INFLATION TARGETING DURING FINANCIAL CRISIS IN VISEGRÁD GROUP COUNTRIES

Transcription

1 INFLATION TARGETING DURING FINANCIAL CRISIS IN VISEGRÁD GROUP COUNTRIES Anna DOBEŠOVÁ and David HAMPEL Abstract: This paper analyses changes of monetary policy impact on inflation during financial crisis in Visegrád group (V4) countries. Main objective of central banks function in V4 countries lies in maintaining price stability. Central bank does not have a direct impact on ultimate goals. Therefore, any monetary policy analysis and assumption of its effectiveness comes out from essential existence of a working transmission mechanism. The aim of this paper is to prove whether a disruption of transmission mechanism, when the inflation rate in future should be influenced by key interest rate, has occurred or not. Next, there is a question whether this relations disruption is a general phenomenon affecting all similar economics systems of Visegrád group countries or it is a phenomenon presented only under local terms. Causality between an inflation rate and base interest rates is tested. The analysis consists in comparing models for before-crisis- and during-crisis-data, respectively. Granger causality is tested for both sub-models and the entire dataset. We found out that causation of monetary policy instruments on a growth of price level has been changed by coming up of financial crisis. In other words, a change of structure was identified. Keywords: financial crisis, Granger causality, inflation targeting, monetary policy, VAR model 1. Introduction Primary statutory objective of central banks in Visegrád Group is to achieve and maintain price stability. This objective is in agreement with European Union legislation, specifically with the Article 127, Treaty on the Functioning of the European Union. To this end, all banks implemented the regime of targeting inflation. Within a framework of inflation targeting strategy, they strive to fulfil publicly announced inflation target through monetary policy instruments. This medium-term target is compared with a medium-term inflation rate forecast. Then central banks determine a character of their monetary policy on the basis of deviations from the target. National Bank of Poland (NBP) adopted the direct inflation target strategy in Since 4 it has followed a continuous inflation target at the level of 2.5 % with a permitted deviation +/- 1 %. Since 1 the inflation targeting has been utilized in Hungarian National Bank (MNB), inflation targets were adopted simultaneously on a yearly basis up to the end of 6. Since 7 explicit medium-term inflation target has been defined. Czech National Bank (ČNB), which implemented the inflation targeting strategy in 1998, changed target band for headline consumer price inflation during 2-5 period to horizontal target, which is currently 2 % with a permissible fluctuation band of +/- 1 percentage point. National Bank of Slovakia (NBS) has worked under inflation targeting regime since 5. The European Central Bank (ECB) took over its mission in 9 with the target less than 2 %. 12

2 A transition to inflation targeting monetary policy became in the world in the 1990s after an unsuccessful monetarist experiment period that came out from an idea of price level proportional to a money supply. Unstable velocity of money flow and instability of money multiplier pointed out to an unclear link between monetary aggregates and price inflation rate. The monetary aggregates were excluded from monetary policy decision making. Nevertheless a Friedman s argument, that inflation is always a monetary phenomenon, has stayed valid. A principle of neutrality of money is accepted generally as well as an assumption that central bank can contribute to an improvement of an economy growth potential only through maintaining a stable price environment. Therefore, a monetary policy operational criterion has become short-term interest rate in relation to inflation forecast in a mainstream economics. An essential assumption of effective monetary policy is an existence of functional transmission mechanism. There is not an exact consensus about its accurate behaviour. Transmission mechanism is a process, in which a change in settings of monetary policy instruments works via changes of intermediary markets on final changes in economy, particularly in inflation with variously long lags. There are several possible channels of transmission mechanism. Traditionally, an interest rate channel is referred, in which a size of base interest rate of central bank impacts on short-term interest rates of interbanking market. In accordance with it, commercial banks set up their deposit and credit rates. Investment behaviour and a demand for goods and services are influenced in this way. This view is stemming from an idea, that central bank determines behaviour of commercial banks and public. Contrary to that, post-keynesians believe that an aggregate demand does not fall down with an ascending price level. According to them, an amount of money in circulation is set by demand for money and is not determined by central bank exogenously. Despite the fact that central banks work with permanently more complex and sophisticated models to achieve stable prices of goods and services, the effectiveness of an inflation targeting is not completely obvious. A lot of authors concern with a rightfulness inflation targeting itself, e. g. [7]. Regardless this issue, there is indisputably clear necessity for an existence of functioning transmission mechanism. Therefore, many authors deal with a modelling of this mechanism for various world economies. There are also a lot of studies for countries in Central Europe for example [1], [9]. However, these papers work with standard techniques and do not reflect potential changes in a size of model parameters with time. There are only few studies, which deal with potential structure modification: recently Darvas [3] has investigated modifications caused by transition to the inflation targeting strategy and by entering to the European Union; alternatively Lyziak et al. [11] have modelled changes in Polish monetary transmission. Holcner and Neubauer in [8] deals with modelling defence specific inflation. The aim of the present paper is to complement the above efforts and to study more thoroughly current changes in transmission mechanism. The paper focuses on contemporary monetary policy problems in Central Europe. There is an assumption that financial and subsequent economic crisis could have an impact on force of monetary policy of V4 countries. Toward this end, we try to find out whether causal relations between changes of key interest rate and inflation rate have been transformed, in other words whether transmission mechanism has been interrupted or not. Modelling for similar Visegrád Four economies can demonstrate a generality of such changes or their local conditionality. 13

