surpluses and deficits. From 1880 to 1913, the United Kingdom had an average surplus of 4½ percent of GDP, and Australia an average deficit of 3½ percent. Indeed, by 1913 the United Kingdom owned so much foreign capital that almost 10 percent of its national income came from payments of net property income from abroad.
These results favor the view that government policy is the main reason for the observed high saving-investment correlation in the recent period. There is, however, a further crucial difference between the gold standard and today—namely, the determination of the exchange rate. The period from 1880 to 1913 saw no changes in the rate of exchange between major currencies, although countries did on occasion suspend convertibility with gold. Under the present system, exchange rates can, and do, vary against each other from day-to-day and show large movements from year-to-year. This is important because, in addition to the normal uncertainty of lending, international lending currently involves the acceptance of exchange rate risk. If a person from country A lends to someone in country B, the sum has to be converted from one currency to another. When it is time to repay, the exchange rate between the two currencies is unlikely to be the same. If the lender has to be paid in his home currency, then the cost to the borrower is uncertain as it depends on the exchange rate. If the repayment is defined in terms of the borrower’s home currency, the lender faces a similar exchange risk. The presence of exchange risk may lower the desire to borrow or lend internationally, despite the dismantling of formal impediments to international borrowing or lending.
Saving and Investment
(As a percentage of GNP)
Chart 1:1966-70 Chart 2:1981-85 Chart 3:1880-1913
Still, the range of possible answers to the Feldstein-Horioka puzzle has now been reduced to two: exchange risk or government policy. Is there any way of differentiating between these two explanations? One way is to compare the performance of countries with fixed exchange rates with another group of countries with floating exchange rates. An example would be countries in the European Exchange Rate Mechanism (ERM), a system that limits deviations from par values, with that of countries outside the mechanism. If the countries in the ERM look similar to those outside it, exchange risk is not important; if they look different, it is. Although results from such an analysis are still somewhat preliminary, countries within the ERM appear to have noticeably lower correspondence between their saving and investment than those outside it. Hence, exchange risk is probably an important influence in explaining the data for the 1980s.
Thus, the high correlations between saving and investment observed in the 1980s appear to be due to a combination of exchange risk and government policy. In addition to solving our puzzle, this matters because the prospects for international capital flows, and the associated current accounts, are important issues for policymakers. Will the 1990s look like the era of the gold standard, with persistent current account imbalances and large flows of capital between countries, or like the 1960s, with almost no movements of capital? Capital flows will probably continue to be larger than in the 1960s or 1970s, but, as long as exchange rates between the major currencies continue to float, there will probably continue to be a link between the two building blocks of the current account, namely, domestic saving and investment. The exception to this rule may well be Western Europe, where fixed exchange rates may facilitate larger movements of capital between countries.