Capital flows and the risk-taking channel of monetary policy

The popular claim that low advanced-economy rates drive cross-border flows and overheat recipient economies is easy to assert and hard to mechanise; the paper reframes it as a question about how monetary policy influences leverage and real exchange rates. The framing rests on Gourinchas and Obstfeld’s (2012) 1973-2010 crisis-prediction evidence that “two factors emerge consistently as the most robust and significant predictors of financial crises, namely a rapid increase in leverage and a sharp real appreciation of the currency,” holding for both emerging and advanced economies throughout their sample, together with Schularick and Taylor’s (2012) emphasis on banking-sector leverage. “Thus, one way to frame the debate on the role of monetary policy in the transmission of global liquidity is to ask how monetary policy influences leverage and real exchange rates” (Section 1). The authors are explicit about the narrower scope they are claiming: the question “is related to the debate on whether monetary policy was ’too loose’ in the run-up to the crisis with respect to the Taylor Rule,” but “our focus is narrower in that we examine the risk-taking channel more explicitly” (fn. 2).

To show that two separate empirical findings – that fed funds cuts dampen the VIX, and that US monetary contractions cause persistent dollar appreciation – “may be seen as two sides of the same coin,” joined by bank leverage. The two findings are Bekaert, Hoerova and Lo Duca’s (2012) VAR result on the policy rate and implied volatility, and Eichenbaum and Evans’s (1995) delayed overshooting result. “We highlight bank leverage as the linchpin in the risk-taking channel of monetary policy that translates lower measures of risk into greater risk-taking, and then to other real and financial variables. Among the variables impacted by a shock to leverage are capital flows and exchange rates” (Section 1). The paper positions itself as complementary to, not competing with, the micro literature on lending standards – Jimenez, Ongena, Peydro and Saurina (2012) on thinly capitalised banks lending to riskier firms at low policy rates, Maddaloni and Peydro (2011) on eroded standards, Dell’Ariccia, Laeven and Suarez (2013) on banks’ own internal ratings: “The existing literature has focused mainly on lending standards using individual loan data. Our complementary approach extends the existing micro studies by addressing the macro dynamics between monetary policy, the financial intermediary sector and the risk-taking channel through vector autoregression (VAR) methods.”

It establishes procyclical leverage and the compression of measured risk in booms – the two facts the rest of the paper is built on. Two features of the plot of two-year changes in Barclays’ equity, debt and risk-weighted assets against two-year changes in total assets, 1992-2010, are highlighted. First, “the relationship between the changes in the total assets and its risk-weighted assets is very flat,” so “the fact that risk-weighted assets change little even as raw assets fluctuate by large amounts indicates the compression of measured risks during lending booms and heightened measured risks during busts. In other words, banks expand their lending when measures of risk point to tranquil conditions” (Section 2). Second, “changes in assets are reflected dollar for dollar (or pound for pound) in the change in debt, not equity,” with the debt regression giving a slope of 0.9974 and an R-squared of 0.9998, so “leverage is thus procyclical; leverage is high when the balance sheet is large, and credit supply and leverage move one-for-one.” The reverse causation is flagged as equally important: “When lending is expanding rapidly, the increased supply of credit is likely to compress risk spreads. To the extent that the VIX index is closely related to such measures of risk, we would expect that shifts in the leverage of the banking sector will have an impact on the VIX index itself.”

Because consolidated European bank balance sheets bundle the wholesale investment bank with a much larger commercial banking unit that manages its balance sheet differently – a substitution the authors justify with an explicit conditional. They state the ideal measure and why it is unavailable: leverage “should ideally be measured as the leverage of the broker dealer subsidiaries of the global banks that facilitated cross-border lending,” and Shin (2012) shows the European global banks were central to banking-sector capital flows before 2008, but “the reported balance sheet data for European banks are the consolidated numbers for the holding company that includes the much larger commercial banking unit, rather than the wholesale investment banking subsidiary alone,” and broker-dealers and commercial banks “differ in important ways in their balance sheet management” (Section 2). The substitution is then stated as an assumption rather than a fact: “To the extent that US broker dealers dance to the same tune as the broker dealer subsidiaries of the European global banks, we may expect to capture the main forces at work.” Leverage is computed from Flow of Funds total liabilities and total liabilities-plus-equity as (equity plus total liabilities) over equity, and the series rises to 2007 before falling abruptly with the crisis; it scatters tightly against the log VIX lagged one quarter, which the paper reads as adding “weight to theories of leverage based on measured risk, such as Value-at-Risk.”

