Liquidity Effects of the Unconventional ECB s Monetary Policy

Transcription

1 18th International Scientific Conference Enterprise and the Competitive Environment, March 5 6, 2015, Brno, Czech Republic Liquidity Effects of the Unconventional ECB s Monetary Policy Michal Dufek 1 and Svatopluk Kapounek 2 1 Department of Finance, Faculty of Business and Economics, Mendel University in Brno, Zemědělská 1, Brno, Czech Republic, michal.dufek105@gmail.com 2 Department of Finance, Faculty of Business and Economics, Mendel University in Brno, Zemědělská 1, Brno, Czech Republic, kapounek@mendelu.cz Abstract In this paper we examine the liquidity effects of the unconventional monetary policy of the ECB. We estimated cointegration vector to identify long-term causal effects of nonstandard measures of monetary policy on the long-term interest rates of sovereign bonds in the selected core euro-area member countries. The pass-through from the official open market operations to the long-term interest rates also depends on the risk premiums and economic activity expectations. Finally, we adopt the technique of error correction to understand the dynamics between the variables studied in the short and long term. The results are discussed in relation to the monetary policy efficiency at the zero lower bound. Keywords: financial crisis, deflation, credit easing, quantitative easing, euro area 1. Introduction The monetary policy of the most word central banks is responsible to maintain of the price stability. In order to the theory of conventional transmission mechanism central banks influence short interest rates on the money market. The short end of the yield curve is fully controlled by the central bank. The inflation and the aggregate demand are consequently affected by manipulation with short term interest rates (according to the conventional transmission mechanism, Bofinger, 2006). However, the success of the monetary policy in the inflation targeting and stimulating of the aggregate demand can be limited when the short term rates hit the zero lower bound. In this situation, central banks can no longer stimulate aggregate demand by further interest rates reductions and must use another monetary policy tools, so called unconventional measures. There are a lot of the economic papers about monetary policy alternatives at zero lower bound. Before the financial crisis in the year 2007 only few studies have presented empirical testing of the potential effectiveness of the unconventional measures. For example, Bernanke and Reinhart have undertaken such research on the Japanese data (Bernanke,

2 Enterprise and the Competitive Environment, March, 5 6, Reinhart, 2004). These studies would help central banks for the choice of the appropriate tools in the zero rates policy times. Bernanke and Reinhart (2004) reached to findings about the likely efficacy of nonstandard policies. Especially, their researchconfirm a potentially important role for central bank communication to try to shape expectations of the future policy actions and about the path of the future short term interest rates. They also find evidence that asset purchase in large volume by central bank would be able to affect the price or yields of the purchased assets. Hiroshi Ugai (2006) surveys the empirical analyses that evaluate quantitative easing policy (QEP) performed by Bank of Japan which has the most experience with the unconventional monetary policy. The Japanese researchers in their papers identify three transmission channels through which unconventional zero interest rates policy and the quantitative easing influence macroeconomic magnitudes such as expectations of future path of short term interest rates, yield curves, risk premium and finally aggregate demand a price development in economy. QEP examine in their research paper Joseph Gagnon, Matthew Raskin, Julie Remache, Brian Sack. More precisely they examine the programme of Federal Reserve System called Large Scale Assets Purchase Programme which was triggered since December 2008 and was running to October The research paper presents evidence that the quantitative easing (asset purchasing) led to economically significant and long lasting decline in longer term interest rates on a wide range of securities even if some of them were not included in the purchase programme. In this paper we examine one of the alternative monetary policy the liquidity effect of the unconventional monetary policy of the ECB. We describe how these policy might work and on the data of several European countries we will test the effectiveness of this monetary policy tool. To obtain this evidence concerning the liquidity effect of the asset purchase and its impact on longer term interest rates we use data from Germany, Greece, France and Austria. The liquidity effect is considered like a significant if the Eurosystem can affect the longer term interest rates, namely yields of the sovereign bonds with the five years maturity. The European Central Bank use several nonstandard measures, through which performs its expansionary monetary policy. Like other world central banks (Bank of Japan, Federal Reserve System and others), ECB especially use asset purchase to stimulate the aggregate demand in the economy. The individual programs of the asset purchase will be described below. The objective of this paper is to identify the relationship between the implementation of the nonstandard measures of the monetary policy and longer term interest rates of sovereign bonds with five years maturity in the selected euro area member countries. The time series are selected so that we can focus on the development of this relationship before (in times when security market purchases and longer term refinancing operations were rarely used) and after the outbreak of the financial crisis. Our theoretical contribution is the analysis of the effectiveness of the nonstandard measures used by central banks during the zero rates episodes. 2. Theoretical background Firstly, we revised methodological approach by James Tobin (1958) called portfolio balance effect (Tobin, 1958). Tobin identified the primary channel through which unconventional measures appear to work is the risk premium on the purchased assets. By purchasing certain assets, central banks reduce the amount of this securities that

