http://orcid.org/0000-0002-1649-6608
![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
Zimbabwe experienced record hyperinflation of 80 billion per cent per month in 2008. This article uses new data from Zimbabwe to investigate money demand under hyperinflation using an autoregressive distributed-lag model for the period 1980–2008. The results produce plausible convergence rates and long-run elasticities, indicating that real-money balances are cointegrated with the inflation rate and signifying an equilibrium relationship between the two series. Evidence is also presented suggesting prices were driven by increases in the money supply rather than by changes in price setting behaviour. The article uses the estimated elasticity on the inflation variable to calculate the maximum level of seigniorage revenue that could be raised in the economy. Actual seigniorage levels increased dramatically after 2000, with inflation eventually exceeding the rate required to maximize this revenue stream. This is discussed in relation to international financing constraints and the collapse of the domestic tax base.
Acknowledgments
Many thanks to Patrick Honohan, Janine Aron, Bent Nielsen, Arto Kovanen, Philip Lane, Luca Onorante, Thomas Conefrey and reviewers at Applied Economics for useful comments.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 Until early 2008, all domestic prices were set in Zimbabwe Dollars (ZWD), the sole legal tender. Throughout 2008, as inflation became increasingly high, ZWDs were substituted for US Dollars (USD) and/or South African Rand (ZAR), this was formalized in February 2009, whilst the ZWD was suspended in April 2009.
2 The population was close to 13 million inhabitants in 2009.
3 Source: World Development Indicators (WDI), 2010.
4 The Fast Track Land Reform (FTLR) programme saw agricultural land appropriated through a disorderly and extralegal process which proved disastrous for output, tax revenue and living standards (Barry, Honohan, and McIndoe-Calder Citation2014).
5 All ZWD denominated data are transformed to reflect the 2006–2008 dollar, i.e. in August 2006 three zeros were removed from the currency and in August 2008 ten zeros were removed from the currency to allow reintroduction of lower denomination notes in order to cope with the very high rates of inflation.
6 Separately, as hyperinflation set in, the Central Statistics Office (CSO) faced increasing delays in publication of its monthly series, which is useful for economic agents only at increasingly high frequencies as inflation accelerates.
7 None of the three series runs for the entire time period 1980–2008, this is addressed by merging the over-lapping periods of each series using monthly log growth rates.
8 The FC series is unpublished (therefore not widely available) and quarterly. Cubic spline interpolation is carried out to obtain a monthly series.
9 The time period over which the empirical estimation is carried out is 1980–2008, as previously discussed.
10 The RealMB series is now denoted in millions of USD (as the nominal money data have been deflated by the exchange rate). The comparable CPI-deflated series has been rebased so that the CPI = 100 in July 1994.
11 OMIR is the implied exchange rate given the price of an Old Mutual Share dual listed on the Harare and London stock exchanges.
12 Further details are available in the appendix.
13 In Zimbabwe the average annual inflation rate in the 1980s was 13 per cent and in the 1990s 30 per cent, i.e. inflation was high for most years before the onset of the 2000s, which was characterized by very high inflation.
14 The null of no cointegration H0: δ2 = α2 = 0 is tested against the alternative of H1: δ2; ≠ 0 α2; ≠ 0 by means of an F -test using the Pesaran, Shin, and Smith (Citation2001) bounds tables.
15 Monthly inflation did reach 33 million per cent in Zimbabwe in late 2008, beyond the estimation period of the article.
16 The hyperinflationary period sub-sample (2006M6-2008M1) does not contain enough data points over which an ARDL model could produce plausible converge parameters.
17 Similarly, during periods of low inflation governments may not be raising as much revenue as possible if they were to increase the money supply even marginally.
18 See Appendix 3 for OLS and AR(1) results on a simple Cagan money demand model. In addition to providing results on a standard Cagan model the appendix provides additional motivation for the use of the new CER price series and for the use of estimation techniques that specifically account for time series behaviour.
19 The F-statistic for the specification in Equation (4.1) is 9.05, above the relevant 2.5 per cent critical value of 8.27, although below the 1 per cent value of 9.63. The null of no-cointegration is rejected (at the 2.5 per cent level).
20 The period from 2000M1 to 2008M1 is used as it is a long enough time period over which to estimate an ARDL ECM. This time period from 2000 is characterized by a large increase in the inflation rate when compared to the pre-2000 average.
21 India’s rate of convergence is −0.47 whilst that of Thailand is −0.01 (Bahmani-Oskooee and Rehman Citation2005). These estimates are for periods of relatively low inflation; however, they are still indicative of the range of estimates that might be expected for money demand functions estimated in this way.
22 In addition, in the early 2000s, the ‘Prescribed Asset Ratio’ (applicable to pension funds and insurance companies) and the ‘Statutory Reserve Ratio’ (applicable to banks) allowed the RBZ access to domestic savings before the authorities resorted to the printing presses.
23 This is calculated as: 1/0.31 * 100, where 0.31 is the long-run elasticity on inflation, as estimated in the ECM (see ).
24 Or 2,417 per cent per year, 201 per cent per month, using simple growth rates.
25 Comparable figures for South Africa indicate that the cost of printing notes and coin comprise between 40 per cent and 60 per cent of total operating costs, this equates to 0.05 per cent of 2009 GDP. In a high inflation situation, the share of printing costs in total operating costs may become high.
26 Tax revenue is proportional in Zimbabwe. As long as the tax base is not altered, inflation should raise increasing amounts of tax. Thus, Tanzi-effects are not pertinent to this discussion. However, this is clearly a partial equilibrium story: given the other large shocks to the economy it is expected that domestic financing options become limited as the productive capacity of the economy declines.
27 It is acknowledged that debt-accumulation as well as borrowing costs inform part of the government’s budget constraint however, the focus here is on revenue flows. As a result, net debt flow rather than total debt owed is the appropriate variable for this article.
28 28 All ZWD denominated data are transformed to reflect the 2006–2008 dollar, i.e. in August 2006 three zeros, in August 2008 ten zeros and in February 2009 twelve zeros were removed from the currency to allow reintroduction of lower denomination notes in order to cope with the very high rates of inflation.
29 From an (unpublished) document compiled to establish (fair) compensation for commercial farmers affected by the Fast Track Land Reform Programme (FTLRP).
30 This series is derived from surveying businessmen in Harare on the actual rates they used to buy and sell foreign currency on the black market, established through correspondence with Techfin, Zimbabwe, summer 2008.
31 As above until December 2008 at which time many businesses closed indefinitely. January–April 2009 represents figures relating to the official interbank rate, as foreign currency was increasingly used in the real economy at this time, ZWD transactions were taking place primarily within the banking system. Data established through correspondence with J. Robertson, an economist in Harare, summer 2008–winter 2010.
32 Budget data are annual from 1980–1997, quarterly from 1998-2005Q2 and monthly from 2006–2007.
33 Aid data are annual.
34 The Zimbabwe African National Union – Patriotic Front (ZANU – PF) is the party of Robert Mugabe, current president of Zimbabwe.
35 Results are robust to changes of several months to the start of this period.