A parsimonious four-variable recursive VAR with two lags, identified by a Cholesky ordering justified on differential speeds of adjustment. The variable set is defended on a stated tradeoff: “the selection of the number of variables follows from the tradeoff between using a parsimonious model to avoid overfitting, while guarding against omitted variable bias that can undermine the interpretation of the results of the VAR,” with the specific choice “motivated by the interaction between measured risks and banking sector leverage” (Section 2.1). The ordering is fed funds target rate, broker-dealer leverage, VIX, dollar REER, with the rationale that “slower moving variables (like the Fed Funds target rate) are better candidates to be ordered before the fast moving variables like REER and other market prices” – the VIX and REER being market prices that “adjust instantaneously to news,” the fed funds rate reflecting “the periodic decision making process at the Federal Reserve,” and book leverage being “of intermediate sluggishness.” The authors flag the residual concern themselves, noting via Stock and Watson (2001) that “some caution is necessary even here… since the realism of the assumptions underlying the recursive identification of shocks may depend on the frequency of the time series.” The lag choice is a deliberate override of the information criteria: “Formal lag selection procedures (Hannan and Quinn information criterion (HQIC) and the Bayesian information criterion (BIC)) suggest one lag. However, the Lagrange multiplier test for autocorrelation in the residuals of the VAR shows that only the model with two lags eliminates all serial correlation in the residuals. We therefore choose two lags.” All eigenvalues are confirmed inside the unit circle, and inference uses 1,000 bootstrap replications with a small-sample adjustment to the disturbance variance-covariance matrix.

Long lags throughout, with economically modest point estimates for leverage and a persistent effect on the dollar. On policy to risk: “the impact of tighter monetary policy is to raise the VIX measure from quarter 4, which corroborates the finding in Bekaert et al. (2012) who find a similar effect on the VIX starting between months 9 and 11” (Section 2.2). On risk to leverage: “an increase in the VIX index lowers bank leverage,” which the paper reads as “indirect support for the proposition that the banking sector’s balance sheet management is driven by risk measures such as Value-at-Risk.” On policy to leverage: “a positive Fed Funds target rate shock leads to a decline in leverage after a fairly long lag of around 10 quarters and remains significant until quarter 17. The impact reaches a maximum response of -0.47 at quarter 12. When measured against the sample average of 21.94 for leverage, the one standard deviation shock to the Fed Funds rate entails a decline in leverage to around 21.5.” On leverage to the dollar: “an increase in bank leverage leads to a fall in the value of the US dollar by 0.42% of the REER index in quarter 3, and with an impact that remains significantly negative for the entire 20 quarters.” The introduction gives the corresponding horizons in summary form: “a decrease in the Fed Fund rate leads to depreciations in the US dollar after about 14 quarters, while an increase in leverage is followed by a depreciation of the US dollar from 3 quarters but which persists for 20 quarters or more.”

As a complement that supplies a mechanism rather than as a new finding about the exchange rate itself. Eichenbaum and Evans (1995) found that a contractionary US monetary shock causes persistent nominal and real dollar appreciation “with an impact that does not occur contemporaneously but which comes between 24 and 39 months after the initial shock depending on the currency pair considered” (Section 2.2; the introduction states it as “at least 24 months”). The paper’s contribution is stated as an addition to that: “Our complementary evidence shows that the impact of monetary policy works through leverage and the VIX.” Notably, the paper does not claim to resolve the puzzle – it locates the transmission in the leverage-VIX block while leaving the deeper question of why UIP fails explicitly open (Section 4.4).

