capital flows and the risk-taking channel of monetary policy
TRANSCRIPT
Author's Accepted Manuscript
Capital flows and the risk-taking channel ofmonetary policy
Valentina Bruno, Hyun Song Shin
PII: S0304-3932(14)00168-8DOI: http://dx.doi.org/10.1016/j.jmoneco.2014.11.011Reference: MONEC2743
To appear in: Journal of Monetary Economics
Received date: 26 March 2013Revised date: 22 November 2014Accepted date: 24 November 2014
Cite this article as: Valentina Bruno, Hyun Song Shin, Capital flows and therisk-taking channel of monetary policy, Journal of Monetary Economics, http://dx.doi.org/10.1016/j.jmoneco.2014.11.011
This is a PDF file of an unedited manuscript that has been accepted forpublication. As a service to our customers we are providing this early version ofthe manuscript. The manuscript will undergo copyediting, typesetting, andreview of the resulting galley proof before it is published in its final citable form.Please note that during the production process errors may be discovered whichcould affect the content, and all legal disclaimers that apply to the journalpertain.
www.elsevier.com/locate/jme
Capital flows and the risk-taking channel of monetary policy1
Valentina Brunoa,, Hyun Song Shinb∗†
a American University, United States; b Bank for International Settlements, Switzerland
November 29, 2014
2
Abstract3
Adjustments in bank leverage act as the linchpin in the monetary transmission mechanism that works through4
fluctuations in risk-taking. In the international context, we find evidence of monetary policy spillovers on cross-5
border bank capital flows and the US dollar exchange rate through the banking sector. A contractionary shock to US6
monetary policy leads to a decrease in cross-border banking capital flows and a decline in the leverage of international7
banks. Such a decrease in bank capital flows is associated with an appreciation of the US dollar.8
Keywords: Bank leverage, monetary policy, capital flows, risk-taking channel9
JEL classification: F32, F33, F3410
∗Corresponding author: Hyun Song Shin, Bank for International Settlements, Centralbahnplatz 2, CH-4002 Basel, Switzer-land. E-mail: [email protected]. Phone: +41 61 280 8080†We are grateful to the editor Urban Jermann and the referee for their comments and guidance. We thank Christopher
Sims, John Taylor, Jean-Pierre Landau, Guillaume Plantin, Lars Svensson and Tarek Hassan for comments on an earlier versionof this paper. We also thank participants at the 2012 BIS Annual Conference, Bank of Canada Annual Research Conference,2013 and 2014 AEA meetings and presentations at the Monetary Authority of Singapore, Bank of Korea and at the CentralBank of the Republic of Turkey. We are grateful to Jonathan Wallen for his capable research assistance. The views expressedhere are those of the authors, not necessarily those of the Bank for International Settlements.
Capital flows and the risk-taking channel of monetary policy 1
1. Introduction1
Low interest rates maintained by advanced economy central banks have led to a lively debate on cross-2
border monetary policy spillovers and the possible transmission channels. We examine one such channel:3
the banking sector. Banks are intermediaries whose financing costs are closely tied to the policy rate chosen4
by the central bank. If funding costs affect decisions on how much exposures to take on, monetary policy5
will then affect the economy through greater risk-taking by the banking sector.6
Borio and Zhu (2012) coined the term “risk-taking channel” of monetary policy to denote the impact7
of monetary policy on the willingness of market participants to take on risk exposures, thereby influencing8
financial conditions and ultimately influencing real economic decisions. Our focus in this paper is on the op-9
eration of the risk-taking channel through the banking sector. We ask how banking sector leverage fluctuates10
in the face of changing financial conditions and track the consequences domestically and internationally.11
Comovement of exchange rates and leverage connect the domestic and global impact of risk taking. Gour-12
inchas and Obstfeld (2012) conduct an empirical study using data from 1973 to 2010 for both advanced and13
emerging economies on the determinants of financial crises. They find that two factors emerge consistently14
as the most robust and significant predictors of financial crises, namely a rapid increase in leverage and a15
sharp real appreciation of the currency. Lund-Jensen (2012) presents similar evidence. These findings hold16
both for emerging and advanced economies. Schularick and Taylor (2012) similarly highlight the historical17