3 VAR modelling is used for finding above-said causality. This approach is a common method for analyses of monetary policy impact on economic variables. This approach is appropriate because it works with lags and central banks derive their decisions from an expected future inflation rate. Moreover, it will be able to determine the causality between monetary policy and real indicators. 2. Material and methods Our sample data include monthly observations in the period from 2:01 to 2012:08. With respect to necessity of an accurate analysis for during-crisis period it was not possible to use quarterly indicators, such as output, because available number of observations would not be sufficient. Therefore, for all countries in our paper Czech Republic, Poland, Hungary and Slovakia three indicators were included: inflation, interbank interest rate and central bank key interest rate. First variable used is P, an inflation rate as an increase in CPI compared with the corresponding month of preceding year, with all items. Data sources were national statistical offices. Next included variable is IR, 3 months interbank interest rate taken from national central banks. In Slovakia there was a transition from BRIBOR to EURIBOR on January 9. Last variable is BR, central bank base rate from an identical source. In Slovakia it was National Bank of Slovakia key rate to the end of 8 and rate of ECB since the time. We have tried to include other variables into model, namely real effective exchange rate and industrial production index. These variables, although they were significant in some lags in particular over-parameterized VAR models, do not carry causality to inflation according to tests we provide. We set the same breakpoint for all datasets as the October 8. In this time the Chow test [2] detects structural change in inflation progress for all V4 countries. Used VAR model is a special case of VARMA model, what is an analogy of Box- Jenkins ARMA model, designated for multivariate time series. Detailed description of this widely used approach is given for example in [10]. There is an important condition for using this model: at least weak stationarity of modelled time series. In this work, stationarity is reached by differencing of time series; another option is outlined for example in [13]. Such attitude should lead to description of inner relations in data; not to relating external characteristics as a common trend. On the other side, descriptive power presented by adjusted coefficient of determination can be low in many cases. When estimating parameters of particular VAR model we follow the parsimony rule, where small number of significant parameters is preferred. When we investigate possible causal relations among variables, we use Granger approach to this problem (see [6]). According this, variable X is said to be in causal relation to variable Y if lagged observations of X can improve prediction of Y t+1 based on lagged values of Y and another vector of variables, say Z. In VAR model, such dependency can be tested via Wald-type test (see for example [10]), where sub-models are compared. All calculations (VAR parameters estimation, tests of Granger causality) were carried out in software Gretl and in computational system Matlab R2014a. 3. Results Figure 1 shows a development of central banks base rates. There is a significant structural change in 8, which marks a beginning of crisis. Classic strategy within wrong status of economy, as it is apparent in Czech and Slovak (respectively euro area) curves, is to decrease interest rates that should bring more money into economy and also a domestic currency depreciation and subsequent export increase. Inflation pressures 14

4 should not be a problem in depressed status of economy. Fig. 1 however demonstrates recent increments of interest rates in Hungary and Poland. In Hungary it is a reaction to indebtedness and effort about domestic currency appreciation so that an imported production would not raise its price and would not cause the growth of inflation. Poland tries to reduce inflation, but there is also an assumption that the inflation would decrease itself as a result of a pressure of retarded economy without raise of central bank rates. Figure 1: Development of base rates of national banks Source: Eurostat (2014). Central bank interest rates dataset Table 1 summarizes results of Granger causality testing in our models. Conclusive causality is highlighted. Results point out to not so powerful monetary policy in V4. Although hypothetical period of the most efficient transmission is 9 14 months in literature, our results do not confirm this hypothesis. We can find models with significant variables in high lags, but the entire VAR model is not acceptable. In another words, we do not conclude, that transmission is a short-time-process, but acceptable VAR models (suitable for testing of Granger causality) were of relatively low orders. These models show importance of used variables, not their exact impacts on different variables. These models have an advantage of precisely visible directions and extents of influences. Table 1 shows both Granger causality for whole data, as well as for before-crisis- and during-crisis-data. If there is not conclusiveness about Granger causality from central banks base rate, it would not be necessarily caused by its nonexistence, but for example its constant level for long periods. 15

