Inflation and seigniorage in Argentina

Polieoy, Planning, and Research WORKING PAPERS Vacroeconomic Adjustment and Growth Country Economics Department The World Bank October 1989 WPS 289 Inflation and Seigniorage in Argentina Miguel A. Kiguel and Pablo Andres Neumeyer In Argentina, increases in inflation appear to be closely linked to government attempts to increase seigniorage (government reve- nues from issuing money). The implication? Any serious stabilization effort requires finding an alternative source of revenue to replace the "inflation tax." The Policy. Planning, and Research Complcx disuributcs PPR Working Papers to disseminate thc fEndings of work in progress and to enoourage the exchange of ideas among Bank staff and all others intcrested in development issues. These papers carry the names of the authors, reflect only their views, and should be used and cited accordingly. Thc ftndings. interpreustions, and conclusions ame the authors own. They should not be attributed to thc World Bank, its Board of Directors, its managemcnt, or any of its member countries.  Plc,Planning, and Res4orch Maroeconomic Adjustment and Growth In their model of the relationship between of atout 7.5 percent of GDP in steady state (this inflation, the inflation tax, and scigniorage, was true for the tablita and pre-Austral periods). Kiguel and Neumeyer analyze the Argentine Between June 1978 ard April 1985, there was a expcrience - for the last decade. clear, positive relation betwcen inflation and the inflation tax for rates of i. flation below 18 To study the robustness of their model under percent. different regimes, they split the study into three periods - each with distinctive rules about the Events are more difficult to interpret at exchange rate, interest rates, and the mobility of inflation rates near and above 20 percent. in the intemational capital flows. 20 percent range, the inflation tax ranged from 7 to 10 percent of GDP. Steady-state seigniorage Argcntina - where increases in inflation is at a maximum 7.5 percent when inflation is appear to be closely linked to government around 20 percent a month. Increases in infla- attempts to raise seigniorage - is a natural tion above 20 percent do not give the go .m- choice for this study because of its persistent ment more inflation tax revenues. The revenue high rates of inflation and fiscal imbalance. from inflation seems to fall unambiguously once Monetization of fiscal deficits becomes a major inflation exceeds 22 percent. force for creating money and inflation in coun- tries with limited access to domcstic and foreign The inflation tax remained close to, and credit. even exceeded, maximum sustainable levels during the first half of the 1980s - and was Kiguel and Neumeyer found fhat inflation in probably the single most important source of Argentina played an important role in generating revenue to the government at that time. The im- public sector revenues. plication: any serious stabilization effort requires finding an altemative source of revenue At the revenue-maximizing rate of inflation, to replace the inflation tax. thcy found, the govemment can get seigniorage This paper is a product of the Macroeconomic Adjustment and Growth Division, Country Economics Department. Copies are available free from the World Bank, 1818 H street NW, Washington DC 20433. Please contact Raquel Luz, room NI l- 057, extension 61588 (43 pages with figures and tables). Thc PPR Working Paper Scrics disseminates thc findings of work under way in the Bank's Policy, Planning, and Research Complex. An objective of the series is to get these findings out quickly, even if presentations are less than fully polished. The findings, interpretations, and conclusions in thcsc papers do not necessarily represent official policy of the Bank. Produced at the PPR Dissemination Center  Inflation and Seigniorage in Argentina by Miguel A. Kiguel and Pablo Andr6s Neumeyer Table of Contents I. Introduction 1 II. Financial Arrangements and Inflation Tax 4 III. Seigniorage and Inflation Tax in Argentina 9 IV. The Demand for Money and the Inflation Tax 1l Laffer Curve V. Implications and Final Reflections 25 References 27 Tables 31 Appendix 35 Figures 37  *Seigniorage is the profit on minting coins, earned by the mint, usually owned or farmed by the sovereign, who has a certain 'droit de seigneur' or monopoly on such profits." "The Contrast ... may well turn on whether the competitors (mints) are interested in short- or long-run gains. In the short run, profits can be maximized by adulteration; in the long-run, by producing to quality standards."1 I. Introduction Public sector deficits occupy a central role in causing inflation in many developing countries. This is especially important in those countries where the government has to rely on the central bank to finauice its fiscal imbalance, due to its limited access to domestic and foreign borrowing. Monetization of fiscal deficits thus becomes the major force for money creation and inflation. Higher rates of inflation, however, do not always provide more resources to the government. There are several reasons for this outcome. First, as we know from the literatuire on the inflation tax, e.g. Friedman (1971), tiere is a revenue maximizing rate of inflation which corresponds to the point where the demand for money is unit elastic. Beyond that point, further increases in inflation will actually reduce the inflation tax revenue in the steady state. Second, there could be changes in inflation which are not of a fiscal nature. The balance of payments theory of inflation, as presented in Liviatan and Pitterman (1985), provides an example of such a case. Liviatan and Pitterman found that in Israel 1 ChArles, P. Kindleberger (1984), A Financial History of Western Europ, London, Ailen & Unwin.  