Chapter 1: Introduction and Literature Review
The credit market is an important subject in today’s macroeconomic world.
Before the credit market was looked at, a traditional model containing the goods market and money market were put together to form the IS-LM model. Under this IS-LM model, a change in monetary policy would only shift the LM curve while the IS curve remained constant. In such case, one can see how output and interest rates would change in response to the monetary policy applied. However, Bernanke and Blinder (1988) introduced the credit channel by way of the credit and commodity (CC) curve to the IS-LM model. The assumption under Bernanke and Blinder (1988) is that under an IS-LM model, bonds and loans are not perfect substitutes, as previous studies assumed that only money and bonds existed or that loans and bonds are perfect substitutes. Once this new assumption was taken into place, the traditional IS-LM model was modified with money, bonds and loans being unique assets, which is shown as the CC curve replacing the IS curve in the traditional IS-LM model and becoming a CC-LM model. When a monetary policy change is implemented, this case would not only shift the LM curve, but as a result will also shift the CC curve. For example, under an IS-LM model, a monetary expansion policy will lead to the LM curve shifting outwards, while the IS curve stays put. This would result in an increase in total output and a decrease in interest rates. When the CC curve is introduced into the model, under the same expansionary policy, the LM curve will shift outward, but the CC curve may also shift due to the policy. It is possible that the CC curve would shift inward, outward, or even stay constant, resulting in interest rates decreasing, remaining constant, or increasing. Under this scenario, total output would increase as a result of the LM curve’s shift, but because of the CC curve’s possible shifts (or no shift), the change in the interest rate is not known. Many future literatures have looked into this effect and test for the resulting changes between interest rates and output.
Bernanke and Gertler (1995) used US data to test this new credit channel to see whether theory matches reality. Bernanke and Gertler look at how output levels would change when a monetary tightening policy is implemented, and all the changes that occur along the way which leads to the resulting output level change.
Bernanke and Gertler use a vector autoregression (VAR) model to determine such effects with monthly data. With this model, they try to emphasize four different facts about the economy when such a monetary tightening policy is implemented, one of which states that while an unanticipated tightening will have a transitory effect on interest rates, it will have a sustained effect on the decline in GDP. When a test is performed with log real GDP, it shows a 4-month lag from when the tightening policy
‧
is implemented to when GDP starts its decline. Also, the results show that interest rates will rise initially, before beginning to decline. As theory shows that interest rates decline only when there is a monetary expansion policy, this case will only occur under Bernanke and Blinder’s 1988 model when the CC curve shifts in a way that allow interest rates to fall below the initial level.
In addition to Bernanke and Gertler, Bernanke and Blinder (1992) used their 1988 model as a starting point to determine which economic variable would be the best predictor to the future trend of the economy. Similar to Sims (1980) and Litterman and Weiss (1985), a VAR model is constructed, showing that the Federal Funds rate has a higher level of predicting power when compared to money supply, which contrast previous theories. To prove this, Bernanke and Blinder, using monthly data from 1959:7 to 1989:12, take MacCallum’s (1983) suggestion to Sims by using money supply M1 and M2, different policy variables (Federal Funds rate, Treasury bill rate, and Treasury bond rate), and different measures of forecasting power to show the better predicting ability. The results show that the Federal Funds rate
outperforms every other interest rate, and that the money supply’s predicting power is limited, with M1 showing no predicting power at all. While Bernanke and Blinder’s results show Federal Funds rate as the best economic predictor, however, because there is no perfect substitute for the Federal Funds rate in Taiwan, the closest alternative for Taiwan would be the overnight rate or the discount rate.
By way of the credit channel, Hulsewig, Mayer and Wollmershauser (2006) look at the bank loan supply and monetary policy transmission in Germany. GDP, CPI, short term rate, and loan rate data from 1991Q1 to 2003Q2 were collected and used to determine how certain economic factors would change when a monetary policy shock was implemented by using a VAR model. Hulsewig et al find that when a shock is implemented, short term rate, loan interest rate, price levels, output levels, and bank loans will decline, similar to De Bondt (2000), and Holtemoller (2003). The output level in particular would drop for about four quarters before gradually returning to the original level, which corresponds to the evolution of the output gap.
Most literature found that a monetary tightening policy would lead to a drop in interest rates. This drop, in turn, would lead to a decrease in overall output levels.
However, previous theories, particularly those involving the IS-LM model, showed that a monetary tightening policy would lead to a rise in interest rates. The results from the literature show that should happen in theory and what does happen in reality are different. As opposed to theory, this situation will occur only in a model that includes the credit channel with dynamics that allow for the interest rate to drop, which is only possible with a CC-LM model. Also, Bernanke and Blinder (1992) used various economic activities to determine the best economic predictor and came up
‧
with the conclusion of the Federal Funds rate being the best, but did not make any references as to whether the Federal Funds rate is the best predictor to overall economic output. However, the lack of using total GDP as an economic factor was covered by Bernanke and Gertler (1995). While Hulsewig, et al (2005) did include total output in their model, they used it to determine the response of bank loans, which is opposite of others that use changes in interest rates or loan demand to determine the changes in total output.
While there have been many previous literatures and studies on this topic, very few have either come from Taiwan, or look at how the credit channel operates in Taiwan. One of the few that did so was done by Wu and Chen1
Also, Ho
(2004). Wu and Chen looked at the lending patterns by individual households and corporations, and
determined which had a bigger impact on total output. They conclude that under the same monetary policy, individual lending has a significant positive impact on total output, while corporate lending did not show a significant impact.
2
This thesis will use data from Taiwan to see the relationship between money supply, interest rates, and output and check whether these movements match theory, or what has already been founded in previous literatures. In contrast to Wu and Chen, this thesis combines both household loans and corporate loans into one, and
determines the impact on total output in Taiwan. In addition, this thesis will also test for the role that monetary policy plays when it comes to changes in the output level and also check whether it matches theory or previous literatures.
(2001) determines the relationship between the overnight money rate and deposit and loan rates by looking at the rate of the resulting changes to the deposit and loan rate when there is a change in the overnight rate. Using Klein’s (1971) assumption of a loan demand curve being negatively sloped and a loan supply curve being positively sloped, and the maximum profit conditions stated by Hannan and Berger (1991) and Cottarelli and Kourelis (1994), the conditions of how often a bank adjusts loan rates. However, when this case was applied to Taiwan, the results were inconclusive as this change was not shown when the overnight rate was present. However, applying Taiwan’s case with the Error-Correction Model, Ho’s results show that when the overnight money rate changes, the deposit and loan rates will lag behind, but will change, and in the opposite direction of the overnight money rate.
1吳中書，陳立修
2何棟欽
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4