Capital Market Integration and Exchange-Rate Regimes in Historical Perspective

James R. Lothian, Fordham University

[PLEASE NOTE THAT SOME SYMBOLS FROM THE EQUATIONS BELOW COULD NOT BE REPRODUCED IN THIS VERSION OF THE SUMMARY--THE CLIOMETRIC SOCIETY]

I would like to thank George Benston, Victor M. Borun, Mark Flannery, Iftekhar Hasan, Kees Koedijk, Clemens Kool, Richard Marston, Paul Wachtel and participants in seminars at the Atlanta Finance Forum, Clemson University, Fordham University, the University of Limburg for their comments on the earlier versions of this paper. I would also like to thank Michael Field, Anthony Hermann and James R. Lothian, III who provided research assistance at various points in the course of this work. The usual caveat, of course, applies.

On one view, international capital markets have become increasingly linked during the past two decades, both as a result of rapid financial and technological innovation, and because of the gradual erosion of governmental controls. On another reading, they are if anything less integrated. According to this line of reasoning, increased volatility - particularly exchange-rate volatility - has substantially increased risk and thus driven international markets farther apart.

The underlying motivation for this paper is to try to distinguish between these two seemingly plausible yet quite contradictory characterizations. To do so I investigate the behavior of interest rates historically, focusing on the time paths and degrees of international convergence of both long-term bond yields and money market interest rates in a sample of panel data that at its longest spans the near two hundred year period 1800-1990, and at its geographically broadest, the United States and nine other countries.

The major empirical finding is the substantial similarity in the degree of convergence of real interest rates under the classical gold standard, the later years of Bretton Woods and the current float. Real interest rates are by no means equal among countries during these subperiods, but the divergences are considerably less than in any other subperiod.

The question that the paper goes on to consider is what forces underlie these results. What were the factors responsible for virtually the same degree of real interest rate convergence in periods so technologically and in many other ways economically diverse as the classical gold standard era, the Bretton Woods years and the current floating rate period? In particular, why have floating exchange rates had so little visible effect?

I. Theoretical Considerations

One of the most important economic relations borrowed by international finance from price theory is the law of one price. Stated in its simplest form (and thus abstracting from risk and transactions costs), the law of one price posits that the prices of identical goods will be the same internationally when expressed in the same currency units. Applied to asset markets in a world of fixed exchange rates, this relation translates into equality between the nominal yields of securities with identical characteristics in different countries:

Rt - R = 0,(1)

where the Rts are the nominal yields on the securities, bonds for our purposes, and the superscript F denotes the foreign country. Applied to goods markets, it translates into equality of inflation rates:

t - = 0,(2)

where the ts are the two countries' inflation rates.

Real yields in the two countries therefore also will be equal. Given (1) and (2) and the equality of actual and anticipated rates of inflation that will prevail in equilibrium, this follows from the open-economy version of the Fisher equation:

_t - _ = (Rt - R) + ( - *),(3)

where the _ts are ex ante real yields and the s are anticipated rates of inflation over the life of the bonds.

To analyze the situation under floating exchange rates, it is useful to expand equation (3) to take account explicitly of variations in the exchange rate:

_t - _ = [(Rt - R) - ] + [( - ( - *)].(4)

The first term in brackets on the right hand side of (4), the difference between the nominal interest-rate differential and the anticipated rate of change of the nominal exchange rate, , is the uncovered interest parity relation; the second term in brackets, the difference between and the inflation differential, is the anticipated rate of change of the real exchange rate.

Under floating exchange rates, inflation rates in the two countries need not, and generally will not, be equal. Nominal yields therefore will differ. But at this level of abstraction there is no reason for real yields to do so. Purchasing power parity and uncovered interest parity both will hold, in which case the right hand side of (4) will be zero.

Hence, in a world of integrated economies in which arbitrage is perfect and there are neither transactions costs nor risk aversion, real yields on financial assets and on other assets will be equal among countries. This will be true regardless of the exchange-rate regime. The regime will be neutral.

