Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The Monetary Transmission Mechanism in the New Economy: Evidence from Turkey (1997-2006)

The Monetary Transmission Mechanism in the New Economy: Evidence from Turkey (1997-2006) This study aimed to test the money base, money supply, credit capacity, industrial production index, interest rates, inflation and real exchange rate data of Turkey during the years 1997 ­ 2006. These were tested through the monetary transmission mechanism and passive money hypothesis, using the vector error correction model-based causality test. Empirical findings showed that the passive money supply hypothesis of the new Keynesian economy is supported in part by accommodationalist views and differs from those of structuralist and liquidity preference theories. However, the monetary transmission mechanism has established that long-term money supply only affects general price levels, while production is influenced by interest rates in the new period of the Turkish economy. Empirical findings show that in this new period, interest transmission mechanisms are at the forefront. JEL: E58, E52, E4, C32 DOI: 10.2478/v10033-007-0011-3 1. Introduction Though advocating similar theories in the use of monetary politics, the New Keynesian and the Monetarist Schools differed in their opinions on whether money is active or passive. While the Monetarist School defends the fact that monetary tools, i.e., money supply, is under the control of the Central Bank, the New Keynesian School argues that, as credit control is not tied to the Central Bank, it does not completely control money supply. Defenders of the New Keynesian School put forward the following evidence in support of these claims (Seyrek and others, 2004): (1) The statistical stochastic aspect in money data and the great errors that result from it determine that money is passive; (2) According to general econometric tests, money stock is passive; (3) The passivity of monetary stock derives from the macroeconomic character of the banking system; (4) The passivity of money stock can be explained with many macroeconomic variables. In addition to credit-money supply, whether money is active or passive is also based on the correlation between money, interest, inflation and productivity. During this process in the new economic period, exchange rates also have their place. The New Keynesian view describing the correlation between money, credit, interest, inflation and exchange rates can be tested through long-term analysis. The econometric methods in long-term analysis are a causality test based on the vector error correction model for cointegrated data or the Granger causality tests for non-cointegrated data. During this study, together with the vector error correction method and the Granger causality test, the monetary transmission mechanism and monetary passivity hypothesis were tested. The second section surveys pertinent literature, the third section outlines the methodology, the fourth section concerns empirical findings, and the fifth section presents the results. *Cifter: PhD Candidate, Econometrics, Marmara University, Istanbul, Turkey, and Financial Reporting Vice President, Deniz Invest-Dexia Group, Istanbul, Turkey, atillacifter@denizyatirim.com *Alper : PhD, Inspector, Market Risk Group, Is Bank, Istanbul, Turkey, alper.ozun@isbank.com.tr 2. Literature Review and Theory There are three main types of monetary transmission mechanism models found in the literature: the interest rate channel, the asset channel and the credit channel (Seyrek and others, 2004). According to the monetary transmission mechanism, money supply is active and, in the short term, monetary tools and increased money supply reduce interest rates. Hence the liquidity effect is only short-term. The drop in interest rates increases credit value. This situation causes a short-term increase in income. In the long term, the increased price in money supply increases its general level and the real value of money stock declines. According to the Monetarist approach, money supply is active during these processes and is controlled by the Central Bank. According to the Keynesian approach, monetary politics tools affect the monetary base first, then the money supply. Following this, the changes in money supply affect interest rates, which in turn affect investments and then revenues. New Keynesian economics argues that money supply is passive*. Rather than the Central Banks' exported money supply, credit money is determined according to the banks' credit preferences. When economic units use credit, deposits created by credit multiply. The passive money hypothesis presumes that causality moves away from credits towards deposits. Credit demands are set by the preferences of the credit applicants and creditor. For this reason, Central Banks do not have control over credits, and therefore, money stocks (Shanmugan and others, 2003). There are three approaches with regard to passive money stock; accommodationalist, structuralist and liquidity preference. According to the accommodationalists (Moree, 1989) credits are the source of money supply and money base, and that money supply and money revenue (GDP) are cointegrated and interdependent. According to the structuralists (Palley, 1996, 1998; Pollin 1991) credits are the source of money supply, money base and money multipliers and that money supply and money revenue (GDP) are cointegrated and interdependent. Finally, according to liquidity preference theorists (Howells, 1995), credits and money supply are cointegrated and interdependent. The monetary transmission mechanism is shown in Diagram No.1 and the New Keynesian Economical Passive Money Theory is shown in Diagram No. 2. In the new economic period, real exchange rates will also be distinct from general price levels. For the New Keynesian economy, the first empirical study on passive money was carried out by Pollin (1991). Pollin (1991), obtained data supporting structuralist views for the USA from 1953 ­ 1988. Vera (2001), obtained findings to support accommodationalist and structuralist views for Spain from 1987 ­ 1998 by applying Granger causality tests using Money Multipliers (according to M1, M2 andM3) and credit data. Nell (2000-01) examined the relationships between money supply, money circulation speed and credit using the vector error correction model for South Africa from 1966 ­ 1997 and found that all new Keynesian approaches (structuralist, accommodationalist and liquid preference theories) were empirically valid. LMB (Monetary Base) Li (Interest Rates) LL (Credits) Lexch (Real Exch. Rate) LIP (Industrial Prod.Ind) LUFE (Inflation) Graph 1. Monetary Transmission Mechanism LMB (Monetary Base) Pall e Poll y(199 in ( 199 6,199 8), 1) Mo ore (19 89 ) 1) 99 (1 in ll Po ), 98 6) 19 99 6, (1 9) 99 lls 98 y(1 we (1 lle Ho re Pa oo M Pa lley (19 Mo or e(1 96 ,19 98 ), P oll in 98 9) (19 91 ) LL (Credits) LIP (Industrial Prod.Ind) Graph 2. Endogenity of Money in New Keynesian Economy Shanmugan, Nair and Li (2003), examined the relationship between money base, money supply, credit and the industrial production index using the vector error correction model and Granger causality test in Malaysia from 1985 ­ 2000 and reached conclusions that support the findings of accommodationalists and liquid preference theorists. Lavoie (2005) tested the passivity of money according to theoretical and empirical literature for Canada and the USA, and reached conclusions that support accommodationalist views. Ahmad and Ahmet (2006) carried out short and long-term tests on the passivity of money supply for Pakistan from 1980 ­ 2003 using the Granger causality test. In the short term, they found that empirical findings supported structuralist and liquidity preference theory, but in the long term found that the money base * The critical evaluation of New Keynesian monetary politics. See Cottrell (1994). set the credit capacity and showed that the Pakistan Central Bank became active in setting money supply.Gunduz (2001) and Seyrek, Duman and Sarikaya (2004) carried out studies on Turkish data. Seyrek and others (2004) found that data for Turkey from 1968 ­ 1996 supported the Keynesian transmission mechanism multi-monetarist hypothesis driven by credit. Gunduz (2001) analysed the monthly macroeconomic data dependent VAR (Vector Autoregressive) model and the bank lending channel roles in Turkey. The findings for the period 1986 ­ 1998 show that the bank lending channel presented limited support for the transmission mechanism. 3. Data and Methodology 3.a.Data Monthly data was used between January 1997 ­ June 2006 for the monetary transmission mechanism and passive money supply test. Due to the fact that the Gross Domestic Product (GDP) was published every three months, the Production Index (PI) was used instead. Because the treasury bond interest rates indicator was not available on a monthly basis before 2002, the 12 month deposit interest rate was used instead. During analyses made for Turkey, IPI was used instead of GNP for national growth and production indicators and deposit interest rates were used instead of treasury bond interest rates. Money Base, Money Supply, Credit Capacity, Industrial Production Index, Interest Rates and Real Exchange Rates were obtained from www.tcmb.gov.tr and inflation rates from www. tuik.gov.tr. Money Base reserves and total Free Market Procedures (FMP) debts have been calculated by the authors. Table 1 shows the unit root tests for the chosen indicators. All series were proven (90%-100%) to contain unit roots. In order to separate the series from unit roots, logarithmic differences have been taken and it has been established that all series are stationary in terms of entry level logarithmic differences (Table 2). L R E MB M1 M2 M2Y M3 M3Y L Exc IP I UFE Augmented Dickey-Fuller Test* 1.35825 {<1.00} 1.70072 {<1.00} 1.59344 {<1.00} 1.80713 {<1.00} 1.08533 {<1.00} 1.02843 {<1.00} 0.86340 {<1.00} 1.31943 {<1.00} 1.58811 {<1.00} -2.6588 {< 1.00} -1.5450 {<0.90} -1.3675 {<0.90} 0.32292 {<0.99} Skewness 0.625191 0.855461 0.884356 0.863718 0.844484 0.395414 0.847275 0.422493 1.29503 0.398535 0.600842 0.178638 0.101448 Kurtosis 0.625191 2.56205 2.84869 2.61482 2.68744 1.99569 2.67047 2.02084 3.86689 2.33168 2.68881 1.864 1.40765 Jarque-Bera statistic 8.95215 14.8155 14.9684 14.8789 14.014 7.76173 14.1555 7.94562 35.4344 5.13939 7.31921 6.73618 12.2396 R: Reserve Money, E: Emission, MB: Monetary Base, L: Credit Capacity, Exc: Real exchange rate_MPI, IP: Industrial Production Index, i: Interest rate_12 Month, MPI: Manufacturer Price Index: * Lag lengths have been identified as 12 maximum according to Schwartz Knowledge Criteria. Values inside brackets are the rejected unit root statistics. a Lag length. Definitions: Reserve Money = Emission + Bank Mandatory Payments + Bank Unbound Opportunities + Fund Calculations + Non Bank Related Deposits Monetary Base = Reserve Money + Open Market Activity Debts Ml = Money in Circulation + Current Deposits at Depositary Banks + Central Bank Deposits M2 = Ml + Fixed Term Deposits at Depositary Banks M2Y = M2 + Foreign Currency Deposit Accounts (TL) 113 Table 1. Level Series, Unit Root Tests and Distribution Specifications LMB Mean Mode Max Min Std. Deviation Multiplier Oblateness J-B Probability Observations Table 2 0.0347 0.0377 0.3384 -0.2467 0.1026 -0.1429 4.0410 5.4877 0.0643 113 LM 2 0.0364 0.0315 0.1497 -0.0532 0.0347 0.5159 3.9321 9.1042 0.0105 113 LL 0.0336 0.0358 0.1531 -0.0772 0.0336 -0.3160 4.9116 19.088 0.0000 113 LIP 0.0045 0.0045 0.2238 -0.2209 0.0810 0.1102 3.7338 2.7642 0.2510 113 LI -0.0112 -0.0050 0.7186 -0.5579 0.1275 1.1458 16.609 896.78 0.0000 113 LEXC 0.0022 0.0053 0.1363 -0.1577 0.0394 -0.6521 6.7538 74.355 0.0000 113 LUFE 0.0275 0.0259 0.1341 -0.0228 0.0236 0.8737 5.7567 50.159 0.0000 113 Logarithmic Difference Series Fundamental Statistical Specifications 3.b.Methodology The vector error correction model-based causality test has been selected for the Passive Money Hypothesis test and the transmission mechanism, which in turn is derived from Money Base, Money Supply, Credit Capacity, Industrial Production Index, Interest Rates, Inflation and Real Exchange Rates. Before the vector error correction model is applied, it must be researched as to whether or not the series contain unit roots. In the literature, unit root-stability identification is generally made by using ADF (Augmented Dickey Fuller Test) and P-P (Philips-Perron) tests. The ADF test was developed by Dickey 0.150 0.125 0.100 0.075 0.050 0.025 0.000 -0.025 -0.050 -0.075 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 and Fuller (1981) and is used together with Equation No. 1: DYt = b1 + b 2t + dYt -1 + a i å DYt -i + e t i =1 m (1) Yt is the first difference in testing the stability of the variable, t the trend variable and is the lag difference term. The `i' lag difference term is added sufficiently for the error term to be a non-correlation series using knowledge criteria. LL X Lm2 Spline k=10.00 Graph 3 M2 and Credits Scatter Diagram (Log differenced series) LMB 0.10 0.00 2000 2005 0.1 0.0 -0.1 2000 2005 0.5 0.0 -0.5 2000 0.15 0.10 0.05 0.00 2000 2005 LUFE Lm2 2000 Lexc LL 2000 LI 0.2 0.0 -0.2 LIP Graph 4 Series Another main unit root test used in the literature is the "Phillips-Peron" (P-P) test developed by Phillips-Perron (1988). The P-P test can be applied using Equation No. 2 DYt = a + cYt -1 + d1DYt -1 + d 2 DYt - 2 + ........ + d p -1DYt - p -1 + e t ltrace ( r ) = -T i = r +1 å In(1 - l ), r = 0,1,2,3,...., n - 1 (4) (2) Yt is the primary difference of Y series, a,c,d1,d2,.....dp-1 the parameters, t is time, p the lag number and t shows error term. H0:c=0 shows that the series is not stationary, H1:c=/0 shows lmax( r ,r +1) = -TIn(1 - l r +1 ) that the series is stationary. Before examining the relationship of data that that is not stationary but at the same level, the series need to be examined to determine whether or not they are integrated. Johansen(1988), Johansen and Joselius (1990) developed the Johansen cointegration test, which is used widely in the literature. In the following model, a non-trend setting and non-restrictive cointegration test containing a stationary term has been preferred (3) H1* (r ) : Õ yt -1 + Bxt = a ( b ' yt -1 ) + r 0 (3) In the Johansen method the cointegration among non-stationary series are identified using trace and maximum eigenvalue statistics (4-5) (5) In the prepared model, if cointegration can be identified between dependent and independent variables, then it can be understood that there is at least one aspect of causality (Granger, 1969). If there is no cointegration between variables, the standard causality test (Granger, 1969) can be applied; and if there is cointegration between variables, then causality can be examined using the vector error correction model (VECM) (Granger, 1988). Engle and Granger (1987) developed the VECM, which is shown in the equation below (6). Dyt = a 0 + å a1i Dyt -i + å a 2i Dc t -i + å a 3 ECt - n + e i i =1 i =1 i =1 n n n (6) The short term causality relationship in the VECM can be tested using the significance of the parameters and the Wald test. The long-term causality relationship can be tested using the ECt-n parameter significance (Shanmugan and others, 2003). 4. Empirical Findings Table No.3 shows the ADF and P-P unit root test results of the logarithmic difference series. All series are stationary to a 99% level of significance, ADF Test t-statistic -16.0154 {<0.01} -4.11103 {<0.01} -4.32343 {<0.01} -6.99893 {<0.01} -8.85429 {<0.01} -5.17817 {<0.01} -4.14952 {<0.01} LM2& LI Ho r=0 Trace Stat 28.5057 {<0.01}* Max Stat 16.3041 {<0.05} 12.2016 {<0.025} r<=1 12.2016 {<0.025} LM2&LIP 4 r=0 55.9499 {<0.01}* 43.3068 {<0.01}* Variables LMB LM 2 LL P-P Test t-statistic -16.5017 {<0.01} -10.6736 {<0.01} -9.62207 {<0.01} -6.85118 {<0.01} -16.5786 {<0.01} -12.6571 {<0.01} -4.27729 {<0.01} r<=1 12.6431 {<0.025} 12.6431 {<0.025} LM2&LUFE 4 r=0 r<=1 LM2&LExc 4 r=0 r<=1 ML2&LL 4 r=0 r<=1 LI&LIP 4 r=0 r<=1 LIP&LUFE 4 r=0 r<=1 LUFE&LExc 4 r=0 r<=1 LExc&LL 4 r=0 r<=1 26.8229 {<0.01}* 21.0262 {<0.01}* 5.79668 {<0.5} 5.79668 {<0.5} 45.6645 {<0.01}* 31.1411 {<0.01}* 14.5233 {<0.01} 25.0972 {<0.01}* 10.1851 {<0.05} 14.5233 {<0.01} 14.9121 {<0.1} 10.1851 {<0.05} Lexc LIP 75.0987 {<0.01}* 57.0977 {<0.01}* 18.001 {<0.01} 18.001 {<0.01} LUFE 63.2395 {<0.01}* 56.6974 {<0.01}* 6.54211 {<0.2} 6.54211 {<0.2} MB: Monetary Base, L: Credit Capacity, Exc: Real Exchange Rate_MPI, IP: Industrial Production Index, i: Interest rate_12 Month, MPI: Manufacturer Price Index * D Lags have been identified as 12 maximum according to Schwartz Knowledge Criteria. Values inside brackets are the rejected unit root statistics. a Lag length. 38.3893 {<0.01}* 32.3643 {<0.01}* 6.02494 {<0.2} 6.02494 {<0.2} Table 3 ADF and P-P Unit Root Tests (Logarithmic difference has been taken)* 40.4699 {<0.01}* 30.6964 {<0.01}* 9.77352 {<0.05} 9.77352 {<0.05} The unrestrictive Johansen cointegration tests demonstrating the passive money hypothesis and the monetary transmission mechanism test can be found in Table Nos. 4 and 5. All series are cointegrated at a secure level of 95-99%. Due to the fact that the series are all cointegrated, the vector error correction model-based causality test has been applied to all hypotheses. LMB& LL Values inside brackets are significance values. Lags have been identified as 12 maximum according to Schwartz Knowledge Criteria. * Hypothesis of H0 is rejected at %1 significance. a Lag length. Table 5. Unrestricted Johansen Cointegration Test (Monetary Transmission Mechanism) Ho r=0 r<=1 Trace Stat Max Stat 45.0642 {<0.01}* 34.9867 {<0.01}* 10.0775 {<0.05} 25.0972 {<0.01}* 10.1851 {<0.05} 10.0775 {<0.05} 14.9121 {<0.1} 10.1851 {<0.05} LM2&LL r=0 r<=1 LM2& LIP r=0 55.9499 {<0.01}* 43.3068 {<0.01}* The causality between credit-monetary base, credit-monetary base-IP, credit-money supply and credit-money supply-IP for the passive money test was examined using the vector error correction model (Table No.6). The results show that there is causality towards credit=>Monetary Base and Credit=>Money Supply. This situation supports in part the views of the accommodationalists in the new Keynesian approach (this is supported completely because there was no Money Supply=>IP causality found). Table No.8 shows the monetary transmission r<=1 12.6431 {<0.025} 12.6431 {<0.025} LMS&LIP 4 r=0 r<=1 r=0 LMB&LL&LIP 4 r<=1 r<=2 r=0 LM2&LL&LIP 4 57.3502 {<0.01}* 49.0907 {<0.01}* 8.25947 {<0.1} 93.6593 {<0.01}* 8.25947 {<0.1} 53.529 {<0.01}* Li (Interest Rates) -0.17 40.1302 {<0.01}* 30.7801 {<0.01}* 9.35016 {<0.05} 9.35016 {<0.05} 63.5928 {<0.01}* 40.9484 {<0.01}* 13.2242 {<0.2} 9.42012 {<0.05} LL (Credits) Lexch (Real Exch. Rate) r<=1 22.6443 {<0.025} r<=2 9.42012 {<0.05} -0 LIP (Industrial Prod.Ind) -0.32 Values inside brackets are significance values. Lags have been identified as 12 maximum according to Schwartz Knowledge Criteria. * Hypothesis of H0 is rejected at %1 significance. a Lag length. -0,96 LUFE (Inflation) Table 4 Unrestricted Johansen Cointegration Test (Endogeneity of Money Hypothesis) Graph 5. Monetary Transmission Mechanism (Turkey) Short-term Effect Wald test: DEPANDENT Var:LMB 8.28649 LL [0.0040] * 8.77713 LL LIP [0.0124] ** Long-term Effects ECt-1 VECM Short-term Long-term DEPANDENT Var:LM2 Short-term Effect Wald test: Long-term Effects ECt-1 VECM Short-term Long-term 0.853448 LL=>LMB [0.005]* 0.861413 LL=>LMB [0.006]* 0.192875 LL=>LMB [0.457] LL,LIP =>LMB LL=>LMB 7.31782 L [0.0068] * DEPANDENT Var:LI 0.276316 L=> Lm2 [0.008]* L=> Lm2 0.0034883 0.0426212 LM 2 [0.9529] DEPANDENT Var:LIP 2.31045 LM 2 [0.1285] [0.131] -0.366596 LM2= > LIP LM2= > LIP [0.953] LM2= > LI LM2= > LI DEPANDENT Var:LM2 4.25987 LL [0.0390] * 3.93154 LL LIP [0.318] DEPANDENT Var:LL 0.0020319 0.0099965 LMB [0.9640] 2.42909 LM 2 [0.1191] 1.8055 LL LMB LIP [0.211] 3.14705 LL 2 LM LIP [0.602] DEPANDENT Var:LIP 2.31045 Lm2 [0.1285] [0.131] -0.366596 LM2=>LIP LM2=>LIP [0.2073] 0.511034 LM2=>LL [0.081]** 0.120232 LIP=>LL LM2,LIP =>LL [0.4055] [0.122] 0.0057303 [0.981] 0.391690 LIP=>LL LMB=>LL LMB,LIP =>LL [0.964] 0.432127 LM2=>LL LM2=>LL LMB=>LL LMB=>LL [0.1400] [0.041]* 0.419147 LL=>LM2 [0.097]** 0.244108 LL=>LM2 LL,LIP =>LMB 0.462158 LL= > Lm2 LL= > Lm2 DEPANDENT Var:LUFE 17.9812 LM 2 [0.0000] ** DEPANDENT Var:LM2 0.0030329 -0.0135914 Lexc [0.9561] DEPANDENT Var:LExc 1.12743 LM 2 [0.2883] DEPANDENT Var:LM2 4.25987 LL [0.0390] ** DEPANDENT Var:LL 2.42909 Lm2 [0.1191] DEPANDENT Var:LIP 4.76484 LI [0.0290] ** DEPANDENT Var:LIP 0.606543 LUFE [0.4361] DEPANDENT Var:LUFE 5.00403 -0.967069 [ 0.027]* Lexc [0.0253] * DEPANDENT Var:LL 0.929813 Lexc [0.3349] DEPANDENT Var:LUFEL 4.33088 LI [0.0374] ** 0.390938 [0.040]* L=> LUFE L=> LUFE -0.281147 [0.337] LExc=> LL LExc=> LL Lexc=> LUFE Lexc=> LUFE [0.438] -0.167911 LUFE=> LIP LUFE=> LIP [0.031]* -0.174685 L=> LIP L=> LIP [0.122] 0.432127 LM2=>LL LM2=>LL [0.041]* 0.462158 LL=>LM2 LL=>LM2 [0.291] -0.248525 LM2=> Lexc LM2=> Lexc [0.956] LExc=> LM 2 LExc=> Lm2 [0.000]* 1.03013 LM2=> LUFE LM2=> LUFE * %1, ** %5 significant level of acceptance respectively. Values inside brackets are t-stats. Lag length is determined as 4. Table 6 . Causality Tests Based on Vector Error Correction Model Endogeneity of Money mechanism vector error correction model test. According to Table No.8, long-term causalities can be found in Diagram No.5. Eight causality directions were identified: Credits=>Money Supply, Interest Rates=>Money Supply, Interest Rates=>Real Exchange Rates (negative), Interest Rates=>Inflation, Interest Rates=>IP (negative), Money Supply=>Inflation, Real Exchange Rates=>Inflation, Inflation=>IP (negative). These results show that money supply is the cause of inflation in the long term (influence factor 1.03), that credits affect money supply (influ- * %1, ** %5 significance level of acceptance respectively. Values inside brackets are t-stats. Lag length is determined as 4. Table 7. Causality Tests Based on Vector Error Correction ModelMonetary Transmission Mechanism ence factor 0.46), that money supply does not affect inflation rates but that interest rates affect money supply (influence factor 0.27) and that real exchange rates affect inflation in a negative and dominant way (influence factor -0.96). Also, it has been found that IP is affected by interest rates but not affected by money supply. This situation conforms neither to the monetary school nor the new Keynesian school views. The Central Bank's choice of interest rates as the main indicator and means of identifying net internal assets after the 2001 crisis is one reason for this situation. Another reason is that in the new economic period factors influenced the real economic activity through credits (consumer credits, business credits and credit cards) and interest rates. Diagram No.6 shows the difference in correlation between money supply and IP and Diagram No.7 shows the difference in correlation between money supply and credits. Because correlation is also under the influence of cyclic effects, causality was tested with the vector error correction model. 0.25 0.2 0.15 0.1 0.05 0.00 -0.05 Correl(LM2,LIP) LIP Poly. (Correl(LM2,LIP)) Poly. (LIP) -0.2 -0.4 -0.6 -0.8 Graph 6. Dynamic Correlation (LM2, LIP, 4 Lags) -0.1 -0.15 -0.2 -0.25 Correl(LM2,LL) Poly. (Correl(LM2,LL)) -0.2 -0.4 -0.6 -0.8 Graph 7. Dynamic Correlation (LM2, LL, 4 Lags) 5. Concluding Remarks This study was conducted to test the money base, money supply, credit capacity, industrial production index, interest rates, inflation and real exchange rate data of Turkey during the years 1997 ­ 2006. These were tested through the monetary transmission mechanism and passive money hypothesis using the vector error correction model-based causality test. Empirical findings show that the passive money supply hypothesis of the new Keynesian economy is supported in part by accommodationalist views, and do not conform to the structuralist and liquidity preference theories. However, according to the monetary transmission mechanism, it has been established that long-term money supply only affects general price levels, and that production is influenced by interest rates in the new economic period. Empirical findings show that in the new economy, period interest transmission mechanisms are brought to the forefront. During the monetary transmission mechanism test, it was decided to leave in theforefront. During the monetary transmission mechanism test, it was decided to leave in the Markov regime variant, which takes into account cyclic effects, a vector error correction model proposed for future studies http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png South East European Journal of Economics and Business de Gruyter

The Monetary Transmission Mechanism in the New Economy: Evidence from Turkey (1997-2006)

Loading next page...
 
/lp/de-gruyter/the-monetary-transmission-mechanism-in-the-new-economy-evidence-from-l4ZYjLPo07
Publisher
de Gruyter
Copyright
Copyright © 2007 by the
ISSN
1840-118X
DOI
10.2478/v10033-007-0011-3
Publisher site
See Article on Publisher Site

Abstract

This study aimed to test the money base, money supply, credit capacity, industrial production index, interest rates, inflation and real exchange rate data of Turkey during the years 1997 ­ 2006. These were tested through the monetary transmission mechanism and passive money hypothesis, using the vector error correction model-based causality test. Empirical findings showed that the passive money supply hypothesis of the new Keynesian economy is supported in part by accommodationalist views and differs from those of structuralist and liquidity preference theories. However, the monetary transmission mechanism has established that long-term money supply only affects general price levels, while production is influenced by interest rates in the new period of the Turkish economy. Empirical findings show that in this new period, interest transmission mechanisms are at the forefront. JEL: E58, E52, E4, C32 DOI: 10.2478/v10033-007-0011-3 1. Introduction Though advocating similar theories in the use of monetary politics, the New Keynesian and the Monetarist Schools differed in their opinions on whether money is active or passive. While the Monetarist School defends the fact that monetary tools, i.e., money supply, is under the control of the Central Bank, the New Keynesian School argues that, as credit control is not tied to the Central Bank, it does not completely control money supply. Defenders of the New Keynesian School put forward the following evidence in support of these claims (Seyrek and others, 2004): (1) The statistical stochastic aspect in money data and the great errors that result from it determine that money is passive; (2) According to general econometric tests, money stock is passive; (3) The passivity of monetary stock derives from the macroeconomic character of the banking system; (4) The passivity of money stock can be explained with many macroeconomic variables. In addition to credit-money supply, whether money is active or passive is also based on the correlation between money, interest, inflation and productivity. During this process in the new economic period, exchange rates also have their place. The New Keynesian view describing the correlation between money, credit, interest, inflation and exchange rates can be tested through long-term analysis. The econometric methods in long-term analysis are a causality test based on the vector error correction model for cointegrated data or the Granger causality tests for non-cointegrated data. During this study, together with the vector error correction method and the Granger causality test, the monetary transmission mechanism and monetary passivity hypothesis were tested. The second section surveys pertinent literature, the third section outlines the methodology, the fourth section concerns empirical findings, and the fifth section presents the results. *Cifter: PhD Candidate, Econometrics, Marmara University, Istanbul, Turkey, and Financial Reporting Vice President, Deniz Invest-Dexia Group, Istanbul, Turkey, atillacifter@denizyatirim.com *Alper : PhD, Inspector, Market Risk Group, Is Bank, Istanbul, Turkey, alper.ozun@isbank.com.tr 2. Literature Review and Theory There are three main types of monetary transmission mechanism models found in the literature: the interest rate channel, the asset channel and the credit channel (Seyrek and others, 2004). According to the monetary transmission mechanism, money supply is active and, in the short term, monetary tools and increased money supply reduce interest rates. Hence the liquidity effect is only short-term. The drop in interest rates increases credit value. This situation causes a short-term increase in income. In the long term, the increased price in money supply increases its general level and the real value of money stock declines. According to the Monetarist approach, money supply is active during these processes and is controlled by the Central Bank. According to the Keynesian approach, monetary politics tools affect the monetary base first, then the money supply. Following this, the changes in money supply affect interest rates, which in turn affect investments and then revenues. New Keynesian economics argues that money supply is passive*. Rather than the Central Banks' exported money supply, credit money is determined according to the banks' credit preferences. When economic units use credit, deposits created by credit multiply. The passive money hypothesis presumes that causality moves away from credits towards deposits. Credit demands are set by the preferences of the credit applicants and creditor. For this reason, Central Banks do not have control over credits, and therefore, money stocks (Shanmugan and others, 2003). There are three approaches with regard to passive money stock; accommodationalist, structuralist and liquidity preference. According to the accommodationalists (Moree, 1989) credits are the source of money supply and money base, and that money supply and money revenue (GDP) are cointegrated and interdependent. According to the structuralists (Palley, 1996, 1998; Pollin 1991) credits are the source of money supply, money base and money multipliers and that money supply and money revenue (GDP) are cointegrated and interdependent. Finally, according to liquidity preference theorists (Howells, 1995), credits and money supply are cointegrated and interdependent. The monetary transmission mechanism is shown in Diagram No.1 and the New Keynesian Economical Passive Money Theory is shown in Diagram No. 2. In the new economic period, real exchange rates will also be distinct from general price levels. For the New Keynesian economy, the first empirical study on passive money was carried out by Pollin (1991). Pollin (1991), obtained data supporting structuralist views for the USA from 1953 ­ 1988. Vera (2001), obtained findings to support accommodationalist and structuralist views for Spain from 1987 ­ 1998 by applying Granger causality tests using Money Multipliers (according to M1, M2 andM3) and credit data. Nell (2000-01) examined the relationships between money supply, money circulation speed and credit using the vector error correction model for South Africa from 1966 ­ 1997 and found that all new Keynesian approaches (structuralist, accommodationalist and liquid preference theories) were empirically valid. LMB (Monetary Base) Li (Interest Rates) LL (Credits) Lexch (Real Exch. Rate) LIP (Industrial Prod.Ind) LUFE (Inflation) Graph 1. Monetary Transmission Mechanism LMB (Monetary Base) Pall e Poll y(199 in ( 199 6,199 8), 1) Mo ore (19 89 ) 1) 99 (1 in ll Po ), 98 6) 19 99 6, (1 9) 99 lls 98 y(1 we (1 lle Ho re Pa oo M Pa lley (19 Mo or e(1 96 ,19 98 ), P oll in 98 9) (19 91 ) LL (Credits) LIP (Industrial Prod.Ind) Graph 2. Endogenity of Money in New Keynesian Economy Shanmugan, Nair and Li (2003), examined the relationship between money base, money supply, credit and the industrial production index using the vector error correction model and Granger causality test in Malaysia from 1985 ­ 2000 and reached conclusions that support the findings of accommodationalists and liquid preference theorists. Lavoie (2005) tested the passivity of money according to theoretical and empirical literature for Canada and the USA, and reached conclusions that support accommodationalist views. Ahmad and Ahmet (2006) carried out short and long-term tests on the passivity of money supply for Pakistan from 1980 ­ 2003 using the Granger causality test. In the short term, they found that empirical findings supported structuralist and liquidity preference theory, but in the long term found that the money base * The critical evaluation of New Keynesian monetary politics. See Cottrell (1994). set the credit capacity and showed that the Pakistan Central Bank became active in setting money supply.Gunduz (2001) and Seyrek, Duman and Sarikaya (2004) carried out studies on Turkish data. Seyrek and others (2004) found that data for Turkey from 1968 ­ 1996 supported the Keynesian transmission mechanism multi-monetarist hypothesis driven by credit. Gunduz (2001) analysed the monthly macroeconomic data dependent VAR (Vector Autoregressive) model and the bank lending channel roles in Turkey. The findings for the period 1986 ­ 1998 show that the bank lending channel presented limited support for the transmission mechanism. 3. Data and Methodology 3.a.Data Monthly data was used between January 1997 ­ June 2006 for the monetary transmission mechanism and passive money supply test. Due to the fact that the Gross Domestic Product (GDP) was published every three months, the Production Index (PI) was used instead. Because the treasury bond interest rates indicator was not available on a monthly basis before 2002, the 12 month deposit interest rate was used instead. During analyses made for Turkey, IPI was used instead of GNP for national growth and production indicators and deposit interest rates were used instead of treasury bond interest rates. Money Base, Money Supply, Credit Capacity, Industrial Production Index, Interest Rates and Real Exchange Rates were obtained from www.tcmb.gov.tr and inflation rates from www. tuik.gov.tr. Money Base reserves and total Free Market Procedures (FMP) debts have been calculated by the authors. Table 1 shows the unit root tests for the chosen indicators. All series were proven (90%-100%) to contain unit roots. In order to separate the series from unit roots, logarithmic differences have been taken and it has been established that all series are stationary in terms of entry level logarithmic differences (Table 2). L R E MB M1 M2 M2Y M3 M3Y L Exc IP I UFE Augmented Dickey-Fuller Test* 1.35825 {<1.00} 1.70072 {<1.00} 1.59344 {<1.00} 1.80713 {<1.00} 1.08533 {<1.00} 1.02843 {<1.00} 0.86340 {<1.00} 1.31943 {<1.00} 1.58811 {<1.00} -2.6588 {< 1.00} -1.5450 {<0.90} -1.3675 {<0.90} 0.32292 {<0.99} Skewness 0.625191 0.855461 0.884356 0.863718 0.844484 0.395414 0.847275 0.422493 1.29503 0.398535 0.600842 0.178638 0.101448 Kurtosis 0.625191 2.56205 2.84869 2.61482 2.68744 1.99569 2.67047 2.02084 3.86689 2.33168 2.68881 1.864 1.40765 Jarque-Bera statistic 8.95215 14.8155 14.9684 14.8789 14.014 7.76173 14.1555 7.94562 35.4344 5.13939 7.31921 6.73618 12.2396 R: Reserve Money, E: Emission, MB: Monetary Base, L: Credit Capacity, Exc: Real exchange rate_MPI, IP: Industrial Production Index, i: Interest rate_12 Month, MPI: Manufacturer Price Index: * Lag lengths have been identified as 12 maximum according to Schwartz Knowledge Criteria. Values inside brackets are the rejected unit root statistics. a Lag length. Definitions: Reserve Money = Emission + Bank Mandatory Payments + Bank Unbound Opportunities + Fund Calculations + Non Bank Related Deposits Monetary Base = Reserve Money + Open Market Activity Debts Ml = Money in Circulation + Current Deposits at Depositary Banks + Central Bank Deposits M2 = Ml + Fixed Term Deposits at Depositary Banks M2Y = M2 + Foreign Currency Deposit Accounts (TL) 113 Table 1. Level Series, Unit Root Tests and Distribution Specifications LMB Mean Mode Max Min Std. Deviation Multiplier Oblateness J-B Probability Observations Table 2 0.0347 0.0377 0.3384 -0.2467 0.1026 -0.1429 4.0410 5.4877 0.0643 113 LM 2 0.0364 0.0315 0.1497 -0.0532 0.0347 0.5159 3.9321 9.1042 0.0105 113 LL 0.0336 0.0358 0.1531 -0.0772 0.0336 -0.3160 4.9116 19.088 0.0000 113 LIP 0.0045 0.0045 0.2238 -0.2209 0.0810 0.1102 3.7338 2.7642 0.2510 113 LI -0.0112 -0.0050 0.7186 -0.5579 0.1275 1.1458 16.609 896.78 0.0000 113 LEXC 0.0022 0.0053 0.1363 -0.1577 0.0394 -0.6521 6.7538 74.355 0.0000 113 LUFE 0.0275 0.0259 0.1341 -0.0228 0.0236 0.8737 5.7567 50.159 0.0000 113 Logarithmic Difference Series Fundamental Statistical Specifications 3.b.Methodology The vector error correction model-based causality test has been selected for the Passive Money Hypothesis test and the transmission mechanism, which in turn is derived from Money Base, Money Supply, Credit Capacity, Industrial Production Index, Interest Rates, Inflation and Real Exchange Rates. Before the vector error correction model is applied, it must be researched as to whether or not the series contain unit roots. In the literature, unit root-stability identification is generally made by using ADF (Augmented Dickey Fuller Test) and P-P (Philips-Perron) tests. The ADF test was developed by Dickey 0.150 0.125 0.100 0.075 0.050 0.025 0.000 -0.025 -0.050 -0.075 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 and Fuller (1981) and is used together with Equation No. 1: DYt = b1 + b 2t + dYt -1 + a i å DYt -i + e t i =1 m (1) Yt is the first difference in testing the stability of the variable, t the trend variable and is the lag difference term. The `i' lag difference term is added sufficiently for the error term to be a non-correlation series using knowledge criteria. LL X Lm2 Spline k=10.00 Graph 3 M2 and Credits Scatter Diagram (Log differenced series) LMB 0.10 0.00 2000 2005 0.1 0.0 -0.1 2000 2005 0.5 0.0 -0.5 2000 0.15 0.10 0.05 0.00 2000 2005 LUFE Lm2 2000 Lexc LL 2000 LI 0.2 0.0 -0.2 LIP Graph 4 Series Another main unit root test used in the literature is the "Phillips-Peron" (P-P) test developed by Phillips-Perron (1988). The P-P test can be applied using Equation No. 2 DYt = a + cYt -1 + d1DYt -1 + d 2 DYt - 2 + ........ + d p -1DYt - p -1 + e t ltrace ( r ) = -T i = r +1 å In(1 - l ), r = 0,1,2,3,...., n - 1 (4) (2) Yt is the primary difference of Y series, a,c,d1,d2,.....dp-1 the parameters, t is time, p the lag number and t shows error term. H0:c=0 shows that the series is not stationary, H1:c=/0 shows lmax( r ,r +1) = -TIn(1 - l r +1 ) that the series is stationary. Before examining the relationship of data that that is not stationary but at the same level, the series need to be examined to determine whether or not they are integrated. Johansen(1988), Johansen and Joselius (1990) developed the Johansen cointegration test, which is used widely in the literature. In the following model, a non-trend setting and non-restrictive cointegration test containing a stationary term has been preferred (3) H1* (r ) : Õ yt -1 + Bxt = a ( b ' yt -1 ) + r 0 (3) In the Johansen method the cointegration among non-stationary series are identified using trace and maximum eigenvalue statistics (4-5) (5) In the prepared model, if cointegration can be identified between dependent and independent variables, then it can be understood that there is at least one aspect of causality (Granger, 1969). If there is no cointegration between variables, the standard causality test (Granger, 1969) can be applied; and if there is cointegration between variables, then causality can be examined using the vector error correction model (VECM) (Granger, 1988). Engle and Granger (1987) developed the VECM, which is shown in the equation below (6). Dyt = a 0 + å a1i Dyt -i + å a 2i Dc t -i + å a 3 ECt - n + e i i =1 i =1 i =1 n n n (6) The short term causality relationship in the VECM can be tested using the significance of the parameters and the Wald test. The long-term causality relationship can be tested using the ECt-n parameter significance (Shanmugan and others, 2003). 4. Empirical Findings Table No.3 shows the ADF and P-P unit root test results of the logarithmic difference series. All series are stationary to a 99% level of significance, ADF Test t-statistic -16.0154 {<0.01} -4.11103 {<0.01} -4.32343 {<0.01} -6.99893 {<0.01} -8.85429 {<0.01} -5.17817 {<0.01} -4.14952 {<0.01} LM2& LI Ho r=0 Trace Stat 28.5057 {<0.01}* Max Stat 16.3041 {<0.05} 12.2016 {<0.025} r<=1 12.2016 {<0.025} LM2&LIP 4 r=0 55.9499 {<0.01}* 43.3068 {<0.01}* Variables LMB LM 2 LL P-P Test t-statistic -16.5017 {<0.01} -10.6736 {<0.01} -9.62207 {<0.01} -6.85118 {<0.01} -16.5786 {<0.01} -12.6571 {<0.01} -4.27729 {<0.01} r<=1 12.6431 {<0.025} 12.6431 {<0.025} LM2&LUFE 4 r=0 r<=1 LM2&LExc 4 r=0 r<=1 ML2&LL 4 r=0 r<=1 LI&LIP 4 r=0 r<=1 LIP&LUFE 4 r=0 r<=1 LUFE&LExc 4 r=0 r<=1 LExc&LL 4 r=0 r<=1 26.8229 {<0.01}* 21.0262 {<0.01}* 5.79668 {<0.5} 5.79668 {<0.5} 45.6645 {<0.01}* 31.1411 {<0.01}* 14.5233 {<0.01} 25.0972 {<0.01}* 10.1851 {<0.05} 14.5233 {<0.01} 14.9121 {<0.1} 10.1851 {<0.05} Lexc LIP 75.0987 {<0.01}* 57.0977 {<0.01}* 18.001 {<0.01} 18.001 {<0.01} LUFE 63.2395 {<0.01}* 56.6974 {<0.01}* 6.54211 {<0.2} 6.54211 {<0.2} MB: Monetary Base, L: Credit Capacity, Exc: Real Exchange Rate_MPI, IP: Industrial Production Index, i: Interest rate_12 Month, MPI: Manufacturer Price Index * D Lags have been identified as 12 maximum according to Schwartz Knowledge Criteria. Values inside brackets are the rejected unit root statistics. a Lag length. 38.3893 {<0.01}* 32.3643 {<0.01}* 6.02494 {<0.2} 6.02494 {<0.2} Table 3 ADF and P-P Unit Root Tests (Logarithmic difference has been taken)* 40.4699 {<0.01}* 30.6964 {<0.01}* 9.77352 {<0.05} 9.77352 {<0.05} The unrestrictive Johansen cointegration tests demonstrating the passive money hypothesis and the monetary transmission mechanism test can be found in Table Nos. 4 and 5. All series are cointegrated at a secure level of 95-99%. Due to the fact that the series are all cointegrated, the vector error correction model-based causality test has been applied to all hypotheses. LMB& LL Values inside brackets are significance values. Lags have been identified as 12 maximum according to Schwartz Knowledge Criteria. * Hypothesis of H0 is rejected at %1 significance. a Lag length. Table 5. Unrestricted Johansen Cointegration Test (Monetary Transmission Mechanism) Ho r=0 r<=1 Trace Stat Max Stat 45.0642 {<0.01}* 34.9867 {<0.01}* 10.0775 {<0.05} 25.0972 {<0.01}* 10.1851 {<0.05} 10.0775 {<0.05} 14.9121 {<0.1} 10.1851 {<0.05} LM2&LL r=0 r<=1 LM2& LIP r=0 55.9499 {<0.01}* 43.3068 {<0.01}* The causality between credit-monetary base, credit-monetary base-IP, credit-money supply and credit-money supply-IP for the passive money test was examined using the vector error correction model (Table No.6). The results show that there is causality towards credit=>Monetary Base and Credit=>Money Supply. This situation supports in part the views of the accommodationalists in the new Keynesian approach (this is supported completely because there was no Money Supply=>IP causality found). Table No.8 shows the monetary transmission r<=1 12.6431 {<0.025} 12.6431 {<0.025} LMS&LIP 4 r=0 r<=1 r=0 LMB&LL&LIP 4 r<=1 r<=2 r=0 LM2&LL&LIP 4 57.3502 {<0.01}* 49.0907 {<0.01}* 8.25947 {<0.1} 93.6593 {<0.01}* 8.25947 {<0.1} 53.529 {<0.01}* Li (Interest Rates) -0.17 40.1302 {<0.01}* 30.7801 {<0.01}* 9.35016 {<0.05} 9.35016 {<0.05} 63.5928 {<0.01}* 40.9484 {<0.01}* 13.2242 {<0.2} 9.42012 {<0.05} LL (Credits) Lexch (Real Exch. Rate) r<=1 22.6443 {<0.025} r<=2 9.42012 {<0.05} -0 LIP (Industrial Prod.Ind) -0.32 Values inside brackets are significance values. Lags have been identified as 12 maximum according to Schwartz Knowledge Criteria. * Hypothesis of H0 is rejected at %1 significance. a Lag length. -0,96 LUFE (Inflation) Table 4 Unrestricted Johansen Cointegration Test (Endogeneity of Money Hypothesis) Graph 5. Monetary Transmission Mechanism (Turkey) Short-term Effect Wald test: DEPANDENT Var:LMB 8.28649 LL [0.0040] * 8.77713 LL LIP [0.0124] ** Long-term Effects ECt-1 VECM Short-term Long-term DEPANDENT Var:LM2 Short-term Effect Wald test: Long-term Effects ECt-1 VECM Short-term Long-term 0.853448 LL=>LMB [0.005]* 0.861413 LL=>LMB [0.006]* 0.192875 LL=>LMB [0.457] LL,LIP =>LMB LL=>LMB 7.31782 L [0.0068] * DEPANDENT Var:LI 0.276316 L=> Lm2 [0.008]* L=> Lm2 0.0034883 0.0426212 LM 2 [0.9529] DEPANDENT Var:LIP 2.31045 LM 2 [0.1285] [0.131] -0.366596 LM2= > LIP LM2= > LIP [0.953] LM2= > LI LM2= > LI DEPANDENT Var:LM2 4.25987 LL [0.0390] * 3.93154 LL LIP [0.318] DEPANDENT Var:LL 0.0020319 0.0099965 LMB [0.9640] 2.42909 LM 2 [0.1191] 1.8055 LL LMB LIP [0.211] 3.14705 LL 2 LM LIP [0.602] DEPANDENT Var:LIP 2.31045 Lm2 [0.1285] [0.131] -0.366596 LM2=>LIP LM2=>LIP [0.2073] 0.511034 LM2=>LL [0.081]** 0.120232 LIP=>LL LM2,LIP =>LL [0.4055] [0.122] 0.0057303 [0.981] 0.391690 LIP=>LL LMB=>LL LMB,LIP =>LL [0.964] 0.432127 LM2=>LL LM2=>LL LMB=>LL LMB=>LL [0.1400] [0.041]* 0.419147 LL=>LM2 [0.097]** 0.244108 LL=>LM2 LL,LIP =>LMB 0.462158 LL= > Lm2 LL= > Lm2 DEPANDENT Var:LUFE 17.9812 LM 2 [0.0000] ** DEPANDENT Var:LM2 0.0030329 -0.0135914 Lexc [0.9561] DEPANDENT Var:LExc 1.12743 LM 2 [0.2883] DEPANDENT Var:LM2 4.25987 LL [0.0390] ** DEPANDENT Var:LL 2.42909 Lm2 [0.1191] DEPANDENT Var:LIP 4.76484 LI [0.0290] ** DEPANDENT Var:LIP 0.606543 LUFE [0.4361] DEPANDENT Var:LUFE 5.00403 -0.967069 [ 0.027]* Lexc [0.0253] * DEPANDENT Var:LL 0.929813 Lexc [0.3349] DEPANDENT Var:LUFEL 4.33088 LI [0.0374] ** 0.390938 [0.040]* L=> LUFE L=> LUFE -0.281147 [0.337] LExc=> LL LExc=> LL Lexc=> LUFE Lexc=> LUFE [0.438] -0.167911 LUFE=> LIP LUFE=> LIP [0.031]* -0.174685 L=> LIP L=> LIP [0.122] 0.432127 LM2=>LL LM2=>LL [0.041]* 0.462158 LL=>LM2 LL=>LM2 [0.291] -0.248525 LM2=> Lexc LM2=> Lexc [0.956] LExc=> LM 2 LExc=> Lm2 [0.000]* 1.03013 LM2=> LUFE LM2=> LUFE * %1, ** %5 significant level of acceptance respectively. Values inside brackets are t-stats. Lag length is determined as 4. Table 6 . Causality Tests Based on Vector Error Correction Model Endogeneity of Money mechanism vector error correction model test. According to Table No.8, long-term causalities can be found in Diagram No.5. Eight causality directions were identified: Credits=>Money Supply, Interest Rates=>Money Supply, Interest Rates=>Real Exchange Rates (negative), Interest Rates=>Inflation, Interest Rates=>IP (negative), Money Supply=>Inflation, Real Exchange Rates=>Inflation, Inflation=>IP (negative). These results show that money supply is the cause of inflation in the long term (influence factor 1.03), that credits affect money supply (influ- * %1, ** %5 significance level of acceptance respectively. Values inside brackets are t-stats. Lag length is determined as 4. Table 7. Causality Tests Based on Vector Error Correction ModelMonetary Transmission Mechanism ence factor 0.46), that money supply does not affect inflation rates but that interest rates affect money supply (influence factor 0.27) and that real exchange rates affect inflation in a negative and dominant way (influence factor -0.96). Also, it has been found that IP is affected by interest rates but not affected by money supply. This situation conforms neither to the monetary school nor the new Keynesian school views. The Central Bank's choice of interest rates as the main indicator and means of identifying net internal assets after the 2001 crisis is one reason for this situation. Another reason is that in the new economic period factors influenced the real economic activity through credits (consumer credits, business credits and credit cards) and interest rates. Diagram No.6 shows the difference in correlation between money supply and IP and Diagram No.7 shows the difference in correlation between money supply and credits. Because correlation is also under the influence of cyclic effects, causality was tested with the vector error correction model. 0.25 0.2 0.15 0.1 0.05 0.00 -0.05 Correl(LM2,LIP) LIP Poly. (Correl(LM2,LIP)) Poly. (LIP) -0.2 -0.4 -0.6 -0.8 Graph 6. Dynamic Correlation (LM2, LIP, 4 Lags) -0.1 -0.15 -0.2 -0.25 Correl(LM2,LL) Poly. (Correl(LM2,LL)) -0.2 -0.4 -0.6 -0.8 Graph 7. Dynamic Correlation (LM2, LL, 4 Lags) 5. Concluding Remarks This study was conducted to test the money base, money supply, credit capacity, industrial production index, interest rates, inflation and real exchange rate data of Turkey during the years 1997 ­ 2006. These were tested through the monetary transmission mechanism and passive money hypothesis using the vector error correction model-based causality test. Empirical findings show that the passive money supply hypothesis of the new Keynesian economy is supported in part by accommodationalist views, and do not conform to the structuralist and liquidity preference theories. However, according to the monetary transmission mechanism, it has been established that long-term money supply only affects general price levels, and that production is influenced by interest rates in the new economic period. Empirical findings show that in the new economy, period interest transmission mechanisms are brought to the forefront. During the monetary transmission mechanism test, it was decided to leave in theforefront. During the monetary transmission mechanism test, it was decided to leave in the Markov regime variant, which takes into account cyclic effects, a vector error correction model proposed for future studies

Journal

South East European Journal of Economics and Businessde Gruyter

Published: Apr 1, 2007

References