Monetary policy rules for Russia

Journal of Comparative Economics 33 (2005) 484–499 www.elsevier.com/locate/jce

Monetary policy rules for Russia ✩ Akram Esanov a , Christian Merkl b , Lúcio Vinhas de Souza c,∗ a CREES, University of Toronto b Kiel University and Kiel Institute for World Economics (IfW) c European Commission and Kiel Institute for World Economics (IfW)

Received 17 March 2005; revised 27 May 2005 Available online 6 July 2005

Esanov, Akram, Merkl, Christian, and Vinhas de Souza, Lúcio—Monetary policy rules for Russia The paper reviews the recent conduct of monetary policy and the central bank’s rule-based behavior in Russia. Using different policy rules, we test whether the Bank of Russia reacts to changes in inflation, the output gap and the exchange rate in a consistent and predictable manner. Our results indicate that, during the period from 1993 to 2004, the Bank of Russia used monetary aggregates as the main policy instrument. Some estimations provide evidence that the Bank of Russia was more concerned with reducing inflation before 1995, while the priorities shifted towards exchange rate stabilization after 1995. Journal of Comparative Economics 33 (3) (2005) 484–499. CREES, University of Toronto; Kiel University and Kiel Institute for World Economics (IfW); European Commission and Kiel Institute for World Economics (IfW).  2005 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved. JEL classification: E52; E61; F33; F41 Keywords: Monetary policy rules; Exchange rate; Central bank; Russia

✩ The research of Akram Esanov and Lúcio Vinhas de Souza was financed by the USAID/IRIS Project No. 220/001.0-03-337 “Analysis of Monetary and Trade Policy Questions for the Russian Federation,” of which Lúcio Vinhas de Souza was the manager. * Corresponding author at: European Commission, Unit ECFIN. D.3, 1, Avenue de Beaulieu, B-1160 Brussels, Belgium. E-mail address: [email protected] (L. Vinhas de Souza).

0147-5967/$ – see front matter  2005 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jce.2005.05.003

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

485

1. Introduction Motivated by the seminal papers of McCallum (1988) and Taylor (1993), much research on the evaluation of monetary policy rules has been undertaken in the last ten years. Many studies investigate the behavior of the monetary authorities in developed economies, using either a simple McCallum–Taylor rule or some straightforward variation, e.g., including lags of the short-term interest rate or output deviations. For developed countries, such rules explain the behavior of central banks rather well and stabilize deviations, either from a target level of inflation or in the output gap, using an interest-rate instrument. Such simple rules have other advantages; they are relatively easy to implement and they are easily understood by economic agents. One advantage of rules is that they do not require a fully specified underlying model of the economy.1 For example, the estimation of the McCallum rule does not entail the modeling of the aggregate demand or the aggregate supply of money, as McCallum (2000) demonstrates. However, for developing and emerging-market countries, the ability of such rules to describe policy behavior is not clear-cut. Given the specific features of emerging-market economies, the adequate policy instrument might not be the short-term interest rate or the monetary base but rather the exchange rate.2 Including the exchange rate in the central bank’s reaction function does not contradict its objectives, if exchange rate stabilization is a precondition for both output stabilization and reducing inflation to a targeted level, as Taylor (2000) discusses. Studies investigating monetary policy rules in emerging-market economies find that central banks follow some rule-based monetary policy and that an open-economy version of the Taylor rule describes much of the variation in short-term interest rates (Calderon and Schmidt-Hebbel, 2003; Minella et al., 2003; Mohanty and Klau, 2003; Taylor, 2001 and Torres Garcia, 2003). Nonetheless, in transition economies, financial markets are even less developed so that the implementation of rule-based monetary policy may face institutional difficulties. Because of greater model specification problems and difficulties in collecting reliable data, little research has been done on monetary policy rules in these economies. This study examines the conduct of monetary policy in Russia from 1993 to 2004. The empirical estimation of alternative rules for monetary policy allows us to test the hypothesis that, in financially less developed economies, monetary targeting rules can provide an effective description of the behavior of the monetary authorities and of their stated objectives in Russia. The rest of this paper is organized as follows. Section 2 describes the evolution of monetary policy instruments and the monetary regime followed by the Russian central bank in chronological order. Section 3 specifies the different empirical models used in evaluating monetary policy rules, while Section 4 presents the results of our empirical estimations. Finally, Section 5 draws some policy implications from the Russian case.

1 However, welfare-based comparisons of the optimality of alternative rules are not possible. 2 Detken and Gaspar (2003) show that a monetary authority for which price deviations matter will also pay

attention to exchange rate developments, even if it does not targeting the latter formally. Therefore, exchange-rate targeting may be observationally equivalent to inflation targeting.

