School of Economics and Information Systems - Faculty of Commerce
Afandi, Akhsyim, Monetary policy transmission mechanism and structural breaks in Indonesia, PhD thesis, School of Economics and Information Systems, University of Wollongong, 2005. http://ro.uow.edu.au/theses/640
This research studies the monetary policy transmission mechanisms in Indonesia. It aims at achieving six objectives. First, it investigates not only the channels associated with monetary-policy-induced changes in the financial markets but also those associated with the monetary-policy-induced changes in the goods market. Second, it assesses and compares the relative strength of each channel of monetary policy transmission, which has not been attempted previously in the case of the Indonesia. This attempt is made possible by the use of a single model to investigate the multiple channels of monetary policy transmission. Third, it examines whether the 1998 financial crisis has changed the course of each channel of monetary policy transmission. Specifically, whether the role played by each channel in the monetary policy transmission weakened or strengthened is investigated. Fourth, to achieve these objectives a structural vector autoregressive (SVAR) model of the Indonesian economy is constructed. However, before these objectives are addressed the issue of whether the monetary policy shocks have plausible macroeconomic effects is dealt with. To that end, a nonrecursive identifying strategy is applied. Likewise, the complexity of the actual monetary policy conduct in Indonesia that makes it hard to single out the monetary policy instruments actually in use in Indonesia is mitigated by devising eight SVAR specifications. Fifth, it addresses the issue of non-stationary data in VAR modelling by conducting unit root test that allows for one and two endogenouslydetermined breaks. Specifically, this study examines the robustness of the Augmented Dickey-Fuller (ADF) unit root test to the presence of one and/or more structural breaks. Given the fact that most time series data have structural breaks, such an attempt has not been done in the case of Indonesia. Finally, it attempts to solve the identification problem associated with bank lending channel using cointegration procedures based on the autoregressive distributive lag (ARDL) model and the impulse response function analysis that have not been applied to an Indonesian case. (One advantage of the ARDL approach is that it is appropriate irrespective of whether the time series data used are integrated of degree one or zero).
SVAR model specification that captures the most plausible effects of monetary policy shocks is selected out of eight specifications of the SVAR model of the Indonesian economy estimated using both monthly and quarterly data from 1984M12- 2003M12 (1984Q4-2003Q4). The eight different SVAR models are estimated out of the fact that in reality there are four potential measures of monetary aggregates: total reserves, base money, narrow money (M1) and broad money (M2) and two potential measures of short-term interest rates: interbank call-money rate (R1) and SBI rate (interest rate on Bank Indonesia bills) (R2) as monetary policy instruments.
Based on the estimated impulse response analyses, the specification that best captures the effects of monetary policy shocks, as expected by theory, is selected. Overall, the monthly eight-variable SVAR specification that uses M2 as measure of monetary aggregate and R1 as short-term interest rate is superior to the other specifications. However, the quarterly version of the same specification is superior to all other specifications because it resolves the exchange rate puzzle (the initial depreciation of exchange rate following a monetary tightening). This model specification is used to empirically assess the relative strength of channels associated with the monetary-policyinduced changes in the financial and goods markets. In regards to the latter, it has shown empirically, relying on the concept of contribution derived from estimated SVAR models, that consumption spending plays a predominant role in the monetary transmission mechanisms. This result is more in line with the findings regarding the United States than with those regarding the Europe and Japan. For the latter countries investment spending is predominant in the monetary transmission mechanisms.
As for the channels associated with the monetary-policy-induced changes in the financial markets, this study finds that bank loans are dominant. The accumulated response of output attributed to this channel is stronger than the exchange rate and asset price channels. A monetary tightening shock leads to a fall in bank credit, with a lag of about three quarters. However, the importance of bank loans has been exaggerated by the credit crunch during the financial crisis period. The role of bank loans is weaker during the period before the financial crisis, although stronger than the other two channels.
The finding on the bank lending channel suffers an identification problem: do the monetary-policy-induced movements in bank loans originate from the supply or demand side? The “credit view” of monetary policy transmission requires that the former is the case. The estimation of error correction models where the cointegating vectors allow for one structural break and are based on the Johansen’s and Autoregressive Distributive Lag (ARDL) procedures find that the former is the case. Further, the impulse response analysis based on both vector error correction (VEC) models of the bank loans market and the eight-variable SVAR corroborates this result irrespective of the origin of shock, monetary policy or otherwise. Likewise, the first requirement of the credit view that a monetary tightening by the central bank is capable of constraining the supply of bank loans is satisfied. Similarly, this analysis also satisfies the second requirement of the credit view that the monetary-induced decline in the supply of bank loans depresses aggregate spending due to the significance of the bank-dependent borrowers. Thus, the bank lending channel is confirmed.
The exchange rate channel is found to be the weakest. The reason is that during and after the financial crisis there are anomalies associated with this channel. The exchange rate channel operates against the conventional wisdom. A monetary tightening shock is accompanied by exchange rate depreciation and subsequently weakening the level of economic activity. These anomalies disappear when the estimation covers observations before the financial crisis only. These anomalies are due to the fact that during the crisis period the exchange rate depreciation was so enormous that monetary policy tightening had little or no effect in arresting the depreciation pressure. Further, since Indonesia is a highly indebted economy the enormous exchange rate depreciation has inflated the foreign-currency-denominated debts and hence greatly worsened firms’ balance sheets, thereby depressing firms’ investment and productive capacity and hence the level of economic activity.
As for the asset price channel, the results reveal that the equity price channel is also weaker than the bank loans channel but slightly stronger than the exchange rate channel. Due to the collapse of stock market and sharp decrease in the use of equity financing during the financial crisis period, the equity price channel of monetary policy transmission had weakened during that period. Further, to the extent that this channel follows Monetarist view or Tobin’s view, the latter seems to play a dominant role. That is, the real effect of this channel is mainly due to investment spending increases spurred by higher Tobin’s q resulting from the increase in stock price following a monetary easing. Finally, this study also confirms the occurrence of credit crunch following the financial crisis.
02Whole.pdf (2751 kB)