Exchange rates and the transmission of global liquidity

Exchange rates move the economy through two channels that pull in opposite directions. The familiar net exports channel means activity picks up when a country’s currency depreciates; the financial channel – which operates when borrowers outside a currency’s home jurisdiction owe money in it – means balance sheets strengthen and activity picks up when the domestic currency appreciates against that funding currency. This paper measures the quantity side of the second channel: how fluctuations in the three major international funding currencies (US dollar, Japanese yen, euro) move cross-border bank lending denominated in those currencies to borrowers outside the respective currency area. The data are the BIS Locational Banking Statistics at quarterly frequency – 106 borrower countries for the dollar and 114 for the yen over Q1 2002 to Q3 2015, and 93 countries for the euro over Q1 2010 to Q3 2015, in each case excluding the currency’s own jurisdiction and any country pegged to it. Four econometric approaches are run in parallel: global time-series regressions, borrowing-country-specific time-series regressions, panel regressions with borrower-country fixed effects, and structural panel VARs. All of them deliver a negative relationship between a funding currency’s value and cross-border lending in it. In the global time-series regressions, a 1 percent dollar depreciation is associated with roughly a 0.63 percentage point contemporaneous increase in the quarterly growth rate of dollar-denominated cross-border lending, with closely comparable estimates for the yen (-0.61) and, in the post-crisis window, the euro (-0.64). The panel estimates let the paper separate two exchange rate concepts and reaches its sharpest conclusion there: the broad dollar index, which the authors read as the credit-supply margin working through global banks’ portfolio value-at-risk, carries a coefficient of about -0.50 – more than twice the -0.22 on the bilateral rate against the borrower’s own currency, and more than three times the -0.15 on the bilateral rate once the index is controlled for. Interbank lending responds more than lending to non-banks, consistent with the “double-decker” core-periphery structure of international banking in Bruno and Shin (2015b). The structural panel VARs, which deliberately order lending ahead of the exchange rate so that FX shocks cannot affect lending contemporaneously – “thus tilting the odds against us finding the results predicted by the theoretical model” – show negative and persistent responses, significant for six to eight quarters for a bilateral shock and over ten quarters for a broad-index shock. The cross-currency comparison is the paper’s second contribution: the dollar’s pattern holds across advanced economies, emerging markets, offshore centres, and all four major emerging-market regions; the yen replicates much of it but with smaller and much less persistent effects concentrated in emerging Asia and offshore centres, which the paper reads as “more of a regional flavour”; and the euro shows nothing before the crisis but acquires a statistically significant negative relationship after 2010, largely through interbank lending and largely for emerging Europe and non-euro-area European advanced economies. Throughout, the claims are framed as interpretation of robust correlations plus a VAR identifying assumption: the authors say the results “conclusively point to a robust negative relationship” and that “we interpret these findings as evidence for the existence of a risk-taking channel of currency fluctuations.”

The Dollar, Bank Leverage, and Deviations from Covered Interest Parity

The full text used here is BIS Working Paper No. 592 (revised July 2017), the freely available version of the paper published in American Economic Review: Insights in 2019. The question it takes up is why apparently risk-free arbitrage opportunities persist in the largest currency market in the world, and its answer begins with an observation about what the textbook argument leaves out: “in textbooks, there are no banks. In practice, though, such arbitrage typically entails borrowing and lending through banks, and the competitive assumption is violated due to balance sheet constraints that place limits on the size of the exposures that can be taken on by banks. Even for non-banks, their ability to exploit arbitrage opportunities rely on banks to provide leverage. Hence, if deviations from CIP persist, it must be because banks do not or cannot exploit such opportunities.” From there the paper documents a “triangular relationship” joining the strength of the dollar, the cross-currency basis and cross-border dollar bank lending, and argues that all three are readings of one thing: the shadow price of bank leverage, for which the dollar spot rate serves as a barometer. The evidence has four parts. First, time-series regressions on the ten most liquid currencies against the dollar (Australian, Canadian and New Zealand dollars, Swiss franc, Danish and Norwegian krone, euro, pound, yen, Swedish krona) over 1 January 2007 to 2 February 2016: a one percentage point appreciation of the broad dollar index is associated with a 2.6 basis point fall in the three-month basis without controls and 2.1 with them, against a 7 basis point standard deviation of daily basis changes; at quarterly frequency the five-year basis coefficient runs -1 to -1.4, so a one standard deviation move in the index (3 percent) implies a 3-4 basis point reduction, and the dollar alone explains 19 percent of the time-series variation. Results are similar and more significant in a post-January-2009 subsample, so they are not a crisis artefact. Second, an asset-pricing result in the cross-section: currency-specific dollar betas correlate with the mean basis at 85 percent for the three-month and 97 percent for the five-year horizon, with a unit increase in beta magnitude corresponding to 11 and 26 basis points of expected CIP-trade return respectively – and with a striking reversal of roles, since “the classical ‘safe haven’ currencies, such as the Japanese yen and the Swiss franc, have the highest exposure to the dollar factor, and high-yielding ‘carry’ currencies, such as the Australian dollar and the New Zealand dollar, have the lowest.” An out-of-sample event study of the 3.9 percent dollar appreciation between 8 and 29 November 2016 confirms it: the basis widened for all G10 currencies, most for the yen (from -70.3 to -90.5 basis points), and the post-election dollar beta correlates with the basis at 98 percent. Third, panel regressions with borrowing-country fixed effects show quarterly growth in dollar-denominated cross-border lending falling with both the broad dollar index and the bilateral rate, jointly and separately, for all sectors and for bank and non-bank borrowers alike – evidence, the authors argue, that the index “has explanatory power over and above the bilateral dollar exchange rate.” Fourth, 51 internationally active G10 banks: a 1 percent broad dollar appreciation goes with a 2 percent decline in bank equity, falling to 0.27 percent once market returns are controlled for, and the interaction with the five-year basis is significantly positive, so banks in currency areas with a more negative basis suffer more. The mechanism offered is the risk-taking channel of Bruno and Shin, in which a weaker dollar flatters dollar borrowers’ balance sheets, reducing tail risk in creditors’ portfolios and freeing capacity under a value-at-risk constraint; this is what the authors call the financial channel of exchange rates, and they emphasise that it “may operate in the opposite direction to the net exports channel.” The triangle is shown to hold for the euro in the post-crisis sample but not for other major currencies, which the authors read as pointing “to the unique role of international funding currencies.”

Stefan Avdjiev | Macro Paper Warehouse

Exchange rates and the transmission of global liquidity

The Dollar, Bank Leverage, and Deviations from Covered Interest Parity