Abstract
This paper compares Hungary’s experience with sterilization with that of other capital inflow episodes. The study focuses on the short-run impact of sterilization on monetary policy. The empirical data indicate that sterilized interventions by the National Bank of Hungary (NBH) were not significant until mid-1994, sometime after the return to power of the former Communist leaders. Thus, in the second half of 1994, the NBH began to demonstrate more firmly its independence by tightening monetary policy. By the beginning of 1995, the direction of fiscal policy had begun to show consonance with the overall aims of monetary policy.
I. Introduction
Fiscal and monetary policies in the transitional economies have had a difficult time overcoming the heritage of central planning. For example, central banks were quick to attempt to implement tight monetary policies and to behave like independent central banks (Siklos (1994)), whereas fiscal policy has seemingly continued to be governed by the soft-budget constraint policy of the past.
The Central and Eastern European countries also adopted different exchange rate regimes at the outset of the transitional phase, and typically experienced real appreciations as well as large inflows of capital in a relatively short time span. 1/ Thus, considering the consequences of these aspects of the macroeconomics of transition is important to understanding the process. Depending on their nature, inflows are vitally important to the process of rebuilding and recovery from the severe economic problems created by the combined effects of central planning and the forced industrialization policy of the Soviets.
Of course, capital flows come in many forms so the composition of the flows is an issue in itself. For example, a fraction of the foreign direct investment (FDI) component of capital flows coming into the transition economies was generated by the different privatization programs initiated under central planning. The fits and starts among programs perhaps helps to explain the varied record of these countries in attracting capital flows from the industrialized world and of their ability to recover from the transition process.
The real exchange rate is clearly influenced by the ability of central banks to sterilize capital flows. Accordingly, an assessment of the role of sterilization in the transition countries is a useful means of determining to what extent capital inflows affected the monetary base or the money supply more generally, and consequently the countries’ ability to control inflation. This study focuses on the short-term impact of sterilization on monetary policy, and Hungary--the use of sterilization by the National Bank of Hungary (NBH)--is a case in point.
Although Poland and the Czech Republic are quickly catching up, Hungary benefited early on from the capital inflows needed to restructure its economy from the distortions induced by central planning and so it has been chosen as the case study for this paper. 1/ As noted, significant capital flows (as a percentage of GDP) have been a feature of the Hungarian experience much longer than in the other transition economies, and the resulting appreciation in Hungary’s real exchange rate has been among the most significant across the transition countries (see the comparative data in Bank for International Settlements (BIS (1995a)). It is likely that foreign exchange market intervention in Hungary, for reasons that will be outlined below, has been the most deliberate among almost all the transition countries, with the possible exception of Czechoslovakia and its successor, the Czech Republic (see Hrnčíř (1995)).
Another reason for choosing Hungary as the case study is that the available time series are not only relatively longer for Hungary but are available for a much wider spectrum of macroeconomic variables than for the other countries. This allows estimation of NBH reaction functions, thereby permitting an evaluation of the effectiveness of sterilization. Further, the IMF and other international organizations--e.g., the European Bank for Reconstruction and Development (EBRD)--have at various times expressed reservation about the direction of Hungary’s macroeconomic policy. In the light of events in Mexico in 1994 and 1995 (see Calvo and Mendoza (1995)), it is also timely to evaluate Hungary’s actual policy stance.
The objective of this paper then is not to assess whether the National Bank of Hungary should engage in sterilization operations, nor to evaluate how difficult it might be to implement such policies, although a brief discussion of the issues is provided. 2/ Rather, it provides an empirical case study of Hungary, which broadly evaluates its dilemma with sterilization, and looks at the record of countries that have experienced similar capital inflow episodes elsewhere in the world, e.g., Latin America and Asia.
Summarizing briefly, the empirical data indicate that sterilized interventions by the NBH were not significant until possibly the middle of 1994, as the NBH began to more firmly demonstrate its independence by tightening monetary policy. By the beginning of 1995, fiscal policy had begun to show signs of change in a direction that was more in line with the overall aims of monetary policy.
The paper is organized in six sections. Section II provides an overview of exchange rate developments in Hungary and briefly contrasts them with the Polish and Czech approaches. Section III discusses the sterilization option, with special reference to economies in transition. Section IV outlines the specification of reaction functions for the National Bank of Hungary and for a model of money demand. Section V presents estimates of NBH reaction functions. A quantitative assessment of the extent to which the inflows of foreign capital were allowed to affect Hungary’s real balances is also provided. The paper does not, however, specifically address the potential implication of incomplete sterilization for, say, investment, the government budget, or foreign debt. Section VI concludes.
II. Exchange Rate Policy in Hungary
Unlike Poland, which adopted a crawling peg (see, for example, Sachs (1993)), or the Czech Republic (and its predecessor Czechoslovakia) which adopted a fixed exchange rate (Hrnčíř and Klacek (1991)), Hungary followed a policy of fixing the exchange rate with irregular devaluations. Moreover, since at least 1991, it has been the stated policy of the NBH to target the real exchange rate. Thus, for example, NBH (1995a, p.144) states:
“Monetary policy intends to stabilize the exchange rate. … it means that the exchange rate policy must aim for the stability of the real exchange rate”. 1/
In late 1994 and early 1995, the government, then made up of former Communists in a coalition government, embarked on a major policy shift in its policies via the appointment of a new Finance Minister and the return of a former central bank Governor (Hírmondó 10 February 1995), both apparently committed to continued economic liberalization and inflation control (Hírmondó 15 March 1995). In March 1995, the NBH Governor announced that, henceforth, Hungary would implement a crawling peg devaluation scheme (NBH 1995, and “Hungary Tries to Show it Means Business”, The Financial Times, 15 March 1995, page 2).
Figure 1 plots two measures of the real exchange rate for the Hungarian Forint (HUF) since 1987. The vertical bars indicate the months during which the NBH and/or the government devalued the currency (a Table in the Appendix provides the chronology of the devaluations). A few considerations have to be kept in mind when interpreting the data in Figure 1. While the calculations are based on the exchange rate vis-à-vis the Deutschmark (DM), the actual exchange rate was pegged to a basket of currencies subject to revisions over time. 1/ Second, and perhaps more importantly, the behavior of the real exchange rate is quite sensitive to whether calculations are based on the CPI or the PPI. As Sándor (1994) and Hochreiter (1995) point out, there are a number of technical and practical difficulties with relying on the CPI measure of the real exchange rate for economies in transition. 2/ The real exchange rate data in Figure 1 are defined in such a manner that a fall in the real exchange rate signals a real appreciation (in contrast with the IMF definition), and for Hungary the foreign price level used was the German CPI or PPI. 3/ While Figure 1 reveals that stabilization in the level of the CPI based real exchange rate was achieved by, roughly, the beginning of 1993, the apparent downward trend in the PPI based real exchange rate continued into 1994. In addition, there does not appear to be a statistically significant relationship between the timing of devaluations and the behavior of any of the real exchange rate measures (Siklos and Ábel (1995)). The NBH is not explicit about which real exchange rate it targets nor within what range, if any, it permits the real exchange rate to fluctuate, and a reading of its annual and monthly reports suggests that the preferred price measure has changed over time. 4/
Hungary: Real Effective Exchange Rates 1987–94
Citation: IMF Working Papers 1996, 086; 10.5089/9781451850833.001.A001
What are the major ingredients contributing to the real appreciation of the exchange rate over time? As noted previously, foreign direct investment in Hungary has been relatively large. Countering or offsetting the impact of FDI (which would include privatization proceeds) on the exchange rate has been Hungary’s relatively large foreign debt payable, naturally, in hard currencies. 1/ Finally, the pent-up demand for consumer goods, durables and nondurables, has produced a growing deficit on the current account. 2/ Concerns over these developments led the NBH to increase its focus in 1994 on the stability of the real exchange rate “… in order to achieve an expansion of exports compatible with the targeted improvement of the current account balance” (NBH 1995c, p. 69). Finally, as noted in the introduction, incompatible fiscal and monetary policies have also contributed to changing the real exchange rate.
Figure 2 shows the behavior of the monetary base and international reserves since 1987. International reserves (right-hand scale) are in billions of $US dollars while monetary base (left-hand scale) data are in billions of HUF. Variable definitions are provided in the notes below the Figure. 3/ Following a sharp rise in the former until late 1992, international reserves became more stable beginning in 1993 and the monetary base followed suit, an indication perhaps that the intensity of sterilized interventions by the NBH may have varied over time. Indeed, a crude measure of the scale of sterilization, namely the ratio of the change in international reserves (valued in HUF) to the change in the monetary base suggests that sterilization was modest in 1991 and 1992 (the ratio is 0.98) while for the 1993–94 (September) period there is an inverse relationship between changes in the two series. 4/ Moreover, Figure 2 hints at a possible change in monetary policy around the middle of 1994. Thus, while the level of international reserves began once again to climb rapidly during the second half of 1993, growth in the monetary base was quite modest. By the end of 1993 international reserves fell sharply while the base level was stable. The seeming policy change is perhaps not surprising in view of the rise in the current account deficit, the rising inflation rate (both actual and expected) and budget deficits, as well as concerns expressed domestically, and especially in international circles, about the perceived direction of fiscal policy, especially in late 1993 and early 1994, heightened by political events such as the reelection of the former Communists in May 1994 and disagreements over monetary policy between the then Finance minister and central bank Governor. 1/
Hungary: The Monetary Base and International Reserves 1987–94
Citation: IMF Working Papers 1996, 086; 10.5089/9781451850833.001.A001
III. To Sterilize or Not to Sterilize: Major Considerations for Economies in Transition
1. Monetary control in EIT
For students of Latin and South American economic history of the last two decades or so (see Rebelo and Végh 1995, and Calvo and Végh 1994), some form of exchange rate pegging has been the policy of choice for those economies that wished to dinsinflate quickly and credibly. The desire to disinflate in the economies in transition was prompted by the same considerations as well as by combination of at least two other factors, among others. First, worries about a monetary overhang, although this proved to be largely unfounded (Bruno 1992). Second, the need to reign in inflation following the immediate impact of price liberalization which policy makers knew would result in very high short-term inflation rates. 2/ The choice of exchange rate regimes was also driven primarily by the need for a credible and clear signal about the pursuit of policies geared toward steady inflation, believed to be an important ingredient in a successful transition to market. 3/ As a result, countries which successfully embarked on a course towards stable inflation experienced significant capital inflows as well as real appreciations of the exchange rate.
Under a rigidly fixed exchange rate and imperfect capital mobility, the inflows of official reserves can be sterilized via open market operations. In EIT, an additional weapon in controlling what would otherwise translate into domestic credit growth is the manipulation of reserve requirements of the banking sector. 1/ In Hungary, open market operations were used early in the transition. As a result, there was a rapid increase in holdings of government bonds, specially in 1992, a year when reserves of foreign exchange also rose rapidly (see Figure 2). 2/ Thereafter, growth in sales of government bonds to the public stalled with the commercial banking system becoming the most important outlet for government debt. Hence the desire on the part of the Hungarian government to find other sources of foreign exchange. The NBH attempted to sell bonds denominated in foreign currencies (e.g., in Swiss Francs or Japanese Yen) but these were not generally well received by international capital markets. 3/
Reserve requirements represent another tool of monetary control in Hungary. They are high (18 percent of deposits; see Kemme 1992, for details) were reduced to 14 percent in 1994 before being raised back to 18 percent in early 1995 in a bid to restrain credit creation) but, as the NBH has itself admitted (NBH 1995), reliance on this policy instrument can also lead to disintermediation. 4/ In 1993 the NBH added forward repurchase transactions to its arsenal of monetary policy tools. Once again, however, in 1995 these repo facilities were severely cut back. 1/
Concerns have also been raised recently not only about the size of capital flows but also about their composition, especially in light of the differences between Latin American and Asian countries (Calvo and Mendoza 1995, Calvo and Végh 1995). Real exchange rates have appreciated more in the former group of countries than in the latter. It has been suggested that, when capital flows are used to finance private consumption or are generated via privatization, these are relatively less productive in an economic sense. In contrast, most capital flows into Asia have been put to more productive uses. Hence, real exchange rate developments in countries in these two regions have been significantly different with Latin American countries experiencing significant real appreciation in their currencies while Asian countries have not (BIS 1995, pp. 113–14).
Privatization proceeds have certainly been one component of inflows into Hungary. 2/ Table 1 presents annual data from 1988–94 by major capital flow component, and clearly shows that revenues derived from privatization represent only a small component of the overall flows. The consequences of the policies of the NBH and of government, were clearly reflected in inflation and the external debt situation, both of which began to deteriorate in 1993 and 1994. As noted earlier, this decline followed a substantial improvement, and may have led to the policy reversal by the former Communists in early 1995 (Table 1).
Hungary: Capital Flows and External Debt, 1988–94
(In millions of U.S. dollars, unless otherwise specified)
Hungary: Capital Flows and External Debt, 1988–94
(In millions of U.S. dollars, unless otherwise specified)
| 1988 | 1989 | 1990 | 1991 | 1992 | 1993 | 1994 | ||
|---|---|---|---|---|---|---|---|---|
| Net medium and long-term capital | 568 | 1192 | -31 | 1668 | -894 | 3066 | 1161 | |
| Net direct investment income | 14 | 187 | 311 | 1459 | 1471 | 2339 | 1146 | |
| Privatization revenues | … | … | 10.6 | 419.5 | 947 | 889 | 92.9 | |
| of which: hard currency sales (in percent) | … | … | 79.1 | 78.4 | 54.8 | 31.2 | 6.6 | |
| Net short-term Capital flows | 288 | -44 | -893 | -617 | 5 | 459 | 960 | |
| Net Debt/GDP ratio (in percent) | … | … | 48.5 | 46.6 | 35.8 | 39 | 45.9 | |
| Debt Service/GDP ratio (in percent) | … | … | 12.6 | 12 | 11.4 | 11 | 12.6 | |
Hungary: Capital Flows and External Debt, 1988–94
(In millions of U.S. dollars, unless otherwise specified)
| 1988 | 1989 | 1990 | 1991 | 1992 | 1993 | 1994 | ||
|---|---|---|---|---|---|---|---|---|
| Net medium and long-term capital | 568 | 1192 | -31 | 1668 | -894 | 3066 | 1161 | |
| Net direct investment income | 14 | 187 | 311 | 1459 | 1471 | 2339 | 1146 | |
| Privatization revenues | … | … | 10.6 | 419.5 | 947 | 889 | 92.9 | |
| of which: hard currency sales (in percent) | … | … | 79.1 | 78.4 | 54.8 | 31.2 | 6.6 | |
| Net short-term Capital flows | 288 | -44 | -893 | -617 | 5 | 459 | 960 | |
| Net Debt/GDP ratio (in percent) | … | … | 48.5 | 46.6 | 35.8 | 39 | 45.9 | |
| Debt Service/GDP ratio (in percent) | … | … | 12.6 | 12 | 11.4 | 11 | 12.6 | |
2. The sterilization dilemma
Before moving to the empirical estimates presented below, it is worthwhile to briefly outline the dilemma facing the economies in transition with capital inflows and the potential necessity for sterilized interventions by their respective central banks. One issue is the credibility of a transitional program, which would be influenced by the availability of new deposit instruments paying market interest rates; this would raise the demand for broad money and, consequently, the equilibrium nominal interest rate, ceteris paribus. In principle then, the central question is whether inflows of capital are used to finance private consumption or to finance productive investment. Capital inflows can lead to real exchange rate appreciations or alternatively to an expansion in the money stock. An expansion of money stock, of course, has inflationary consequences with additional consequences for inflation expectations and money demand. The former will rise and the latter variable’s behavior is, of course, crucially dependent on how it is measured.
Another complication in the context of the economies in transition is the fragility of the financial system. Thus, higher interest rates might simply attract funds into the market for government Treasury bills which are perceived to be less risky than commercial bank deposits. 1/ One might also suspect that these higher rates would attract so-called “hot” money in search of relatively higher (real) returns. There are a variety of reasons to doubt that this would take place, at least in a transitional economy such as Hungary. First, the NBH has had little success in marketing bonds denominated in foreign currencies. Second, the informational and other transactions costs may be too prohibitive, and the uncertainties of the ex post real returns too great, to entice capital inflows of the short-term variety. Instead, it will be argued that the resort to sterilized intervention was deemed necessary by the NBH to demonstrate its credibility and independence from the government particularly in light of the concerns expressed by the international community in 1994 about the direction of fiscal and monetary policies following the return of the former Communists to power. 2/
It is also questionable whether sharply higher domestic interest rates are necessarily the outcome of the central bank’s attempt to sell bonds to the public or the commercial banks. The information and transactions costs of acquiring foreign assets are still considerable and not only because of restrictions on capital outflows and foreign exchange transactions. Moreover, given the poor state and relative immaturity of the domestic commercial banking system, government bonds have always represented a more attractive investment to all sectors of an essentially captive market so that significantly higher rates may not be necessary to absorb new government bond issues.
IV. A Reaction Function for the National Bank of Hungary and the Behavior of Money Demand
This section estimates a reaction function for the NBH, and explores the behavior of money demand. The aim is to discern the scope for sterilized interventions by the NBH, the short-term impact of sterilization and its implications for the conduct of monetary policy in Hungary.
1. Specification and estimation of a reaction function 1/
A conventional reaction function is specified (e.g., see Cumby and Obstfeld 1983). 2/ The monetary authorities are viewed as sterilizing the monetary effects of the balance of payments by changing their domestic assets according to the following:
where DA represents domestic assets of the central bank, Z is a vector of variables also believed to affect monetary policy (see below), while the coefficient a represents the degree of sterilization, and t is the time subscript. The current account of the balance of payments is denoted by CA, K equals net capital flows, and CA+K is the balance of payments, that is, the net change in foreign assets (ΔFA, is usually proxied by changes in international reserves). 3/ Obviously, when measuring in domestic currency units one needs to adjust equation (1) to exclude valuation increases due to periodic devaluations. An additional consideration arises with changes in reserve requirements. As noted earlier, changes in reserve requirements also represent a feature of monetary policy, and this can be particularly true of the transition economies (Kemme (1992) and Sahay and Végh (1995)). In what follows then, we define the monetary base (the measure used to proxy central bank credit creation) in adjusted terms, namely corrected for the impact of changing reserve ratios over time. 4/ Since changes in the monetary base are then captured by a combination of changes in the adjusted base, and changes in the domestic and foreign assets of the central bank, Cumby and Obstfeld (1983) suggest the following measure of monetary policy to which the central bank reacts:
where MP represents monetary policy while BA represents the adjusted monetary base (DA has already been defined and is adjusted for changes in reserve requirements). Combining equation (1) and equation (2) we can write a reaction function for the NBH in regression form:
In equation (3), where FA represents the foreign assets of the central bank (proxied here by international reserves) sterilization neutralizes the impact of inflows by offsetting effects on domestic credit. When 0 < α1 < 1 sterilization is less than perfect, and nonexistent when the coefficient is zero. Finally, a positive value for αl indicates that internal monetary control or balance is being sacrificed for external balance.
The remaining variables in equation (3) capture other influences which have a bearing on monetary policy (i.e., the vector Zt in equation (1)). These include: the rate of change in output (y), the rate of change in the real effective exchange rate (ϵ), government deficits (DEF), and a variable to capture irregular devaluations of the Forint (DEV). 1/ Since some of the data are not seasonally adjusted additional dummies to capture the impact of seasonality are also necessary. All the series were also examined for nonstationarity via unit root testing. Output, the real exchange rate, DEF were all found to have a unit root in log levels (details not shown) so equation (3) are estimated using first differences of the series. In addition, tests for cointegration were applied. One would not reject the null of a single cointegrating vector among the variables in equation (3). 2/
An additional consideration is that since most of the right hand side variables in equation (3) are endogenous, estimation via Ordinary Least Square (OLS) may be inappropriate. 3/ Two-stage least squares is used to estimate equation (3) with the constant lagged values of the endogenous variables (y, ϵ, DEF, MP), the (lagged) Hungarian (RHun) -German (RGer) interest rate differential, and the (lagged) gap between borrowing and lending rates as instruments. 4/
Clearly, capital flows, the Hungarian budget deficit and, therefore, monetary policy could be influenced by the interest rate differential with Germany. The gap between borrowing and lending rates at commercial banks in Hungary reflects a combination of factors which can significantly impinge on the behavior of the variables in equation (3). The gap is a function of the costs of obtaining liquidity from the NBH and/or foreign sources, is representative of the perceived risks of commercial lending to the so-called enterprise sector, and serves as an indicator of the potential for disintermediation in the banking system. As such, a rise in the gap will signal a potential contraction in domestic credit to the enterprise sector if the gap is due to a rise in the lending rate relative to the borrowing rate. Similarly, a contraction in deposits following a reduction in the borrowing rate relative to the lending rate will result in a contraction of commercial loans which are viewed as being riskier than holding government debt. Thus, while the gap will affect monetary policy, the direction of the effect is unclear.
2. Specification of a money demand function
It is useful to remind observers of the development of economies in transition that monetary policy in the prereform period was entirely passive to the central plan and, therefore, endogenous. Indeed, separate monetary circuits, as they are called, existed for the enterprise and household sectors. The reason is that the enterprise sector could obtain credits directly from the central banks (or, more likely, from the specialist banks operating as agents for the central bank) while credit to the household sector was either unavailable or severely restricted. Thus, the banking system did not act as an intermediary and there was no need for a conduit between savings and investment (see Ábel and Szekely (1992) for a more complete description). During the transitional phase monetary policy, while no longer passive, nevertheless remains somewhat hobbled by its centrally planned heritage (Ábel and Siklos 1994). Therefore, another element in the discussion about which monetary aggregate should be the target of sterilized interventions (see the discussion in Sahay and Végh 1995) is whether the newly freed enterprise and household sectors money demand functions are similar. Figure 3 plots disaggregated measures representing the money stock (using Hungary’s broadest definition, comparable to M3 in other countries) held by the enterprise and household sectors, respectively. There are three features worthy of discussion. First, the time series behavior of the two series appears to be quite different. 1/ indeed, one cannot reject the null that real balances held by the enterprise sector are stationary (i.e., the series does not contain a unit root based on the augmented Dickey-Fuller statistic; real balances held by households, by contrast, have a unit root). Second, the fraction of the total money stock in the hands of the household sector has been rising over time. Third, the sharp rise in money demand by households appears to end abruptly in 1993 and, in fact, the downward trend is not reversed until the middle of 1994, that is, precisely when the NBH’s policy seems to have changed. Figure 3 suggests then a loss of credibility in the macroeconomic policies around the beginning of 1993.
Hungary: Household and Enterprise Sector Real Balances
Citation: IMF Working Papers 1996, 086; 10.5089/9781451850833.001.A001
Due to the time series behavior of real balances held by the household sector, it is investigated via a fairly conventional money demand function except that the proposed specification is augmented with variables to capture the potential impact of capital flows.
where m is the logarithm of household real balances (i.e., nominal balances deflated by the CPI) and all other variables have been defined previously. All series, except DEV, are in first differences to ensure their stationarity (see also the discussion in the following section). 1/
3. Data sources
Monthly data from 1991 to the end of 1994 from the National Bank of Hungary’s Monthly Reports were employed for the monetary, fiscal, and some of the exchange rate variables (i.e, m, FA, ϵ, DEF, DEV). Data for industrial production, domestic credit, consumer and producer prices are from the International Monetary Fund’s International Financial Statistics CD-ROM, as well as the NBH data source. All series, before differencing, are in domestic currency units (e.g., billions of HUF) or in percent (e.g., interest rate and DEV).