evidence on financial vulnerability, especially that associated with the size of the banking sector. Thus, one18
way to frame the debate on international monetary policy spillovers is to ask how monetary policy influences19
bank leverage, cross border flows and the exchange rate.20
The response to this question ties together two strands in the empirical literature. In a domestic21
context, Bekaert, Hoerova and Lo Duca (2013) conduct a vector autoregression (VAR) study that shows a22
close relationship between the policy rate chosen by the Federal Reserve (the target Fed Funds rate) and23
measured risks given by the VIX index of implied volatility on US equity options. They show that a cut in24
the Fed Funds rate is followed by a dampening of the VIX index. In particular, they show that a loosening25
of monetary policy reduces the volatility risk premium in stocks after about three quarters with a persistent26
effect lasting from more than two years.27
Meanwhile, in an international context, an earlier paper by Eichenbaum and Evans (1995) found that28
a contractionary shock to US monetary policy leads to persistent appreciation in the US dollar both in29
nominal and real terms, with a maximum impact that does not occur until at least 24 months after the30
shock. This finding has been dubbed the delayed overshooting puzzle due to its apparent contradiction with31
Capital flows and the risk-taking channel of monetary policy 2
the instantaneous appreciation implied by uncovered interest parity (UIP).1
Our contribution is to show that the findings of Bekaert et al. (2013) and Eichenbaum and Evans (1995)2
may be seen as two sides of the same coin. Banking sector leverage is the linchpin that connects these two3
sets of results. By tying together these two sets of findings under a common framework, our analysis sheds4
light on the links between the domestic and international dimensions of monetary policy transmission.5
The investigation is guided by the model of cross-border banking developed in our earlier paper on global6
liquidity (Bruno and Shin (2014)), that incorporates known institutional features of cross-border banking7
to construct a “double-decker” model of global banks. In this setting, regional banks borrow in US dollars8
from global banks in order to lend to local corporate borrowers. Thus, bank-to-bank credit plays a key role9
in propagating capital flows. The regional banks could either be the regional branches or subsidiaries of10
the global banks or locally owned banks. In turn, the global banks finance cross-border lending to regional11
banks by tapping US dollar money market funds in financial centers.12
[INSERT FIGURE 1 HERE]13
In our model, when the US dollar interest rate falls, the spread between the local lending rate and the14
US dollar funding rate increases. Such lower dollar funding costs have spillover effects on global financial15
conditions because they lead to greater cross-border liabilities and consequently more permissive credit16
conditions in recipient economies.17
Figure 1 shows evidence of the acceleration of bank capital flows through the bank leverage channel.18
The left panel of Figure 1 plots cross-border debt liabilities classified by type of counterparty. Cross-border19
liabilities where both the creditor and debtor are banks grew rapidly in the years before the 2008 crisis. The20
amounts involved were economically significant before the crisis. The right hand panel of Figure 1 shows21
cross-border bank-to-bank liabilities as a proportion of all private credit. At the peak in 2007, bank-to-bank22
cross-border liabilities accounted for 20% of total private credit.23
In this paper, we conduct VAR exercises to shed light on the domestic and international transmission24
channels of monetary policy through the bank-to-bank lending explained in Bruno and Shin (2014) and25
illustrated in Figure 1. Two sets of results trace the interactions of the domestic and international channels26
of monetary policy transmission.27
In the domestic dimension, a decline in US dollar bank funding costs results in an increase in bank28
leverage through the mitigation of volatility risk. In fact, banking sector leverage is closely tied to risk29
measures, such as the VIX index as also illustrated in Figure 2.30
[INSERT FIGURE 2 HERE]31
Capital flows and the risk-taking channel of monetary policy 3
The left hand panel of Figure 2 plots the leverage series of the US broker dealer sector from 1995Q4.1
Leverage increases up to 2007, and then falls abruptly with the onset of the financial crisis. The right panel2
of Figure 2 shows how US broker dealer leverage is closely associated with the risk measure given by the VIX3
index of the implied volatility in S&P 500 stock index option prices from Chicago Board Options Exchange4
(CBOE). The dark squares in the scatter chart are the observations after 2007 associated with the crisis5
and its aftermath.6
The scatter chart adds weight to theories of bank leverage based on measured risk, such as Value-at-Risk7
presented in Adrian and Shin (2010, 2014). Adrian and Shin (2014) document that Value-at-Risk (VaR) is8
an important empirical determinant of financial intermediary leverage. In turn, VaR relative to total assets9
(“unit VaR”) is shown to capture market risk measures, such as the implied volatility of the equity option10
prices as given by the VIX. In our setting, the unit VaR series reflects the risk environment of the recent past,11
which financial intermediaries take into account when they decide on their risk exposure. The findings in12
Adrian and Shin (2014) imply that financial intermediaries can be seen as following a rule of thumb whereby13
they manage their balance sheets actively so as to maintain VaR equal to their equity in the face of rapidly14
changing market conditions. At the same time, Bekaert et al. (2013) have shown that risk aversion is closely15
tied to monetary policy actions. The VIX can therefore be seen as the empirical link between monetary16
policy and bank leverage decisions. In turn, these leverage decisions have macro implications.17
The close relationship between leverage and the VIX Index also provides a point of contact between18
Gourinchas and Obstfeld (2012) who point to the importance of leverage and Forbes and Warnock (2012)19
who have highlighted the explanatory power of the VIX index for gross capital flows.20
In the international dimension, we find evidence of monetary policy spillovers in cross-border capital21
flows through the banking sector. An expansionary shock to US monetary policy increases cross-border22
bank capital flows through higher leverage of international banks. Such an increase in cross-border flows is23
associated with a depreciation of the US dollar over a prolonged period of time. In keeping with the earlier24
findings in Eichenbaum and Evans (1995), a decline in the Fed Funds rate leads to a depreciation in the U.S.25
dollar after about 14 quarters, while an increase in leverage is followed by a depreciation of the US dollar26
from 3 quarters but which persists for 20 quarters or more.27
Taken together, the key contribution of our paper is to show that banking sector leverage is a candidate28
channel for the transmission of monetary policy to exchange rate changes and cross-border bank capital29
flows. Doing so provides additional context to the delayed overshooting puzzle. In particular, our results30
suggest that the delayed overshooting effect is amplified by the banking channel. We point to the influence31
on US dollar exchange rates of not only the monetary policy channel, but also of the channel working through32
Capital flows and the risk-taking channel of monetary policy 4
the behavior of financial intermediaries in the face of fluctuating risk. Given the close relationship between1
leverage and measured risks, our findings bridge the domestic channel of monetary policy studied by Bekaert2
et al. (2013) and the international channel of monetary policy in Eichenbaum and Evans (1995).3
Further corroboration of the importance and relevance of the banking channel in monetary transmission4
comes from the strength of the empirical results during our sample period (1995 - 2007) chosen to coincide5
with the era of global banking expansion. For the same VAR exercises for the sample period 1986 to 19956
that excludes the period of rapid banking sector growth, the key impulse response relationships that drive7
our main conclusions no longer hold. This difference in findings across the two periods provides further8
evidence in support of the bank-to-bank transmission mechanism during the period running up to the 20089
crisis.10
Our VAR exercises complement micro studies of the risk-taking channel, showing how credit standards11
are influenced by the central bank policy rate. For instance, Jimenez, Ongena, Peydro and Saurina (2012)12
using data from Europe find that a low policy rate induces thinly capitalized banks to grant more loans to13
ex ante riskier firms. Maddaloni and Peydro (2011) find that low rates erode lending standards for both14
firms and households. Using US survey data, Dell’Ariccia, Laeven and Suarez (2013) find that low interest15
rates are associated with riskier lending according to the internal ratings used by the banks themselves.16
Compared to these micro studies on lending standards, our complementary approach provides a backdrop17
to the individual loan decisions by showing how risk-taking overall is affected by monetary policy. Our18
paper is part of recent active research on how exchange rates are influenced by financial factors which inject19
medium term forces in addition to traditional macroeconomic fundamentals. A good example is Gabaix20
and Maggiori (2014), who develop a framework for the determination of exchange rates in financial markets21
via capital flows and the risk-bearing capacity of financiers.22
Our VAR exercise also complements the study of investor portfolio flows in Foot and Ramadori (2005),23
who use VAR methods to decompose the transitory price impact of flows from the permanent component of24
price impact. Their results show that the impact of investor order flow on the exchange rate is persistent.25
In particular, investor order flows are related to short-medium term variation in expected currency returns,26
while macroeconomic fundamentals better explain long-term returns and values.27
The findings in this paper paint a consistent picture of the fluctuations in “global liquidity” and what28
role monetary policy has in moderating global liquidity. By identifying the mechanisms more clearly, we29
hope that policy debates on the global spillover effects of monetary policy may gain a firmer footing. The30
BIS report on global liquidity (BIS (2011)) served as a catalyst for further work in this area, and our paper31
can be seen as one component of the analytical follow-up to this report.32
Capital flows and the risk-taking channel of monetary policy 5
2. Evidence from VARs1
Our empirical investigation consists of recursive vector autoregressions (VAR) examining the dynamic2
relationship between the CBOE VIX index of implied volatility on the S&P index options, the real Feds3
Funds target rate of the US Federal Reserve, and a proxy for the leverage of global banks. The real Fed4
Funds target rate is computed for the end of the quarter as the target Fed Funds rate minus the CPI inflation5
rate.6
The leverage of the US broker dealer sector from the US Flow of Funds series published by the Federal7
Reserve serves as our empirical proxy for global bank leverage.1 Since the focus is on investigating spillover8
effects on cross-border bank-lending, our empirical counterpart for global bank leverage should ideally be9
measured as the leverage of the broker dealer subsidiaries of the global banks that facilitate cross-border10
lending. Shin (2012) shows that the European global banks were central in banking sector capital flows in11
the years before the crisis of 2008. However, the reported balance sheet data for European banks are the12
consolidated numbers for the holding company that includes the much larger commercial banking unit, rather13
than the wholesale investment banking subsidiary alone. For the reasons discussed in Adrian and Shin (2010,14
2014), broker dealers and commercial banks will differ in important ways in balance sheet management. To15
the extent that US broker dealers dance to the same tune as the broker dealer subsidiaries of the European16
global banks, we may expect to capture the main forces at work.17
In the VAR model, we also include the US dollar exchange rate (in log differences) as a prelude to our18
more detailed examination of the international dimension of the risk-taking channel and cross-border effects.19
The US dollar exchange rate is measured by the Real Effective Exchange Rate (REER) of the US dollar,20
which is a trade-weighted index of the value of the dollar, obtained from the IMF’s IFS database. An21
increase in REER indicates an appreciation of the US dollar relative to its trade-weighted basket of other22
currencies.23
The sample consists of quarterly data from the fourth quarter of 1995 to the last quarter of 2007 in order24
to examine the workings of the risk-taking channel on the up-swing of the global liquidity cycle. The fourth25
quarter of 2007 marks the beginning of the financial crisis, and our empirical results turn out to be sensitive26
to the zero lower bound on the policy rate after the crisis, as explained below.27
Our chosen sample period coincides with the period when banking sector growth was strong in the run-up28
to the 2008 crisis and conforms most closely to the model of Bruno and Shin (2014), where the transmission29
of financial conditions occurs through the fluctuations in lending across the global banking system.30
1Leverage is obtained from (1) “total liabilities” (FL664190005.Q) and (2) “total liabilities and equity” (FL664194005.Q) ofthe US broker dealer sector from the Flow of Funds. Leverage is defined as 2/(2 – 1).
Capital flows and the risk-taking channel of monetary policy 6
2.1. Set-up1
The selection of the number of variables carefully considers the tradeoff between using a parsimonious2
model to avoid overfitting, while guarding against omitted variable bias that can undermine the interpretation3
of the results of the VAR. Sims (1980) and Stock and Watson (2001) describe the tradeoffs that are entailed4
in the selection of variables in the VAR. In our case, the selection of variables is motivated by the interaction5
between measured risks and banking sector leverage. Our interest is focused especially on the way that6
monetary policy interacts with measured risks and the risk-taking behavior of banks. To capture the core7
mechanisms that involve financial intermediaries, the model includes both the VIX index and the broker-8
dealer leverage variable.9
We identify the impact of shocks by writing the vector autogression in recursive form. For the data10
series {yt} consisting of the vector yt of the variables of interest, the system is specified as11
A(L)yt = εt (1)12
where A(L) is a matrix of polynomial in the lag operator L, and εt is a vector of orthogonalized distur-13
bances. For the four variable VAR, the Cholesky restrictions result in the following exclusion restrictions on14
contemporaneous responses in the matrix A to fit a just-identified model:15
A =
a11 0 0 0
a21 a22 0 0
a31 a32 a33 0
a41 a42 a43 a44
(2)16
The ordering of the variables imposed in the recursive form implies that the variable with index 117
is not affected by the contemporaneous shocks to the other variables, while variable 2 is affected by the18
contemporaneous shock to variable 1, but not variables 3 and 4. In general, the recursive form implies19
that a variable with index j is affected by the contemporaneous shocks to variables with index i < j, but20
not by the contemporaneous shocks to variables with index k > j. Thus, slower moving variables (like21
the Fed Funds target rate) are better candidates to be ordered before fast moving variables like REER and22
the VIX Index, which adjust instantaneously to news.2 The Fed Funds target rate reflects the periodic23
decision making process at the Federal Reserve and the slowly evolving implementation of monetary policy.24
The order of our VAR model is also consistent with the mechanism in Bruno and Shin (2014), where a25
decrease in the cost of borrowing or other risk measures is reflected in banks’ balance sheet management.26
2Some caution is necessary even here, as explained in Stock and Watson (2001), since the realism of the assumptionsunderlying the recursive identification of shocks may depend on the frequency of the time series.
Capital flows and the risk-taking channel of monetary policy 7
Specifically, the adjustment of broker dealer (book) leverage will reflect the speed of the balance adjustment1
of market-based intermediaries and may be characterized as intermediate sluggishness.2
Formal lag selection procedures (Hannan and Quinn information criterion (HQIC) and the Bayesian3
information criterion (BIC)) suggest one lag. However, the Lagrange multiplier test for autocorrelation in4
the residuals of the VAR shows that a model with two lags eliminates all serial correlation in the residuals.5
Therefore the VAR is specified with two lags. A stable VAR model requires the eigenvalues to be less than6
one and the formal test confirms that all the eigenvalues lie inside the unit circle. The results display7
bootstrapped bias-corrected confidence intervals based on 1000 replications with a small-sample degree-of-8
freedom adjustment when estimating the error variance-covariance matrix.9
2.2. Evidence from Impulse Response Functions10
Figure 3 presents the impulse response functions from our four variable recursive VAR with 90 percent11
confidence bands.3 The ordering of the four variables is (1) Fed Funds target rate (2) broker dealer leverage12
(3) VIX and (4) US dollar REER. Figure 3 is organized so that the rows of the matrix indicate the variable13
whose shock we are following and the columns of the matrix indicate the variable whose response we are14
tracking. Each cell of the tables graphs the impulse responses over 20 quarters to a one-standard-deviation15
variable shock identified in the first column.16
[INSERT FIGURE 3 HERE]17
Figure 4 groups the key panels for the narrative. Consider first the impact of a shock to the Fed Funds18
target rate, interpreted as a monetary policy shock and examine the impact of the shock on the leverage19
of the US broker dealer sector. The right panel of Figure 4 shows that a positive Fed Funds target rate20
shock leads to a decline in leverage after a fairly long lag of around 10 quarters and remains significant until21
quarter 17. The impact reaches a maximum response of -0.47 at quarter 12. When measured against the22
sample average of 21.94 for leverage, a one standard deviation shock to the Fed Funds rate entails a decline23
in leverage to around 21.5.24
The other panels reveal related aspects of the mechanism. The left panel of Figure 4 shows that tighter25
monetary policy raises the VIX Index from quarter 4, which corroborates the finding in Bekaert et al. (2013)26
who find a similar effect on the VIX Index starting between months 9 and 11.27
Our distinctive finding is the middle panel in Figure 4, which shows that an increase in the VIX index28
lowers bank leverage. This panel provides indirect support for the proposition that the banking sector’s29
3Results are robust to 95% bootstrapped confidence intervals.