5 Table 1 Granger causality in VAR models (tabulated are p-values, H0: Granger causality doesn t exist) CZ SK Direction Whole to from Direction Whole to from BR P 0,130 0,208 0,498 BR P 0,729 0,635 0,049 IR P 0,776 0,032 0,804 IR P 0,465 0,563 0,874 P BR 0,000 0,008 0,813 P BR 0,626 0,926 0,334 IR BR 0,064 0,000 0,001 IR BR 0,000 0,000 0,047 P IR 0,000 0,003 0,357 P IR 0,016 0,107 0,313 BR IR 0,000 0,078 0,043 BR IR 0,896 0,402 0,000 Lag Lag HU PL Direction Whole to from Direction Whole to from BR P 0,136 0,015 0,141 BR P 0,716 0,651 0,729 IR P 0,238 0,057 0,072 IR P 0,715 0,086 0,774 P BR 0,390 0,165 0,020 P BR 0,026 0,010 0,219 IR BR 0,000 0,000 0,000 IR BR 0,006 0,011 0,423 P IR 0,356 0,130 0,013 P IR 0,000 0,046 0,007 BR IR 0,016 0,044 0,210 BR IR 0,005 0,101 0,017 Lag Lag Results for the Czech Republic indicate weak monetary policy impact to price stability. Interest rate managed to influence inflation only during before-crisis period. Effect of monetary policy disappeared during crisis. This result, together with low lag, corresponds to findings that Darvas [3] reported for the Czech Republic. On the contrary, in Slovakia we can see conclusive monetary policy influence only during crisis. It is necessary to refer to an aspect of Slovak entrance to EMU on January 9. This phenomenon corresponds to more powerful monetary transmission in euro area compared to any Central European country [3]. In Hungary there is a significant dependence of inflation on interest rate during beforecrisis period; however it has disappeared during crisis similarly as in the Czech Republic. By contrast no causality was discovered in Poland. Parameters of Polish model have changed during crisis, as confirmed also by Liziak et al. [11], but contrary to Darvas [3] no dependence was found. Darvas concludes that the most powerful monetary transmission in Central Europe is in Poland. 4. Discussion There are impacts of central banks instruments on inflation rate in the models, but it is not a global phenomenon. This is in accordance with other studies ([4], [5]). On the other hand, Krušec [9] recently used different methods and discovered powerful causality between the above-said variables. However, we are limited by small extent of data for crisis period. Our results lead to the conclusion that effects of monetary policy have disappeared during crisis in the Czech Republic and in Hungary. Model parameters changed in Poland; however the causality was not proved. Contrarily, monetary policy in Slovakia became powerful since 8. There could be a lot of factors that was able to cause these changes in monetary policy effectiveness during crisis. 16

6 There was a significant grow up of value added tax in the Czech Republic, in Hungary and partially in Poland, see Table 2. This taxflation caused by governmental discretionary provisions could lead to an overlap of the causality between interest rates and inflation goal. Yet, more relevant factor can be aggregate demand insensitiveness to interest rate. This phenomenon, which is cited in literature (e.g. [12]), happens during crisis, when pessimistic expectations can drown out an impact of interest rate. Monetary expansion is then not able to influence investments and consumption in economy. Table 2 Changes in value added tax (in percents) Standard Reduced Standard Reduced Czech Republic Hungary 20 5, , 18 Slovakia Poland 23 0, 3, , 5, 8 Source: TAX RATES: retrieved 7. February 2013, (available at An important aspect can be also an openness of economy. In this respect, there was a significant increase of exports shares on GDP during crisis in the Czech Republic and in Hungary, too (see Table 3.). With higher economy openness the monetary policy has a lower ability to affect export GDP components. Czech Republi c Table 3 Export of goods and services (percent of GDP), Hungary Poland Slovakia Source: Eurostat (2014). GDP and main components dataset. 5. Conclusions There is a debate as to whether monetary policy can be affected by crisis in economy. Crisis can destabilize structure of economy and monetary policy, too. Moreover, even if conditions in economy stay unchanged, expectations of economics subjects can be transformed. This occurrence itself can lead to the substantial change in monetary policy effectiveness. We used VAR modelling of V4 countries data to study this phenomenon. Our results, indeed, show different effects of monetary policy in particular V4 countries. Testing of Granger causality and comparison of subsequent results of before-crisis- and during-crisis-models led to the finding, that monetary policy impacts to inflation disappeared during crisis in the Czech Republic and in Hungary. Influence of monetary policy was not proved in our models in Poland, at all. There was an opposite effect in Slovakia, where monetary transmission has become more powerful since 8. Our results demonstrate that crisis can affect monetary policy power in a significant way. But our findings are not equal for each V4 country. One of the most important factors, which can cause this process, can be substantial grow up of taxation in Hungary 17