Enough that the authors argue the effects are economically and not merely statistically significant, with the notable asymmetry that leverage shocks matter more for the exchange rate than policy shocks do. “Monetary policy shocks account for almost 30% of the variance of VIX and between 10% and 20% of the variance of BD leverage at horizons longer than 10 quarters. On the other hand, we see that monetary policy shocks are less important drivers of the variance of US dollar exchange rate as given by REER.” In the other direction, “BD leverage shocks account for more than 20% of the variance of the exchange rate and for almost 40% of the variance of the Fed Funds rate at horizons longer than 10 quarters. They also count for about 20% of the variance of VIX at horizons longer than 15 quarters” (Section 2.2.1). The conclusion drawn is consistent with the paper’s thesis and is stated carefully: “Our variance decomposition reveals a considerable degree of interactions between the variables in our model, and point to the importance of the leverage cycle of the global banks as being a key determinant of the transmission of monetary policy shocks.”

Broadly yes, with one partial exception the paper flags rather than hides. Three alternative stance measures are used. A backward-looking Taylor rule residual – constructed assuming a 2 percent natural real fed funds rate and 2 percent inflation target with the output gap as the percentage deviation of IFS real GDP from CBO potential – reproduces the pattern: “A positive interest rate shock leads to an appreciation of the US dollar after a lag of 10 quarters, and the mechanism is consistent with a decline in banking sector leverage after around 7 quarters,” with “greater measured risks after two quarters” (Section 2.2.2). The nominal effective fed funds rate, reflecting actual transaction prices rather than the target, “confirm[s]” the earlier conclusions, unsurprisingly since the two differ only in “small, high frequency deviations.” M1 growth is the partial exception: the exchange rate and leverage conclusions hold, but “the impact on the VIX dissipates more quickly than for the other monetary shock measures,” for which the authors offer a conjecture rather than a result – “greater search for safe assets during periods when markets become turbulent, as investors seek out bank deposits rather than riskier claims. Further empirical investigations may reveal more the reasons for the differences.” On ordering, the online appendix reports swapping REER and VIX, and placing the fed funds rate last – an alternative the authors credit to a suggestion from Chris Sims, designed “to investigate within-quarter policy responses of the Fed Funds rate to VIX or bank leverage” – with key results “qualitatively unchanged” in both. Using the nominal rather than real effective exchange rate, and adding US industrial production growth, also leave the conclusions intact.

That the relationships break down at the zero lower bound, so the findings are about the boom, not the aftermath. “Our sample period stops in 2007. The crisis period presents special challenges in the VAR estimation, especially since the post-crisis period is associated with the Fed Funds rate pressed against the zero lower bound… The VAR using an extended sample period that encompasses the zero lower bound period show markedly weaker VAR impulse responses, and many of the impulse response functions associated with shifts in the Fed Funds target rate fail to show significant effects. All the evidence points to a structural break in the relationships driving our key macro variables” (Section 2.2.3). The authors note Bekaert et al. (2012) find a similar break and that shifts in autoregressive slope parameters may have offsetting effects on impulse responses, then state the scope condition outright: “For this reason, the results reported in this paper should be seen as applying mainly for the boom period preceding the onset of the crisis.”

