«Abstract This paper examines the effect of financial frictions on the strength of the monetary transmission mechanism. Credit channel theory implies ...»
Financial Frictions and the Strength of Monetary Transmission
This paper examines the effect of financial frictions on the strength of the monetary transmission
mechanism. Credit channel theory implies that the transmission mechanism of monetary policy
should be stronger in countries with high levels of financial frictions, all else equal. The
intuition is that in these countries, external finance premiums are more sensitive to firms’ financial leverage. By affecting asset prices, therefore, monetary policy has greater impact on external finance premiums and output. We test this theoretical prediction by estimating SVAR models on cross-country data to generate indicators for the strength of monetary transmission.
We find a positive relationship between various measures of financial frictions and the strength of monetary transmission, supporting the predictions of credit channel theory.
JEL Classification: E44; F31; F41 Keyword(s): monetary transmission, financial frictions, bankruptcy costs.
1. Introduction According to the credit channel theory of the monetary transmission mechanism, frictions in credit markets that generate a wedge between the costs of raising funds externally and internally, the external finance premium, help explain the effect of monetary policy on real variables. For example, the cost of monitoring in credit markets suggests poorly-collateralized borrowers will pay a higher premium for external funds than larger, more-collateralized borrowers. The credit channel of monetary policy is a mechanism through which the impact of monetary policy shocks on the real economy is amplified through its effect on external finance premiums. In particular, by affecting this wedge countercyclically, monetary policy has an additional impact on real variables beyond its standard effect through the cost of capital.
The credit channel mainly operates through two conduits: the balance sheet channel, in which monetary policy affects borrowers’ net worth and debt collateral, and the bank lending channel, in which policy impacts the level of intermediated credit (c.f. Bernanke and Gertler, 1995 for a review of the credit channel). These channels have been incorporated into general equilibrium models through costly-state-verification to enhance their empirical relevance (c.f. the financial accelerator model of Bernanke, Gertler and Gilchrist,1999, hereafter BGG). A key result from these models is that the strength of both channels and therefore the broader credit channel increases with the level of financial frictions.1 In particular, in financial systems where financial frictions such as the cost of monitoring (state verification cost) are more pronounced, monetary policy has a larger impact on external premiums through the credit channel.2 1 More recent studies identify the bank capital and risk taking channels as alternative transmission mechanisms. Under the bank capital channel, the strength of banks’ balance sheets (instead of the borrowers’) is the main focus. Under the risk taking channel, banks search for higher yield in response to a reduction in risk aversion. For both of these channels, higher financial frictions are argued to prompt higher borrowing premiums for banks. Blum and Hcllwig (1995) and Van Den Heuvel (2002) provide a detailed discussion of the bank capital channel. Borio and Zhu (2008) and Rajan (2005) do the same for the risk taking channel.
2 The amplification mechanism in the costly-state-verification models can be summarized as follows: shocks that affect borrowers’ net worth also affect their borrowing premiums. This in turn decreases investment, output and asset prices further
as a distinct alternative to the other monetary transmission mechanisms, such as the more traditional cost of capital channel (Bernanke and Gertler, 1995). Rather, it is argued to be a mechanism in which frictions in credit markets amplify the effect of monetary policy on real economic activity. In this paper, we test whether higher levels of financial frictions are consistent with stronger monetary transmission (hereafter MTS) and the amplification mechanism defined by the credit channel theory.
In conducting this analysis we use cross-country data. Using cross-country data is preferable to comparing MTS within a specific country at different time periods. The reason is that financial frictions are relatively stable over time, especially compared to monetary policy.3 In our baseline model, we use bankruptcy recovery rates, the proportion of a firm’s value creditors can recover from a defaulting firm, as an indicator of financial frictions. This variable provides a close match to the source of financial frictions in costly-state-verification models.
Our paper therefore represents, to the best of our knowledge, a first attempt at testing the relationship between financial frictions and MTS implied by costly-state-verification models.
We begin our empirical investigation by generating proxies for MTS in each country. To do so, we obtain the maximum amplitude of output responses to a monetary policy shock (a 100 basis points interest rate shock). Impulse responses are obtained from a structural vector autoregressive (SVAR) model, and monetary policy shocks are identified using the strategies of Kim (1999) and Hoffman (2007) for G-8 and non-G-8 countries, respectively.4 For most of the countries, the impulse responses do not show any evidence of the price and liquidity puzzles (the decreasing net worth and borrowing premiums, and creates an adverse feedback mechanism in credit markets. Since the response of borrowing premiums to the changes in borrowers’ net worth is positively related to the costs of bankruptcy (the source of financial frictions), the adverse feedback mechanism is amplified when there are higher levels of financial frictions.
3 See for example Djankov et al. (2007, 2008) for empirical evidence that support the stability of financial frictions.
4 See the discussion by Elbourne and de Haan (2006) on the usefulness of estimating structural VARs in comparing the monetary transmission mechanisms across countries.
validation for our approach.
Next, we use our measure of MTS in a pooled regression of up to 56 countries from 1984 through 2008 to test the effect of recovery rates on the MTS variable. Our results reveal a negative and statistically significant relationship between recovery rates and MTS that is robust to alternative specifications. Specifically, the coefficient of the recovery rate variable (in our benchmark model) implies that a one percentage point increase in recovery rates leads to a 0.02 percentage point drop in the output response to a monetary policy shock. A value of 0.02 implies that if Malaysia were to increase its recovery rate to the level of the UK’s, it would reduce the response of output from 0.72% to 0.13%. Thus this study provides cross-country evidence on the importance of financial frictions in explaining the transmission mechanism of monetary policy.
