Exports and productivity tend to move together in developed market economies, but does this reflect a real causal link or just that exports are a component of GDP? Recent developments in trade theory under imperfect competition suggest that exports do have a causal effect on productivity, but the sign of this effect depends on the type of competition in the domestic market, ease of entry and exit, and whether in response to a trade disturbance market structure will change so as to promote improvements in productivity. In Discussion Paper No. 362, Research Fellow Dalia Marin attempts to resolve this issue empirically for four developed economies: the United States, Japan, United Kingdom and West Germany.
Recent empirical work on trade has employed calibration techniques, because of inadequate data for econometric estimation and hypothesis testing. Marin takes a quite different approach, using aggregate data and vector autoregression (VAR) techniques commonly used in applied macroeconomics. Whereas calibration is theory-driven, in the sense that assumptions about economic behaviour have to be made in order to construct a model, VAR techniques are not shaped by theory except in the choice of time-series variables. Marin uses quarterly data from 1960-87 for each country on manufacturing exports, manufacturing output per employee, manufacturing terms of trade and OECD output. The latter two variables reflect export growth from improvements in price competitiveness or growth in the world economy, while the terms of trade also detect possible linkages between the real exchange rate and productivity, as emphasized by recent `hysteresis' models of trade.
Marin first establishes that each series contains a stochastic trend. Cointegration regressions reveal that, for all countries except the United Kingdom, exports, productivity and the terms of trade share common trends and that exports, the terms of trade and world output are all positively related to productivity in the long run. Having characterized the trend properties of the data, she then examines causality. One variable `Granger-causes' another if forecasts of the second variable based only on its own history are improved by using the history of the first variable in that forecast. The tests show that, for all four countries, labour productivity is Granger-caused by exports. The hypothesis of export-led growth cannot be rejected for any of these countries, Marin concludes: an `outward-looking' regime seems to improve productivity performance in developed market economies as well as LDCs.
Her tests also reveal a significant causal link in both the US and the UK data, with improvements in the terms of trade positively associated with productivity growth. One explanation is that increases in the real exchange rate might induce rationalization of the import-competing sector, increasing average productivity both because of exit of low-productivity firms and through economies of scale, as incumbent firms take over the market shares of exiting firms.

Is the Export-Led Growth Hypothesis
Valid for Industrialized Countries? Dalia Marin

Discussion Paper No. 362, January 1990 (IT)