486

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

2. The development of monetary policy in Russia The dissolution of the Soviet Union at the end of 1991 did not immediately lead to the establishment of a wholly Russian monetary authority capable of conducting an independent and effective monetary policy.3 Until mid-1993, some of the former republics of the Soviet Union used the ruble, the Russian national currency, and the central banks of those republics conducted their own policy simultaneously with the Bank of Russia. Only after 1993 did the Bank of Russia start to conduct its own independent monetary policy, although its scope was limited by the policy choice to finance a large budget deficit caused mainly by a dramatic decline in output. A loose monetary stance continued until mid-1995, when the Russian economy exhibited signs of stabilization. A new banking was passed in that year providing some degree of legal independence to the Bank of Russia in conducting monetary policy.4 These positive developments allowed the central bank to adopt a tighter monetary policy and to introduce a pegged exchange rate regime with a crawling band against the US dollar, from July 1995 onwards, to replace the previous dirty float. As a result of these measures, inflation was reduced. Furthermore, because of favorable developments in the local securities market, direct credit to the government decreased significantly. The Bank of Russia began to conduct monetary policy through indirect instruments, e.g., interest rates targets and reserve requirements. However, the Asian financial crisis of 1997 caused a negative shock throughout emerging-market economies. This external shock decreased investment confidence in Russia, resulting in capital outflows that forced the Bank of Russia to defend the band. During exchange market interventions in November 1997, the Bank of Russia lost over $6 billion of its liquid reserves, which was equal to two thirds of the total reserves at that time. However, the exchange band was defended successfully. After renewed attacks prior to August 1998, the government was forced to default on its domestic debt obligations. The ruble was devalued and the exchange rate band was abandoned, leading to the adoption of a dirty float regime. After 1998, de facto targeting of the nominal exchange rate seems to have been the adopted policy.5 The sharp depreciation of the ruble led to rapid acceleration in inflation. Although ruble-denominated debt was re3 The Central Bank of the Russian Federation, hereafter the Bank of Russia, was founded on July 13, 1990 based on the Russian Republic Bank of the State Bank of the Soviet Union. On December 2, 1990, the Supreme Soviet of the RSFSR passed the law “On the Central Bank of the RSFSR (Bank of Russia),” which made the Bank of Russia a legal entity and declared it to be the main bank of the Russian Federation. 4 Nevertheless, the Bank of Russia retains some functions that are not associated traditionally with a central bank. For example, in spite of being a banking supervisor and regulator, the Bank of Russia has a majority stake in Russia’s largest bank, Sberbank Rossii. With 21,000 branches, Sberbank holds 23 percent of the all banking assets, 70 percent of the household deposits, and 20 percent of the corporate deposits in Russia. Until late 2002, the Bank of Russia also held an ownership stake in the second largest state owned bank, the VTB. Moreover, acting as an agent for the Ministry of Finance, the Bank of Russia set up and manages the government securities market, i.e., the GKO market. 5 Contrary to the conclusion in Dabrowski et al. (2002), the choice of a more flexible exchange rate regime might have been welfare improving for Russia due to the shock-absorbing properties of such regimes, conditional on the quality of institutions and on the consistency of the policy mix, as Vinhas de Souza and Ledrut (2003) claim. Given the higher propensity of commodity-based economies to be buffeted by external shocks, which are increased in harder exchange rate regimes, the flexible regime may be more appropriate for Russia.

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

487

Table 1 Stated objectives and targets of the Bank of Russia 1999

2000

2001

2002

2003

2004

M2 aggregate growth rates as an intermediate target: 18–26% growth. Reduction of the inflation rate to 30%.

M2 aggregate growth rates as an intermediate target: 21–25% growth. Reduction of the inflation rate to 18%.

M2 aggregate growth rates as an intermediate target: 27–34% growth. Reduction of the inflation rate to 12–14% a year.

GDP growth: −1% to −3% GDP fall. Exchange rate: in 1999 the exchange rate was not a formal monetary policy target.

GDP growth: 1.5%.

GDP growth: 4–5%.

M2 aggregate growth rates as an intermediate target: 22–28% growth. Reduction of inflation to 12–14% a year range (“core inflation” concept introduced). GDP growth: 3.5–4.5%.

M2 aggregate growth rates as an intermediate target: 20–26% growth. Reduction of inflation to 10–12% (core inflation should be kept within the 8.0–8.5% range). GDP growth: 3.5–4.5%.

M2 aggregate growth rates as an intermediate target: 19–25% growth. Reduction of inflation to 8–10% (or 7–8% core inflation), to 6.5–8.5% in 2005 and to 5.5–7.5% in 2006. GDP growth: 3.5–4.5%.

Exchange rate: in 2000 the exchange rate was not a formal monetary policy target.

Exchange rate: in 2001 the exchange rate was not a formal monetary policy target.

Exchange rate: nominal exchange rate targeting?

Exchange rate: “The Bank of Russia believes that the ruble’s REER may safely rise by 4 to 6% in 2003.”

Exchange rate: “the REER of the ruble may rise by 3–5%. The Bank of Russia will try to stop it from rising by more than 7%.”

Source: Bank of Russia (various years).

structured, investor confidence continued to decline due to increased political uncertainty. In the presence of private capital outflows, the Bank of Russia attempted to preserve the financial system and fulfill its obligation of lender of last resort by injecting liquidity into the banking system through a reduction in reserve requirements and the extension of considerable new credits. However, base money declined significantly in real terms, reflecting the sharp decline in output and the increased use of non-monetary forms of payment. As a consequence of the renewed inflationary pressures in 1999, one of the main stated objectives of the Bank of Russia was to reduce inflation, initially to 30 percent, while keeping the output decline in the range of 1 to 3 percent. Table 1 presents the stated objectives and instruments of the Bank of Russia in the post-1998 period. To achieve its objective, the central bank tightened monetary policy by reducing net credit to the banking system. Inflation fell sharply and the exchange rate stabilized. Furthermore, fiscal performance improved significantly due to the approval of a new package of fiscal measures and improvements in revenue collection. In addition world energy prices increased, resulting in trade surpluses, renewed capital inflows and a resumption of growth in Russia for which over 50% of its exports are energy-related products. As a result, the real exchange rate became one of the main targets of monetary policy. The continuing strength of the balance of payments and the Bank of Russia’s reluctance to permit real appreciation of the ruble put increasing pressure on monetary policy. The Bank of Russia placed more weight on exchange rate stability by apparently setting