2 inflation accelerated at times when the economy was facing serious external imbalances. Balance of payments problems triggered a mazi-devaluation which very quickly moved the economy to a higher inflationary *plateaus. If there is inertia in the inflation process which is accommodated through monetary and exchange rate policies, inflation could remain at the higher plateau even in the absence of a change in the budget deficit. The relationship between inflation and money financed budget deficits is illustrated in figure 1, where we show data on seigniorage, and inflation rates for Argentina, Bolivia, Brazil, Israel and Mexico. Seigniorage represents the amount of resources that the government gains from printing money and is measured here as a percentage of GDP. These figures indicate that there are two different types of relationships between seigniorage and inflation. In Brazil, Israel and Mexico seigniorage has been relatively stable over the years while inflation has displayed a tendency to rise. In these three countries the fiscal approach does not seem to provide a convincing explanation of the evolution of inflation. In Argentina and Bolivia, on the other hand, increases in inflation appear to be closely linked to attempts by the government to raise seigniorage. In this paper, we will investigate the relationship between inflation, the inflation tax, and seigniorage on the basis of the Argentine experience of the last decade. The persistent high rates of inflation and the continuously large fiscal isibalances observed in Argentina makes this country a natural choice for a case study on this topic. The paper will be organized as follows. In section II we present the basic analytical framework and discuss what is the appropriate measure of  seigniorage for Argentina. The framework is similar to other models of inflation finance (e.g. Bailey (1956), Friedman (1971), Calvo (1978), Anand and Van Wijnbergen (1989), etc.), but we adjust it to incorporate the major stylized facts of the Argentine financial system. This section establishes that due to the structure of reserve requirements and the charge. and compensations that the central bank imposes on the various deposits in the financial system, Ml is the basis for the inflation tax. This discussion is continued in section III where we examine the behavior of the inflation tax and seigniorage in Argentina from 1978 till 1985. In section IV we conduct an empirical study of the demand for money based on monthly data for the period 1979-85. Our results are of interest for two reasons. First, we split the sample in three different periods for estimation purposes, each of them having distinctive rules for the exchange rate, interest rates and on the mobility of international capital flows. This enabled us to study the robustness of our estimated parameters to regime changes. Second, we were able to overcome the simultaneity bias that arises because of the correlation between the opportunity cost of holding money and monetary shocks. This was possible because during the first period the money stock was truly endogenous and interest rates were determined by the preannounced rate of devaluation and by arbitrage conditions. For the third period we found that the stock of money and rate of inflation were cointegrated and therefore we could obtain consistent estimators of the money demand function's parameters. We conclude in section V with a discussion of the implications of our empirical results for the inflationary process in Argentina.  4 II. Financial Arrangements and Inflation Tax A. Seigniorage and Inflation Money creation is an important source of public sector revenue in many developing countries. The analytical literature on this subject (e.g. Friedman (1971), Calvo (1978), Bruno (1988), Bruno and Fischer (1986), Dornbusch and Fischer (1986), etc.) usually considers a closed economy, where money creation is driven by fiscal needs. In this paper we will follow the presentation used in Dornbusch and Fischer (1986), which will be modified to introduce a banking system. The money supply process is captured in equation (1) (1) AH - Pg where H represents the monetary base (i.e. the liabilities of the central bank), P is the price level and g is the monetized portion of the deficit. AH denotes the change in the monetary base over time. AH/P denotes the real amount of resources that the government receives from printing money, sometimes referred to as seigniorage. When there is a banking system, total money supply (M) will be given by (2) M - kH  5 where k is the money multiplier. The monitary base can be held as currency (C) by the public or used to satisfy thi reserve requirements on bank deposits (D). Defining the reserve reqtu -cement ratio as r, then k - (1 + cl/(c + r), where c is the currencT-deposit ratio. The stock of base money is H - (1/k)M. Real money balances (m) are defined as (3) m - M/P -kH/P - kh where h - H/P. Differentiating (3) with respect to time yields (4) m- M/P -mm - k(AH/P - lh) = k(g - fh) where m - dm/dt, g represents seigniorage and wh is the inflation tax. In the long run equilibrium (i.e. when real money balances are constant) seigniorage is equal te the inflation tax. The monetary base (i.e. the stock of central bank liabilities) represents the base for the inflation tax. The standard presentation of the inflation tax model is completed with the specification of the money demand function. In Cagan's model it is given by (5) md = khd , Ae-aP  6 where A is a constait, p is the expected rate of inflation." If we assume that expectations are rational then p - f. The basic structure of the mode?. is summarized in figure 2. The m