"Essays on International Business Cycles", PhD thesis, Economics Department, University of Chicago, 1991

PhD dissertation, 1991, Economics Department, University of Chicago. (Thesis committee: Michael Woodford, advisor; José Scheinkman; John Huizinga.) Models of the international economy which assume complete asset markets predict that consumption co-moves closely in different countries as this structure of asset markets allows agents in different countries to 'pool' the country-specific risks which they face (see Scheinkman (1984), Leme (1984)). Examples in this class of models include the recent international Real Business Cycle models of, among others, Backus, Kehoe & Kydland (1989), Baxter & Crucini (1989), Stockman & Tesar (1991). The first essay in this thesis (chapter II) tests the implications for the trend behavior of consumption of models of the international economy which assume complete asset markets. In a world where consumptions and real bilateral real exchange rates of different countries follow unit root processes, such models predict that (under certain assumptions about preferences) consumptions and bilateral real exchange rates are cointegrated for any given pair of countries. The paper presents statistical tests which suggest that data on consumptions and real exchange rates for the US, Japan, France, Britain, Italy and Canada during the period 1971-1987 are inconsistent with this prediction. These findings suggest that models with limitations on international asset markets might be needed to explain the international covariation of consumption. The second essay in this thesis (chapter III) presents a Real Business Cycle model in which limitations on international capital markets exist in the sense that only debt contracts are available for international capital flows. Simulations of the model suggest that it can explain the low cross-country correlations observed in detrended consumption data, and that for 'realistic' cross-country correlations of the exogenous shocks. The second essay argues also that a model which allows for additive technology shocks is better able to explain the observed positive correlations of investment and output across countries than standard business cycle theories in which multiplicative shocks to total factor productivity are the only source of economic fluctuations. One possible interpretation of the additive shocks is as shocks to government consumption. The dissertation (pp.12-13) shows (inter alia) that efficient international risk sharing countries requires that ratios of Home and Foreign marginal utilities of aggregate consumption are proportional to the relative price of Home vs. Foreign consumption: U’(C)/U’(C*)=kP/P*, with C,C*: aggregate consumption of Home and Foreign households, respectively; P,P*: Home and Foreign consumption prices indices (expressed in a common currency); U’ is the marginal utility of consumption, and k>0 is a date-and-time invariant coefficient (that reflects countries’ relative wealth). With constant relative risk aversion (identical for both countries), this implies: - s log(C/C*)=log(P/P*), where s>0 is the coefficient of relative risk aversion. Thus, in an efficient world, relative Home vs. Foreign consumption would be closely linked to the (CPI-based) real exchange rate; a country whose real exchange rate depreciates would experience faster consumption growth than the rest of the world. The dissertation shows that this prediction is strongly rejected by the data (pp. 22-27). My 1995 Journal of International Money and Finance paper (see above) is based on these results. Two years after my dissertation, David Backus and Gregor Smith (Journal of International Economics, 1993) published a paper that derives the same risk sharing condition; these authors likewise conclude that this condition is rejected empirically. The literature sometimes refers to the risk sharing condition as the 'Backus-Smith condition', and describes the empirical failure of that condition as the 'Backus-Smith puzzle'. My dissertation independently--and earlier--derived that condition, and it provided empirical tests that are complementary to those of Backus and Smith.