The issue to be investigated is how apt a characterization this is. A number of studies using data for the floating-rate period alone have concluded that capital market integration is considerably less than complete, judged both on the basis of real-interest equalization and by other criteria (see Frankel, 1992, for a succinct summary). The major reason why in the view of many researchers is that uncovered interest parity has been subject to systematic violation. Of particular concern to some economists has been the effect of exchange-rate volatility on international capital markets (e.g. McKinnon, 1990). This volatility and the risk that it engenders, they argue, has decreased the flow of capital from one country to another and hence widened the differentials among real interest rates internationally.

II. Data and Overview of Empirical Results

In the empirical analysis I use annual data for long-term bond yields (either government or high-grade corporate) and for a variety of short-term, money market interest rates, for samples of ten countries and seven countries respectively. The full ten-country sample is made up of Belgium, France, Germany, the Netherlands, Sweden, the United Kingdom, the United States, Canada, Italy, and Norway; the seven-country sample excludes the last three. The sample period varies according to data availability, but at its longest - long-term rates for France, the United Kingdom, and the United States - spans the 190 years from 1800 to 1990.

As a proxy for the ex ante real interest rate I use the spreads between the nominal rates and the contemporaneous rates of inflation. The price-level data used to construct these estimates are for GNP (or GDP) deflators beginning in 1870 and linked to WPIs (or CPIs) before that date. For the money market rates, the errors introduced by this procedure are arguably less of a problem than for the long-term bond yields. In any event, there is no other simple way to model expectations and then to compare behavior across regimes, since the expectations generating process is itself likely to vary with the regime.

Here I only present results for short-term rates. I have, however, obtained qualitatively similar results for long-term rates and report these in the conference paper. Figure 1 and Table 1 summarize the basic findings. Plotted in Figure 1 are cross-country standard deviations of quinquennial averages of the annual short-term real interest rates for the various country groups. These provide measures of the divergence (or alternatively, convergence) of real interest rates internationally. Shown in Table 1 are subperiod averages of these cross-country standard deviations.

Figure 1: Cross-Country Standard Deviations of Real Short-Term Interest Rates

In both the chart and the table, two things stand out. First are the differences in short-term real rates across countries that we see throughout the sample period. In no instance is real-rate convergence complete. The second has to do with the variations in the degree of real-interest convergence over time, in particular the similarity between the classical gold-standard period, the later years of the Bretton-Woods period, and the current floating-rate period. Real rates are more nearly equal during these three subperiods than in any of the other subperiods. And contrary to the argument that increased risk has widened the gaps between real interest rates under floating, there is actually some decrease in the cross-country divergence in real interest rates between the floating-rate period and the Bretton-Woods period. We see the same thing, moreover, when we subdivide the Bretton-Woods and floating-rate periods and compare the two later, more stable, subperiods within each. Real-rate divergences are less in the later subperiods for both Bretton Woods and the float than in the corresponding earlier subperiods, but are also less in the later subperiod for the float than in the later subperiod for Bretton Woods.

Table 1. Period Averages of Cross-Country Standard Deviations of Quinquennial

Averages of Annual Ex-Post Real Short-Term Interest Rates

_____________________________________________

Period

Number of Countries in Average

------------------------------------------------

3

4

5

6

7

-----------------------------------------------------------------------------------------

Pre Gold Standard

1835-1850

3.16

1855-1875

4.73

4.07

3.89

3.58

Classical Gold Standard

1880-1915

1.39

1.72

1.65

1.73

1.19

Interwar Period

1925-1940

1.30

4.84

4.45

5.28

5.08

Bretton Woods

1950-1970

1.30

2.00

1.98

3.91

3.63

1950

2.42

3.38

3.64

12.32

11.32

1955-1970

1.02

1.65

1.56

1.81

1.71

Floating Rates

1975-1990

1.43

2.05

1.99

1.85

1.78

1975-1980

2.05

2.64

2.63

2.39

2.26

1985-1990

.81

1.45

1.36

1.30

1.29

_____________________________________________

Note: The quinquennial average for the five-country series begins with the period ending in 1860; for the six-country series with the period ending in 1870; and for the seven-country series with the period ending in 1905. The quinquennial average for Germany for the period ending in 1925 is missing from all series.