3 Enterprise and the Competitive Environment, March, 5 6, holds the private sector while at the same time are central banks increasing the amount of short term, risk free bank reserves held by the private sector. QEP (more precisely the purchasing programs) push the prices of the purchased assets up and hence lower its yield. In the second step we follow the contribution of Krugman (1998), Eggertson and Woodford (2003), Bernanke and Reinhart (2004), that there are some alternatives of the expansionary monetary policy on the zero lower bound of the short term interest rates. For securities is the most important component of the risk premium. The risk premium is also called term premium and reflects unwillingness of economic agents (investors) to bear the interest rate risk which is connected with holding assets that have long term of maturity (or duration). We follow model of Gagnon, Raskin, Remache and Sack (2010) and Bernanke, Reinhart, Sack s model (2004) where are included following variables to capture term premium variation to the business cycle and fundament ucertainty: unemployment gap measured as the difference between the unemployment rate and the Congressional Budget Office s estimate of the natural rate of unemployment, core CPI inflation, 6-month realized daily volatility of the on-the-run 10-year Treasury yield a proxy for interest rate uncertainty, publicly-held Treasury securities with at least one year to maturity, treasury securities held in the Federal Reserve s SOMA portfolio with at least one year to maturity, U.S. debt securities held by foreign official agencies with at least one year to maturity. 3. Currently used methods The empirical analysis identifies a long-term relationship given by the open market operations and long term refinancing operations of the ECB at the internal banking market to influence portfolio of the banks (maturity, duration or yield curves of the securities) to affect long-term yield curves represented by the 5-year government bond yields (Bonds_5Y) in formula (1):, (1) where CDS represents insurance paid by buyers against a credit event on the underlying government debt, VIX is Chicago Board Options Exchange Market Volatility Index, ES represents economic sentiment indicator (ESI) which is calculated by Directorate general for economic and financial affairs (DG ECFIN), IP is industrial production index, P is harmonized index of consumer prices, L is level of loans, LTRO represents long term refinancing operations and OMO represents open market operations. Finally ε denotes the residual term and p represents lag. The datasets are provided by public available database Eurostat, the dataset regardins implied volatility (VIX) is available at website of the Chicago Board of Exchange, the dataset of credit default swaps we obtained in the Bloomberg s database. We used monthly data in the period 2005M M08 at levels without seasonally adjustment. To identify the long-term causality and short-term adjustments we applied cointegration analysis. To estimate multiple cointegrating relations we applied Johansen s approach and its maximum likelihood approach avoiding conditional estimates (Johansen 1988 and Johansen 1991). To pre-test stationarity of the variables we applied Augmented Dickey Fuller test (ADF test) as a test for unit root in the form:

4 Enterprise and the Competitive Environment, March, 5 6, y t y p t 1 1 i y t i t (2) i 1 where α represents deterministic term (constant), p the lagged difference terms, Δyt i, are used to approximate the ARMA structure of the errors, and the value of p is set so that the error εt is serially uncorrelated. The optimal lag length of the AR-model is obtained on the basis of Akaike and Bayesian information criterion. The criterions are selected according to the parsimony optimality. The rank of the multiple cointegrating relations r is tested by trace and maximum eigenvalue tests at specific lag structure. The optimal lag order is identified by unrestricted VAR model based on the Akaike information criterion. The tested model with deterministic trend is specified in formula (3): y t AB y q B y t 1 i t 1 t, (3) i 1 whereyt is an n-dimensional time series of variable i, matrixbirepresents cointegrated relations and combination matrices A and B represents error-correction term AB yt 1, B yt 1 measures the deviation from the stationary mean at time t 1 and A is a matrix of adjustment speeds. This approach reflects that all variables are possibly endogenous. All zero-rows in matrix A indicate a variable that is weakly exogenous with respect to the coefficients in matrix B. According to Johansen s approach, such a variable may affect other variables, but does not adjust to disequilibrium in the cointegrating relation. To identify such variables we applied Johansen s constraint test on adjustment speed of the selected variables. 4. Results According to the Johansen s methodology we pre-tested all variables in the system to assess their order of integration. The results of Augmented Dickey-Fuller test at the levels and the first differences are presented in the Table 1. In the case of first differenced variables we rejected unit root processes at 0.01 significance level, except the loans in France, where we concluded stationarity at 0,1 significance level. Because the all variables are integrated of the same order, we estimated the model and determined the cointegration rank. The applied trace test and maximum eigenvalue test assesses null hypotheses of cointegration rank less than or equal to r. The Table 2 shows that there should exists 2 or less cointegrating relationships in Austria, 1 in France, 2 in Germany and 3 in Greece, at the 0.05 significance level. Subsequently, we tested if cointegrating relationship forms a stationary linear combination of the all first-order integrated variables, thus, if certain combinations of variables suggested by economic theory are stationary. The composition of the deterministic term is presented in the form. The results showed that although trace test and the maximum eigenvalue test assume multiple cointegrated relations we cannot confirm more than 1 cointegrating relationship in Austria and France. We also rejected existence of the cointegrating relationship in Greece.

5 Enterprise and the Competitive Environment, March, 5 6, Table 1: Unit root stationarity ADF test Variables Levels First differences Order of ADF stat Lags ADF stat Lags integration Bonds_5Y * *** 0 1 CDS_Spreads * *** 0 1 VIX * *** 1 1 Economic_Sentiment * *** 0 1 Industrial_Production * *** 4 1 HICP * *** 3 1 Loans * *** 0 1 Longer_Term_Ref_Operations * *** 0 1 Open_Market_Operations * *** 0 1 Bonds_5Y * *** 1 1 CDS_Spreads * *** 0 1 VIX * *** 1 1 Economic_Sentiment * *** 0 1 Industrial_Production * *** 6 1 HICP * *** 0 1 Loans * *** 2 1 Longer_Term_Ref_Operations * *** 0 1 Open_Market_Operations * *** 0 1 Bonds_5Y * *** 0 1 CDS_Spreads * *** 0 1 VIX * *** 0 1 Economic_Sentiment * *** 0 1 Industrial_Production * *** 1 1 HICP * *** 0 1 Loans * *** 1 1 Longer_Term_Ref_Operations * *** 0 1 Open_Market_Operations * *** 0 1 Bonds_5Y * *** 0 1 CDS_Spreads * *** 1 1 VIX * *** 1 1 Economic_Sentiment * *** 0 1 Industrial_Production * *** 11 1 HICP * *** 4 1 Loans * *** 0 1 Longer_Term_Ref_Operations * *** 0 1 Open_Market_Operations * *** 0 1 Notes: *, ** and *** denote significance at the 10, 5 and 1% level. Austria France Germany Greece

6 Enterprise and the Competitive Environment, March, 5 6, Table 2: Johansen rank test for cointegrated relations Austria France Trace test Maximum Maximum Trace test eigenvalue test eigenvalue test lag r = *** ** *** *** r = *** ** *** r = ** r = * r = r = r r = 1 r = 1 p-value 1) Germany Greece Trace test Maximum Maximum Trace test eigenvalue test eigenvalue test lag r = *** *** *** *** r = *** ** *** *** r = *** * *** ** r = * *** r = * r = r r = 2 r = 1 p-value 1) Notes: *, ** and *** denote significance at the 10, 5 and 1% level. 1) we test stationarity of cointegrating relation The Figure 1 shows evidence of cointegrating relations represented combination of all variables in different models. Obviously, there are relatively stationary relationship in Germany and Austria, possibly in France we can identify trend stationary process, but we cannot confirm theoretical assumption in the case of Greece. Normalized cointegrating relations with respect to the bond yields we presented in the Table 3. We can see that the coefficients are close to zero in the case of Greece as well. The cointegrating relationship among the variables in France is also complicated because we assumed intercepts and linear trends in the cointegrating relations and quadratic trends in the data. This model may produce good in-sample fits but would be very poor for out-of-sample forecast. Moreover, the quadratic trends in the all variables are questionable. For these reasons we will discuss only models for Austria and Germany in the next parts of this paper.