488

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

ranges for the real exchange rate from 2003 and seemingly accepting any inflationary consequences. This policy slowed the real appreciation of the ruble and reduced inflation, even though the pace of disinflation was slower than the formal target set by the authorities.6 Starting in late 2002, the interest rate policy of the Bank of Russia changed. The profitability of deposit operations fell considerably, while the time structure shifted from short towards longer-term deposits. The Bank of Russia began to hold auctions for one-month and three-month deposits replacing a previous stand by facility for the same maturities. A unified fixed interest rate of 3 percent was set for all deposits with a maturity of up to one week and longer-term deposits were terminated. Moreover, the set of available monetary policy tools was extended and refinancing of commercial banks became more effective.

3. The specification of the empirical model Since 1991, the Russian economy experienced both sharp fluctuations in the main macroeconomic variables and deep structural changes. Given this unstable nature of the economic environment in Russia, the task of estimating a monetary policy rule is complicated. No single policy-rule equation is likely to capture fully all the aspects of central bank behavior during this period. Therefore, we estimate different types of rules. Recent literature on monetary policy rules distinguishes two types of rules, namely interest-rate based rules and monetary-based rules. These two types are referred to as the Taylor rule and the McCallum rule, respectively.7 The key difference between them is due to the choice of the instrument in the central bank’s reaction function in response to changes in macroeconomic conditions. The Taylor rule, which uses a short-term nominal interest rate as an instrument, is used widely in monetary policy estimations because of its simplicity. However, the McCallum rule, which uses the growth rate of monetary base as an instrument, figured prominently in monetary policy formulation before the nineties.8 To address the issue of the adequacy of those rules for emerging markets, researchers use modified versions of them. A general consensus that monetary policymakers in emerging economies are more concerned about exchange rate movements than their counterparts in mature economies has emerged, e.g., Williamson (2000), due to primarily to the degree of exchange rate pass-through to prices. Hence, the exchange rate has been incorporated in the central bank’s reaction function. Moreover, Ball (1998) suggests that, in a small open economy, the central bank should use a weighted average of the nominal interest rate and the exchange rate as an instrument. However, this type of hybrid rule has not been 6 The policy relevance of such concerns with real appreciation of the exchange rate is doubtful because it is unclear if the Russian ruble is above its long run equilibrium value or if it is recovering from undershooting (IMF, 2003). 7 Razzak (2001) shows that the McCallum and Taylor rules are co-integrated as expected. 8 For example, the German central bank, the Deutsche Bundesbank, announced M3 as an intermediate target but did not use it as an instrument or operational target. Although Clarida and Gertler (1996) question the reliance of the Bundesbank on monetary aggregates, Gerberding et al. (2004) claim that, once the real time bias is taken into account, the Bundesbank was indeed a true monetary targeter.

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

489

popular among empirical researchers because of the uncertainties involved in determining the weights.9 In his seminal work, Taylor (1993) estimates the following policy rule to describe the Federal Reserve’s behavior in setting short-term interest rates: it = πt + 0.5yt + 0.5(πt − 2) + 2,

(1)

where it is the short term interest rate, πt is inflation over the four previous quarters, and yt is the percent deviation of real GDP from a linear trend. The inflation target and the equilibrium real interest rate are both set at 2 and assumed to be constant over time. The policy maker is assumed to care equally about deviations of inflation and output from target. Because it is difficult to define stable long-run target values for inflation and the real interest rate in Russia, we estimate the above relationship by running a regression without pre-specifying the other parameters of the model. Therefore, following Taylor (2001), we estimate the modified open-economy Taylor rule given by it = β0 + β1 πt + β2 yt + β3 xrt + β4 xrt−1 + β5 it−1 + ut ,

(2)

where it is the short-term interest rate, πt is a measure of inflation, yt is a measure of deviations from GDP trend growth, xrt is the growth of the real effective exchange rate, and it−1 is a term used to capture interest-rate smoothing, which is behavior observed from many monetary authorities. Finally, ut is a white noise error term. The notation t − 1 indicates the lagged values of the variables and β0 is the constant term. The expected signs of the parameters are as follows: β0 , β2 , and β5 > 0; β1 /(1 − β5 ) > 1, β3 < 0, and β4 < 0. As discussed in Section 2, the short-term interest rate has not been the most important instrument in conducting monetary policy in Russia.10 Uncertainty in measuring real expected interest rates, shallow financial markets, and large shocks to investment or net exports may make monetary aggregates a preferred instrument. These may have been important concerns in Russia, especially during the nineties. The original McCallum rule can be expressed as bt = x ∗ − νt + 0.5(x ∗ − xt−1 ) + ut ,

(3) x ∗

where bt is the rate of change of the monetary base in percent per year, is the target rate of change of nominal GDP in percent per year, νt is the rate of change of base velocity in percent per year, averaged over the previous four years, and x is the rate of change of nominal GDP in percent per year. For this rule, the target value of nominal GDP growth is calculated as the sum of the target inflation rate and the long-run average rate of growth of real GDP. We take M1 as monetary aggregate used as the monetary policy instrument in Russia. Some studies attempt to explain inflation in Russia using monetary aggregates, e.g. 9 Ball-type rules are hybrid rules found in the literature using the Monetary Conditions Index (MCI), as Freedman (1996) discusses. An MCI is an indicator of the stance of monetary policy that considers not only an output target but also the influence of the exchange rate on inflation. 10 Currently, the Bank of Russia adopts officially the money supply aggregate M2 as an intermediate anchor to policy, as Table 1 indicates.

