Consumption and macroeconomic policies: Theory and evidence from developing countries

Abstract

This paper examines determinants of private consumption in a sample of developing countries. The empirical model includes income, a proxy for the cost of consumption, and the exchange rate. Anticipated movements in these determinants are likely to trigger adjustment in planned consumption, while unanticipated changes determine random transitory adjustment in consumption. Fluctuations in private consumption are mostly random with respect to unanticipated changes in income and, to a lesser extent, the exchange rate. Consumption increases during cyclical expansion of income and decreases in the face of an unanticipated increase in the cost of consumption. Exchange rate fluctuations have mixed results on private consumption. As for the effects of domestic policies, fiscal policy has a limited, and sometimes negative, effect on private consumption. Monetary growth, in contrast, stimulates an increase in private consumption. This evidence supports recent calls to decrease the size of government and enhance the role of monetary policy in stimulating private activity in developing countries.

Key words:

• Consumption
• macroeconomic policies
• developing countries

Acknowledgements

The authors thank an anonymous referee for helpful comments on an earlier draft. The views in this paper are those of the authors and should not be interpreted as those of the International Monetary Fund.

Notes

∗ Other related references include Canner and Fergus (1987), Carmichael et al. (1999), Cashin and McDermott (1998), Dejuan and Seater (1999), Friedman (1957), Fuhrer (2000), Girardin et al. (2000), Hahm (1998), Kandil (2001), Krugman and Baldwin (1987), Leong and McAleer (1999), Mathieson and Rojas-Suarez (1990), Park and Rodrigues (2000), Rodrik (2000), Scott (1996), Zeldes (1989), Zhang and Wan (2002).

¹ This channel focuses on the relative prices of tradables and non-tradables. Private consumption may not be affected if consumers substitute imported goods for consumption of domestically produced goods. Other researchers have focused, however, on the contractionary effect of currency depreciation on real income and, therefore, private consumption. According to Diaz-Alejandro (1963), devaluation transfers real income from workers to producers of exports and non-tradables. The latter group has a smaller marginal propensity to consume. Along the same lines, Krugman and Taylor (1978) and Barbone and Rivera-Batiz (1987) have formalized several channels of the contractionary effects of devaluation on private consumption.

² For related references, see Heller and Starr (1979), Reinhart and Vegh (1995), Sarno and Taylor (1998), Hussein and de Mello (1999), and Yin and Wan (2002).

³ As the exchange rate is measured in real terms, it captures the effects of fluctuations in real growth in the largest economy of the world, as a proxy for world activity.

⁴ For details, see Kwiatkowski et al. (1992 which tests the null-hypothesis of stationarity). To select lags for the KPSS test, we follow the suggestions of Newey and West (1994). Non-stationarity indicates that the series follows a random walk process. Upon first-differencing, the resulting series is stationary. Table A1 in the appendix summarizes the results of non-stationarity. The results are robust with respect to alternative tests for the null-hypothesis of non-stationarity.

⁵ We test for the endogeneity of the explanatory variables in the model (see Table A3). Given evidence of endogeneity, the forecast equations account for lagged values of variables proven to be statistically significant. We also test for structural break in the forecast equations. Given evidence of structural break, we introduce dummy variables in the forecast equations. Having accounted for the dummy variables, the estimated reduced form equations do not exhibit any structural break.

⁶ In theory, consumption is endogenous with respect to domestic and external variables that appear on the right-hand side. It is possible, therefore, that they have a common stochastic trend in the long-run.

⁷ As long as there exists at least one cointegrating vector, it is necessary to control for this long-run relationship in the empirical model using stationary data. The error correction term captures deviation around the long-run trend, i.e. the lagged value of the residual from regressing the non-stationary dependent variable on the non-stationary variables in the model. For details, see Pesaran et al. (2001).

⁸ Historically, Syria has enjoyed a relatively moderate price inflation, or even negative inflation some time, compared to other developing countries. Accordingly, anticipated inflation indicates a pick up in aggregate demand and an increase in the growth of private consumption.

⁹ In a separate experiment, we estimate nominal consumption as a function of nominal government spending, nominal money, and the nominal exchange rate. Using this model specification, there is evidence of a significant crowding out of private consumption in the face of higher government spending, both anticipated and unanticipated in Egypt. Higher government spending increases the budget deficit and raises the interest rate. Higher deficit increases uncertainty and, therefore, incentives for savings. Similarly, higher interest rates increase incentives for savings and decrease consumption.