7 Enterprise and the Competitive Environment, March, 5 6, Figure 1: Figure caption Table 3: Cointegrating vectors estimations Variables (B) Austria France Germany Greece CDS_Spreads VIX Economic_Sentiment Industrial_Production HICP Loans Longer_Term_Ref_Operations Open_Market_Operations In the long-run relationship we identified negative effects of economic activity on the bond yields in Austria and the first cointegrating vector in Germany. These results could be explained by consequences of debt crisis and government bond sales. On the contrary, theoretical assumptions were confirmed by the second cointegrated vector in Germany. In that case we identified strong positive impact of the economic activity on the bond yields and negative impact of the prices and the economic sentiment indicator. We also suggest positive relationship between the loans and bond yields. Important result provided long-run relationship between the open market operations and bond yields because the results confirmed our suggestion that the expansionary monetary policy affected long-run interest rates with possible effects on the economic growth and investment activities. This cointegrating relation was confirmed by the all identified models (except France and Greece). The Table 4 provides the error-correction term estimations under the cointegration rank restrictions. According to these results we can discuss disequilibrium in cointegrating relations and adjustment speeds of variables to disequilibrium. In this sense we identified strong disequilibrium in long-run relationship caused by economic activity and subsequent short-term adjustment of the open market operations and CDS spreads (second cointegrating relation in Germany). The impact of economic activity on

8 Enterprise and the Competitive Environment, March, 5 6, the long-term cointegrated relations is significant also in the case of Austria and the first model in Germany. In these booth cases we can see positive adjustment of long term refinancing operations and CDS spreads. Very small estimations of the adjustment speeds of the other variables imply that these variables are not responsive to last period s equilibrium error. Table 4: VECM parameters estimations A (adjustment speeds) Austria France Germany Bonds_5Y CDS_Spreads VIX Economic_Sentiment Industrial_Production HICP Loans Longer_Term_Ref_Operations , Open_Market_Operations Austria France Germany B (cointegrating relations) Bonds_5Y CDS_Spreads VIX Economic_Sentiment Industrial_Production HICP Loans Longer_Term_Ref_Operations Open_Market_Operations Table 5: Johansen constraint test on weak exogeneity Austria France Germany Bonds_5Y CDS_Spreads VIX Economic_Sentiment Industrial_Production HICP Loans Longer_Term_Ref_Operations Open_Market_Operations Notes: p-value estimation of probability of rejecting the null of the each variable adjustment with respect to the other variables in the model To provide better insight into the dynamic adjustment process in cointegrated relations we tested constraints on adjustment speed of the selected variables and try to reject the null that the selected variable is weakly exogenous with respect to other variables in the system (Table 5). Our results confirmed that the deviation from the

9 Enterprise and the Competitive Environment, March, 5 6, equilibrium in the system will lead to the open market operations adjustment in the future that eventually eliminates the disequilibrium in Germany. The weak exogeneity was rejected for economic activity and consumer prices at the 0.05 significance level.in the case of Austria we can reject weak exogeneity of bond yields and economic activity at the 0.05 significance level. 5. Discussion and Conclusions In this paper we examined the liquidity effects of the unconventional monetary policy of the ECB in the case when short term interest rates hit the zero lower bound. We estimated cointegration vector to identify long term causal effects of nonstandard measures of monetary policy on the long the long term interest rates of sovereign bonds in the selected core euro area member countries Austria, Germany, France and Greece. We started with the model creation which is based on the Gagnon, Raskin, Remache, Sack (2010) and Bernanke, Reinhart, Sack (2004). We applied cointegration analysis and weak exogeneity test to identify strong positive impact of the economic activity on the bond yields and negative impact of the prices and the economic sentiment indicator. Important result provided long run relationship between the open market operations and bond yields because the results confirmed our suggestion that the expansionary monetary policy affected long run interest rates with possible effects on the economic growth and investment activities. This cointegration relation was confirmed by identified models in Germany and Austria. Our results confirmed that the deviation from the equilibrium in the system will lead to the open market operations adjustment in the future that eventually eliminates the disequilibrium in Germany. Acknowledgements The results introduced in the paper have been funded with the support from the Czech Science Foundation via grant No. P403/ S Financial Crisis, Depreciation and Credit Crunch in CEECs. References BABA, N., NAKASHIMA, M., SHIGEMI, Y. and UEDA, K. The Bank of Japan's Monetary Policy and Bank Risk Premiums in the Money Market. Tokyo, Dostupné z: BERNANKE, B. S. and SACK, R. Monetary Policy Alternatives at the Zero Bound: An Empirical Assessment. Washington D. C., Dostupné z: Bloomberg [online] [cit ]. Dostupné z: BOFINGER, P. Monetary policy: goals, institutions, strategies, and instruments. Oxford: Oxford University Press, 2006, xxi, 454 s. ISBN

