Bridging Canadian Business Lending and Market-Based Risk Measures

Introduction

In recent years, the Bank of Canada has used several indicators (or measures) that rely on data from the financial market to monitor risk in Canada’s economy and financial system. These market-based risk indicators are like canaries in a coal mine—they can provide advance notice of economic conditions by:

• reacting quickly to new information (i.e., they are real-time indicators)
• capturing the expectations of market participants (i.e., they are forward-looking indicators)

Forward-looking indicators tell us a great deal about future economic activity (Allen, Bali and Tang 2012; Gilchrist and Zakrajšek 2012; and Leboeuf and Hyun 2018). In this note, we ask whether these forward-looking indicators can predict bank lending to businesses because such lending may affect economic growth. Specifically, we look at the ability of three market-based risk measures to predict how much banks are lending to non-financial firms in Canada.

We find that all three market-based risk measures predict changes in business loans. But we show that one measure of the perception of risk in the Canadian banking system tells us more than the others about future growth in business loans. Further, sudden increases in our measure of Canadian banking risk are associated with a weaker outlook for loans to business and real gross domestic product (GDP).

Advance notice of potential changes in business loans is particularly important to the Bank of Canada given the central role of credit in the transmission of monetary policy. By actively monitoring a wide range of these forward-looking indicators, the Bank of Canada can build economic projections and make appropriate decisions on monetary policy.

Market-based risk measures are correlated but also diverge at times

This note compares how three market-based risk measures explain future development in business loans:

• the excess bond premium (EBP)—a measure derived from bond prices that is often associated with investor sentiment or risk preference in the corporate bond market (Leboeuf and Pinnington 2017)
• the market-based indicator of banking system risk (MBI)—a composite index of five measures derived from stock prices that captures different types of risk in the Canadian banking system (MacDonald and Van Oordt 2017)
• the Chicago Board Options Exchange Volatility Index (VIX)—a widely used volatility measure derived from options on the S&P 500 index

An increase in these market-based risk measures signals that market participants are seeing growing downside risks in the financial system. Chart 1 shows that the three measures move together, but these correlations vary over time.

Chart 1: Market-based risk measures are highly correlated but diverge at times

Sources: Bloomberg, Haver Analytics and Bank of Canada calculations Last observation: December 2018

Table 1: Correlation between our market-based risk measures

MBI is an important leading indicator of business loans

We want to know whether market information can help forecast the amount banks are lending to businesses. Our business-lending variable includes non-mortgage loans, mortgage loans and bankers’ acceptances extended to Canadian non-financial firms.

Following Gilchrist and Zakrajšek (2012), we first estimate a baseline predictive regression of the growth of business loans using past growth and two variables reflecting the stance of monetary policy: the real overnight rate and the term spread (i.e., the difference between the Canadian 10-year nominal government bond yield and the 3-month treasury bill rate). We then compare the precisions of forecasts at two horizons, 3 and 12 months, by adding our market-based risk measures to the predictive regressions. The Appendix provides details on the data sources and the methodology.

Chart 2 shows the R-squared to compare the proportion of the variation in business loans explained by each specification.

Chart 2: MBI is highly informative about the outlook for business loans

Adjusted R-squared of regressions with different market-based risk measures

Sources: Bloomberg, Haver Analytics and Bank of Canada calculations Last observation: December 2018

Two findings emerge:

• All three market-based risk measures predict bank lending to businesses. Supplementing the baseline regression with market-based risk measures improves forecasts of future lending at horizons of 3 and 12 months.
• The MBI is more informative than the EBP and the VIX about future growth of business loans. Supplementing the 12-month horizon baseline regression with MBI increases the adjusted R-squared by 41 percentage points (from 38 percent to 79 percent), compared with increases of 28 and 32 percentage points for the EBP and the VIX, respectively.

All results hold if we restrict the estimation sample to the post-crisis period (2010 onwards).

This predictive content of the MBI for the growth of bank lending to business could reflect the fact that most of the loans Canadian banks extend are to Canadian non-financial corporations (e.g., Canadian non-financial corporations account for 71 percent of the outstanding stock of business loans in 2018). As a proxy for banking system risk, the MBI could capture the capacity or willingness of Canadian banks to extend financing to firms.

Increases in MBI are associated with a contraction in business loans and GDP

We then investigate how business loans and real GDP behave following a sudden increase in the MBI. This exercise is particularly useful since the MBI is available in real time and predicts future growth of business loans.

Following Leboeuf and Hyun (2018), we use a standard vector autoregression (VAR) that describes the Canadian economy in terms of economic and financial variables. The economic variables are:

• the Canadian real GDP
• the consumer price index for Canada

The financial variables are:

• the S&P/TSX equity market Composite Index
• the 10-year Government of Canada yields
• the real overnight rate
• the MBI

We estimate the model using monthly data from January 1999 to December 2018. Additional details on the VAR specification can be found in the Appendix.

Using the model, we analyze how both bank lending to business and GDP respond when MBI rises unexpectedly by one standard deviation. This exercise can show how business loans and economic activity respond in Canada to a sudden increase in market participants’ perception of risk in the banking system. Chart 3 shows the response of each variable within our specified model. We find that when the MBI increases suddenly, macroeconomic conditions subsequently deteriorate: business loans and GDP decrease by a total of 2 and 0.2 percentage points, respectively, over three years.

Chart 3: Responses of business loans and GDP to a one-standard deviation increase in MBI

Chart 3: Responses of business loans and GDP to a one-standard deviation increase in MBI

Notes: The charts depict the response to a one-standard-deviation orthogonalized shock to the MBI. Responses of GDP and business loans have been cumulated. The 95% confidence interval is shown with dashed lines.
Sources: Bloomberg, Haver Analytics and Bank of Canada calculations

Conclusion

How market participants see the level of risk in the financial system can provide advance information on future economic developments. Our findings reveal that the perception of risk in the Canadian banking system (MBI) tells us more about the future growth of business loans than the EBP and the VIX. Market-based risk measures are therefore an important addition to a practitioner’s tool box for predicting Canada’s economic outlook.

We conclude with two caveats. First, of course, market-based risk measures can be volatile and sometimes inaccurate. So these measures will complement and cannot replace other leading indicators of Canadian economic activity.

Second, lending to households also contributes to economic activity, beyond the business loans that we consider. Moreover, Canadian businesses now rely less on bank loans and depend more heavily on financial markets (i.e., bond issuance) as their primary source of funding. Future research could thus investigate whether market-based risk measures also lead consumer credit, mortgage credit and corporate bond issuance.

Appendix

Equation 1 shows the specification of the reduced-form regressions. We estimate the model using monthly data from January 1999 to December 2018. Lags of the dependent variable are chosen using the Akaike information criterion. The results are robust to changes in the number of lags and to the exclusion of the financial crisis. We can provide detailed regression tables.

Equation 1: Predictive regression specification

\(ΔLending_{t+h}\) \(=\,α\) \(+\,\displaystyle\sum_{i=1}^{N}γ_{i} ΔLending_{t-i} \) \(+\,β_{1} Term\,\,spread_{t}\) \(+\,β_{2} Overnight\,\,rate_{t}\) \(+\,β_{3} Risk\,\,measure_{t}\) \(+\,ε_{t+h}\)

Equation 2 shows the specification of our VAR. We estimate the model using monthly data from January 1999 to December 2018. Impulse responses are orthogonalized using Cholesky decomposition in the order of the endogenous variables listed in Table A-1. The identifying assumption is that real GDP and inflation respond with a lag, while the S&P/TSX, Government of Canada bonds and short-term rates respond contemporaneously to an unexpected increase in the MBI. Two lags are selected based on the Akaike information criterion. We also include West Texas Intermediate (WTI) crude oil prices as an exogenous variable to account for the importance of the oil sector in the Canadian economy.

Equation 2: Vector autoregression specification

\(y_{t}\) \(=\,Φ_{0}\) \(+\,\displaystyle\sum_{i=1}^{2}Φ_{i} y_{t-i} \) \(+\,Θx_{t}\) \(+\,ε_{t,}\)

where \(y_{t}\) is the vector of endogenous variables, \(\,Φ_{0}\) is a constant vector, \(\,Φ_{i}\) are coefficient matrices of lag i, \(\,x_{t}\) is the exogenous variable and \(\,ε_{t}\) is a vector of white noise innovations.

Table A-1: Variables and data sources

References

1. Allen, L., T. G. Bali and Y. Tang. 2012. “Does Systemic Risk in the Financial Sector Predict Future Economic Downturns?” The Review of Financial Studies 25 (10): 3000–3036.
2. Gilchrist, S. and E. Zakrajšek. 2012. “Credit Spreads and Business Cycle Fluctuations.” American Economic Review 102 (4): 1692–1720.
3. Leboeuf, M. and D. Hyun. 2018. “Is the Excess Bond Premium a Leading Indicator of Canadian Economic Activity?” Bank of Canada Staff Analytical Note No. 2018-4.
4. Leboeuf, M. and J. Pinnington. 2017. “What Explains the Recent Increase in Canadian Corporate Bond Spreads?” Bank of Canada Staff Analytical Note No. 2017-2.
5. MacDonald, C. and M.R.C. van Oordt. 2017. “Using Market-Based Indicators to Assess Banking System Resilience.” Bank of Canada Financial System Review (June): 29–41.

Acknowledgements

We thank Guillaume Bédard-Pagé, Jean-Sébastien Fontaine, Maarten van Oordt and Virginie Traclet for helpful comments and suggestions. Finally, we are grateful to Carole Hubbard and Colette Stoeber for editorial assistance.