490

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

Pesonen and Korhonen (1998) and Dabrowski et al. (2002). However, our Granger causality tests indicate that, at least in the short-run of up to seven months, Granger causality goes from prices to monetary aggregates only and not the other way around. Ball (1998) argues that interest-rate-based Taylor rules are inefficient. He stresses that monetary policy affects the economy through the exchange rate as well as through interest rate channels. Ball constructs a simple model having an open-economy IS curve, a Phillips curve, and a link between the interest and exchange rates. Rearranging terms yields the following policy rule: wit + (1 − w)xrt = αyt + β(πt + δxrt−1 ) + ut ,

(4)

where w is a weight that depends on the calibration of the model, δ is the effect of a one percent exchange rate appreciation on inflation, and both α and β depend on calibrations of the model. The calibration parameters that we use are based on Ball (1998). For a robustness test, we use different weights and check their effect on the estimated coefficients.

4. Empirical results Data for Russia must be treated cautiously; their availability is limited and phenomena such as dollarization and the barter economy may lead to a biased picture. Some authors, e.g. Falcetti et al. (2000) think that the decline in output due to the transitional recession was overestimated during the first years.11 In our empirical estimations, we use monthly data covering the period from 1993 to 2004 for reasons of data availability. The sources are the International Monetary Fund’s International Financial Statistics database, the website of the Bank of Russia, the monthly database of the Vienna Institute for International Economic Studies (WIIW), and the Russian European Centre for Economic Policy (RECEP). We use data on short-term interest rates, i.e., refinancing rates, consumer price inflation, monetary aggregates, different exchange rate measures, i.e., the dollar exchange rate, the nominal effective exchange rate, and the real effective exchange rate. We estimate he output gap using a standard Hodrick–Prescott (HP) filter. Our output data, i.e., industrial production as a proxy for GDP, are from RECEP and WIIW, deflated by the monthly consumer price inflation because no monthly GDP deflator is available. First, we estimate an open-economy version of the Taylor rule in levels for the entire sample in Table 2, column 1. With the exception of the estimated output gaps, the level series are non-stationary but a Johansen test indicates the existence of a co-integrating relationship. All the coefficients have the expected signs, although only the lagged interest rate term is significant. The long-run response of the central bank is calculated as β LR =

βinf 1 − β5

(5)

11 Åslund (2001) estimates that, although official 1995 GDP was measured at only 60.2 percent of the 1989 GDP in Russia, the actual figure should be 94 percent of the same base when illegal and under-reported activities, among other things, are taken into account. Using Åslund’s estimates, the decline in Russian GDP is minimal.

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

491

Table 2 Testing a Taylor rule for Russia Sample Intercept Inflation Output gap (ex post data)

1994:01–2004:09 0.023795 (0.019173) 0.036188 (0.022862) 0.317487 (0.399659)

1999:12–2004:09 0.008916 (0.009134) 0.294724 (0.120822)** 0.017302 (0.062893)

Output gap (“real time” data) Changes in REER Changes in REER (−1) Interest rate (−1) R2 Adjusted R 2 Durbin–Watson statistics

−0.002638 (0.002548) −0.000581 (0.001260) 0.903455 (0.044768)*** 0.95 0.94 2.40

0.000347 (0.001860) −0.000324 (0.001757) 0.721798 (0.105607)*** 0.96 0.95 2.67

1994:01–2004:07 0.026018 (0.019667) 0.030181 (0.019374)

−0.013758 (0.414115) −0.001630 (0.002480) −5.47E–05 (0.001304) 0.904126 (0.048422)*** 0.95 0.94 2.37

1999:12–2004:07 0.005347 (0.007922) 0.319497 (0.122481)**

0.137718 (0.097299) 0.000392 (0.001776) 0.000110 (0.001805) 0.715572 (0.103414)*** 0.96 0.95 2.68

Notes: (i) The symbol (−1) indicates a one-period lag. (ii) Standard errors are in parentheses. ** Significance at the 5% level. *** Idem., 1%.

where β LR is the long-run response on inflation and βinf is the estimated coefficient for year-to-year inflation, i.e., β1 . We calculate a long-run response of about 0.37 so that the Taylor principle, i.e., β LR > 1, does not hold. According to our estimations, the central bank reacts to a one percent increase of inflation with less than a one percent increase in the short-term nominal interest rate and hence, causes a decrease in the real interest rate. The behavior of the Russian monetary authorities need not remain constant over time, especially in light of the initial transitional recession, the original extreme shallowness of the Russian financial system, and the series of external shocks that buffeted the country. To test for a change in policy, we estimate the same open-economy Taylor rule for different time periods in the sample. For a shorter sample that includes only the second half of our sample corresponding roughly to the post-1998 stabilization cum growth years, we find that the Taylor principle is satisfied because the long-run response is 1.06. In addition, the coefficients β1 and β5 are significant and have the expected signs as Table 2, column 2 indicates. Nevertheless, the other coefficients remain insignificant. These unsatisfactory results might be related to a situation in which the real time output data are significantly different from the ex post data, as Orphanides (2001) discusses. Therefore, we construct a real-time series by taking the yearly output data published in the Annual Reports of the Bank of Russia. Based on these data, we construct a monthly series by interpolating and re-basing the available industrial production monthly series from the WIIW.12 However, when we rerun the previous regressions using this real-time 12 Appendix Figure 1 provides evidence on the differences between the original output gap and the real-time

series.

492

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

output gap, we observe no substantial changes as Table 2, columns 3 and 4 indicate.13 Moreover, adding a dummy for post-1998 period in the full-sample regressions results in an insignificant coefficient and does not change qualitatively or quantitatively the other results. However, adding a dummy for the post-1995 period yields a significantly positive coefficient but the other results are again not changed qualitatively or quantitatively.14 Therefore, our estimation results suggest that a simple Taylor rule with modifications does not describe well the interest-rate setting behavior of the Bank of Russia, although the latter part of the sample provides somewhat more promising evidence. Second, we consider the McCallum rule. Now the expected signs of the estimated coefficients are reversed because a decrease in the monetary aggregate indicates monetary tightening and, therefore, an increase in the interest rate. In our estimates of the original McCallum rule specified in Eq. (3), the coefficients are all statistically insignificant.15 Moreover, this regression has a major statistical disadvantage because it requires discarding a large number of observations so as to average the velocity of money over the fouryear period. Due to this drawback, we estimate a modified McCallum rule in which the interest-rate instrument from of a Taylor-type rule is replaced by changes in a real monetary aggregate.16 In addition, we include seasonal dummies for December and January because the Russian money supply shows seasonal spikes in these months.17 We also add another lag to inflation to control for autocorrelation and estimate the following equation: 
 log(M1t ) = β0 + β1 (πt ) + β2 (πt−1 ) + β3 yt + β4 (xrt ) + β5 (xrt−1 )
 + β6  log(M1t−1 ) + β7 DJAN + β8 DDEC + ut

(6)

where M1t is the monetary aggregate series and DJAN and DDEC are seasonal dummies. As the results reported in Table 3, column 1 indicate, a modified McCallum rule explains the behavior of the Bank of Russia better than simple interest-rate-based rules. The estimated coefficients have the expected signs,18 although the measure of the output gap 13 The standard literature uses the gap as a measure of excess output relative to a stable long- run trend, which is relevant in a mature economy because excess output causes concerns about future inflation. However, the Bank of Russia may respond in a significantly positive manner to output growth, i.e., by increasing money after an output increase, if it assumes that this output increase is the result of technological improvements. In other words, the interest rate should not change after a permanent increase in output but the money supply should increase to accommodate the real shock. 14 These results are not shown but they are available from the authors upon request. 15 These results are also not shown but they are available from the authors upon request. 16 McCallum (2000) estimates a similar rule with what he calls a hybrid target variable. The usage of real monetary aggregates indicates that the monetary authorities have information on the inflation rate at a higher frequency than the monthly series used in our estimations. Until fairly recently, the Russian statistical office, Goskomstat, published weekly inflation estimates. Although they are no longer published, such estimates are still internally produced so that it is reasonable to assume that the Bank of Russia has access to such information. 17 According to Dabrowski et al. (2002), this effect is attributable to technical and accounting measures. 18 In this specification, the lagged coefficients have opposite signs from the expected ones. To check for residual autocorrelation, we employ a coefficient restriction test of the equality of the relative coefficients of the lagged terms and find its existence accepted marginally. However, both the Durbin–Watson statistic and the Breusch– Godfrey test reject the existence of autocorrelation.

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

493

Table 3 Testing a McCallum rule for Russia Sample

1994:01–2004:09

1999:12–2004:09 1994:01–2004:07

1999:12–2004:07

Intercept

0.018213 (0.003476)*** −0.315564 (0.119098)*** 0.237381 (0.112822)** −0.044667 (0.083492)

0.019969 (0.004101)*** −0.258940 (0.126153)** 0.177813 (0.111622)*

0.015365 (0.003253)*** −0.325535 (0.111341)*** 0.151864 (0.102715)

Inflation Inflation (−1) Output gap (ex post data) Output gap (“real time” data) Changes in dollar exchange rate Changes in dollar exchange rate (−1) Growth rate of M1(−1) Seasonal dummy for January Seasonal dummy for December R2 Adjusted R 2 Durbin–Watson statistics

−0.020131 (0.006820)*** 0.008076 (0.003402)** 0.165383 (0.059966)*** −0.124062 (0.019219)*** 0.082157 (0.010802)*** 0.72 0.70 2.11

−0.210828 (0.129778)* −0.023341 (0.006545)*** 0.006880 (0.003603)* 0.156038 (0.053864)*** −0.125342 (0.018718)*** 0.084752 (0.011040)*** 0.72 0.70 2.13

0.014441 (0.003165)*** −0.329694 (0.120664)*** 0.180016 (0.116178) −0.036433 (0.092252)

−0.019155 (0.006882)*** 0.008627 (0.005360)* 0.151308 (0.074372)** −0.120374 (0.017646)*** 0.085327 (0.011075)*** 0.71 0.69 2.00

−0.240878 (0.127581)* −0.021411 (0.006290)*** 0.008084 (0.004647)* 0.144721 (0.066941)** −0.121388 (0.017665)*** 0.086623 (0.011187)*** 0.72 0.70 2.01

Notes: (i) The symbol (−1) indicates a one-period lag. (ii) Standard errors are in parentheses. * Significance at the 10% level. ** Idem., 5%. *** Idem., 1%.

is still statistically insignificant.19 When we re-run regressions using the interpolated real time output gap, the estimated coefficients always have the expected signs and are statistically significant, as reported in Table 3, column 2.20 In addition, when we compare the McCallum model estimated using Eq. (6) with the Taylor model estimated using by Eq. (2), with either real time or ex post output gaps, standard goodness of fit measures like the standard error of the regression and the sum squared residuals indicate that the McCallum model outperforms the Taylor one. Given the absence of explicit inflation targeting in Russia, the Bank of Russia has defined informal desired inflation ranges since 1999. However, the central bank has missed virtually all of its targets to date. Hence, we estimate a gap model as defined in Mohanty 19 When we use nominal and real GDP as an alternative to the output gap, the estimated coefficients do not change considerably. 20 Because we use an approximation to real-time data, the results do not necessarily indicate that the Bank of Russia was concerned with output stabilization but only that this may have been the case. Further evidence requires actual real-time data, which are not available to us.

494

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

and Klau (2003), using a Generalized Method of Moments (GMM) estimator.21 In addition to correcting for endogeneity, this model allows us to use an HP filter to measure of inflation instead of a targeted level. Hence, we have:
  log(M1t ) = β0 + β1 (πt − πtrendt ) + β2 (πt−1 − πtrendt−1 ) + β3 yt + β4 (xrt − xrtrendt ) + β5 (xrt−1 − xrtrendt−1 )
 + β6  log(M1t−1 ) + β7 DJAN + β8 DDEC + ut

(7)

where πtrendt is the HP filter of the inflation rate and xrtrendt is the HP filter of the change of the log exchange rate. We include seasonal dummies for December and January. The results reported in Table 3, column 3 are similar to the ones obtained from the estimation of Eq. (6). Once again, when we use the real time output gap, the estimated coefficients always have the expected signs and are statistically significant, as Table 3, column 4 indicates. Overall, our estimation results suggest that the Bank of Russia has been targeting monetary aggregates in its policy decisions. Drobyshevsky and Kozlovskaya (2002) and Vdovichenko and Voronina (2004) present a similar interpretation of Russian monetary policy. At times of high inflationary pressure, i.e., when the positive output gap calculated on the basis of the constructed real-time data is positive, the Bank of Russia responded by reducing monetary aggregates in real terms. However, at times of exchange rate appreciation, its policy response was expansionary monetary policy. Moreover, these results are not sensitive to model specification and no major statistical problems are observed. Estimation results for the open-economy Ball model are mixed and unstable, as Table 4 reports. They appear to suffer from severe and persistent autocorrelation, unless we attach a 100% weight to the real effective exchange rate (REER). In general, the lower the weight attributed to the REER, the weaker are the overall regressions’ results. However, attaching a 100% to the REER is unrealistic given the central bank’s limited foreign exchange reserves. Nevertheless, these results may indicate actual targeting of the exchange rate by the Bank of Russia during most of the sample. The Russian economy experienced different shocks during different time periods over the sample years. Hence, we investigate whether the Bank of Russia responded differently in different sub-periods. Since the money-based model performs best in the previous estimations, we adopt it and test further for changes in the different time periods as we did above for the Taylor rule. We separate the sample into the period before and after mid1995, which marks the introduction of the exchange-rate-targeting regime. As with our test for the Taylor rule, other alternative time dummies did not yield significant results. We estimate the following equation:
  log(M1t ) = β0 + β1 (πt ) + β2 (πt−1 ) + β3 (πt ∗ D1995 ) + β4 yt + β5 (xrt )
 + β6 (xrt−1 ) + β7 (xrt−1 ∗ D1995 ) + β8  log(M1t−1 ) + β9 DJAN + β10 DDEC + β11 D1995 + ut

(8)

21 A GMM estimator is a robust type of estimator that does not require information about the distribution of the disturbances vector but does requires a set of instrumental variables that are orthogonal to the residuals. The instruments that we use include lagged values of the CPI, the output gap, and the exchange rate.

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

495

Table 4 Testing a Ball rule for Russia

Intercept Output gap Output gap (−1) Output gap (−2) Inflation + 0.5 ∗ (real effective exchange rate (−1)) Inflation (−1) + 0.5 ∗ (real effective exchange rate (−2)) R2 Adjusted R 2 Durbin–Watson test statistics

Exchange rate weight = 1

Exchange rate weight = 0.5

Exchange rate weight = 0

8.440386 (4.340481) 49.57056 (29.85972)* 16.37530 (15.96622) −7.574612 (14.21208) 2.182535 (0.252367)*** −0.314541 (0.227062) 0.94 0.94 1.93

4.883806 (2.226620)** 22.58852 (15.37102) 7.119564 (8.532535) −4.229031 (7.471790) 1.111315 (0.130461)*** −0.183286 (0.117846) 0.94 0.94 1.87

1.327226 (0.326119)*** −4.393518 (1.764125)** −2.136175 (2.167044) −0.883450 (1.818182) 0.040095 (0.020022)** −0.052031 (0.019394)*** 0.25 0.22 0.19

Notes: (i) The symbol (−1) indicates a one-period lag. (ii) Standard errors are in parentheses. * Significance at the 10% level. ** Idem., 5%. *** Idem., 1%.

where D1995 is a dummy variable that is one for the period before mid-1995 and zero otherwise. The estimation results reported in Table 5, column 1 suggest that the Bank of Russia conducted different monetary policies before and after 1995. The estimated coefficients indicate that the Bank of Russia was more concerned with reducing inflation before 199522 but that its priorities shifted towards exchange rate stabilization in the subsequent period. These findings are consistent with official announcements and are robust to the use of both different monetary aggregates and different measures of the output gap, as Table 5, column 2 corroborates. The interaction of the time dummy with all the variables does not change the results either qualitatively or quantitatively, as Table 6, columns 1 and 2 indicate.

