The Share of Systematic Variations in the Canadian Dollar—Part III

Introduction

Fontaine and Nolin (2016), Part I in this series, show that more than 50 per cent of the Canadian dollar variations are now systematic. Systematic variations are well-understood in the equity market. In the simplest case—the capital asset pricing model—the covariance between a stock and a market index determines the share of systematic variations. In the case of currencies, the covariance of an exchange rate with other exchange rates determines this share.

Since 2000, the share of systematic variations has increased dramatically for many currencies, including the Canadian dollar. Why that is the case is a puzzle. In this note, we draw a parallel between the share of systematic variations in exchange rate and international bank lending. We find that when a country’s currency has a larger share of systematic variations, lending flows by international banks to that country become more sensitive to global bank lending—they also become more systematic.

This parallel is most visible for large commodity exporters, including Canada. Lending flows into countries that are large commodity exporters have become more sensitive to global lending flows. At the same time, the currencies of commodity exporters have larger shares of systematic variations. This result holds when we account for the response of exchange rates to oil or commodity prices (see Fontaine and Nolin (2017), Part II in this series)

Recent research suggests that exchange rates can reflect a compensation that financial intermediaries earn for bearing currency risk when they absorb imbalances between cross-border financial flows (e.g., Gabaix and Maggiori 2015). Indeed, cross-border lending around the world has tripled since 2000, giving a greater role to financial intermediaries (e.g., Shin 2016). Our findings are consistent with this theoretical channel and provide further evidence that financial intermediation across countries creates a new channel between the real economy and exchange rates.

Systematic variations in cross-border lending

We first document that lending flows from international banks to individual countries have become more systematic over time. The evidence is based on locational banking statistics compiled by the Bank for International Settlements. The share of systematic variations is the R² from a regression of lending flows to each country where the explanatory variable is the global lending flow. The lending flow for each country is the change in cross-border claims from banks located in other countries (denominated in all currencies) weighted by the country’s gross domestic product (GDP). The global lending flow is the average change in lending flows for various countries. We provide more details about this measure in the appendix.

As an example, Chart 1 reports the increased share of systematic variations in cross-border lending for Australia, Canada, New Zealand and South Africa. The R²s were all close to zero from 2000 to 2004 but range between 20 and 50 per cent since 2013.

Chart 1: The sensitivity of commodity exporters' cross-border lending activities to global cross-border lending activity.

Sources: Bank of International Settlements, International Monetary Fund, and Bank of Canada calculations

Systematic variations in currencies and bank lending are related

The share of systematic cross-border lending parallels that of systematic exchange rate variations. Chart 2 provides a simple illustration. It compares the increase in systematic exchange rate variations with the increase in systematic variations in cross-border lending for 15 countries between 2000 and 2017. In this chart, the share of systematic exchange rate variations is measured following Verdelhan (2016) and Fontaine and Nolin (2017), using the R² from a regression based on currency portfolios to capture different sources systematic variations (see also the appendix). The result in Chart 2 is robust to the period used to compute changes.

The dashed line in Chart 2 represents a simple linear regression, with an R² of 41 per cent and a coefficient close to 1. There is a close relationship between exchange rates and bank lending. Chart 2 also shows that the countries with the largest increases are commodity exporters. In contrast to most countries, cross-border lending flows have become less systematic in Japan; this change has mirrored the decline in the share of systematic exchange rate variations.

In the appendix, we also estimate the relationship in Chart 2 with a panel regression. The results confirm that a larger share of systematic cross-border lending is strongly correlated with a larger share of systematic exchange rate variations. The panel regression accounts for country fixed effects; the appendix provides detailed results.

Chart 2: Changes in the share of systematic exchange rate variations (x-axis) versus changes in the share of systematic variations in cross-border lending (y-axis)

Note: Change is the difference between the level of the share of systematic variations in 2017Q1 and 2000Q1 (or later if data are not available).

Line fit: y = 0.87x + 0.01, R² = 0.41.

Sources: Bank of International Settlements, International Monetary Fund and Bank of Canada calculations

Channels between two systematic variations

Our findings are consistent with the economic channel in Bruno and Shin (2015), where the depreciation of the US dollar improves the balance sheets of borrowers in each country and increases the willingness of international banks to lend. The findings are also consistent with the channel in Gabaix and Maggiori (2015), where depreciation of the US dollar raises expected returns on US dollar assets held by intermediaries and increases their willingness to bear currency risk when absorbing US dollar imbalances. The first channel attributes the parallel variations in lending and exchange rates to balance sheet improvements in each country: this is a default risk channel. The second channel attributes the parallel variations to the compensation for risk offered to intermediate cross-border lending: this is a currency risk channel. These channels are not mutually exclusive.

Empirically, Shin (2016) documents a similar but distinct link between the level of global cross-border lending and the level of the dollar portfolio (or some other index of the strength of the US dollar). We add to this result. The response of each currency’s exchange rate to changes in the dollar portfolio parallels the response of cross-border flows in this country to global lending flows.