Two further items of interest are the noticeably higher cross-country standard deviations in the interwar years than in earlier or later periods, and the decline in these measures of real-interest divergence during the course of the nineteenth century. The nineteenth century decline is consistent with the hypothesis that the world did in fact gradually become more integrated under the classical gold standard. The sizable divergences in the interwar years provide yet another illustration of the aberrant economic behavior that in other ways characterized that era. I suspect that they also in part reflect the heavily interventionist stance in matters of international trade and finance that many governments adopted during that era.

Results of dummy variable regressions confirm the impressions formed from the tabular and graphical evidence. Differences among subperiods in the cross-country standard deviations of real rates are statistically significant, but this is due totally to the war-time, interwar and early nineteenth century observations. Differences among the gold, Bretton-Woods and floating-rate periods are not statistically significant in either instance. The degree of convergence of real interest rates in the various countries, therefore, appears to have been approximately the same in all three of these periods.

III. Other Evidence

The question is what underlies these results. Are they consistent with other evidence and with what might be expected on the basis of theory? Or might they be simply statistical artifacts?

One bit of evidence that they are actually behavioral phenomena comes from examination of historical data on gross foreign assets that are available for a subsample of the countries studied here. These alternative measures of capital-market integration show a time pattern that in the main is consistent with the results that I have just described. In the classical gold standard period, gross foreign assets of these countries, expressed both in 1914 dollar amounts and as a fraction of GNP, grew continually, and by the start of World War I stood at well over 50 per cent of the countries' combined GNP. In the interwar years, the data show substantial declines on both bases, but after World War II that pattern is reversed. By 1984, foreign assets had reached record amounts when expressed in 1914 dollars, but given the substantial growth in real income in these years still remained well below their highs as a fraction of nominal GNP recorded in 1914.

Additional corroborative evidence comes from Jeffrey Williamson's recent study of real wage behavior among countries (1995). Using panel data for the years 1830-1988, Williamson reports an almost identical time pattern for real-wage convergence across countries to that shown here for real interest rates.

A third, and in many ways most convincing, type of evidence comes from examination of the institutions surrounding the major countries' financial markets. What emerges from comparison of the gold-standard and current floating-rate periods is how little things have changed on net between then and now. It is clear from qualitative evidence derived from a variety of sources - practitioner oriented textbooks, the financial press and other contemporary accounts - that capital markets and the foreign exchange markets during the classical gold standard years had many of the earmarks of markets in our own era. Corporate finance and trading both had a strong international focus. The mechanics of international arbitrage were well understood and the practice of such activities common. Commercial banking in all of the major countries was, or as in the case of the United States, soon became internationalized.

A Rip van Winkle who went to sleep in the City of London or on Wall Street in 1914 and awoke eighty or so years later, probably would be shocked by the technological changes that had occurred, the fact that, for example, the time lag in receiving on-the-market price quotes had been reduced to seconds rather than minutes or hours, but he would certainly feel more at home today than if he had awakened three or four decades earlier. For many of the developments in banking and financial markets that now go under the headings of "internationalization" and "globalization" were simply part of a return to the status quo ante - a "re-internationalization," and "re-globalization" following the long period of international capital- and goods-market fragmentation that began with World War I and continued with only brief interruption through the 1950s.