7 and in Czech Republic, respectively. Another aspect can be an openness of economy. Moreover, pessimistic expectations of economics subject during crisis can lead to the insensitivity to central bank actions. However, there can be much more factors. Our dataset for crisis period and our number of countries is too small for making general conclusions about determinants that inflict different monetary policy power during crisis. That is why we are going to continue our investigation. 6. Summary This paper focuses on implications of the current financial crisis within the framework monetary policy of Central European countries. The article studies changes in monetary transmission channel using vector autoregression models. Regime of targeting inflation is currently used as a monetary policy regime for achievement of the inflation target. Changes in setting of monetary policy instruments should be able to inflict causal changes on intermediary markets and via these markets on target markets. Monetary transmission depends on monetary policy, but also on a structure of economy. Financial crisis is able to affect both these phenomena. Therefore, it is not sufficient to study whole models, but it is necessary to compare the causality for before-crisis- and duringcrisis-data. Our results point out to changed impact of monetary policy on inflation before and during crisis. We found causality before crisis in the Czech Republic and in Hungary, but during crisis monetary policy became ineffective there. Contrarily, the opposite effect appeared in Slovakia where monetary transmission was more powerful since 8. This can be associated with an entrance to the European Union. However, our tests proved no conclusive influence of monetary policy instruments to inflation in Poland. There can be a variety of determinants inflicting the changes in power of monetary policy during crisis. But due to small dataset we could not bring formal analysis of them. Yet we outlined some of the most important factors for studied countries. These can be (i) substantial tax increase, (ii) changes in openness of economies and (iii) pessimistic expectations of economics subjects. In general, our findings conclude about relatively low effectiveness of inflation targeting strategy in V4, especially during crisis. Moreover, the reverse causality is confirmed: an influence of the inflation on central bank instruments. References [1] BURIAN, S., BRČÁK, J., 2012: The Basic Transmission Mechanisms of Monetary Policy in the Czech Republic. International Advances in Economic Research. 18(1), ISSN: [2] CHOW, G.C., 1960: Tests of Equality Between Sets of Coefficients in Two Linear Regressions. Econometrica. 28(3), ISSN: [3] DARVAS, Z., 2012: Monetary transmission in three Central European economies: Evidence from time-varying coefficient vector autoregressions. Empirica. 40(2), ISSN: [4] European Forecasting Network (EFN), 4. The Euro Area and the Acceding Countries. EFN Economic Outlook Spring Report 4 [online], Paris. Available from 18

8 [5] GANEV, G., MOLNAR, K., RYBINSKI, K., WOZNIAK, P., 2: Transmission of Monetary Policy in Central and Eastern Europe. Report Center for Social and Economic Research. 52, ISSN: [6] GRANGER, C.W.J., 1969: Investigating Causal Relations by Econometric Models and Cross-spectral Methods. Econometrica. 37(3), ISSN: [7] HLAVÁČ, P., MUNZI, T., 2011: Inflation Targeting and its Impact on the Nature of the Money Supply and the Financial Imbalances. Politická Ekonomie. 59(4), ISSN: [8] HOLCNER V., NEUBAUER J., 2014: Modelling Expected Defence Specific Inflation In SIMOS, T. E., 12th International Conference of Numerical Analysis and Applied Mathematics 2014: ICNAAM 2014, AIP Conference Proceedings. Melville, New York: American Institute of Physics. In print. [9] KRUŠEC, D., 2011: Is Inflation Targeting Effective? Monetary Transmission in Poland, the Czech Republic, Slovakia, and Hungary. Eastern European Economics. 49(1), ISSN: [10] LÜTKEPOHL, H., KRÄTZIG, M.: Applied time series econometrics. New York: Cambridge University Press, 9. ISBN: [11] LYZIAK, T., PRZYSTUPA, J., STANISLAWSKA, E., WRÓBEL, E., 2011: Monetary Policy Transmission Disturbances During the Financial Crisis: A Case of an Emerging Market Economy. Eastern European Economics. 49(5), ISSN: [12] MANKIW, G., ELMENDORF, D.W., SUMMERS, L.H., 9: Brooking Papers on Economic Activity: Fall 8. Washington, D.C.: The Brookings Institutions. ISSN: [13] NEUBAUER, J., 8: Kointegrační analýza modelu inflace v České republice. Forum Statisticum Slovacum. 4(6), ISSN: