Channels of Monetary Policy Transmission in Indonesia: A VAR model analysis

Channels of Monetary Policy Transmission in Indonesia: A VAR model analysis Iwan Evander Syema Abstract This study examines the various channels of monetary policy transmission mechanism in Indonesia. The interest rate, exchange rate, asset price and credit channels will be analysed using a VAR short-run restriction model for quarterly data, ranging from 2000Q1 to 2018Q1. The results show that the interest rate (cost of capital) channel is the most effective channel of monetary transmission in stabilising output and price level. Meanwhile, the exchange rate and asset price are the least effective channels of monetary policy, which operate via trade pass-through and Tobin’s q asset valuation channel. The credit channel is also found to be weak in controlling the price level. JEL Codes: C32, E52, E58 Keywords: Vector autoregression, monetary transmission mechanism, Indonesia.  1. Introduction The primary mandate of the central bank of Indonesia, Bank Indonesia (BI), is to achieve currency stability, which is reflected in the price of goods and services (inflation) and price of foreign currency (exchange rate). The pre-eminence of understanding the different channels of monetary transmission came following the 1997 Asian Financial Crisis (AFC). Mishkin (2014) identifies the common five types of transmission channels, namely, interest rate, exchange rate, asset price, credit and inflation expectations channels (see Figure 1 – Appendix B). Monetary policy transmission has been discussed and used extensively in a number of empirical literature using vector autoregression (VAR) models. The most remarkable VAR study was conducted by Taylor (1995), describing the impact of the monetary policy shock on the real GDP. Christiano et al. (1999) employed VAR approaches for identifying the impact of an exogenous disturbance to monetary policy. Likewise, Erceg and Levin (2002) found that the optimal policy rule can be closely approximated by using the weighted average of aggregate wage and price changes. In the context of Indonesia, the following studies have been conducted. Former BI’s governor, Goeltom (2005), shows that the strongest channel of monetary policy transmission is through asset price channel followed by exchange rate channel. In the study conducted by Fuddin (2014), the credit channel is found to be effective in explaining the changes in the price level and output in Indonesia. Similarly, Ridhwan et al. (2011) found that the credit channel of monetary is more profound in explaining inflation after the 1997 AFC. In this research, the VAR approach is used, which focuses primarily on the reduced-form relationship between monetary policy and output-price level responses. Some of the key macroeconomic variables which are included in analysing the monetary policy channel are the real interest rate, real effective exchange rate, asset price and domestic bank credit. The results show that the most significant channel of monetary policy in Indonesia is the interest rate channel. The exchange rate channel is more profound in explaining the changes in price level. Meanwhile, the asset price and credit channels appear to be significant in explaining the sources of shocks in output only. This paper is constructed as follows. First, the VAR model is constructed to identify the monetary policy shocks considered in this paper. The response of the inflation and output to the money supply shock will be analysed based on the impulse responses from each channel of transmission. Finally, the paper concludes by summarising the findings of this research. 2. Data and Model 2.1 The VAR Model This paper follows the identification strategy provided by Sims (1980), and the detailed VAR model follows a case study in India by Aleem (2010). The VAR model is described as follows: The form of VAR(p) is: (1) Where is the vector of endogenous variables and is the vector of exogenous variables, and are polynomial coefficients, is the vector of innovations or shocks. Indonesia, a developing country, is an open-economy and understanding the sources of international shocks is important as it can influence the domestic economy. Foreign shocks such as world oil price (oil), federal fund rates (ffr) and the United States’ GDP (yus) are included in the model. The matrix for the VAR model can be written as follows: To conduct the analysis, each endogenous variable will be incorporated in the VAR model. Each channel will have a different variable of interest based on the type of the channel. Since data for inflation expectation is not sufficient, there will be only four channels that will be analysed in this study. The model uses the following variables: (i) real output (yid); (ii) consumer price index (cpi); (iii) broad money supply (m2); (iv) market interest rate (irate); (v) real effective exchange rate (reer); (vi) asset price (idx); and (vii) domestic credit (credit). This study used the broad money supply as an instrument for monetary policy shocks because BI is targeting the broad money supply (Goeltom 2005). Interest rate channel: Exchange rate channel: Asset price channel: Credit channel: 2.2 Data Description The data description can be found in Table 1 (Appendix A). In this study, the author uses data period from 2000Q1 to 2018Q1. All the variables are seasonally-adjusted using Census X-13 to remove the cyclical effects in the variables. Thereafter, the level variables are transformed to the natural logarithmic form to reduce the skewness of the variables. This method is done to make the interpretation become more intuitive. 2.3 Identification Strategy Identification is necessary to pin down how macroeconomic variables respond to unexpected or surprise changes in monetary policy. For instance, when the policy interest rate is high, it does not necessarily mean that it increases output. It could be that since output is high, inflation tends to rise, causing interest rate to go up. This channel should be investigated using appropriate identification approach. Therefore, the Cholesky ordering of endogenous variables in the VAR model is output, cpi, the variable of interest and m2. The identification matrix can be obtained through Cholesky decomposition. The output does not react contemporaneously to shock to other variables in the system. Price level reacts simultaneously to output shock but is not affected by the money supply shock and the variables of interest market interest rate (irate), real effective exchange rate (reer), asset price (idx) and domestic credit (credit)concurrently. The variables of interest react contemporaneously to output and price level shocks in the short-run. Finally, the broad money supply, which is the proxy instrument for policy rate is influenced by all shocks in the system. The identification can be illustrated as follows: Interest rate channel: = Exchange rate channel: = Asset price channel: = Credit channel: = 2.4 Lag Length Selection The information criteria are useful for choosing the optimal lag length. Lag length can be chosen based on the largest information criteria such as final prediction error (FPE), Akaike information criterion (AIC); Schwarz information criterion (SC), Hannan-Quinn information criterion (HQ) and other information criteria. Table 2 (Appendix A) shows the results of the information criteria for each model. The optimal lag length selection is relevant in having a better model for analysis and forecasting. 2.5 Data Stationarity The variables in the system should be stationary in order to test for statistical significance and to compute for long-run moments. One way to check for the stationarity of the model is by using the characteristic inverse roots of the VAR model. By applying eigen decomposition analysis, the stability of the VAR system can be ensured as long as the absolute values of the eigenvalues are less than unity. Figure 2 (Appendix A) shows the graph of the eigenvalues from each system. 3. Results Analysis and Findings The results of the impulse response functions analysis in this study contribute to the existing body of research in the transmission mechanism of monetary policy. Based on the analysis using short-run restriction VAR models, the results can be summarised as follows: Indonesia m2 yid cpi irate reer asset credit Expected + + + - - + + Fung (2002) + + + - + N/A N/A This study + + + - - + + 3.1 Interest Rate Channel The impulse response functions (see Figure 3 – Appendix B) depicts the impact of a positive shock to the money supply in the interest rate channel. According to Mishkin (2014), the impact of a positive shock to the money supply causes the real market interest rate to fall, which also means that the cost of capital also decreases. The real interest rate falls for ten quarters in response to one standard deviation positive shock in the money supply. The price level increases initially then fall in the sixth quarter. This result shows that there is no “price puzzle” in Indonesian economy after 2000 where BI started implementing inflation targeting – an indication that the shift to inflation targeting by BI is effective. Also, the price level responds negatively to aggregate demand which means that the inflation is caused by the supply side. Meanwhile, output reacts negatively to the positive shock of the money supply in the first period before it keeps on increasing from second quarter onwards. This situation can be explained by the increasing spending on investments as a result of a decrease in the real interest rate. This increase in investment spending leads to an increase in aggregate demand and output. The variance decomposition (see Table 3 - Appendix A) shows that 7.67 percent of shocks in output after four quarters are due to shocks in the money supply and becomes more significant in the long run. Meanwhile, 19.31 percent of shocks in output is caused by the real market interest rate after four quarters which means that the interest rate channel is quite significant. Increase in money supply also affect the price level but not significant. However, the real interest rate seems to be more significant in price determination which accounted for 7.84 percent after four quarters. 3.2 Exchange Rate Channel The impulse response functions (see Figure 4 – Appendix B) depicts the impact of a positive shock to the money supply in the exchange rate channel. A fall in the real interest rate causes domestic currency to be less attractive compared to the foreign currency, which causes depreciation of domestic currency. This depreciation makes domestic goods cheaper than foreign goods, hence, net export and output increases. In the results, the real exchange rate depreciates for more than eight quarters in response to one standard deviation positive shock in the money supply. The price level steadily increases by 0.3 percent in the following periods. Meanwhile, output reacts positively to the positive shock of the money supply, which keeps on increasing until being steady at 0.1 percent. The variance decomposition (see Table 4 - Appendix A) shows that 0.83 percent of shocks in output after four quarters are due to shocks in the money supply. Meanwhile, 0.86 percent of shocks in output is caused by the real exchange rate after four quarters which means that the exchange rate channel is not significant in output. However, the real exchange rate appears to be more significant in explaining price changes which accounted for 12.93 percent after four quarters. Similarly, the price level is an essential source of shocks in the real exchange rate which accounted for 14.70 percent after four quarters. In general, the real exchange rate is a significant factor to explain the sources of shocks in the price level. 