How is the Hyperinflation in Zimbabwe Different?

Abstract

This chapter addresses two alternative reasons for hyperinflation in Zimbabwe: was it caused by speculation in the private sector or did reduced foreign aid result in the government using seigniorage to finance the deficit? A long-run money demand equation is estimated with an ARDL model, which is then used to test for self-perpetuating price movements and to study the interaction between money and prices. We find that prices were not self-perpetuating. Finally, the revenue-maximising level of seigniorage is calculated, and the scale of the reduction in ODA is compared to the fall in domestic financing. We find that authorities converged on a seigniorage-maximisation path only towards the end of the period of interest. We cannot reject that ODA had been expected to grow substantially from year 2000, and that it was this shortfall that the government financed by borrowing from the Reserve Bank of Zimbabwe.

Appendix: Empirical Motivation

Estimation is carried out on the log-linearised Cagan demand for money Eq. (4.11). Table 7.6 provides empirical motivation for the use of the novel CER price series and for augmenting the baseline Cagan model to take explicit account of both exogenous shocks and time series effects. Real money balances are regressed on inflation and exogenous shock variables using OLS and AR(1) estimation techniques. Columns (1) and (2) use the CPI series to measure inflation and deflate M1 in the dependent variable. The expected inverse relationship between inflation and RealMB is found only for the inflation rate measured using the CPI series when estimated via an AR(1) process (Column (2)). The change in sign on the inflation rate between Columns (1) and (2) and an AR(1) term of one in Column (2) indicates the need for an alternative measure of prices and that the persistence of the dependent variable needs to be modelled explicitly. The CER series is used in the remainder of the empirical work. In all specifications, inflation is negative and highly significant: accelerating price rises reduce holdings of real money balances. Introduction of an AR(1) term results in reduced precision on the exogenous variables indicating the trade-off between accounting simultaneously for time series behaviour and exogenous shocks. We include a dummy variable (Tail) for the hyperinflation period 2006M09 to 2008M1¹² and interact this with the inflation rate in Column (5) to establish whether the hyperinflationary period is driving the relationship of interest between RealMB and prices in Zimbabwe. The Tail dummy is not significant; however, the interaction term is both significant and positive. During periods of very high inflation in the country, increases in the inflation rate reduce the holdings of RealMB by a smaller amount than at other times. The interaction term suggests that this effect dampens the relationship between money and prices by about 40 per cent and may indicate a weakening in the relationship between money and prices as agents switch to holding alternate currencies as both a store of value and a medium of exchange. This is confirmed in Columns (6) and (7) which estimate a money demand relationship for the period up to the start of hyperinflation (1980 to mid-2006) and for only the hyperinflationary period (mid-2006 to early 2008), respectively. The coefficient on the inflation rate remains significant and negative for both sub-samples, although it is smaller in Column (7), for the reasons suggested above.

Table 7.6 Empirical motivation