Much of this recent reinternationalization appears ultimately to have been driven by the very forces which, if they had operated in vacuo, would very likely have produced the opposite result -further fragmentation of the world's financial markets as opposed to greater integration. The principal reason why this did not happen, I believe, is that the institutions surrounding these markets evolved in response to what was then going on in the world economy. Indeed, the move to floating exchange rates itself was part of this process. As monetary policy in the United States, the reserve-currency country under Bretton Woods, became more expansive countries necessarily had to let their exchange rates float (Darby and Lothian, 1983). The increases in the volatility of nominal economic variables, the greater divergences in such variables among countries and the fluctuations of nominal exchange rates that accompanied these developments produced additional feedback effects. As theory had predicted (Telser and Higinbotham, 1977) financial practices themselves changed. New markets, new instruments, and new corporate finance techniques that enabled economic agents to cope better with the increased volatility and risk were developed. The organized and over-the-counter markets in foreign exchange futures and options, in interest-rate futures and options and in swaps of various sorts are all examples. At the same time, regulations and other governmentally imposed impediments to the functioning of financial markets increasingly fell by the wayside, either de facto or de jure, as the societal costs of such restrictions mounted in the new more volatile world. The Regulation Q ceilings on deposit interest rates in the United States and the capital controls that were commonplace in many countries are prominent examples.

On net, therefore, floating exchange rates have left no visible trace on international capital markets internationally. In this respect at least, Adam Smith's dictum that "there is much ruin in a nation" appears to have been borne out.

Endnotes

1The principal sources of the interest-rate data were Homer (1977), and of the price-level data Mitchell (1975). Most of the updates to these series came from the International Financial Statistics and companion CD ROM.

2The sources of the foreign asset data were Woodruff (1967, p. 150, Table IV/I) prior to 1960 and Dunning and Cantwell (1987) from 1960 to 1984. Michael Bordo graciously provided GNP data.

3See Cameron (1991), Deutsch (1933), Michie (1987), Myers (1931), Spalding (1938) for discussions of various aspects of financial markets in the gold-standard era.

References

Cameron, Rondo, "Introduction," in Rondo Cameron and V.I. Bovykin, eds., International Banking, 1870-1914, Oxford and New York: Oxford University Press, 1991.

Darby, Michael R. and James R. Lothian, "Conclusions," in Michael R. Darby, James R. Lothian, et al., The International Transmission of Inflation, Chicago: University of Chicago Press for the NBER, 1983.

Deutsch, Henry, Arbitrage in Bullion, Coin Bills, Stocks, Shares, and Options, London: Sir Isaac Pitman & Sons, Ltd., Third Edition (revised by Otto Weber), 1933.

Dunning, John and John Cantwell, IRM Directory of Statistics of International Investment and Production, New York: New York University Press, 1987.

Frankel, Jeffrey A.,"Measuring International Capital Mobility: A Review," American Economic Review, May 1992, 82: pp. 197-202.

Homer, Sydney, A History of Interest Rates, 2nd ed. New Brunswick, NJ: Rutgers University Press, 1977.

International Monetary Fund, International Financial Statistics, various issues.

Lothian, James R., International Banking, Capital Market Integration and Exchange Rate Regimes in Historical Perspective," unpublished paper, Fordham University, 1994.

McKinnon, Ronald I. "Interest Rate Volatility and Exchange Risk: New Rules for a Common Monetary Standard." Contemporary Policy Issues, April, 1990, 7: pp. 1-17.

Michie, R.C., The London and New York Stock Exchanges, 1850-1914 , London: Allen and Unwin, 1987.

Mitchell, B. R., European Historical Statistics, 1750-1970 New York: Columbia University Press, 1975.

Myers, Margaret, The New York Money Market, New York: Columbia University Press, 1931.

Neal, Larry, "Integration of International Capital Markets: Quantitative Evidence from the Eighteenth to the Twentieth Centuries," Journal of Economic History, June 1985, 45: pp. 219-26.

William F. Spalding, The London Money Market, 6th ed., London: Sir Isaac Pitman and Sons, 1938.

Telser, Lester G. and Harlow N. Higinbotham, "Organized Futures Markets: Costs and Benefits," Journal of Political Economy, October 1977, 85: 969-1000.

Williamson, Jeffrey G., "The Evolution of Global Labor markets since 1830: Background Evidence and Hypotheses," Explorations in Economic History, April 1995, 32: 141-196.

Woodruff,William, The Impact of Western Man: A Study of Europe's Role in the World Economy, 1750-1960, New York: St. Martin's Press, 1967.