Capital flows and the risk-taking channel of monetary policy

Bank leverage is the linchpin of a risk-taking channel through which monetary policy travels across borders: in a pre-crisis quarterly VAR a tighter US policy rate raises the VIX, lowers broker-dealer leverage, appreciates the dollar and shrinks cross-border bank flows, and an accompanying contracting model delivers the result that bank leverage rises with the expected appreciation of the borrower’s currency. The empirical work is a recursive VAR on quarterly data from 1995Q4 to 2007Q4 in the real fed funds target rate, the log VIX, the leverage of the US broker-dealer sector from the Flow of Funds, and the log change in the dollar’s real effective exchange rate, estimated with two lags and 90 percent bootstrapped confidence bands from 1,000 replications. Three links appear. A positive fed funds shock raises the VIX from quarter 4, consistent with Bekaert, Hoerova and Lo Duca’s finding of an effect between months 9 and 11. A rise in the VIX lowers broker-dealer leverage. And a positive fed funds shock lowers leverage after a lag of around 10 quarters, remaining significant to quarter 17, with a maximum response of minus 0.47 at quarter 12 – against a sample average leverage of 21.94, a decline to about 21.5. Leverage in turn moves the exchange rate: an increase in broker-dealer leverage lowers the dollar’s real effective exchange rate by 0.42 percent by quarter 3, with an effect that stays significantly negative across the whole 20-quarter horizon, which the paper offers as a complement to the delayed overshooting puzzle of Eichenbaum and Evans (1995). Adding the first difference of the BIS series for dollar liabilities of banks outside the US shows that higher broker-dealer leverage raises cross-border bank flows after 11 quarters, peaking at 17, and that a fed funds tightening lowers those flows from quarter 8 to quarter 17. Variance decompositions show monetary policy shocks accounting for almost 30 percent of VIX variance and 10 to 20 percent of leverage variance beyond 10 quarters, while leverage shocks account for over 20 percent of exchange rate variance and almost 40 percent of fed funds variance. The theory then rationalises this with a contracting problem in which a bank funds dollar loans from the wholesale market and its local borrowers hold local-currency assets: moral hazard over the correlation of the loan portfolio yields a unique solution with a binding leverage constraint, zero bank default, and the paper’s main proposition that leverage is increasing in expected currency appreciation. Two scope conditions are load-bearing and the authors state both. The sample stops in 2007 because extending it through the zero lower bound produces “markedly weaker VAR impulse responses,” with many fed funds responses insignificant, so “the results reported in this paper should be seen as applying mainly for the boom period preceding the onset of the crisis.” And the amplification story relies on capital inflows coinciding with appreciation, which conflicts with uncovered interest parity; the paper notes UIP’s empirical failure but says plainly that “uncovering the precise mechanism for the failure of UIP is beyond the scope of our paper.”

Valentina Bruno | Macro Paper Warehouse

Capital flows and the risk-taking channel of monetary policy