Ideally, one might have wished to rely on, say, a measure of the output gap instead of the rate of change in output, especially in equation (3). Given that economies in transition are undergoing radical changes in economic structure it would seem particularly difficult to attempt to estimate anything resembling a potential output series at this stage. Output is proxied by industrial production since the data are sampled at the monthly frequency.
Next, since this study is concerned with the short-run impact of sterilization on monetary policy it is also important not to specify what might look like a long-run relation. All the relevant series are stationary in first differences based on the augmented Dickey-Fuller test.
In addition to the definition of monetary policy in equation (2) changes in monetary policy are also proxied, as in other such studies, by the change in domestic credit which is the sum of credits to the government, enterprises (large and small) and household sectors. These data are also from the NBH. Finally, the real exchange rate is measured in terms of German prices (i.e, e [pGer/pHun], where e is the nominal exchange rate) and converted into an index with a 1987 base year.
V. Estimation Results
1. Reaction functions of the National Bank of Hungary
As noted previously, the existing literature gave relatively little thought to the statistical properties of the time series under consideration. Thus, estimation of short-run relationships can be misspecified if a long-run equilibrium relationship exists between the central variables of interest and it is not explicitly recognized in the estimated equation. The reason is that equation (3) (and equation (4)) is a short-run equation which omits a long-run equilibrium condition which might be present between MP, FA, y, and ϵ. There is, as Engle and Granger (1987) have pointed out, a natural connection between the long-run, statistically represented by the condition of cointegration, and the short-run represented by the addition of an error correction term in equation (3) which shall be labelled EC. 1/ Four cases were considered in cointegration testing, depending upon whether the real exchange rate (ϵ) is calculated on a CPI or PPI basis and whether domestic credit (DC) is used to proxy MP (as in Obstfeld 1983).
The results (not shown) 2/ clearly reveal that when DC is used to proxy MP there is no evidence of any long-run relationship between the variables (i.e., the null of no cointegration cannot be rejected). When MP, as defined in equation (2) is used instead, the tests reveal that a single vector or linear combination can characterize the long-run relationship between the four variables considered. 3/ It is unclear what might explain the differences in the results except that the domestic credit series is from the IMFs IFS while the other proxy for MP is from Hungarian sources. The estimates of the cointegrating vectors suggest that higher industrial production implies a tighter monetary policy (i.e., MP falls), and the same is true when there is an appreciation of the real exchange rate (i.e., ϵ falls). 4/ Indeed, the long-run coefficient on FA is not statistically different from -1 (the null is distributed as a χ2 with one degree of freedom with a test statistic of 0.97 (.33 sig. level)) which would imply complete sterilization, that is, the complete neutralization of capital flows in the long-run. Similar cointegration test results are obtained when the real exchange rate is measured on a PPI basis, except that the sterilization coefficient is much larger (the null that α1=-1 in the long-run is rejected; χ2 (1)= 5.96 (.02 sig. level)) which would suggest an overreaction to capital inflows by the NBH. It was pointed out earlier that the underlying behavior of the. PPI and CPI series are quite different from each other.
Estimates of NBH reaction function (3) augmented, where needed, with the appropriate long-run restrictions based on the results of cointegration tests, are given in Table 2. The estimates shown are based on two-stage least squares (TSLS) with the list of instruments used provided in the notes to the Table. 1/ Regressions were generated according to whether the real exchange rate was measured using the CPI (columns labelled (1)) or the PPI (columns labelled (2)). Separate tests were also generated by including the error correction term for the full sample case only. 2/ Moreover, because heteroskedasticity was found in most of the estimated models, all t-ratios are based on heteroskedasticity-consistent standard errors. 3/
Hungary: Reaction Functions of the National Bank of Hungary, 1991–94: Two-Stage Least Squares
Hungary: Reaction Functions of the National Bank of Hungary, 1991–94: Two-Stage Least Squares
| Dependent Variables | MP | DC | ||||||
|---|---|---|---|---|---|---|---|---|
| Samples | ||||||||
| Independent Variables | 1991:01–1994:12 | 1991:01–1994:04 | 1991:01–1994:12 | 1991:01–1994:12 | ||||
| (1) | (2) | (1) | (2) | (1) | (2) | (1) | (2) | |
| ΔFA | -.96 (.90) |
-.69 (1.89)* |
-.77 (.49) |
-.40 (.98) |
-.11 (.08) |
-.42 (.78) |
-.20 (.81) |
-.28 (.99) |
| Δy | .94 (.16) |
1.91 (.40) |
3.30 (.22) |
2.88 (.38) |
-2.54 (.34) |
-1.05 (.25) |
-1.19 (.42) |
-.82 (.27) |
| Δϵ | -3.06 (.11) |
-4.13 (.71) |
-1.45 (.05) |
-5.34 (.76) |
19.44 (.50) |
7.00 (.66) |
-1.78 (.51) |
-1.72 (.46) |
| ΔDEF | -.001 (.44) |
-.001 (.79) |
-.007 (.37) |
-.002 (1.47)+ |
.0003 (.08) |
-.001 (.24) |
.001 (1.13) |
.001 (.96) |
| DEV | 18.58 (2.57)* |
17.44 (2.98)* |
30.86 (1.41)+ |
21.91 (2.91)* |
-4.45 (.59) |
-2.55 (.47) |
-.65 (.29) |
-2.05 (.68) |
| JAN | 31.41 (.94) |
29.87 (1.04) |
-8.97 (.09) |
14.82 (.61) |
15.00 (.30) |
8.16 (.20) |
10.99 (.41) |
16.39 (.52) |
| DEC | 57.01 (1.02) |
54.96 (2.41)* |
75.91 (.79) |
63.07 (2.16)* |
71.09 (1.09) |
83.93 (1.80) |
62.27 (1.73)* |
77.30 (1.73)* |
| Constant | 8.40 (.30) |
.03 (.002) |
1.94 (.04) |
-9.50 (.56) |
28.89 (1.09) |
24.65 (1.26) |
12.34 (2.43)* |
12.50 (2.04)* |
| R2 adj. | .27 | .24 | .08 | .08 | .11 | .21 | .07 | .09 |
| S.E. Regr. | 47.44 | 45.75 | 94.62 | 39.39 | 59.49 | 35.90 | 32.05 | 30.42 |
| F (sig. lev.) | 2.41 (.04)* |
4.12 (.002)* |
.56 (.08)* |
3.38 (.01) |
.84 (.56) |
1.77 (.13) |
2.91 (.02)* |
2.49 (.04)* |
| ECt-1 | .24 (.20) |
-.06 (.38) |
-- | -- | … | … | … | … |
Hungary: Reaction Functions of the National Bank of Hungary, 1991–94: Two-Stage Least Squares
| Dependent Variables | MP | DC | ||||||
|---|---|---|---|---|---|---|---|---|
| Samples | ||||||||
| Independent Variables | 1991:01–1994:12 | 1991:01–1994:04 | 1991:01–1994:12 | 1991:01–1994:12 | ||||
| (1) | (2) | (1) | (2) | (1) | (2) | (1) | (2) | |
| ΔFA | -.96 (.90) |
-.69 (1.89)* |
-.77 (.49) |
-.40 (.98) |
-.11 (.08) |
-.42 (.78) |
-.20 (.81) |
-.28 (.99) |
| Δy | .94 (.16) |
1.91 (.40) |
3.30 (.22) |
2.88 (.38) |
-2.54 (.34) |
-1.05 (.25) |
-1.19 (.42) |
-.82 (.27) |
| Δϵ | -3.06 (.11) |
-4.13 (.71) |
-1.45 (.05) |
-5.34 (.76) |
19.44 (.50) |
7.00 (.66) |
-1.78 (.51) |
-1.72 (.46) |
| ΔDEF | -.001 (.44) |
-.001 (.79) |
-.007 (.37) |
-.002 (1.47)+ |
.0003 (.08) |
-.001 (.24) |
.001 (1.13) |
.001 (.96) |
| DEV | 18.58 (2.57)* |
17.44 (2.98)* |
30.86 (1.41)+ |
21.91 (2.91)* |
-4.45 (.59) |
-2.55 (.47) |
-.65 (.29) |
-2.05 (.68) |
| JAN | 31.41 (.94) |
29.87 (1.04) |
-8.97 (.09) |
14.82 (.61) |
15.00 (.30) |
8.16 (.20) |
10.99 (.41) |
16.39 (.52) |
| DEC | 57.01 (1.02) |
54.96 (2.41)* |
75.91 (.79) |
63.07 (2.16)* |
71.09 (1.09) |
83.93 (1.80) |
62.27 (1.73)* |
77.30 (1.73)* |
| Constant | 8.40 (.30) |
.03 (.002) |
1.94 (.04) |
-9.50 (.56) |
28.89 (1.09) |
24.65 (1.26) |
12.34 (2.43)* |
12.50 (2.04)* |
| R2 adj. | .27 | .24 | .08 | .08 | .11 | .21 | .07 | .09 |
| S.E. Regr. | 47.44 | 45.75 | 94.62 | 39.39 | 59.49 | 35.90 | 32.05 | 30.42 |
| F (sig. lev.) | 2.41 (.04)* |
4.12 (.002)* |
.56 (.08)* |
3.38 (.01) |
.84 (.56) |
1.77 (.13) |
2.91 (.02)* |
2.49 (.04)* |
| ECt-1 | .24 (.20) |
-.06 (.38) |
-- | -- | … | … | … | … |
The results suggest that a significant amount of sterilization took place during the 1993–94 period but only when the PPI based real exchange rate measure is used and not when DC is used to proxy the NBH’s monetary policy. Given the potential importance of events in 1994, due primarily to the return to power of the former Communists, it seems desirable to determine whether the results could be sensitive to sample choice. 4/ Sterilization was found to be possibly a feature of the most recent sub-sample only (i.e., 1994). This represents a little bit of evidence to the effect that the reelection of the former Communists signalled a somewhat tighter monetary policy, and not the looser monetary policy they advocated during the election campaign. 5/ This interpretation would also be consistent with the earlier description (see Figure 2) of the relationship between international reserves and the monetary base. Assuming that a PPI based real exchange rate and MP are the preferred variables there is both evidence of sterilization, with a coefficient somewhat higher than those obtained for many, but not all, other countries (e.g., see Rivera-Batiz and Rivera-Batiz 1994, p. 413), and of a tightening of monetary policy during the second half of 1994. The results are also broadly consistent with the view that the NBH permitted real appreciations of the HUF, at least until recently when stabilizing the real exchange rate became the dominant objective of monetary policy. This might also explain why extensive sterilization in Hungary is a relatively recent phenomenon.
Turning to some of the other coefficients we find no significant reaction by the NBH to any of the fundamentals considered, except DEV which, in effect, captures the monetization of the exchange rate devaluations and reflects the continued softness of the government budget constraint during the period considered. 1/ Finally, notice that the error correction terms are statistically insignificant thereby suggesting the possible lack of cointegration. This result is no doubt partly explained by the combination of a relatively small sample and high volatility in the time series in question. This also explains the relatively low coefficient of determination in the regressions although, for the most part, they are all statistically significant.
2. Estimates of household sector money demand
Once again cointegration tests were constructed for the series in (4) in levels (with the exception of the DEV variable). One cannot reject the null of a single cointegrating vector at the 5 percent level. The estimated vector is mh + .003 RHum [.002] -.035 RGer [.009]-.0004 FA [.00004] + .41 y [.061] with all the long-run coefficients statistically significant at the 5 percent level (standard errors in brackets), except for the Hungarian interest elasticity coefficient. By contrast, long-run household money demand is highly sensitive to the German interest rate and to foreign exchange reserves. Table 3 provides estimates of (4) augmented with the appropriate error correction term. The estimates reveal that, in the shortrun at least, households do not respond to Hungarian interest rates but do react negatively to change in the German rate (lagged) while changes in international reserves are insignificant. 2/ Recursive estimates of the lagged German elasticity coefficient reveal that the significance of this coefficient is largely confined to the post- 1994 election period which is, again, consistent with a policy of sterilized intervention by the NBH. The lack of response to domestic interest rates is a feature of poorly developed domestic capital markets and may help explain why the NBH felt that it was possible to sterilize with modest consequences for domestic interest rates. Nevertheless, the interest rate differential between Hungary and Germany did grow sharply in 1994, although the levels reached were well below those attained during much of 1992.
Hungary: Household Sector Money Demand Estimates
Dependent Variable: (log) Household Sector Real Balances
Hungary: Household Sector Money Demand Estimates
Dependent Variable: (log) Household Sector Real Balances
| Independent Variables | Coefficient | Standard Error |
|---|---|---|
| Constant | 1.590 | .42* |
| ΔRHun | -.003 | .003 |
| ΔRHunt-1 | .003 | .003 |
| ΔRGer | .035 | .020 |
| ΔRGert-1 | -.056 | .021* |
| ΔFA | .0001 | .0001 |
| ΔFAt-1 | -.00019 | .00015 |
| Δy | -.001 | .001 |
| Δyt-1 | -.001 | .001 |
| DEV | -.003 | .003 |
| ECt-1 | -.481 | .129* |
| Diagnostics: | ||
| Autcorr. (Q(5)) | 2.36 (.80) | |
| ARCH (1) | .35 (.56) | |
| Whit | .25 (.99) | |
| R2 adj. | .31 | |
| RSS | .03 | |
| F (sig. level) | 3.09 (.01)* | |
| Reset (1) | 1.18 (.28) | |
Hungary: Household Sector Money Demand Estimates
Dependent Variable: (log) Household Sector Real Balances
| Independent Variables | Coefficient | Standard Error |
|---|---|---|
| Constant | 1.590 | .42* |
| ΔRHun | -.003 | .003 |
| ΔRHunt-1 | .003 | .003 |
| ΔRGer | .035 | .020 |
| ΔRGert-1 | -.056 | .021* |
| ΔFA | .0001 | .0001 |
| ΔFAt-1 | -.00019 | .00015 |
| Δy | -.001 | .001 |
| Δyt-1 | -.001 | .001 |
| DEV | -.003 | .003 |
| ECt-1 | -.481 | .129* |
| Diagnostics: | ||
| Autcorr. (Q(5)) | 2.36 (.80) | |
| ARCH (1) | .35 (.56) | |
| Whit | .25 (.99) | |
| R2 adj. | .31 | |
| RSS | .03 | |
| F (sig. level) | 3.09 (.01)* | |
| Reset (1) | 1.18 (.28) | |
The error correction term is highly significant and has the correct sign suggesting that the underlying equilibrium relationship exists and that disequilibria are reversed in just over two months time. Examination of the recursive residuals do not reveal any outliers in particular and the equation passes all the diagnostic tests. Nevertheless, these residuals (not shown) do reveal that while expected money demand is often above actual money demand before the middle of 1994, a reversal takes place following the elections of that year. This outcome is also consistent with the loss of credibility hypothesis and its recovery by mid-1994. A Chow test for a break in the relationship in May 1994 rejects the null hypothesis of coefficient stability (Log likelihood = 13.18 [.07 sig. level]).
Overall, the results are consistent with the increased credibility of the NBH’s policy by the middle of 1994, which raised the demand for an opportunity for the central bank to intensify its sterilized interventions. However, while household money demand does not appear to be responsive to domestic interest rates so far, it remains an open question whether the rising interest rate differential between Hungary and Germany can be sustained for long without jeopardizing the credibility of present policies, as Calvo (1991) feared for other countries faced with the sterilization dilemma. It is also to be noted that the Hungarian government has, so far, missed all its budget deficit projections by a wide margin and has only since early 1995 vowed to stabilize the budget. The other ingredients in the current macroeconomic policy being pursued by the NBH may make this rather more difficult than anticipated.
VI. Conclusions
This paper has considered the performance of monetary policy in Hungary, which during the 1990s has experienced large capital inflows under its system of quasi-fixed exchange rates with limited capital mobility. Despite high levels of foreign indebtedness, net inflows of capital into Hungary have been important enough to inquire about the policy actions taken by the National Bank of Hungary in the face of these inflows. Of particular concern is the NBH’s attempt to sterilize the inflows, a policy which, even when exchange rates are fixed and despite several irregularly timed devaluations in the period considered, is akin to implementing independent monetary policy. The results in this paper suggest that only during the second half of 1994 did the NBH successfully sterilize to neutralize capital inflows in order to pursue its objective of targeting the real exchange rate of the forint. No sterilization effect could be detected when a subsample, which excludes the period when the former Communists returned to power, was considered. The same result held true when the real exchange rate was measured with consumer prices rather than producer prices, or when domestic credit was used to proxy monetary policy actions. Nevertheless, descriptive evidence suggests that a policy change did take place sometime during 1994 when the NBH moderated monetary base growth via sterilization, in line with its more restrictive monetary policy. It was suggested that this may not, at first, have been the intended policy of the newly-elected government in May 1994.
Despite the short sample period 1991–94, and other statistical problems raised in the paper, the reaction function estimates presented here are consistent with a policy change by the NBH in mid-1994. Thus, the present era of restraint, formally in early 1995 with the appointment of a new Finance Minister and the return of a former central bank Governor appears to predate these appointments by several months.
Similarly, estimates of money demand by the household sector reveal sensitivity to foreign interest rates and only a weak response at best to changes in foreign exchange reserves. Moreover, the estimated money demand function also appears to be consistent with the view that significant sterilization is a relatively recent phenomenon.
Hungary: Schedule of Forint Devaluation Rates, 1989-94
Hungary: Schedule of Forint Devaluation Rates, 1989-94
| Year | Date (day/month) | Size of Devaluation | Reference Basket |
|---|---|---|---|
| 1989 | 21 March | 5 | weighted average of several currencies |
| 14 April | 6 | ||
| 18 July | .5 | ||
| 25 July | .1 | ||
| 29 July | .1 | ||
| 1 August | .07 | ||
| 8 August | .53 | ||
| 15 August | .2 | ||
| 15 December | 10 | 50 percent $US - 50 percent ECU [from 8/12] | |
| 1990 | 31 January | 1 | |
| 6 February | 2 | ||
| 20 February | 2 | ||
| 1991 | 7 January | 15 | |
| 8 November | 5.8 | ||
| 1992 | 6 March | 1.9 | |
| 23 June | 1.6 | ||
| 9 November | 1.9 | ||
| 1993 | 12 February | 1.9 | |
| 26 March | 2.9 | ||
| 7 June | 1.9 | ||
| 9 July | 3 | ||
| 29 September | 4.5 | [from 2/8] 50 percent $US - 50 percent DM | |
| 1994 | 3 January | 1 | |
| 16 February | 2.6 | 70 percent DM - 30 percent $US | |
| 16 June | 1.5 | [from May] 70 percent ECU-30 percent $US |
Hungary: Schedule of Forint Devaluation Rates, 1989-94
| Year | Date (day/month) | Size of Devaluation | Reference Basket |
|---|---|---|---|
| 1989 | 21 March | 5 | weighted average of several currencies |
| 14 April | 6 | ||
| 18 July | .5 | ||
| 25 July | .1 | ||
| 29 July | .1 | ||
| 1 August | .07 | ||
| 8 August | .53 | ||
| 15 August | .2 | ||
| 15 December | 10 | 50 percent $US - 50 percent ECU [from 8/12] | |
| 1990 | 31 January | 1 | |
| 6 February | 2 | ||
| 20 February | 2 | ||
| 1991 | 7 January | 15 | |
| 8 November | 5.8 | ||
| 1992 | 6 March | 1.9 | |
| 23 June | 1.6 | ||
| 9 November | 1.9 | ||
| 1993 | 12 February | 1.9 | |
| 26 March | 2.9 | ||
| 7 June | 1.9 | ||
| 9 July | 3 | ||
| 29 September | 4.5 | [from 2/8] 50 percent $US - 50 percent DM | |
| 1994 | 3 January | 1 | |
| 16 February | 2.6 | 70 percent DM - 30 percent $US | |
| 16 June | 1.5 | [from May] 70 percent ECU-30 percent $US |
References
Ábel, I., J.P. Bonin, and P. L. Siklos, “Crippled Monetary Policies in Transforming Economies: Why Central Bank Independence Does Not Restore Control,” in P.L. Siklos (ed.), Varieties of Monetary Reforms: Lessons and Experiences on the Road to Monetary Union (Boston, Mass: Kluwer Academic Publishers, 1994), pp. 367–82.
Ábel, I., J.P. Bonin, and P. L. Siklos, and I. Szekely, “Monetary Policy and Separated Monetary Circuits in a Modified CPE (The Case of Hungary),” Acta Oeconomica, Vol. 44 (3-4), pp. 392–428.
Ábel, I., J.P. Bonin, and P. L. Siklos, and I. Szekely, and P.L. Siklos, “Constraints on Enterprise Liquidity and its Impact on the Monetary Sector in Hungary,” Comparative Economic Studies, 36 (Spring 1994) pp. 13–22.
Argy, V., International Macroeconomics (London: Routledge, 1994).
Banerjee, A., J. Dolado, J. W. Galbraith, and D.F. Hendry, Co-integration, Error-Correction and the Econometric Analysis of Non-Stationary Data (Oxford: Oxford University Press, 1993).
Bank for International Settlements, 65th Annual Report (Basle, 1995).
Bank for International Settlements, Economic Indicators for Eastern Europe (June 21, 1995a).
Bruno, M., “Stabilization and Reform in Eastern Europe: Preliminary Evaluation,” IMF Working Paper 92/30 (May 1992).
Bruno, M., and W. Easterly, “Inflation Crises and Long-Run Growth,” Working Paper, Washington: The World Bank, 1995.
Calvo, G., “The Perils of Sterilization,” IMF Staff Papers, Vol. 38 (December 1991), pp. 921–26.
Calvo, G., and C. Végh, “Inflation Stabilization and Nominal Anchors,” Contemporary Policy Issues, Vol. 12, (1994) pp. 35–45.
Calvo, G., C. Végh, and E. Mendoza, “Reflections on Mexico’s Balance of Payments Crisis: A Chronicle of a Death Foretold,” (unpublished manuscript, Centre for International Economics, University of Maryland, 1995).
Calvo, G., and E. Mendoza, R. Sahay, and C. Végh, “Capital Flows in Central and Eastern Europe: Evidence and Policy Options,” IMF Working Paper 95/57 (May 1995).
Claessens, M.P., M. Dooley and Warner, “Portfolio Capital Flows: Hot or Cold?,” The World Bank Economic Review, (January 1995), pp. 153–74.
Csillik, P., “Az 1994. évi monetáris politika tapasztalatai” [The Experience of Monetary Policy in 1994], Bankszemle (9/10) (1994) pp. 9–18.
Cumby, R.E., and M. Obstfeld, “Capital Mobility and the Scope for Sterilization: Mexico in the 1970s,” in P.A. Armella, R. Dornbusch, and M. Obstfeld (eds.) Financial Policies and the World Capital Markets: The Problem of Latin American Countries (Chicago: The University of Chicago Press, 1983).
Engle, R.F., and C.W.J. Granger, “Co-Integration and Error-Correction: Representation, Estimation and Testing,” Econometrica 55 (March 1987) pp. 251–76.
Frankel, J.A., “Sterilization of Money Flows: Difficult (Calvo) or Easy (Reisen)?”, IMF Working Paper 94/159 (December 1994).
Frankel, J.A., “Comment,” in P.A. Armella, R. Dornbusch, and M. Obstfeld (eds.), Financial Policies and the World Capital Market: The Problem of Latin American Countries (Chicago: University of Chicago Press 1983), pp. 269–276.
Hallwood, C.P., and R. MacDonald, International Money and Finance, Second Edition (Oxford: Blackwell, 1994).
Hendry, D.F., Dynamic Econometrics (Oxford: Oxford University Press, 1995).
Hochreiter. E., “Central Banking in Economies in Transition,” in R.J. Sweeney, and C. Wihlborg (Eds.), Establishing Monetary Stability in Emerging Market Economies (Boulder: The Westview Press, 1995), pp. 127–44.
Hochreiter. E., “Necessary Conditions for a Successful Pursuit of a Hard Currency Strategy in an Economy in Transition,” in T.D. Willett, R.J. Seveenly and C. Wihlborg (eds)., Trade and Currency Policies for Emergin Market Economies, (Boulder, Col.: The Westview Press, forthcoming 1996).
Hrnčíř, M., and J. Klacek, “Stabilization Policies and Currency Convertibility in Czechoslovakia,” European Economy, Special Edition No. 2, (1991) pp. 17–40.
Kemme, D., The Reform of the System of Money, Banking and Credit in Central Europe, unpublished manuscript, (1992).
Lane, T.D., “Inflation Stabilization and Economic Transformation in Poland: The First Year,” Carnegie-Rochester Conference Series on Public Policy, Vol. 36, (July 1992), pp. 105–55.
Mosolygó, Zs., “Késés a monetaris politikaban” [Delays in Monetary Policy], Bankszemle, (1994) pp. 28–34.
National Bank of Hungary, Accounts for the Year 1994 (Budapest, 1995).
National Bank of Hungary, (1995a), “Excerpts from the 1994 Monetary and Credit Policy Guidelines,” Budapest.
National Bank of Hungary, (1995b), Az 1995. Évi Monetáris Politikai Irányelvek [The Guidelines for Monetary Policy in 1995], Budapest.
Newey, W., and K. West, “A Simple Positive-Definite, Heteroscedasticity and Autocorrelation Consistent Covariance Matrix,” Econometrica, Vol. 55, (1987) pp. 703–708.
Obláth, G., “Exchange Rate Policy and Exchange Rate Regimes: A Comparison of Experiences in Hungary and Some of the Other Economies in Transition,” presented at the ACE Conference on the Macroeconomics of Recovery in East-Central Europe, Budapest, June 1995.
Obstfeld, M. “Exchange Rates, Inflation, and the Sterilization Problem,” European Economic Review, Vol. 21, (1983) pp. 161–189.
Rebelo, S., and C.A. Végh, “Real Effects of Exchange Rate Based Stabilizations: An Analysis of Competing Theories,” in S. Fischer and O. Blanchard (eds.), Macroecopomics Annual 1995 (Cambridge, Mass: The MIT Press, 1995), pp. 125–73.
Rivera-Batiz, F.L., and L.A. Rivera-Batiz, International Finance and Open Economy Macroeconomics, Second Edition (New York: MacMillan, 1994).
Sachs, J., Poland’s Jump to the Market Economy (Cambridge: The MIT Press, 1993).
Sahay, R., and C. Végh, “Inflation and Stabilization in Transition Economies: A Comparison with Market Economies,” IMF Working Paper 95/8 (January 1995).
Sándor, Gy., “Financial Liberalization in Central and Eastern Europe and its Impact on the Exchange Rate,” in J.P. Bonin and I.P. Szekely (eds.), The Development and Reform of Financial Systems in Central and Eastern Europe (London: Edward Elgar, 1994), pp. 161-178.
Siklos, P.L., and I. Ábel, “Fiscal and Monetary Policies in the Transition: Searching for the Credit Crunch,” in T.D. Willett, R.C.K. Burdekin, R.J. Sweeney, and C. Wihlborg (eds.), Establishing Monetary Stability in Emerging Market Economies (Boulder: The Westview Press, (1995)), pp. 237–68.
Siklos, P.L., and I. Ábel, and I. Ábel, “Monetary Policy Strategies in Transition: An Evaluation of the Hungarian and Czech Experiences,” in P. Gáspár (eds.), The Macroeconomics of Recovery in East-Central Europe (Budapest: MTA, (1995a)).
Siklos, P.L., and I. Ábel, “Real Exchange Targeting in a Transitional Economy and the Sterilization Problem: The Hungarian Experience,” in T.D. Willett, R.J. Sweeney, C. Wihlborg, (Eds)., Trade and Currency Policies for Emerging Market Economies (Boulder, Col.: The Westview Press, forthcoming 1996).
Várhegyi, É., Bankok Versenyben [Banks in Competition], (Budapest: Pénzügykutató Rt., 1995).
The author is a Professor in the Department of Economics, Wilfrid Laurier University, Waterloo, Ontario, Canada N2L 3C5. He is grateful to the International Monetary Fund, where portions of this paper were prepared during a stay as Visiting Scholar in August 1995. He is also grateful to Peter Dittus, Edi Hochreiter and Werner Riecke of the National Bank of Hungary for useful conversations about the issues covered in the paper. An earlier version of this paper was invited for presentation at the Association for Comparative Economic Studies Conference, San Francisco, January 5–7, 1996. Comments by Ronald McKinnon on a previous draft are appreciated. All remaining errors are the author’s responsibility.
See Table 18.1 in Hallwood and MacDonald (1994), for a convenient summary of most transitional economies, as well as Calvo, Sahay and Végh (1995), and Sahay and Végh (1995). A word of caution is in order here: according to Classens, Dooley and Warner (1995), “labels” such as long-term and short-term capital flows (or FDI for that matter) do not provide much information about the time series properties of these flows in a sample of industrialized and emerging market economies--i.e., the volatility, persistence and predictability of the flows.
For example, in 1992 the flow of FDI (figures are in millions of U.S. dollars) into Hungary was $1,337, for the Czech Republic $1,210, and $830 for Poland. In the previous year, Hungary had attracted $1,338 in FDI, while the Czech Republic received only $200, and Poland $470 (See Siklos and Ábel (1995).
Calvo, Sahay and Végh (1995) tentatively answer this question in the negative but even they recognize the inherent difficulties, from a policy perspective, in addressing a seemingly straightforward question. Frankel (1994) also discusses this question and argues, as have others, that the decision whether to sterilize ultimately depends on the nature of the shocks affecting the economy.
One should not overemphasize the clarity of this stated policy goal. For example, NBH (1995, p. 5) states: “The main goals of monetary policy in 1994--a current account deficit substantially lower than in the preceding year.…” while not inconsistent with a policy of real exchange rate targeting need also not be congruent with such an objective. Obláth (1995) also suggests that the NBH’s exchange rate policy consisted at times of mixed signals.
Until December 8, 1991, the exchange rate was based on a composition of currencies based on trading patterns in the preceding year. Between December 9, 1991 and August 1, 1993, the basket was made up 50 percent each of ECUs and U.S. dollars. From August 2, 1993 to May 15, 1994, the basket was changed to 50 percent each of U.S. dollars and deutsche marks. Since May 16, 1994, the basket was redefined as 70 percent ECUs and 30 percent U.S. dollars. A comparison with IMF, BIS (1995a), and see Hochreiter (1995, 1996) data, which evaluates real exchange rates using a basket of currencies for 21 industrial countries, reveals patterns almost identical as the ones shown in Figure 1.
Among the various reasons cited by Hochreiter (1995) for economies in transition in general and by Sándor (1994) for Hungary, in particular, are the following: the CPI includes services, usually nontradable goods whose prices behave rather differently from those for tradeable goods and the level and scope of price liberalization which affects the CPI and PPI measures differently. Thus, CPI may reflect more the impact of exchange rate movements on aggregate demand and inflation rather than the true state of competitiveness.
Other versions of the real exchange rate (e.g., IMF, BIS (see Hochreiter 1994)) reveal similar patterns.
For example, the March 1995 issue of the NBH’s Monthly Report (page 18) uses the so-called domestic sales price index. Previous issues would also report and describe CPI and the PPI real exchange rate behavior.
NBH (1995) figures reveal that net debt as a percent of GDP was as follows (with years in parenthesis): 48.5 percent (1990), 46.6 percent (1991), 35.8 percent (1992), 39.0 percent (1993), and 45.9 percent (1994). Meanwhile, however, the import coverage ratio has improved substantially. The data are as follows: 2.3 (1990), 5.3 (1991), 5.2 (1992), 7.1 (1993), and 7.2 (1994). The latter indicator is expressed in months.
Starting from an average monthly surplus of US$27 million dollars in 1992, the current account balance went into deficit in 1993 (average monthly balance of -US$288.42 until Sept. 1994). Data are from NBH Monthly Report (various issues).
The monetary base excludes so-called “voluntary deposits held with the NBH” for which precise data were unavariable. However, as the nominal interest rate on such “excess” reserves is only 5 percent (1 percent for insurance companies) it is unlikely that such deposits play a significant role in the overall monetary base.
The ratio calculation for 1993–94 omits an outlier for the month of June 1993.
For fuller details see among others, Mosolygó (1994), Csillik (1994), and NBH (1995b).
Sahay and Végh (1995) develop a very appealing model, based on the cash-in-advance methodology, to show that inflation following the end of central planning in EIT is a function of the choice of nominal anchors following price liberalization. In particular, they point out the importance of wage policies in such countries relative to market economies, again a peculiar legacy of the era of central planning.
An important and unresolved question in this respect is the appropriate level of inflation at which steady economic growth will occur. See “World Bank Discussion on Second-Generation Transition Issues”, Transitions 6 (May-June 1995), pp. 1–6, and Bruno and Easterly (1995).
Industrialized economies are increasingly dropping the use of reserve requirements. Germany and the United States are notable exceptions, although this instrument has become a relatively less important tool of monetary policy over time.
Ábel, Bonin and Siklos (1994) explain how, in a financial system with weak banks and a narrow set of financial instruments, government bonds which pay positive real ex post interest rates were especially attractive to the public.
Part of this increase is also attributable to the more aggressive attempt by the NBH to raise foreign exchange via foreign bond issues. Thus, in 1993, the NBH raised US$600 million, 170 billion Japanese yen, 2,600 million deutsche mark, 1,000 million Austrian shillings, 1,000 million French francs, and 300 million Swiss francs. In 1994, the NBH borrowed 10 billion Spanish pesetas, DM 1,550 million, 165 billion yen, 3,000 million Luxembourg francs, 2,550 million Austrian shillings, $250 million, and 150 million Dutch (Netherlands) guilders. The simple mean interest rate at issue for the 1993–94 period was 7.657 percent with over half of the interest rates at issue in the 8 to 10 percent range (see NBH (1995c), p. 237). In contrast, commercial enterprises have been relatively more successful in international capital markets, a development bemoaned by the NBH, for reasons to be pointed out below.
Given inflation rates of 20–30 percent and nominal interest rates of 25–35 percent during the period in question, interest rates of 2–8 percent on HUF reserves and 11–18 percent on foreign exchange rate reserves clearly implied a substantial opportunity cost to the banking sector. For the complete details about the development of reserve requirements in 1994 and early 1995, see NBH (1995c, p. 98).
These repos proved especially popular in September 1993 when, for “technical reasons”, repo rates were lower than the return on government Treasury bills. This, of course, produced an excess demand (relative to the daily limits) for use of the repo facility. See Mosolygó (1994).
Moreover, savings rates in Hungary have been declining so that this does not appear to be a potential remedy (via the reduction in interest rates it would create) for Hungary, at least in the short-run.
This has indeed been a problem for Hungary with important consequences for the rate at which the financial system has matured.
Therefore, unlike the Mexican case (Calvo and Mendoza 1995), the NBH’s resort to sterilized intervention was a demonstration of a contractionary monetary policy. It is unlikely, however, the events in Mexico served as a catalyst in Hungary’s case because, as was mentioned earlier, changes in the key variables under study took place several months before the Mexican crisis.
This section draws on Siklos and Ábel (1996).
Argy (1994, pp. 398–400), Rivera-Batiz and Rivera-Batiz (1994, pp. 411–14) briefly survey the literature.
Equation (1) can also be related to a structural model of the type developed by, say, Obstfeld (1983).
Fortunately, the data in NBH Monthly Reports makes it easy to construct the series since it provides data adjusted for changes in reserve requirements.
This series takes on a positive value (i.e., the size of the devaluation in percent) in the month when a devaluation takes place and zero otherwise. In months when more than one devaluation took place the devaluations for that month were simply added together. A dummy variable, which captures political instability in Hungary (see Siklos and Ábel 1995), was also considered. This variable proved to be statistically insignificant and was dropped from the final specification.
See Siklos and Abel (1996) for the details.
Frenkel (1983) points out that OLS estimates of (3) may be biased but that the direction of the bias in unknown.
These series were also found to have a unit root and also enter the specification in first differences. Both Calvo and Mendoza (1995) and Sahay and Vegh (1995) emphasize the importance of interest rate differentials as a determinant of capital flows. A structural model would be ideal under the circumstance but the length of the sample justifies resort to single equation methods. Obstfeld (1983) uses nonlinear least squares (with a correction for serial correlation) although he also points out that a more “thorough” analysis requires the application of two-stage least squares (op. it., p. 173, No. 17).
Siklos and Ábel (1996) explore this question in greater detail.
Output is proxied by industrial production instead of, say, income from employment because the data are unavailable at the monthly frequency. Others (e.g., Lane (1992)) have had to resort to a similar, if imperfect proxy for y.
For additional details on the relationship between cointegration and error correction, see Banerjee, Dolado, Gabraith and Hendry (1993), or Hendry (1995).
Siklos and Ábel (1996) describe both the estimation technique and present detailed results for Hungary.
At a slightly more generous significance level there is evidence of two cointegrating vectors but subsequent testing (results not shown) rejects the presence of a second vector.
Note that MP and FA are measured in billions of HUF.
We did experiment with the list of instruments such as trying longer lags but none of these attempts altered our conclusions.
The EC term was included in both the Ordinary Least Squares and TSLS versions of the model with no impact on the conclusions.
Obstfeld (1983) and Cumby and Obstfeld (1983) also adjust for first order serial correlation in their applications to German and Mexican data. In the present study, we also experimented with versions of (3) with an AR(1) correction factor estimated via maximum likelihood but found little impact on the coefficient estimates for sterilization. In any event, not much support was found for serial correlation in the residuals. Finally, estimates were also produced using the Newey-West (1987) procedure with no impact on the conclusions.
Given the date of the election (May 1994) and the size of the model a structural test for a “break” in the relationship could not be carried out.
Estimates of (3) with the addition of an electoral dummy variable were also generated but proved to be statistically insignificant. See Siklos and Abel (1995) for the view that the events of the second half of 1994 were a reflection of the growing independence of the NBH vis-à-vis the government.
The statistical significance of the DEC variable may also be capturing fiscal influences as this is the year end when the deficit seems to increase substantially (as well as seasonality of monetary policy because of the Christmas season).
The sum of the coefficients is statistically insignificant as is a likelihood ratio test for the exclusion of FA and its lagged value from the regression. However, examination of the recursive coefficients on lagged changes in international reserves suggest that they had an increasingly smaller (negative) impact on money demand over time and there is a break in May 1994.