As an additional test of the relationship between recovery rates and MTS, we generate an alternative measure of financial frictions and determine whether this measure is related to MTS.
As mentioned above, credit channel theory predicts that as the level of financial frictions increases, the external finance premium becomes more sensitive to firm leverage, i.e. leverage sensitivity increases. We can therefore use leverage sensitivity as a proxy for financial frictions.
Using financial market data to predict financial frictions is an alternative to survey based measures of financial frictions and is a contribution of this paper. To capture leverage sensitivity in each country, we estimate, using firm level data, the effect of firm leverage on corporate bond spreads. Using the estimated leverage sensitivities, we reinvestigate the relationship between financial frictions and MTS and again find support for the model’s predictions.
A key implication of our results is that lowering the level of financial frictions may weaken the ability of central banks to affect economic activity. While MTS in the United States
countries depending on the characteristics of a country’s financial institutions. In countries with greater financial frictions, lenders may be more sensitive to a shift in the health of the borrowers’ balance sheets.5 Our paper complements a broader literature on the effect of institutions on the effectiveness of monetary policy. Cecchetti (1999), for example, using data from 11 European countries, finds that nations with legal origins more protective of creditor rights have weaker MTS. Mishra et al., (2010) argue that central bank independence (CBI) can augment MTS. The argument has also been made that financial market development reduces MTS (Elbourne and de Haan, 2006). We incorporate this broader literature by including as controls measures of legal origin, CBI, and financial market development. We find that financial frictions have an independent effect on MTS after controlling for these variables.
The rest of the paper is organized as follows: in Section 2 we generate our indicator of MTS. In Section 3 we estimate the effect of the bankruptcy recovery rate, our main indicator of financial frictions, on MTS. Sections 4 and 5 present robustness tests. Section 6 concludes.
2. Approximating MTS We begin our empirical analysis by approximating MTS in different countries. It is important to note here that our goal in this section is to generate a proxy for MTS that is comparable across countries. This task is difficult, however, since variables that represent the stance of monetary policy or economic output are likely to vary across countries. Nevertheless, we use the same definitions for monetary policy (the money market rate) and output (the monthly production index) for each country to facilitate a cross-country comparison of MTS.
5 Indeed, a significant number of studies point to a stronger monetary transmission outside of the U.S. (Angelopoulou and Gibson, 2009; Arena et al., 2006; Atta-Mensah and Dib, 2008; Braun and Larrain, 2005; Gambacorta, 2005).
responses to an unanticipated tightening of monetary policy and then used forecast error variance decompositions (FEVD) obtained from these models for inference. MTS in these studies (c.f.
Christiano et al., 1996; Kim, 1999; Kim and Roubini, 2000) is then measured by the percentage of variations in output explained by monetary policy. Most of these studies analyze one country or a group of similar countries. Alternatively, Ceccheti (1999) uses the maximum response of output and inflation as a measure of monetary policy effectiveness. We choose to follow this latter approach and approximate MTS by the maximum amplitude of output responses to a 100 basis points shock to interest rates. This method is preferable to FEVDs for our cross country analysis since the countries in our sample are at different stages of development and their economies face different degrees of shocks.
SVAR Models To derive our measure for MTS, we estimate a SVAR model for each country. Each country specification includes the following variables: the industrial production index, ipt, the consumer price index (CPI), cpit, the monetary aggregate (M1), mt, the short term interest rate (the money market rate), rt, and the world export price index (measured in local currency), wxpit.
This SVAR model can be represented by the following vector: Yt = [ip t, cpit, mt, rt, wxpit ]'.
These variables are widely used in the open economy literature. The industrial production index and the CPI measure overall economic activity and prices, respectively. We include the monetary aggregate to separate money demand and money supply shocks. The interest rate variable captures the monetary authority’s reaction to the other variables in the model. The world export price index is included to account for the monetary responses to external price shocks and to external developments that affect the exchange rate. This variable helps us identify monetary
1984:1-2008:5, and are obtained from the International Financial Statistics (IFS) database for the 56 countries listed in Appendix A.6,7,8 Identification The identification of monetary policy shocks in SVAR models is difficult, and the literature seems to be far from a consensus on this subject. Moreover, comparing the effects of monetary policy across countries (as we do in this paper) amplifies the level of difficulty.
We follow two strategies to check the robustness of our analysis. First, we choose an identification strategy similar to Kim (1999) for G-8 countries and Hoffmann (2007) for others.
The appeal of these strategies has been their ability to find responses to monetary policy shocks that are fairly similar to the predictions of theory. Furthermore, using different identification strategies enhances our ability to capture the effects of monetary policy in economies that are dissimilar. The drawback to this approach, however, is that differences across countries may be generated artificially by the differences in model specification. Therefore, as a robustness test, 6 The industrial production index was available for a majority of the countries in our data set. When these data were not available, we used other indicators for monthly economic activity. These indicators (by country) are provided in Appendix A. The money market rate and monetary aggregate data for 11 countries (Austria, Chile. Colombia, Cyprus, Czech Republic, Hong Kong, Hungary, Israel, Luxembourg, Nigeria, Russia) were obtained through their central bank websites.
7 This appendix also provides the IFS definitions of the macroeconomic variables used in our analysis.