10 Enterprise and the Competitive Environment, March, 5 6, COOTNER, P. H., FRIEDMAN, M. and SCHWARTZ, A. J. A Monetary History of the United States ISBN / Eurostat. Eurostat_National Accounts [online] [cit ]. Dostupné z: GAGNON, J., RASKIN, M., REMACHE, J. and SACK, B. Large-Scale Asset Purchases by the Federal Reserve: Did They Work? March Dostupné z: HEINZ, F. F. and SUN, Y. Sovereign CDS Spreads in Europe The Role of Global Risk Aversion, Economic Fundamentals, Liquidity, and Spillovers. January Dostupné z: JÍLEK, J. Finance v globální ekonomice. 1. vyd. Praha: Grada Publishing, c2013, 557 s. Finanční trhy a instituce. ISBN JOHANSEN, S Statistical Analysis of Cointegrating Vectors. Journal of Economic Dynamics and Control, 12: JOHANSEN, S. 1991: Estimation and Hypothesis Testing of Cointegrating Vectors in Gaussian Vector Autoregressive Models. Econometrica, 59: KIYOTAKI, N. and MOORE, J. Liquidity, Business Cycles, and Monetary Policy. April Dostupné z: pdf/kimo.pdf KROGSTRUP, S. and SUTTER, B. Liquidity Effects of Quantitative Easing on Long-Term InterestRates Dostupné z: working_paper_2012_02/source/working_paper_2012_02.n.pdf KRUGMAN, P. MIT. IT S BAAACK! JAPAN S SLUMP AND THE RETURN OF THE LIQUIDITY TRAP Dostupné z: OKINA, K. and SHIRATSUKA. VENICE SUMMER INSTITUTE Policy Commitment and Expectation Formations: Japan s Experience under Zero Interest Rates Dostupné z: ODA, N. and UEDA, K. The Effects of the Bank of Japan s Zero Interest Rate Commitment and Quantitative Monetary Easing on the Yield Curve: A Macro-Finance Approach Dostupné z: wps_2005/data/wp05e06.pdf ROMER, C. D. and ROMER, D. H. The Missing Transmission Mechanism in the Monetary Explanation of the Great Depression. American Economic Review: Papers & Proceedings 2013, 103 (3): [online]. 2013, 103(3) [cit ]. Dostupné z: Statistical Data Warehouse. ECB. European Central Bank [online] vyd [cit ]. Dostupné z: reports.do?node= TAYLOR, J. B. The Monetary Transmission Mechanism: An Empirical Framework. Journal of economic perspectives. Volumes 9, Number 4, Fall Pages [online]. 1995, Volumes 9 [cit ]. Dostupné z: ecmon_readings/taylor_jep95.pdf

11 Enterprise and the Competitive Environment, March, 5 6, TOBIN, J. Liquidity Preference as Behavior Towards Risk Dostupné z: UGAI, H. Effects of the Quantitative Easing Policy: A Survey of Empirical Analyses. July Dostupné z: data/wp06e10.pdf WOODFORD, M. Interest and prices: foundations of a theory of monetary policy. Woodstock, Oxfordshire [England]: Princeton University Press, c2003, xv, 785 p. ISBN