a 0 scheaule, from equation (4), is a rectangular hyperbola showing the combinations of I and h such that seigniorage equals inflation tax. The md schedule depicts the pairs of h and i such that the money market clears. There are two stationary equilibria, points A and B, at which both conditions are satisfied simultaneously. The characteristics of the model and its stability properties are discussed at some length in Bruno and Fischer (1986), Dornbusch and Fischer (1986), Evans and Yarrow (1981), and Kiguel (1989). A clear implication of the model is that if the economy starts at the low inflation equilibrium (point A), and that print is stable, an increase in the budget deficit (shown by an upward shift in the m - 0 schedule) will lead to a permanent increase in the rate of inflation. A second important implication is that there is a maximum amount of seigniorage that the government can extract without destabilizing inflation. This corresponds to point C in figure 2, where the demand for money is tangent to the m- O schedule. Seigniorage in excess of that amount cannot be financed in a stable way. In that case, under plausible assumptions regarding thTe adjustment in the money market, there will be a continuous acceleration in inflation (see Kiguel (1989)). Notice that the continuous increase in inflation will occur in spite of a constant level of seigniorage. 2 Cagan's model represents the traditional way to analyze this problem. One possible extension of the model could be based upon the demand for money recently used in Eckstein and Leiderman (1989).  7 B. Remuneration of Reserve Requirements and Seigniorage The analysis needs to be modified in those cases where the central bank pays interest on bank reserves. This practice has be*n adopted in many high inflation countries (e.g. Argentina, Mexico, etc.) as a way to reduce the costs of financial intermediation. For simplicity, we can assume that the central bank p ys an interest rate Mi) on bank reserves, and that i - i. Under the fractional banking system being considered total deposits (D) are (6) D - 1/(c + r)H. We define d - DIP. In our example, seigniorage will be (7) AH/P - rrd - AC/P - g; in other words, the government collects the inflation tax on currency, while it returns to the private sector the tax on deposits through interest payments on reserves. An additional difficulty for the interpretation of the results arises if we extend the model to an open economy. In that case seigniorage can be used either to finance the budget deficit or to accumulate international reserves. This element was very important in Mexico during 1987,3 where seigniorage levels were relatively large, as can be seen from figure l1E, 3 A similar phenomenon is observed in Chile and Argentina during the period of the predetermined exchange rates (the Tablita). In both episodes money creation was linked to accumulation of international reserves by the central bank.  8 while the operational deficit of the consolidated public sector was negligible. It follows from the abo"e discussion that a correct calculation of the government's revenue from money creation requires a careful examination of the structure of the financial system and of the regulations on reserve requirements. We now turn to the Argentine case. On June 1, 1977 a financial reform introduced a fractional reserve banking system and liberalized interest rates. The central bank paid interest on the required reserves on time deposits to compensate anrks for the cost of these "immobilized' funds. At the same time, it charged conmercial banks interest on the fraction of the stock of demand deposits (on which banks did not pay interest) that they were able to lend. In other words, the central bank taxed away the seigniorage levied by commercial banks on demand deposits, while it compensated them for the required reserves on time deposits.4 Given that the interest rate paid and charged on reserves was roughly the same, the inflation tax (in steady state) was given by (8) rtax - f(cc + rddd + rtdt) + i(l-rd)dd - irtdt 4 This system.of taxes and subsidies WV:b recorded through the Monetary Regulation Account (MRA, in Spanish Cuenta de Rebalacion Monetaria). Two reasons were invoked for the creation of the MRA in June, 1977: (a) Paying interest on the legal reserves required for time deposits was a mechanism designed to eliminate the distortionary effect of a high legal reserve requirement on interest rates. (b) Taking away the inflation tax on commercial banks demand deposits provided an instrument to avoid an 'unfair' advantage of the latter over other financial institutions (financieras and savings and loans associatiors), that were not allowed to accept demand deposits. For a complete description of the Monetary Regulation Account, see En&ayos Economicos, No 31, September 1984.  9 where cc, dd and dt are resrectively currency, demand and time deposits in real terms, and rd and rt are the reserve requirements on demand and time d&posits. If we assume that the interest rate paid on reserves is equal to the rate of inflation, we can rewrite (8) as (8') rtax - w(cc + dd) . Hi, which is usually defined as the sum of currency plus demand deposits, thus becomes the basis for the inflation tax (irtax). This set up appears to be appropriate in studying the inflation tax in Argentina. A casual look at the evidence indicates that the central bank sets the interest rate on bank reserves at roughly the same levels as the rate of inflation. The choice of Ml appears to be robust to the various institutional changes that took place in the period under study5. III. Seigniorage and Inflation Tax in Argentina In the previous section we established that Hi is the relevant monetary aggregate to measure inflation tax and seigniorage. In this section we will present our estimates of these variables for the period under study and a brief interpretation of the stylized facts. There are a number of technical difficulties that arise when one 5 After the July 1982 financial reform the legal reserve requirement for demand deposits was usually above 90Z.  10 attempts to obtain accurate measures of inflation tax and seigniorage. An important part of the problem is that the government obtains seigniorage and collects the inflation tax on a continuous basis while our estimates are based on discrete observations. This concern can be very difficult to orercome whan inflation is high (in three digit levels).6 In this paper we adopted a methodology to calculate the inflation tax and seigniorage that satisfies some basic consistency criteria and yields results that are compatible with the existing literature and the empirical evidence. In a discrete time version, inflation tax and seigniorage (S) are given by (9) S - (Mt - Mlt_)/GDPt (10) ltax - S - (Mlt/PtYt - Mt-l/Pt-lYt) where GDPt is the nominal gross national product, Yt is real gross domestic product in period t and Pt is the price level at the end of period t. These definitions ensures that S - itax in the steady state. The results of our calculations of monthly seigniorage and of the inflation tax using equations (9) and (10) from 1977 to 1987 are presented in figure 3.7 We also included the corresponding inflation rates to illustrate the relationship between them. 6 For an excellent discussion of some of the problems see Rodriguez (1985) and Bressiani-Turroni (1937), Appendix to Chapter 3. Some of the difficulties in measuring seigniorage are also addresaed in Cukierman (1988). 7 A 5 period moving average was calculated for seigniorage to compensate for seasonal fluctuations in the variables.  11 This figure indicates that seigniorage has been an important source of revenue in Argentina, exceeding 3 percent of GDP for most of the period. There is also a marked increase in seigniorage between 1982 and 1985, which was accompanied by an increase in the rate of inflation. Seigniorage fell from mid-1985 on (after the Austral plan) and basically remained at pre- 1982 levels. Changes in seigniorage were very significant in five occasions. There were two sharp reductions in seigniorage, the first, at the beginning of 1981, resulted from capital outflows in anticipation of large devaluations (i.e. the end of the tablita period); the second, in late 1984, resulted from the implementation of tight money. There were also three large increases in seigniorage, the first, in the second half of 1982, was caused by the monetization of domestic debt under Cavallo; the second, in late 1983 and early 1984 resulted from a large increase in the budget deficit; and the last, in mid-1985, was driven by a remonetization during the early stages of the Austral plan. Of special interest is the acceleration in inflation that started in 1982 and was brought to a halt by the Austral plan in mid-1985. This acceleration was taking place at a time when seigniorage was relatively high (around 6 percent of GDP) but constant. One plausible interpretation of this episode, consistent with our discussion in section II.A, is that the amount of seigniorage was excessive in the sense that it could not be financed by any stable rate of inflation. Instead, it had to be financed in an unstable fashion through increasingly higher inflation rates. IV. The Demand for Money and the Inflation Tax Laffer Curve  12 In this section we will investigate 7rhether seigniorage levels were in effect excessive in Argentina based on an estimation of the demand for money. Using Cagan's demand for money function we will attempt to determine the value of the revenue maximizing rate of inflation and the corresponding level of seigniorage. There is no agreement, based on the existing literature on th. demand for money in Argentina, regarding the revenue maximizing rate of inflation. Fernandez and Mantel (1985), for example, estimated that this rate is in the 20 percent per month range, Rodriguez (1988)8 calculated numbers that are closer to 30 percent per month, while Demaestri and Duefias (1978) suggested that the rate is closer to 7 percent. Melnick (1988) incorporates a ratchet effect in the demand for money and estimates the revenue maximizing rate of inflation at 22Z when inflation exceeds previous levels and at 29Z when it does not. In this vection we will investigate the characteristics of the demand for money in Argentina based on monthly data from 1979 to 1985. For estimation purposes we will divide the sample into three clearly differentiated periods. The first one, from January 1979 to January 1981, corresponds to the interval in which the government preannounced the value of the exchange rate (the otablita' period). The authorities started the preannouncement of the exchange rate on December 20, 1978 and the regime continued in place until February 1981, when a 10 percent unscheduled devaluation was effected. During this time there were no controls on 8 in Bruno et al, Inflation and Stabilization  13 capital flows and hence the quantity of money was endogenously determined. Interest rates (which represent the opportunity cost of holding money) were essentially determined by arbitrage conditions and closely followed the international interest rate plus the expected rate of depreciation of the exchange rate (see Blejer 1982). There was a second transitional period, from February 1981 until June 1982, characterized by continuous changes in the structure of the financial markets and a lack of a rule for the exchange rate. There were two maxi- devaluations in 1981 (30 percent deval-