5. Conclusion In this paper, we examine the conduct of monetary policy in Russia from 1993 to 2004. We estimated three types of monetary policy rules, namely the Taylor rule, the McCallum rule, and the hybrid Ball rule. Our regression results indicate that a simple Taylor rule and its different variants in which the short-term interest rate is used as a policy instrument, do a poor job in describing the behavior of the Bank of Russia, although estimations that include only the latter part of the sample perform better in this regard. The McCallum rule, for which the policy instrument is a monetary aggregate, fits the data best. The estimated 22 Of course, average inflation was substantially higher before 1995 than afterwards.

496

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

Table 5 Testing time-varying responses with a McCallum rule for Russia Intercept Inflation Inflation (−1) Inflation ∗ dummy for the period before 1995 Output gap (“ex-post” data)

0.014193 (0.004355)** −0.127181 (0.202817) 0.104531 (0.144261) −0.701036 (0.222196)*** 0.112010 (0.088992)

Output gap (“real time” data) Changes in dollar exchange rate Changes in dollar exchange rate (−1) Changes in dollar exchange rate ∗ dummy for the period before 1995 Growth rate of M1(−1) Seasonal dummy for January Seasonal dummy for December Dummy for before 1995 R2 Adjusted R 2 Durbin–Watson statistics

−0.267017 (0.112104)** 0.031396 (0.065289) 0.233412 (0.138334)* 0.175978 (0.065062)*** −0.124772 (0.012323)*** 0.088408 (0.010245)*** 0.070995 (0.020784)*** 0.74 0.71 1.98

0.016416 (0.004623)*** −0.117178 (0.207793) 0.055210 (0.152196) −0.549886 (0.227181)**

−0.087614 (0.101224) −0.319339 (0.112457)*** 0.037564 (0.067692) 0.278140 (0.141725)* 0.183515 (0.066807)*** −0.127736 (0.012661)*** 0.087183 (0.010810)*** 0.052491 (0.018945)*** 0.73 0.70 1.88

Notes: (i) The symbol (−1) indicates a one-period lag. (ii) Standard errors are in parentheses. * Significance at the 10% level. ** Idem., 5%. *** Idem., 1%.

coefficients of this specification are significant and remain largely unchanged across different equation specifications. Hence, our results indicate that, from 1993 to 2004, the Bank of Russia used monetary aggregates as its main policy instrument. Given that the Bank of Russia has adopted officially a monetary aggregate as an intermediate anchor and that its main instrument of monetary policy is deposit auctions, this result is consistent with the literature. We find evidence of a structural break in the time series in 1995 with the introduction of an exchange rate pegging regime. Before 1995, the Bank of Russia was more concerned with inflation reduction, while afterwards the primary objective was exchange rate stabilization. Our estimations using a Ball rule in which a weighted average of the interest rate and the exchange rate is used as a policy instrument yield mixed results. Depending on the choice of the weights, the estimated coefficients change and most of them are insignificant.

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

497

Table 6 Testing time-varying responses with a McCallum rule for Russia Intercept Inflation ∗ dummy for the period before 1995 Inflation (−1) ∗ dummy for the period before 1995 Inflation ∗ dummy for the period before 1995 Output gap (“ex-post” data) ∗ dummy for the period before 1995 Output gap (“real time” data) ∗ dummy for the period before 1995 Changes in dollar exchange rate ∗ dummy for the period before 1995 Changes in dollar exchange rate (−1) ∗ dummy for the period before 1995 Changes in dollar exchange rate ∗ dummy for the period before 1995 Growth rate of M1(−1) ∗ dummy for the period before 1995 Seasonal dummy for January Seasonal dummy for December Dummy for before 1995 R2 Adjusted R 2 Durbin–Watson statistics

0.017264 (0.004102)*** −0.090130 (0.182441) 0.228461 (0.258306) −0.893811 (0.259193)*** 0.193528 (0.155605)

−0.325320 (0.095717)*** 0.086952 (0.097724) 0.270409 (0.134890)** 0.242277 (0.103778)** −0.106932 (0.011149)*** 0.087291 (0.010289)*** 0.070887 (0.030473)** 0.74 0.71 1.92

0.018036 (0.004263)*** −0.110248 (0.187075) 0.331439 (0.251539) −0.802339 (0.260830)***

0.004649 (0.148057) −0.315921 (0.098178)*** 0.105592 (0.100482) 0.244624 (0.137368)** 0.259578 (0.105238)** −0.109850 (0.011464)*** 0.088168 (0.010982)*** 0.042648 (0.022163)* 0.73 0.70 1.86

Notes: (i) The symbol (−1) indicates a one-period lag. (ii) Standard errors are in parentheses. * Significance at the 10% level. ** Idem., 5%. *** Idem., 1%.

Russia’s reliance on money targeting is in sharp contrast with the recent experience of other emerging market countries for which interest rate rules describe well the behavior of the monetary authority, e.g., Mohanty and Klau (2003), Minella et al. (2003), and Torres Garcia (2003). Of course, our results are backward looking in the sense that they reflect relationships that exist to this point in the data. As the experience of other emerging market countries indicates, the evolution of forward-looking behavior among Russian economic agents aided by the development of stronger institutions, especially the strengthening of the credibility of the Bank of Russia and the improvement of its policy instruments, along with the deepening of Russia’s financial markets may enable the Bank of Russia to pursue an interest-rate policy rule.

498

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

Appendix Fig. 1. Differences between ex post and real-time output gaps.

Acknowledgments We thank Felix Hammermann, Thomas Kick, Nienke Oomes, Franziska Schobert, Elena Rumyantseva, Rainer Schweickert and Oleg Zamulin, two anonymous referees and the participants of seminars held at the College d’Europe, the IfW, the Bank of Finland, the NES/CEFIR and the University of Lodz for useful comments. The usual disclaimers apply.

References Åslund, Anders, 2001. The myth of output collapse after communism. Working paper series No. 18. Carnegie Endowment, Washington, DC. Ball, Laurence, 1998. Policy rules for open economies. Working paper No. 6760. NBER, Cambridge. Bank of Russia. Various years. Annual Report. Bank of Russia, Moscow. Calderon, César, Schmidt-Hebbel, Klaus, 2003. Macroeconomic policies and performance in Latin America. Journal of International Money and Finance 27 (7), 895–924. Clarida, Richard, Gertler, Mark, 1996. How the Bundesbank conducts monetary policy. Working paper No. 5581. NBER, Cambridge. Dabrowski, Marek, Paczynski, Wojciech, Rawdanowicz, Lukasz, 2002. Inflation and monetary policy in Russia: Transition experience and future recommendations. Studies and Analyses, No. 241. CASE, Warsaw. Detken, Carsten, Gaspar, Vitor, 2003. Maintaining price stability under free-floating: A fearless way out of the corner? Working paper series No. 241. ECB, Frankfurt A.M. Drobyshevsky, Sergey, Kozlovskaya, Anna, 2002. Inside aspects of Russian monetary policy. Working paper No. 45R. IET, Moscow. Falcetti, Elisabetta, Raiser, Martin, Sanfey, Peter, 2000. Defying the odds: Initial conditions, reforms and growth in the first decade of transition. Working paper No. 55. EBRD, London. Freedman, Charles, 1996. The use of indicators and of the monetary conditions index in Canada. In: Bank of Canada (Ed.), The Transmission of Monetary Policy in Canada. Bank of Canada, Ottawa, pp. 67–79.

A. Esanov et al. / Journal of Comparative Economics 33 (2005) 484–499

499

Gerberding, Christina, Worms, Andreas, Seitz, Franz, 2004. How the Bundesbank really conducted monetary policy: An analysis based on real-time data. Discussion paper series No. 25/2004. Bundesbank, Frankfurt A.M. International Monetary Fund, 2003. Russian Federation: Selected issues. Country report No. 03/146. IMF, Washington, DC. McCallum, Bennett, 1988. Robustness properties of a rule for monetary policy. Carnegie–Rochester Conference Series on Public Policy 29, 173–204. McCallum, Bennett, 2000. Alternative monetary policy rules: A comparison with historical settings for the United States, the United Kingdom and Japan. Working paper No. 7725. NBER, Cambridge. Minella, Andre, Springer de Freitas, Paulo, Goldfajn, Ilan, Kfoury Muinhos, Marcelo, 2003. Inflation targeting in Brazil: Constructing credibility under exchange rate volatility. Journal of International Money and Finance 27 (7), 1015–1040. Mohanty, Madhusudan, Klau, Marc, 2003. Monetary policy rules in emerging market economies, issues and evidence. Working paper. Presented at the Kiel Workshop for Monetary Policy and Macroeconomic Stabilization in Latin America, Kiel Institute for World Economics, Kiel, 11–12 September 2003. Orphanides, Athanasios, 2001. Monetary policy rules based on real-time data. American Economic Review 91 (4), 964–985. Pesonen, Hanna, Korhonen, Iikka, 1998. The short and variable lags of Russian monetary policy. Review of Economies in Transition 4, 59–72. Razzak, Weshah, 2001. Is the Taylor rule really different from the McCallum rule? Discussion paper No. 2001/07. Reserve Bank of New Zealand, Auckland. Taylor, John B., 1993. Discretion versus policy rules in practice. Carnegie–Rochester Conference Series on Public Policy 39, 195–214. Taylor, John B., 2000. Using monetary policy rules in emerging market economies. Revised paper. Presented at the 75th anniversary conference at the Banco de Mexico, Mexico City. Taylor, John B., 2001. The role of the exchange rate in monetary policy rules. American Economic Review 91 (2), 263–267. Torres Garcia, Alberto, 2003. Monetary policy and interest rates: Evidence from Mexico. Working paper. Presented at the Kiel Workshop for Monetary Policy and Macroeconomic Stabilization in Latin America, Kiel Institute for World Economics, Kiel, 11–12 September 2003. Vdovichenko, Anna, Voronina, Victoria, 2004. Monetary policy rules and their application in Russia. Working paper series No 04/09. EERC, Moscow. Vinhas de Souza, Lúcio M., Ledrut, Elisabeth, 2003. Modelling alternative paths to EMU for the Accession Countries. In: Vinhas de Souza, Lúcio, van Aarle, Bas (Eds.), The European and the New EU Member States. Macmillan & Co., Basingstoke/London, pp. 39–53. Williamson, John, 2000. Exchange rate regimes for emerging markets economies: Reviving the intermediate option. Mimeo. Institute for International Economics, Washington, DC.