RDP 2005-06: Credit and Monetary Policy: An Australian SVAR 1. Introduction

Credit is an important macroeconomic variable that both helps to drive economic activity and is also dependent on economic activity. Expenditure financed by credit growth will boost output, but at the same time strong output growth can stimulate the demand for credit to finance further expenditure (such as investment). This paper estimates a small structural vector autoregression (SVAR) model to examine the relationships between credit and other key macroeconomic variables.

The SVAR methodology is used in this paper because it can account for endogenous relationships, and can summarise the empirical relationships without placing too many restrictions on the data. While a SVAR model is compatible with many different economic theories, the estimates can be sensitive to the set-up of the model. For this reason, this paper gives special attention to the robustness of the results.

Particular attention is paid to the interaction between credit and monetary policy. This relationship has mostly been viewed through the lens of the ‘credit channel’, whereby monetary policy changes cause financial institutions to alter the volume of loans that they issue.[1] This is a story about the supply of credit. Changes in monetary policy also affect the demand for credit. Both sides of the story suggest that tighter monetary policy will be associated with weaker credit growth. On the other hand, the interest rate responds to the macroeconomic consequences of credit growth. Rapid growth of credit could prove to be inflationary, and so elicit a response by the central bank. In this case stronger credit growth will be associated with higher interest rates.

Australian credit has been the subject of many studies, though none have used an economy-wide model to investigate the simultaneous relationships of aggregate credit with economic activity and monetary policy. In an early contribution, Bullock, Morris and Stevens (1989) examined the ability of various financial indicators to lead real private demand. They found that lending and credit aggregates appeared to lag rather than lead changes in activity. Stevens and Thorp (1989), using more rigorous statistical techniques, were largely supportive of these findings. In later work, Tallman and Chandra (1996, 1997) argued that financial aggregates held little or no predictive power for other macroeconomic series.

However, Blundell-Wignall and Gizycki (1992) suggested that the conclusions of Bullock et al (1989) and Stevens and Thorp (1989) may not hold after the financial reforms of the 1980s. For the period 1984–1991, they showed that business credit led business investment, while overall credit was found to have a two-way relationship with GDP. While Bullock et al and Stevens and Thorp did cover some of the deregulation period, Blundell-Wignall and Gizycki contended that it was not sufficient to fully capture the change in dynamics.

Blundell-Wignall and Gizycki additionally argued that credit rationing had not been important in Australia because the supply of loans had consistently exceeded demand. This suggests that a supply-driven credit channel may not have been particularly strong in Australia. Suzuki (2004) supported this view for bank loans using a VAR.

In contrast, Tallman and Bharucha (2000) argued that supply considerations can be important, at least at a more disaggregated level and during particular episodes. They found that after the distress of the early 1990s recession, the major banks pulled back on risky commercial lending. They noted that this reallocation was more marked in the banks that were under the most financial stress.

Most international studies considering the relationship between credit and monetary policy have focused on the credit channel, and so have been limited to bank credit or other components of the aggregate. Two notable studies are Romer and Romer (1990) and Bernanke and Blinder (1992). Both found that contractionary monetary policy shocks reduced the level of US bank loans, albeit with a considerable lag of about 6–12 months. In fact, both studies found that immediately after the shock the level of bank loans actually increased. A possible explanation is that firms initially borrowed to smooth the impact of a downturn. A lagged response of credit was also found for the Netherlands by Garretsen and Swank (1998).

However, a lagged response is by no means a universal finding. For example, Safaei and Cameron (2003) found an immediate impact of monetary policy on Canadian bank credit. For the US, Gertler and Gilchrist (1993) found a similar instantaneous response, contrasting with Bernanke and Blinder (1992) and Romer and Romer (1990).

The remainder of this study is structured as follows. In Section 2 there is a discussion of the variables included in the model, and the assumptions made to identify the SVAR. The results are presented in Section 3. The preferred model is outlined and then used to examine the consequences of shocks to the interest rate and to credit. The robustness of the findings to alternative specifications is considered. This is followed by some concluding remarks in Section 4.


Bernanke and Gertler (1995) and Mishkin (1996) provide a more extensive discussion of the credit channel. [1]