That the dollar’s role in cross-border banking dwarfs the euro’s, sterling’s and the yen’s, and that the offshore dollar banking system is comparable in size to the US commercial banking sector. From BIS locational banking statistics, “it is clear from the Figure that the US dollar plays a much more prominent role in cross-border banking than does the euro, sterling or yen,” and scaling against US chartered commercial banks shows that “US dollar assets of banks outside the US exceeded $10 trillion in 2008Q1, and briefly overtook the US chartered commercial banking sector in terms of total assets. So, the sums are substantial. It is as if an offshore banking sector of comparable size to the US commercial banking sector is intermediating US dollar claims and obligations” (Section 3). The plumbing is documented in the surrounding literature the paper cites: branches and subsidiaries of foreign banks in the US borrow from money market funds and channel funds to headquarters, and in the run-up to the crisis “roughly 50% of the assets of U.S. prime money market funds were obligations of European banks” (fn. 7). The five-variable VAR then adds the first difference of the BIS Table 5A series for dollar liabilities of banks located outside the US, ordered between leverage and the market prices because “capital flows reflect the speed of balance adjustment of the intermediaries.” The estimated chain is that a fed funds tightening lowers leverage from around quarter 10 with a peak at quarter 12; higher leverage raises BIS bank flows “after 11 quarters and reaching its maximum impact after 17 quarters”; and the composite effect is that a tightening produces “a decrease in the capital flows in the banking sector starting in quarter 8 and remaining significant until quarter 17” (Sections 3.1). These results are summarised as three Empirical Features: a declining dollar funding rate is followed by dollar depreciation; by rising bank leverage and rising BIS-measured capital flows; and accelerating banking flows are followed by dollar depreciation.

A bank funds dollar loans from the wholesale market and lends dollars to local borrowers whose project payoffs are in local currency; the bank has no mismatch but the borrowers do. “The banks in our model have well diversified loan portfolios consisting of loans to many local borrowers. Although the bank does not have a currency mismatch, the local borrowers do have a currency mismatch. They borrow in US dollars, but invest in projects whose outcome is denominated in local currency” (Section 4.1). The authors note two reasons borrowers might do this: for exporters, borrowing in foreign currency while holding local-currency assets is “one way for exporting companies to hedge their future dollar export receivables,” and “even for non-exporters, borrowing in foreign currency is a means toward speculating on currency movements.” Individual credit risk follows Merton (1974), with default when the terminal dollar value of the project falls below the notional dollar debt; portfolio credit risk follows the Vasicek (2002) single-factor extension, which the paper notes is “the workhorse credit risk model for banks, and has been adopted by the Basel Committee for Banking Supervision (2005) as the backbone of international bank capital rules” (fn. 8). The consequence of the mismatch is the model’s engine: “the distance to default is increasing in [expected appreciation] reflecting the stronger balance sheet of borrowers following currency appreciation when they have borrowed in dollars.” The model choice is motivated by Woodford’s (2010) call for “models in which intermediation plays a role, but in which intermediation is modeled in a way that better conforms to current institutional realities.”

The bank can choose a portfolio with more correlated defaults, which is worse in expected repayment but better for the bank because greater dispersion raises the option value of limited liability; keeping leverage low is what deters it. “The good portfolio consists of loans which have a probability of default epsilon, and low pairwise correlation of default across loans. The bad portfolio consists of loans with a higher probability of default epsilon plus alpha… as well as a higher pairwise correlation of default. For the bank, however, the greater correlation in defaults across loans generates greater dispersion in the outcome density for the loan portfolio, which is associated with a higher option value of limited liability of using debt financing” (Section 4.2). In the analytically convenient limit the good portfolio has zero correlation and the bad portfolio pairwise correlation rho. The incentive-compatibility constraint reduces to a comparison between the additional option value from the bad portfolio and the greater expected payoff of the good one, so “incentive compatibility is maintained by keeping leverage low enough that the higher option value to default does not exceed the greater expected payoff of the good portfolio.” Lemma 1 establishes a unique threshold, proved using the Breeden-Litzenberger result that the state price density is the second derivative of the option price with respect to strike, which makes the option-value difference single-peaked.

