Research proposal. Authored by: Hayk Sargsyan

Transcription

1 Research proposal Authored by: Hayk Sargsyan

2 Outline 1. Introduction a. Research Questions b. Literature review 2. Proposed Method a. Overview of research design b. Data and model variables 3. Proposed Analyses 4. Discussion a. Limitations b. Potential contributions to the literature 5. References 6. Appendices 1

3 1. Introduction a) Research Questions To implement a successful monetary policy and make reasonable decisions concerning to the policy instruments, monetary authorities must have a thorough understanding of the timing and effects of their policies on the real economy and inflation. The process that describes the effects of changes in monetary policy on aggregate demand and inflation is called transmission mechanism of monetary policy. Well-functioning transmission mechanism is a guarantee of a successful monetary policy. This makes monetary policy transmission mechanism one of the most studied areas of monetary economics. Armenia began to implement independent monetary policy since early 90s. The Central bank of Armenia adopted the strategy of monetary targeting as the method of monetary regulation since 1994, which was replaced by the strategy of inflation targeting in Under the framework of inflation targeting strategy the Central bank of Armenia adopted by low the official quantitative target of 4 percent within tolerance band for the inflation rate for a one year horizon, and through an effective monetary policy tries to achieve this target. Among the other criteria of a successful implementation of monetary policy in Armenia, one of the crucial conditions is the clear understanding of the process through which monetary policy shocks propagate to the real economy. Modern literature suggests different channels of monetary policy transmission mechanism. Mishkin (1996) provides brief discussions of the main channels of monetary policy transmission mechanism. As in many other developing countries similar to Armenia the main channels of monetary transmission in Armenia are interest rate channel and exchange rate channel. During these years the Central bank of Armenia made essential steps towards the achievement of developed and effective monetary policy framework. But still the overall macroeconomic situation and 2

4 the stage of financial development do not let the Central bank to achieve effective propagation of monetary policy changes to the real economy and inflation. After adopting inflation targeting the Central bank managed to achieve the targeted inflation rate only four times within the period 2006 to This result was relatively one of the worst among the countries which have adopted inflation targeting monetary policy strategy 1. Although most of the times of this failures the causes came mainly from supply side or external shocks, which were not under the control of the Central bank. Regardless, this is also a signal of a weak monetary transmission in Armenia. The evidence can also be found in the literature. There are several research papers which identify the weakness of monetary transmission in Armenia. For example, in their working paper Dabla-Norris and Floerkemeier (2006) found that the influence of monetary policy to the economic activity and inflation in Armenia were still limited, as important channels of monetary transmission were not fully functional. They mention that in particular, the interest rate channel remained weak, even though there was some evidence of transmission to prices of changes in repo rate. In another, more recent working paper Bordon and Weber (2010) concluded that the impact of the policy rate on prices in Armenia remain weak although there was evidence that it strengthened after the regime switch in In these papers the authors also try to discover the main reasons and sources of the problem, and identify the constraints of effective monetary transmission in Armenia, but goals of these papers do not include quantitative measure of the influence of disturbing factors on the effectiveness of monetary transmission. The central objective of this paper was the disclosure of the main determinants of a wellfunctioning monetary policy transmission mechanism in Armenia. The paper initially examines the historical developments of monetary policy pass-through coefficient in Armenia, investigates the main developments during the separate periods of time, and identifies the main disturbing factors of efficient monetary policy transmission in Armenia. The paper goes further, and also empirically estimates the 1 The calculation base is the database from IMF: 3

5 influence of those factors on the monetary policy pass-through coefficient, and makes policy recommendations towards the improvements in the monetary policy propagation to the real economy in Armenia. b) Literature review The interest in this research topic is understandable, and there can be found a lot of relevant literature. For example, Cechetti (1999) examined differences in the banking system characteristics across the countries of the European Union, and concluded that the differences in financial structure are an immediate cause for national asymmetries in the monetary policy transmission mechanism in these countries. Furthermore, he argued that the differences in financial structure across the countries of Europe are a consequence of their dissimilar legal structures, and so he led to the possibility that it is the legal system in a country that forms the basis for the structure of financial intermediation and, hence, for the impact of monetary policy on output and prices. Aysun et al (2010) found a positive relationship between the level of financial frictions and the effects of monetary policy shocks on the economy. Based on the predictions of the financial accelerator model, and by using firm level data, they measure the sensitivity of bond spreads to financial leverage of firms that issue these bonds. The results indicated a positive relationship between financial frictions and monetary policy pass-through. The next appropriate paper was suggested by Medina Cas et al (2011), who examined the key factors that influence the strength of the interest-rate transmission mechanism in Central American countries. They employed a panel regression analysis, the results of which suggested that the interest rate pass-through has a negative and statistically significant relationship with dollarization, and a positive and statistically significant relationship with exchange rate flexibility and financial system development. They also tried bank concentration ratio as a determinant of the effectiveness of monetary transmission. 4

6 Although bank concentration also had a negative relationship with the interest-rate transmission mechanism, its statistical significance was not confirmed. Mishra and Montiel (2012) examined the results of a large number of studies that have estimated the effects of monetary policy in low-income countries. They concluded that a wide range of empirical approaches used by different researchers fail to yield a considerable evidence of effective monetary transmission in low-income countries, and only relatively more institutionally developed countries such as some Central and Eastern European transition economies appear to have some evidence for effective monetary transmission. Another relevant study was presented by Mishra et al (2012). Cross-country evidence on the effectiveness of the monetary transmission mechanism led them to conclude that at the limited degree of financial development, the transmission mechanism in low income countries mainly dominated by the bank lending channel, and the strength and reliability of the monetary pass-through depends critically on the effectiveness of this channel. They also argued that the transmission from central bank monetary policy actions to bank lending rates in the low income countries is both weak and unreliable. Another relevant research in the given topic was introduced by Saborowski and Weber (2013), who found that exchange rate flexibility, banking sector concentration, liquidity ratios, along with nonperforming loans ratios and financial dollarization are important determinants of interest rate passthrough. They proved that in more developed markets monetary policy shocks almost fully propagate to retail lending rates. In contrast, the pass-through in developing countries is significantly lower at around percent. This is mainly explained by the existence of flexible exchange rate regimes, lower liquidity and NPL ratios, and more developed financial systems in advanced economies. 5

7 2. Proposed Method a) Overview of research design The final goal of this research is to identify the main disturbing factors for the monetary policy propagation to real economy and inflation in Armenia, and measure the influence of those factors on the effectiveness of monetary pass-through. Led by this purpose, first of all we will need to find a proxy variable which will describe the effectiveness of monetary policy pass-through in Armenia, and will be used as an outcome variable in the final regression analyses. Estimation of a proxy variable for the effectiveness of monetary transmission mechanism is not an easy task. For this purpose we will employ a time varying structural vector auto regression model, the outcome of which will give us a proxy for the coefficient of the effectiveness of monetary transmission in Armenia. The existence of the estimated coefficient of monetary policy pass-through will provide us the opportunity to perform a detailed analysis of the historical development of monetary policy transmission mechanism in Armenia. At the last stage we will identify the main determinants of the effectiveness of monetary transmission in Armenia and use them in a vector error correction model to estimate the long-run and short-run relation with the monetary policy pass-through coefficient. Estimation will provide us in-depth understanding of the influence of those factors on the effectiveness of monetary policy and we will be able to suggest policy recommendations for the improvements of monetary transmission in Armenia. b) Data and model variables At the first stage of the analysis, in order to estimate historical time series of monetary passthrough coefficient (MTS) a small time varying structural vector auto regression model (TVP_VAR model) will be employed 2. The model will include three key variables: inflation rate, economic growth 2 The choice of a small model is mainly due to the attempt to reduce the number of estimated parameters. 6

8 rate, and a short-term nominal interest rate. The sample will include the database from the first quarter of 1998 to the second quarter of Quarterly CPI inflation rate will be taken as an inflation rate. Quarterly growth rate of seasonally adjusted real GDP will be considered as an economic growth rate. Economic growth rate will be seasonally adjusted in order to escape the bias generated by the huge seasonality of Armenian economic development. The both database are sourced from the National statistical service of the Republic of Armenia 3. At last, as a short-term interest rate the weighted average of interbank overnight loan rate will be taken. The choice of this rate is mainly due to the fact that it is considered as the operational target of the monetary policy in the Central bank of Armenia. As the Central bank introduced this toolkit since the second quarter of 2010, up to that point the weighted average of interbank repo rate will be used. The sources of this database are the Central bank of Armenia 4 and Nasdaq OMX 5. At the second stage of the analysis the estimated MTS variable will be used in a VECM model, along with the variety of variables which can act as potential determinants of the effectiveness of monetary policy pass-through in Armenia. Theory suggests a variety of potential constraints to an effective monetary policy transmission mechanism, from which initially the followings will be considered as control variables in the model for Armenia: Financial dollarization: The literature lacks in the research field which explains how financial dollarization influences the effectiveness of monetary policy. Although there is a common view among economists that dollarization makes monetary policy more complicated and less effective. The literature emphasizes that in highly dollarized financial systems, the Central bank has only limited control over the interest rates. The reason comes from the fact that foreign currency funding is linked to external factors which cannot be directly influenced by the monetary authorities. Thus, money supply mainly follows the

9 behavior of agents holding foreign and domestic-currency denominated assets. This understandably complicates the Central bank`s ability to control inflation. Additionally, Leiderman et al (2006) mention that in highly dollarized financial markets there is a fear of floating capital that can be damaging for effective interest rate transmission. For the purposes of this research, financial dollarization will be measured as the ratio of foreign currency deposits to broad money (M2X). Broad money in this contest is the sum of M2 and all foreign currency deposits (including accounts) of residents (real sector entities) 6. The source for the given database is the calculations of the Central bank. Like the most other developing countries, the dollarization rate historically was very high in Armenia. Only the period between 2004 and the start of the global financial crisis was the period of steadily decreasing dollarization rate. Armenia managed to achieve the historical minimum of the dollarization rate at the end of At that time the dollarization rate was about 19 percent. But with the crisis the dollarization rate increased dramatically and returned to its previous level of about 50 percent. Financial development: Although economic theory has recognized the role of financial development in the effectiveness of monetary transmission and the subject has been widely discussed in the literature over the last decades, empirical studies examining this relation remain limited. The effectiveness of monetary policy propagation to the real economy is crucially affected by the degree of development in financial markets, as monetary policy is implemented mainly through operations in these markets. As was mentioned in Cottarelli and Kourelis (1994) developed financial systems typically offer a greater variety of financial products, which increases the competition in these markets. Increased competition constrains the profit margins and makes market rates more responsive to policy rate shocks. During the empirical analyses the value of private credit as the share of GDP will be used as a measure of financial development in Armenia. The source is the Central bank calculations. This is a 6 Detailed calculation can be found in the manual on compilation of monetary and financial statistics in Armenia: ( ) 8

10 preferred measure of financial development in the recent literature. For example, Ross et al (2000) mention that higher levels of private credit indicate higher levels of financial services and therefore greater financial intermediary development. Exchange rate flexibility: Exchange rate flexibility contributes to the strength of monetary policy pass-through. A lack of exchange rate flexibility indicates that the changes of policy rate are not aiming to influence market rates. Changes in the policy rate influences the capital flows, which under flexible exchange regimes leads to currency changes. Under fixed exchange regimes such capital flows are not fully sterilized, which affects the liquidity and results pressures on the interest rates. Since exchange rate flexibility is a measure that is not directly observable, there is no clear answer how to measure it. Theory suggests several main methods of measuring exchange rate flexibility. Most empirical studies have used the standard deviation of the moving average of the logarithm of the exchange rate as a measure of exchange rate flexibility. Nevertheless, this method deserves some criticism for the failure to capture the potential effects of high and low peak values of the exchange rate. To solve this problem, for example, Reinhart and Rogoff (2004) composed two sets of estimated equations. The first contained the standard deviation of the moving average of the logarithm of the real effective exchange rate as a measure of volatility and the second contained a dummy variable capturing only high and low values of the exchange rate. They defined some certain thresholds above and below which the exchange rate was considered as volatile. During this research a dummy variable based on the standard deviation of exchange rate within a quarter will be used. The source is the author`s calculations. Health of financial system: The existence of bad loans in the financial system increases the risk and makes the financial institutions more cautious while giving new loans. This makes extra barriers for effective monetary policy, as banks with weak balance sheets may react to expansive monetary shocks by accumulating liquidity rather than giving credit at lower rates. As a result, monetary policy changes may thus have only a limited impact on market rates. 9

11 The share of non-performing loans (NPL) in total loans will be used as a measure of health of financial system in Armenia. The source is the staff calculation of the Central bank of Armenia. 3. Proposed Analyses The given research will implement two-step analyses. The first stage will include the estimation of a proxy for the effectiveness of monetary transmission (MTS variable) in Armenia. The literature provides two relevant approaches for approximation of MTS. By the first approach many authors (Christiano et al., [1996], Kim, [1999], and Kim and Roubini, [2000]) estimate VAR models to measure the responses to an unanticipated shock of monetary policy and then use forecast error variance decompositions (FEVD) to obtain the coefficients of MTS. The strength of monetary transmission in these studies is measured by the percentage of variations in output or CPI inflation explained by monetary policy. Alternatively, impulse response analyses in VAR models can be used to obtain the values of MTS. For example, Ceccheti (1999) implements impulse response analyses in his VAR model and uses the maximum amplitudes of output and inflation responses to a one standard deviation shock to interest rates as a measure of monetary policy effectiveness. The same approach is also used by Aysun et al (2010). This research will use the later method to obtain the proxy for the monetary pass-through coefficient in Armenia. A small quarterly model with three variables (interest rate, inflation rate and economic growth) will be estimated for Armenian economy. The theoretical base for the model is taken from Primiceri (2004). The model will be a multivariate time series model with both time varying coefficients and time varying variance covariance matrix of the additive innovations. The fluctuations in the coefficients capture the possible nonlinearities or time variation in the lag structure of the model. The fluctuations in the multivariate variance covariance matrix capture possible heteroskedasticity of the shocks and nonlinearities in the simultaneous relations among the variables of the model. The time variation both in the coefficients and the variance covariance matrix allows the data to determine whether 10

12 the time variation of the linear structure derives from changes in the size of the shocks or from changes in the propagation mechanism. The heteroskedastic TVP-VAR model takes the following form: 7 Where is a vector of observed endogenous variables 8, is an vector of time varying coefficients that multiply constant terms, are matrices of time varying coefficients, are heteroskedastic unobservable shocks with time-varying variance covariance matrix. Timevarying variance covariance matrix can be presented in the following form: Where is the lower triangular matrix of time-varying covariances with ones on the diagonal: [ ] And is the diagonal matrix of the error time-varying standard deviations: [ ] It is crucial for a time varying structural VAR model that the matrices and vary over time. A constant implies that an innovation to the variable has a time invariant effect on the variable. This contradicts to the fundamental concept that was proposed about the simultaneous interactions among variables. A constant violates the assumption of heteroskedasticity. 7 t=1 T is the observation number, i=1 L is the lag order. 8 n=1 N is the number of endogenous variables in the model. 11

13 For estimating the model parameters we need to define as the arranged row vector of the model parameters, as the arranged row vector of the free lower-triangular elements of the time-varying covariances, and, where, as the arranged row vector of the diagonal matrix of the error time-varying standard deviations. The timevarying parameters of the model are assumed to follow the random walk process dynamics: (3) We assume that all the innovations in the model are jointly normally distributed and have the following variance covariance matrix: ( ) ( ) ( ) Where,, ( ), ( ), ( ). In order to evaluate the posterior distributions of the parameters of interest we need to use Bayesian methods of estimate. So, initial priors of the parameters are required to be defined. The priors for the, and will be assumed to be distributed as independent inverse-wishart. The priors for the initial states of the time varying coefficients, simultaneous relations and log standard errors (, 12

14 and ) will be assumed to be normally distributed. To calibrate priors Kalman filter estimation technology will be used, and after several robustness check the most probable priors will be chosen. The numerical estimation of the parameters will be done by an efficient Markov chain Monte Carlo algorithm 9. At the next step an impulse response analyses will be applied in order to estimate the responses of CPI inflation to the one standard deviation shocks to interest rates during the whole historical range. The maximum amplitude of the responses will be considered as the rate of monetary transmission mechanism, and as a result historical time series of MTS variable will be obtained. During the next stage the estimated time series of the monetary pass-through coefficient will be used in a VECM model. The analyses will provide the short-run and long-run influences of the given determinants on the effectiveness of monetary transmission mechanism in Armenia. The results will provide the opportunity to understand how monetary policy effectiveness changes due to the changes of different factors, and suggest policy recommendations. 4. Discussion a) Limitations The main limitation of the research is the short history of the time series. For the estimation of MTS variable the length of the used dataset is 66. For VECM model there are some time series whose lengths are 46 quarters. Short history of the used dataset may generate some bias during the estimation. The next limitation for the research is the choosing of appropriate priors for the TVP_VAR model while estimating the monetary pass-through coefficient. It is crucial to give such a priors to the model, which will not affect significantly on the results of the estimated coefficients. 9 I am very thankful to Jouchi Nakajima for technical assistance while carrying out MCMC simulation. 13

15 Another limitation of the research is connected with the second stage. Although we have chosen almost all the main determinants of the effectiveness of monetary policy pass-through suggested by the literature, we understand that there can be a lot of country specific factors which may have a significant influence on the monetary transmission. The difficulty or in some cases even the impossibility to estimate the impact of those factors on the monetary transmission in Armenia may reduce the explanatory power of the final model. b) Potential contributions to the literature Current research has several potential contributions to the literature. First of all this is the first attempt to estimate the historical development of the monetary policy passthrough in Armenia. Obtaining the historical time series of the coefficient of monetary pass-through will help as not only to understand the evaluation of the effectiveness of monetary policy during the years, but it also will make possible to identify the main reasons of the weak pass-through, and make partial analyses of the developments during different periods. Another important contribution to the literature is the disclosure of the main determinants which influence on the effectiveness of monetary transmission in Armenia. Up to this point the literature only hypothetically has suggested the possible hindering factors of the efficient monetary transmission in Armenia. This research tries to quantitatively measure the influence of those factors. This will help us to answer several questions which may have a very crucial policy implementations: We will be able to measure the effect of changes in the level of dollarization in Armenia on the effectiveness of monetary pass-through. This will help us to answer the question whether it is worth to take actions against the dollarization from monetary policy point of view. 14

16 Another important implication is that with the help of this analyses we will be able to understand how much will monetary policy changes propagate to real economy at the given stage of financial development. We will be able also to understand how we can improve the effectiveness of monetary policy transmission in Armenia. 15

17 References Aysun, Uluc, Ryan Brady, and Adam Honig Financial frictions and monetary transmission strength: A cross - country analysis. Working Paper R. (August 2009, revised June 2010). Berger, Allen N., and Timothy H. Hannan The Price-Concentration Relationship in Banking. Review of Economics and Statistics. (May). Bordon, Anna R., and Anke Weber The Transmission Mechanism in Armenia: New Evidence from a Regime Switching VAR Analysis. IMF Working Paper no 10/270. (November). Cecchetti, Stephen G Legal Structure, Financial Structure, and the Monetary Policy Transmission Mechanism. Economic Policy Review of Federal Reserve Bank of New York. (July). Christiano, Lawrence, Eichenbaum Martin, and Charles Evans The effects of monetary policy shocks: Evidence from the flow of funds. The Review of Economics and Statistics. (February). Cottarelli, C., A. Kourelis Financial structure, bank lending rates, and the transmission mechanism of monetary policy. IMF Working Paper no 94/39. (March). Dabla-Norris, Era, and Holger Floerkemeier Transmission Mechanisms of Monetary Policy in Armenia: Evidence from VAR Analysis. IMF Working Paper no 06/248. (November). Kim, Soyoung Do monetary policy shocks matter in the G-7 countries? Using common identifying assumptions about monetary policy across countries. Journal of International Economics. (August). Kim, Soyoung, and Nouriel Roubini Exchange rate anomalies in the industrial countries: A solution with a structural VAR approach. Journal of Monetary Economics. (June). Leiderman, L., R. Maino, and E. Parrado, Inflation targeting in dollarized economies. IMF Working Paper no 06/157. (June). 16

18 Medina, S. Cas, Alejandro Carrión-Menéndez, and Florencia Frantischek The policy interest-rate pass-through in Central America. IMF Working Paper no WP/11/240. (October) Mishkin, Frederic S. (1996). The channels of monetary policy transmission: Lessons for monetary policy. National Bureau of Economic Research. (February). Mishra, Prachi, Peter J. Montiel, and Antonio Spilimbergo Monetary transmission in low income countries. IMF Working Paper no 10/223. (October). Mishra, Prachi, and Peter Montiel How effective is monetary transmission in low-income countries? A survey of the empirical evidence. IMF Working Paper no 12/143. (June). Primiceri, Giorgio E Time Varying Structural Vector Autoregressions and Monetary Policy. Northwestern University. (July). Reinhart, C. and K. Rogoff The Modern History of Exchange Rate Arrangements: Reinterpretation. The Quarterly Journal of Economics Ross, Levine, Norman Loayza, and Thorsten Beck Financial intermediation and growth: Causality and causes. Journal of monetary economy. Saborowski, Christian, and Sebastian Weber Assessing the determinants of interest rate transmission through conditional impulse response functions. IMF Working Paper no 13/23. (January). Saxegaard, Magnus Excess liquidity and effectiveness of monetary policy: Evidence from Sub-Saharan Africa. IMF Working Paper no 06/115. (May)

19 Appendices Chart 1: Inflation target and actual inflation levels for years Chart 2: The development of financial dollarization in Armenia. 18

20 Chart 3: The financial development in Armenia. Chart 4: Exchange rate flexibility in Armenia. 19

21 Chart 5: The developments of bad loans in Armenia. 20

22 Chart 6: Example of obtaining pass-through coefficient. The responses at the simultaneous periods of shocks are positive which contradicts to literature, but there can be found a lot of researches where the same results are obtained. Therefore, the first periods responses will be ignored and as the coefficient of monetary policy pass-through will be taken up to the second quarters cumulative responses. 21