3.3 Asset Price Channel The impulse response functions (see Figure 5 – Appendix B) describes the impact of a positive shock to the money supply in the asset price channel. The asset price channel can work through two possible conditions: a wealth effect on consumer spending (Modigliani 1971) and Tobin’s q asset price valuation (Tobin 1969). According to Mishkin (1995), the impact of a positive shock to the money supply causes an increase in the asset price. Modigliani (1971) explained that the positive effect on asset price on consumption and output is possible because of the wealth effect (i.e., higher level of wealth when asset price increases). The asset price and price of goods rise at the second period in response to one standard deviation positive shock in the money supply. Meanwhile, similarly to interest rate channel, output reacts negatively to the positive shock of the money supply in the first period before it keeps on increasing from second quarter afterwards. The variance decomposition (see Table 5 - Appendix A) shows that 1.65 percent of shocks in output after four quarters are due to shocks in the money supply and becomes more significant in the long run. Meanwhile, 3.19 percent of shocks in output is caused by the asset price after four quarters. Money supply is an important source of changes in the price level which accounted for 8.82 percent. Moreover, the asset price is significantly explained by the income effect which accounted for 10.75 percent after four quarters. These results confirm that the asset price channel appears to be significant. 3.4 Credit Channel The impulse response functions (see Figure 6 – Appendix B) depicts the impact of a positive shock to the money supply in the credit channel. According to Mishkin (2014), the credit channel can work through two sub-channels: the bank lending channel and the balance-sheet channel. The impact of a positive shock to the money supply causes an increase in bank deposits, which increase the amount of money that a bank can lend out. Hence, investment is expected to be higher and in turn, increase in output. In the results, the domestic credit rises for more than eight quarters in response to one standard deviation positive shock in the money supply. The price level gradually increases by 0.1 percent in the following periods. Meanwhile, output reacts positively to the positive shock of the money supply, which keeps on increasing until being steady at less than 0.1 percent. The variance decomposition (see Table 6 - Appendix A) shows that 0.86 percent of shocks in output after four quarters are due to shocks in the money supply. Meanwhile, 2.81 percent of shocks in output is caused by the domestic credit after four quarters which means that the domestic credit is quite significant in output. Also, the domestic credit is not significant in explaining price changes which accounted for less than 0.05 percent after four quarters. However, money supply seems to be an important source of shocks in both price and domestic credit level which accounted for 0.65 percent and 1.93 percent after four quarters respectively. In general, domestic credit is important in explaining the sources of shocks in output. 4. Conclusion This paper employs VAR models to help identify the transmission mechanism of monetary policy to output and price levels in Indonesia using quarterly data from 2000Q1 to 2018Q1. From the IRFs analysis, the interest rate channel is more pronounced in the monetary transmission mechanism to the real output and price level in the economy. While the other channels such as exchange rate, asset price and credit channel are found to be weak and less effective. The IRFs rely on its reduced form equations. The path of adjustments do not depend on the whole set of the endogenous variables but only in the subset of those variables. Therefore, it will be more accurate to use a more complex model like dynamic stochastic general equilibrium (DSGE) model to understand how adjustment mechanisms in monetary policy work. This can be the future extension topic of this paper.  