Modelling inflation in Ethiopia : an empirical investigation

Abstract

Summary

In this thesis I have tried to empirically evaluate two basic models of inflation-the triangular model of inflation of Gordon (1982, 85) and the P* model of inflation of Hallman (1991) - using annual data on Ethiopian economy of the period 1974-2005.

In chapter two, I have presented the overall structure of the economy during the period under study. Examination of the composition of GDP through out the past four decades shows only a slight decrease in the role of agriculture and a slight increase in the share of the service sector while manufacturing sector is stagnant. But fundamentally the economy is still agrarian: Agriculture still contributes half of the GDP and employs close to 90%of the labor force.

Chapter three presents the formal models of inflation and the estimation results based on these models. It is shown that the triangular model of inflation, which incorporates the workings of both the labor and product markets results in the usual expectation augmented Phillips curve type relation of unemployment and inflation. Estimation of this model, however, shows that the main variable driving inflation in the model- output gap - has the opposite effect of what is theoretically expected. One reason for such a result may be the presence of agricultural output in the GDP series. But estimation, after netting out the agricultural GDP, shows that the coefficient still remains negative. This puts a big question mark on the importance of the Phillips curve framework in explaining inflation in Ethiopia.

I, then, went on investigating the role of money in the inflation process. This is done in two ways: First inflation is regressed on past money to see if money granger causes inflation. The result indicates that past money growth has important predictive power on inflation. Second, I used a more formal monetary model of inflation-P* model of inflation. The basic idea of this model is that as long as velocity of money is stationary and money is neutral in the long run, current money growth will show up in future inflation. Estimation results indicate that: the main variable deriving inflation in this model-past pricegap- has the expected sign; it is robust to the price index employed as a dependent variable; and passes most of the diagnosis tests.

In addition, attempt has been made to see what variables explain fluctuations in food price. The reason for estimating food prices alone is that it accounts for a large portion of the consumption basket of the poor who are vulnerable for such fluctuations. The result indicates that most of the price change in food is a result of the performance of the agricultural sector which is heavily influenced by weather shocks. However, the growth of money-a substitute for demand shocks- and currency depreciation- substituted to represent supply shocks such as input prices ( eg. Fertilizer prices)-have important explanatory power as well.