International intertemporal solvency in industrialized countries: evidence and implications.

• Author: Liu, Peter C.

Introduction

Recent large current account deficits among industrialized countries have caused concern for policy makers. Some believe, for example, that the current buildup of claims on the U.S. by foreigners violates her solvency condition with respect to the rest of the world. Recent research [22] suggests that, for a country's intertemporal solvency condition to hold, the change in the country's obligations to the rest of the world (i.e., her current account deficit) must be stationary. Stationarity tests on current accounts were performed for the U.S. by Trehan and Walsh [22] and Wickens and Uctum [23], for the U.S. and Canada by Otto [15], and for twenty-three countries by Gundlach and Sinn [10]. These papers present evidence that current accounts are non-stationary for several major industrialized countries, including the U.S., the U.K., Canada, Germany, and Japan.

These findings do not favor international intertemporal solvency. For this reason, the issue merits further investigation. In this paper, we test for the stationarity of current account deficits of the "Group of Seven" industrialized countries. Preliminary results confirm many of Gundlach and Sinn's [10] findings. However, several events during the 1980s might have caused discrete breaks in current account balances. Events potentially linked with current account breaks include: the reduction in world inflation, tax reduction and reform in the U.S. and other industrialized countries, attempts at international macroeconomic coordination (i.e., the Plaza Accord), and shifts in the after-tax rate of return on capital in many countries. As shown in recent research on gross national product [16; 2], if such breaks are not accounted for, stationarity tests will be biased in favor of accepting the null hypothesis of non-stationarity. Our results suggest strong evidence of discrete breaks in the U.S. and Japan, and some evidence favoring breaks in the U.K., Germany, and Canada. The break in the U.S. current account appears to occur in 1983, while breaks in the other countries appear to occur somewhat later, between late 1984 and 1987. More importantly, once these breaks are incorporated, the evidence is more favorable toward the stationarity of current accounts in these countries. These findings thus support intertemporal solvency for these countries.

Our results can also be related to the high correlation of savings and investment found in many countries. This observation was due originally to Feldstein and Horioka [5]. Their conclusion was that capital is not mobile across international borders.(1) While several other interpretations have been proposed, no one dominates. An extension of our findings is that high savings-investment correlations may be related to intertemporal solvency. Since savings minus investment equals the current account surplus plus the government budget deficit, a high savings-investment correlation may simply result from the fact that a country is in intertemporal balance externally and domestically. Satisfying these two solvency conditions will result in a high correlation between savings and investment, regardless of capital mobility.

The paper is organized as follows. In section II, the foreign intertemporal solvency condition and its empirical implications are presented. In section III, results of stationarity tests on the current accounts from the "Group of Seven" countries are performed. These results coincide with those from Trehan and Walsh [22], Gundlach and Sinn [10], and Wickens and Uctum [23]. In section IV, results of nonstationarity tests, with the break determined by the pre-test procedure, are presented. To determine break date, a procedure suggested by Christiano [2] is used. In section V, we link our findings to the savings-investment correlation issue. In section VI, a summary and some conclusions are presented.

2. Foreign Intertemporal Budget Balance: A Framework for Testing

   The international, intertemporal solvency condition implies that, over an infinite horizon, the net present value of the stock of one country's claims on the rest of the world must tend to zero:

   lim [f.sub.t]/[[1 + r].sup.t] = 0 where t [approaches] [infinity] (1)

   where f is the stock of net claims, expressed in constant units of the home currency, and r is the world real interest rate, adjusted for inflation and exchange rate changes.(2)

   An analog of condition (1), but with public debt, applies to governments. Several recent papers explore the empirical implications of the government's intertemporal solvency condition.(3) However, except for Trehan and Walsh [22] and Wickens and Uctum [23], few papers have explored the empirical implications of the international intertemporal solvency condition (1).

   Our goal is to transform equation (1) into a testable expression. To do so, the first step is to write out the one-period budget constraint for the country, in constant units of the home country currency:

   [X.sub.t] - [M.sub.t] + [r.sub.t][f.sub.t-1] = [f.sub.t] - [f.sub.t-1] (2)

   where X and M are real exports and imports of goods and non-factor services. For simplicity, we assume that [r.sub.t] is stationary with mean r : [r.sub.t] = r + [v.sub.t], where r is the average real interest rate and [v.sub.t] is a mean-zero random error. However, as discussed later in this section, this assumption is not essential for the solvency test employed by this paper. Repeated substitution of expression (2) over an infinite horizon yields an infinite-horizon budget constraint.

   (1 + r)[f.sub.t-1] [summation of] ([M.sub.t+k] - [X.sub.t+k])/[(1 + r).sup.k] where k=0 to [infinity]

   + lim [f.sub.t+k]/[(1 + r).sup.k] where k[approaches][infinity] + [summation of] [v.sub.t+k]/[(1 + r).sup.k] where k=0 to [infinity] (3)

   An additional simplifying assumption, following several authors [8; 22], is that X and M follow random walks with drifts [[Mu].sup.x] and [[Mu].sup.m], respectively: [Mathematical Expression Omitted], and [Mathematical Expression Omitted](4). In this case, identity (3) may be reexpressed in terms of expected values:

   C[A.sub.t] = [Alpha] + lim E[r[f.sub.t+k]/[(1 + r).sup.k]] where k[approaches][infinity] + [w.sub.t] (4)

   where the current account C[A.sub.t] [equivalent] [X.sub.t] + r[f.sub.t-1] - [M.sub.t], [Alpha] [equivalent]...