References Aleem, A 2010, ‘Transmission mechanism of monetary policy in India’, Journal of Asian Economics, vol. 21, no. 2, pp. 186–197. Christiano, LJ, Eichenbaum, M & Evans, CL 1999, ‘Chapter 2 Monetary policy shocks: What have we learned and to what end?’, in Handbook of Macroeconomics, vol. 1, Elsevier, pp. 65–148, viewed 27 October 2018, <http://www.sciencedirect.com/science/article/pii/S1574004899010058>. Erceg, CJ & Levin, AT 2002, Optimal Monetary Policy with Durable and Non-Durable Goods, SSRN Scholarly Paper, ID 366585, 1 December, Social Science Research Network, Rochester, NY, viewed 28 October 2018, <https://papers.ssrn.com/abstract=366585>. Fung, BSC 2002, ‘A VAR analysis of the effects of monetary policy in East Asia’, viewed 29 October 2018, <https://www.bis.org/publ/work119.htm>. Goeltom, MS 2005, ‘The transmission mechanisms of monetary policy in Indonesia’, no. 35, p. 24. Mishkin, FS 1995, ‘Symposium on the Monetary Transmission Mechanism’, Journal of Economic Perspectives, vol. 9, no. 4, pp. 3–10. ― 2014, The Economics of Money, Banking and Financial Markets, 11 edition, Pearson, Boston. Modigliani 1971, ‘Monetary Policy and Consumption: Linkages via Interest Rate and Wealth Effects in the FMP Model, Consumer Spending and Monetary Policy: The Linkages’, in Journal of Finance and Economics, vol. 5, Conference Series No. 5, pp. 219–232, viewed 29 October 2018, <http://pubs.sciepub.com/jfe/5/5/4/index.html>. Ridhwan, MM, Henri L.F. de Groot, Piet Rietveld & Peter Nijkamp 2011, ‘The Regional Impact of Monetary Policy in Indonesia’, p. 29. Sims, CA 1980, ‘Macroeconomics and Reality’, Econometrica, vol. 48, no. 1, pp. 1–48. Taylor, JB 1995, ‘The Monetary Transmission Mechanism: An Empirical Framework’, Journal of Economic Perspectives, vol. 9, no. 4, pp. 11–26. Tobin, J 1969, ‘A General Equilibrium Approach To Monetary Theory’, Journal of Money, Credit and Banking, vol. 1, no. 1, pp. 15–29.  Appendix A: Tables Table 1: Data Description Variables Descriptions Units Sources Real GDP ID (yid) Real Gross Domestic Product for Indonesia based constant price at 2010 IDR billion CEIC / Central Bureau of Statistics Consumer Price Index (cpi) Index for all goods and services Index at 2010 = 100 Federal Reserve Economic Data (FRED) Broad money (m2) The money supply including all elements of M1, cash and checking deposits. IDR billion Bank Indonesia Market interest rate (irate) The interbank lending rate will be used for the proxy of the market interest rate percent Bank Indonesia Real effective exchange rate (reer) The weighted average of Rupiah in relation to an index of US dollar, adjusted for the effects of inflation IDR/USD index at 2010=100 Bank for International Settlements (BIS) Indonesia composite stock market price index (idx) The primary stock market index in Indonesia which is used to quantify the value changes in the stock prices. Index at 10 August 1982=100 Indonesia Stock Exchange Domestic credit (credit) Domestic credit is provided by domestic banks, all other sectors of the economy and non-residents IDR billion Bank for International Settlements (BIS) Real GDP US (yus) Real Gross Domestic Product for the US chained at 2012 Dollars USD billion Federal Reserve Economic Data (FRED) Effective Federal Funds Rate (ffr) The federal funds rate is the interest rate at which depository institutions trade federal funds (balances held at Federal Reserve Banks) with each other overnight. percent Board of Governors of the Federal Reserve System (US) Oil price (oil) Crude Oil Prices: West Texas Intermediate (WTI) Dollars per Barrel U.S. Energy Information Administration  Table 2: Lag Length Selection Interest Rate Channel:  Lag LogL LR FPE AIC SC HQ 0  303.7951 NA   2.80e-09 -8.341887 -7.823833 -8.136357 1  642.2786  598.4782  2.45e-13 -17.68924 -16.65313 -17.27818 2  678.0157   59.04388*   1.40e-13*  -18.26133*  -16.70716*  -17.64474* 3  687.4559  14.50237  1.73e-13 -18.07119 -15.99897 -17.24907 4  702.0183  20.68276  1.86e-13 -18.02952 -15.43925 -17.00187 Exchange Rate Channel:  Lag LogL LR FPE AIC SC HQ 0  120.9515 NA   4.11e-07 -3.352354 -2.950928 -3.193966 1  562.3411  787.7106   8.54e-13*  -16.44126*  -15.50460*  -16.07169* 2  576.7940  24.01417  9.04e-13 -16.39366 -14.92177 -15.81291 3  582.2113  