Consumption and real exchange rates in dynamic economies with non-traded goods

The full text used for this summary is Queen’s Economics Department Working Paper No. 1252 (January 1993), the freely available version of the paper published in the Journal of International Economics in November 1993. The paper takes on a device that had become international macroeconomics’ all-purpose explanation. Nontraded goods had been invoked to account for large and persistent deviations from purchasing power parity, for the fact that cross-country consumption correlations are “considerably less than one, and are similar to cross-country output correlations,” for sizable international real interest differentials, and elsewhere for home bias in portfolios and for high savings-investment correlations. The authors’ stated interest is not in whether the device can generate any one of these facts but in what it implies about the relations among them: “While nontraded goods have been suggested as an explanation for many features of international macroeconomic data, the focus of this paper is on general equilibrium restrictions.” They build a stochastic exchange economy extending Lucas (1984) to many agents – I countries, each a single consumer endowed with the single traded good and with its own nontraded good, complete markets, finite horizon and finitely many states – and derive Proposition 2: with isoelastic period utility, the bilateral real exchange rate and the consumption ratio are monotonically related state by state, and hence in moments. In growth rates this gives the tight restriction that gamma times the change in the log consumption ratio equals the change in the log real exchange rate. Three implications follow that can be tested without ever observing the nontraded endowments: a pair of countries with a more variable (or higher-mean, or more persistent) consumption-ratio growth rate should have a real exchange rate with the same property; the two growth rates should have identical dynamics; and their time-series cross-correlation should be unity for every pair. The authors emphasise how few auxiliary assumptions this requires – no restriction on parameter values, no detrending choices, no laws of motion for endowments, and no need “to identify specific categories of consumption goods as traded or nontraded.” Applied to quarterly seasonally adjusted real private consumption and its deflator for Australia, Canada, France, West Germany, Japan, Sweden, the United Kingdom and the United States over 1971-1990, all 28 pairs, the predictions fail. Scatterplots that theory says should lie on upward-sloping lines through the origin are “cloud-like,” with rank correlations of -0.263 for standard deviations, -0.466 for first-order autocorrelations and 0.074 for means, against a standard error of 0.192; only the negative autocorrelation figure is significant, so “there certainly is no evidence of positive rank correlation.” The persistence result is a sign reversal, not merely a weak fit: all 28 real exchange rate growth rates are positively autocorrelated, while 27 of the 28 consumption-ratio growth rates are negatively autocorrelated. The cross-correlation that theory puts at unity averages 0.045, with a range of -0.08 to 0.17. The authors report their own robustness checks against themselves, including that the means figure really rests on only 7 independent observations by transitivity so that “including all 28 points biases the case in favour of the theory; even so, no significant positive relation can be detected.” Per capita annual data give rank correlations of -0.114, -0.045 and 0.170 and an average correlation of -0.056 (range -0.63 to 0.21); restricting consumption to nondurables and services also yields no support. One positive finding survives: real exchange rates are more variable and have larger absolute mean growth than consumption ratios, “which facts are consistent with gamma greater than 1.” The conclusion lists candidate repairs – taste shocks, wealth effects through non-homothetic preferences, measurement error from fixed-weight indexes, incomplete markets, and pricing to market with spatial segmentation – without endorsing one, noting that taste shocks would deliver the opposite-signed correlation but that the volatility prediction would still fail “unless there is considerable heterogeneity across countries.”

Gregor W. Smith | Macro Paper Warehouse

Consumption and real exchange rates in dynamic economies with non-traded goods