Proposition 2 gives a unique solution with zero bank default and risk-free funding; Proposition 3 states that leverage is increasing in expected currency appreciation – and the authors are careful about which parts are special to the setup. On Proposition 2: “The solution entails zero probability of default for the bank and it borrows at the risk-free rate… The common thread [with Geanakoplos (2009) and Fostel and Geanakoplos (2012)] is that actual default does not happen precisely because the contract addresses the possibility of default” (Section 4.3). The generality is then bounded explicitly: “The fact that [the funding rate] is the risk-free rate derives from the feature of our model that the good portfolio consists of loans that are i.i.d. However, the uniqueness result is general, and depends only on the fact that [one curve] cuts [the other] once from below. The qualitative features of our model are preserved when the good portfolio entails positive probability of default by the bank, provided that the funding rate incorporates the bank’s option value of default.” Proposition 3 follows from comparative statics on the default probability, which is decreasing in expected appreciation while the incentive-compatible leverage threshold is increasing as default risk falls. Its interpretation is stated with the business cycle in mind: “Propositions 2 and 3 tell us that bank leverage is procyclical over the cycle – that leverage is high and bank borrowing is large when fundamentals are strong. The additional feature flagged by Proposition 3 is that expected currency appreciation is one possible channel through which the probability of default can decline.” The paper presents this as rationalising, not testing, Gourinchas and Obstfeld’s (2012) leverage-appreciation association.

On an empirical regularity that contradicts uncovered interest parity, which the paper uses but does not explain. The loop is: cheaper funding raises lending, which is financed by capital inflows, which coincide with appreciation of the recipient currency, which lowers borrowers’ default probability, which permits higher leverage and more lending (Section 4.4). The link the authors cannot derive is the second one. Under UIP a low-interest currency should be expected to appreciate, but “in our VAR exercise, Empirical Feature 1 corroborates the finding in Eichenbaum and Evans (1995) that the US dollar tends to depreciate over a protracted period when the US dollar funding cost declines. Such a finding is also consistent with the finding in Fama (1984) that not only does the simple version of UIP fail empirically, but the interest rate differential term appears with the opposite sign. Although uncovering the precise mechanism for the failure of UIP is beyond the scope of our paper, the importance of Empirical Feature 1 is that it has the potential to generate an amplification mechanism.” The authors also acknowledge the awkwardness of Empirical Feature 3: dollar depreciation following accelerating cross-border banking activity “implies an upward-sloping demand response and may seem counterintuitive at first,” and they defend it by appeal to the literature rather than to theory – “the theme of strong currency appreciation amid surging capital inflows is a familiar one in the literature on emerging market crises,” citing Calvo, Leiderman and Reinhart (1993) on Latin America in the early 1990s. One further condition is quantitative: the amplification works through the variance of the loan portfolio’s realised value, which is maximised at a default probability of 0.5 and increasing in the correlation parameter, so the mechanism as stated requires the default probability to sit below 0.5. The paper also notes that the initiating shock need not be monetary at all – “the shock could be purely a domestic one, such as an improvement in domestic fundamentals” – since in this model any increase in bank lending entails the financing inflows that start the loop. The size of the effect turns on how much measured risk the boom itself suppresses: “The magnitude of the risk-taking channel can be substantial when [the sensitivity of default risk to lending] is large, indicating a substantial drop in the credit risk of lending due to the credit boom itself. In other words, the credit boom acts to suppress measured credit risk, rather than make apparent immediately the greater risks inherent in lending.”

That the dollar, rather than the nationality of the intermediary, is the organising variable for global financial conditions – and that this is an agenda rather than a conclusion. “Our empirical results have highlighted the role played by the US dollar as the currency that underpins the global banking system, even if the intermediaries are non-US intermediaries,” which “lends support to studies that have emphasized the US dollar as a bellwether for global financial conditions,” citing Lustig, Roussanov and Verdelhan and Maggiori (Section 5). The results are placed against the Calvo-Leiderman-Reinhart push-pull distinction, with Bruno and Shin’s (2012) companion panel work having “verified the role of global factors associated with the leverage of the banking sector as being a key determinant of cross-border capital flows.” The closing claim is deliberately modest: “The results in our paper suggest that further research on the impact of the risk-taking channel of monetary policy may yield insights into the transmission of global liquidity conditions across borders,” with the paper framed as “one component of the analytical follow-up” to the BIS (2011) global liquidity report.