8.334213  1.28e-12 -16.06804 -14.06091 -15.27610 4  605.3862   32.80144*  1.06e-12 -16.28881 -13.74645 -15.28568 Asset Price Channel:  Lag LogL LR FPE AIC SC HQ 0  474.2374 NA   2.00e-11 -13.28224 -12.76419 -13.07671 1  805.6523  585.9799  2.15e-15 -22.42470  -21.38860* -22.01365 2  837.9317   53.33129*   1.36e-15*  -22.89657* -21.34241  -22.27998* 3  849.1717  17.26714  1.59e-15 -22.75860 -20.68638 -21.93648 4  863.1684  19.87942  1.74e-15 -22.70053 -20.11026 -21.67289 Credit Channel:  Lag LogL LR FPE AIC SC HQ 0  302.1048 NA   8.49e-10 -9.536825 -8.978333 -9.318368 1  551.9771  433.1121   3.51e-13*  -17.33257*  -16.21559*  -16.89566* 2  562.3048  16.52429  4.31e-13 -17.14349 -15.46802 -16.48812 3  568.3739  8.901324  6.19e-13 -16.81246 -14.57849 -15.93863 4  593.3650   33.32152*  4.83e-13 -17.11217 -14.31971 -16.01988 * indicates lag order selected by the criterion ; LR: sequential modified LR test statistic (each test at 5 percent level) ; FPE: Final prediction error; AIC: Akaike information criterion; SC: Schwarz information criterion; HQ: Hannan-Quinn information criterion.  Table 3: Variance Decomposition for Interest Rate Channel Variance Decomposition of LYID:  Period S.E. LYID LCPI IRATE LM2  1  0.003893  100.0000  0.000000  0.000000  0.000000  4  0.006387  69.87482  3.131630  19.31443  7.679116  8  0.009262  36.32532  2.976835  31.20594  29.49190  12  0.011147  25.42025  9.514406  26.75393  38.31141  16  0.012451  20.87433  14.67376  23.48573  40.96618  20  0.013516  18.22003  17.23522  22.06120  42.48354 Variance Decomposition of LCPI:  Period S.E. LYID LCPI IRATE LM2  1  0.010640  0.634526  99.36547  0.000000  0.000000  4  0.021100  2.781783  86.49189  7.845026  2.881305  8  0.025660  2.193006  81.26373  14.47945  2.063817  12  0.026002  2.807067  80.50051  14.58705  2.105373  16  0.026310  2.865332  78.82861  15.27944  3.026624  20  0.026609  2.802515  77.82048  15.28847  4.088536 Variance Decomposition of IRATE:  Period S.E. LYID LCPI IRATE LM2  1  0.742393  2.421306  13.14593  84.43276  0.000000  4  2.242353  1.088570  15.53915  79.25972  4.112567  8  2.802621  6.166341  13.54294  69.33630  10.95442  12  3.062965  8.027560  20.95647  61.21287  9.803100  16  3.155612  7.651026  21.29767  60.57544  10.47587  20  3.172979  7.577318  21.23383  60.09015  11.09871 Variance Decomposition of LM2:  Period S.E. LYID LCPI IRATE LM2  1  0.011735  1.254724  0.812791  3.435916  94.49657  4  0.017550  3.017081  7.434814  6.822729  82.72538  8  0.021253  16.88569  5.866642  9.276768  67.97090  12  0.027534  14.98376  3.757095  22.41668  58.84246  16  0.032917  11.14385  6.189752  24.06320  58.60320  20  0.036872  9.230249  11.07727  21.88107  57.81141  Cholesky Ordering: LYID LCPI IRATE LM2  Table 4: Variance Decomposition for Exchange Rate Channel Variance Decomposition of LYID: Period S.E. LYID LCPI REER LM2  1  0.004709  100.0000  0.000000  0.000000  0.000000  4  0.008562  98.23860  0.056156  0.866982  0.838262  8  0.011189  94.74542  0.422333  2.526722  2.305524  12  0.012870  91.94818  1.103045  3.411607  3.537164  16  0.014027  89.80125  1.967524  3.570526  4.660701  20  0.014864  87.92264  2.911606  3.387993  5.777762 Variance Decomposition of LCPI: Period S.E. LYID LCPI REER LM2  1  0.010285  0.557177  99.44282  0.000000  0.000000  4  0.020144  8.990118  76.26389  12.93631  1.809690  8  0.029831  16.48043  50.14773  28.47047  4.901367  12  0.036648  18.71514  38.63923  35.46921  7.176410  16  0.041032  18.82635  33.55629  38.69430  8.923048  20  0.043798  18.18413  31.18909  40.27588  10.35090 Variance Decomposition of REER: Period S.E. LYID LCPI REER LM2  1  3.275771  2.209299  10.61276  87.17794  0.000000  4  4.895918  1.994728  14.69857  83.00738  0.299318  8  5.194567  2.270984  18.28841  78.87413  0.566481  12  5.319362  4.652783  19.12158  75.65865  0.566985  16  5.466189  7.569145  18.62038  73.25665  0.553816  20  5.590825  9.804089  17.95910  71.66450  0.572310 Variance Decomposition of LM2: Period S.E. LYID LCPI REER LM2  1  0.012198  3.722169  2.155218  