Solving parts of the puzzle

Parts I and II of this series left us with a puzzle: the share of systematic exchange rate variations has increased for most commodity exporters, yet the exposures to oil and commodity prices offer a limited explanation. The results in Part III offer a partial solution to this puzzle. We find that cross-border lending to commodity exporters became more sensitive to global cross-border lending activity. The effect extends beyond commodity exporters and helps explain the change in the share of systematic exchange rate variations for many other currencies.

The puzzle is only partially solved, however. New questions arise. Cross-border lending flows became more systematic for nearly all countries. But why? Why has cross-border lending by large commodity exporters is more sensitive than most other countries? We hope this offers an avenue for fruitful research.

Appendix

The dollar portfolio has long positions with equal weights in all global exchange rates relative to the US dollar. To measure systematic variations, we update results in Fontaine and Nolin (2017) based on the following regression:

$$Δs_{t+1}=α+β(i_t^*-i_t )+γ(i_t^*-i_t )Carry_{t+1}+δCarry_{t+1}$$ $$+τDollar_{t+1}+ρOil_{t+1}+ε_{t+1}.$$

The share of systematic exchange rate variations is the R² from this regression. The cross-border banking claims data are from the locational banking statistics compiled by the Bank for International Settlements, and they are measured by the location of reporting banks (Table A6.1). The series are quarterly from 1980 to 2017. The share of systematic variations in cross-border banking claims is the R² of the following regression estimated in 10-year rolling windows:

$$y_{i,t+1}=α_i+ β_i y_{w,t+1}+ ε_{i,t+1},$$

where the country $$i$$ lending flow $$y_{i,t+1}$$ is the quarterly change of cross-border banking claims in country $$i$$ from all reporting banks, normalized to a mean of zero and standard deviation of 1. The global lending flow $$y_{w,t+1}$$ is the average quarterly change of cross-border banking claims in 17 countries (Australia, Brazil, Canada, Chile, India, Japan, South Korea, Mexico, New Zealand, Norway, Philippines, Singapore, South Africa, Sweden, Switzerland, Thailand and the United Kingdom), where each country’s flows are scaled by their respective GDP. The results are similar when scaling flows by population.

Table 1: Panel regression results

Table 1: Panel regression results
Constant Share of systematic
cross-border lending
R2 (within) R2 (between) R2 (overall) Rho
Share of systematic exchange rate variations 0.18*** 0.54*** 0.29 0.01 0.11 0.65

Note: Sample is composed of 1,092 observations covering 15 currencies. *** p <0.01.

Sources: Bank of International Settlements, International Monetary Fund, Thomson Reuters and Bank of Canada calculations

References

1. Bruno, V. and H. S. Shin. 2015. “Capital Flows and the Risk-Taking Channel of Monetary Policy.” Journal of Monetary Economics, 71: 119–132. Available at https://www.sciencedirect.com/science/article/pii/S0304393214001688
2. Fontaine, J.-S. and G. Nolin. 2016. “The Share of Systematic Variations in the Canadian Dollar—Part I.” Bank of Canada Staff Analytical Note No. 2016-15. Available at https://www.bankofcanada.ca/2016/11/staff-analytical-note-2016-15/
3. Fontaine, J.-S. and G. Nolin. 2017. “The Share of Systematic Variations in the Canadian Dollar—Part II.” Bank of Canada Staff Analytical Note No. 2017-01. Available at https://www.bankofcanada.ca/2017/02/staff-analytical-note-2017-1/
4. Gabaix, X. and M. Maggiori. 2015. “International Liquidity and Exchange Rate Dynamics.” The Quarterly Journal of Economics, 130 (3): 1369–1420. Available at https://academic.oup.com/qje/article/130/3/1369/1933306
5. Shin, H. S. 2016. “The Bank/Capital Markets Nexus Goes Global.” Speech to the London School of Economics and Political Science. Available at https://www.bis.org/speeches/sp161115.pdf
6. Verdelhan, A. 2017. “The Share of Systematic Variation in Bilateral Exchange Rates.” Journal of Finance, 73 (1): 375–418. Available at https://onlinelibrary.wiley.com/doi/abs/10.1111/jofi.12587

Avis d’exonération de responsabilité

Les notes analytiques du personnel de la Banque du Canada sont de brefs articles qui portent sur des sujets liés à la situation économique et financière du moment. Rédigées en toute indépendance du Conseil de direction, elles peuvent étayer ou remettre en question les orientations et idées établies. Les opinions exprimées dans le présent document sont celles des auteurs uniquement. Par conséquent, elles ne traduisent pas forcément le point de vue officiel de la Banque du Canada et n’engagent aucunement cette dernière.

Sujet(s) : Taux de change
Code(s) JEL : F, F3, F31