27.96794  66.15468  4  0.020846  2.107303  6.054481  24.37720  67.46101  8  0.026846  6.711449  10.98284  19.88979  62.41592  12  0.031481  11.43727  14.41123  17.00185  57.14965  16  0.035392  14.61020  16.69299  15.32069  53.37612  20  0.038783  16.58386  18.26399  14.34596  50.80620  Cholesky Ordering: LYID LCPI REER LM2  Table 5: Variance Decomposition for Asset Price Channel Variance Decomposition of LYID: Period S.E. LYID LCPI LIDX LM2  1  0.004383  100.0000  0.000000  0.000000  0.000000  4  0.007032  92.46741  2.687585  3.192348  1.652656  8  0.009071  82.33949  2.073578  8.688392  6.898540  12  0.010968  72.32839  3.123238  13.92429  10.62408  16  0.012646  64.99650  5.040392  17.11878  12.84433  20  0.014132  59.48772  7.115119  19.16000  14.23716 Variance Decomposition of LCPI: Period S.E. LYID LCPI LIDX LM2  1  0.011933  0.243392  99.75661  0.000000  0.000000  4  0.021436  0.617897  89.49672  1.063527  8.821859  8  0.026075  0.532270  81.81934  7.796618  9.851772  12  0.028211  0.498317  80.21287  9.216902  10.07191  16  0.029307  0.484943  79.53119  9.811529  10.17234  20  0.029889  0.480958  79.17770  10.11605  10.22529 Variance Decomposition of LIDX: Period S.E. LYID LCPI LIDX LM2  1  0.102219  11.07300  0.969112  87.95788  0.000000  4  0.158817  10.75208  1.756993  86.14071  1.350218  8  0.163584  10.14723  6.854441  81.29229  1.706037  12  0.166177  9.887476  8.573339  79.48189  2.057295  16  0.167650  9.750625  9.547491  78.44705  2.254836  20  0.168525  9.676770  10.10122  77.84711  2.374901 Variance Decomposition of LM2: Period S.E. LYID LCPI LIDX LM2  1  0.016061  1.185146  3.778642  3.193534  91.84268  4  0.022952  8.787830  3.788296  11.86394  75.55994  8  0.030507  23.24790  2.674553  23.34257  50.73498  12  0.035148  28.72334  4.276190  23.89642  43.10405  16  0.039376  30.85945  6.187618  24.58667  38.36626  20  0.043133  31.54053  8.098200  25.07802  35.28326  Cholesky Ordering: LYID LCPI LIDX LM2  Table 6: Variance Decomposition for Credit Channel Variance Decomposition of LYID: Period S.E. LYID LCPI LCREDIT LM2  1  0.003032  100.0000  0.000000  0.000000  0.000000  4  0.005603  96.24726  0.073696  2.812844  0.866204  8  0.007446  93.77926  0.715893  3.237590  2.267258  12  0.008746  91.24958  1.879666  3.302944  3.567813  16  0.009796  88.73741  3.304074  3.292009  4.666509  20  0.010695  86.38227  4.787512  3.257821  5.572398 Variance Decomposition of LCPI: Period S.E. LYID LCPI LCREDIT LM2  1  0.010999  1.171561  98.82844  0.000000  0.000000  4  0.019236  2.935354  96.38552  0.027938  0.651186  8  0.023517  4.361958  93.91480  0.027221  1.696025  12  0.025444  4.995750  92.55445  0.037354  2.412442  16  0.026387  5.208688  91.90441  0.042826  2.844073  20  0.026867  5.239316  91.61553  0.044302  3.100857 Variance Decomposition of LCREDIT: Period S.E. LYID LCPI LCREDIT LM2  1  1.527640  0.008412  0.662962  99.32863  0.000000  4  1.591089  4.339904  0.696857  93.02595  1.937289  8  1.610827  6.134028  0.691420  90.92188  2.252676  12  1.617361  6.828907  0.687989  90.23384  2.249259  16  1.621150  7.231068  0.691117  89.83332  2.244498  20  1.624125  7.523314  0.703917  89.51827  2.254501 Variance Decomposition of LM2: Period S.E. LYID LCPI LCREDIT LM2  1  0.009537  5.528801  1.359330  5.628398  87.48347  4  0.015071  3.261917  6.164366  3.024729  87.54899  8  0.018557  10.20525  13.17978  2.115265  74.49970  12  0.021465  18.65382  17.57393  1.957739  61.81451  16  0.024014  25.19223  19.98210  1.983319  52.84235  20  0.026244  29.96330  21.32789  2.039472  46.66934  Cholesky Ordering: LYID LCPI LCREDIT LM2 Appendix B: Figures Figure 1: Transmission Mechanism of Monetary Policy  Figure 2: Stationarity of the VAR Model Note: No root lies outside the unit circle. Therefore, VAR satisfies the stability condition. Figure 3: Impulse Response Functions for Interest Rate Channel Figure 4: Impulse Response Functions for Exchange Rate Channel Figure 5: Impulse Response Functions for Asset Price Channel Figure 6: Impulse Response Functions for Credit Channel 21