Capital flows and the risk-taking channel of monetary policy 8
balance sheet management is driven by risk measures such as Value-at-Risk, as argued by Adrian and Shin1
(2010, 2014). Thus, the conjunction of the first two panels tells the story underlying the final panel - of2
how an increase in the US dollar bank funding costs results in a decline in bank leverage.3
[INSERT FIGURE 4 HERE]4
Finally, in anticipation of our examination of the international dimension to the risk-taking channel,5
Figure 3 shows that a positive Fed Funds target rate shock leads to an appreciation of the US dollar after a6
fairly long lag. This result is consistent with the “delayed overshooting puzzle” found in Eichenbaum and7
Evans (1995) who find that a contractionary shock to US monetary policy leads to persistent appreciation8
in nominal and real US dollar exchange rates, with an impact that does not occur contemporaneously but9
which comes between 24 and 39 months after the initial shock depending on the currency pair considered.10
The incremental contribution of our findings is to show that intermediary leverage amplifies the transmission11
of monetary policy to exchange rate changes.4 We will return to this aspect in Section 3.12
Our sample period stops in 2007 due to the crisis, a structural break. The crisis period presents special13
challenges in the VAR estimation, especially since the post-crisis period is associated with the Fed Funds14
rate pressed against the zero lower bound (see Liu, Waggoner and Zha (2011) and Kilian (2011)). The15
VAR using an extended sample period that encompasses the zero lower bound period show markedly weaker16
VAR impulse responses, and many of the impulse response functions associated with shifts in the Fed Funds17
target rate fail to show significant effects. All the evidence points to a structural break in the relationships18
driving our key macro variables.19
Bekaert et al (2013) also find a similar structural break, suggesting that shifts in the autoregressive slope20
parameters may also have offsetting effects on the impulse response functions. Specifically, they find that21
the impact of monetary policy is quantitatively weaker when the sample period is extended to encompass22
the crisis period. They argue that such a weaker statistical power is due to both measurement problems23
and high volatility of the VIX. They also find that monetary policy reacts significantly to uncertainty in24
some cases when the post crisis period is included. These findings are consistent with David and Veronesi25
(2014), who find that a rise in option-implied volatility predicts a subsequent decline in the interest rates26
in a bivariate VAR that includes the financial crisis period. For these reasons, the results reported in this27
paper should be seen as applying mainly for the boom period preceding the onset of the crisis.28
4When omitting both the VIX and the broker-dealer leverage variables, the effect of Fed fund rate on USD exchange rate issignificant in the medium term. However, it is no longer significant in the long term. This evidence is in contrast to the four-variable VAR that includes VIX and bank leverage, where bank leverage has an effect on USD exchange rate in the mediumterm and monetary policy has an effect in the long term. This result shows that the delayed overshooting effect driven bymonetary policy is amplified by the banking channel.
Capital flows and the risk-taking channel of monetary policy 9
By using the VXO index, which is a precursor of the VIX index, it is possible to extend the sample back1
to 1986. The VARs using this earlier sample show much smaller impact of banking sector leverage, and2
the impulse responses are not statistically significant. In the context of banking sector developments at the3
relevant time, these findings confirm that the bank-to-bank transmission mechanism operated strongly only4
during the period running up to the 2008 crisis. The earlier sample period of 1986 - 1995 does not show5
a similarly active role for the global banking system in the cross-border transmission of monetary policy.6
Results are reported in Figure A.1 of the online Appendix.7
Regulatory barriers prevented U.S. commercial banks from expanding their operations into securities8
activities (Section 20 of the Glass–Steagall Act of 1933). Regulatory barriers gradually disappeared from9
1987 when the Federal Reserve authorized bank holding companies to establish subsidiaries, which could10
underwrite corporate debt securities up to a 5% revenue limit. A major milestone was passed in 1995 when11
the House Banking Committee approved a bill that would repeal Glass–Steagall. In 1999 the Gramm-Bliley12
Act, also known as the Financial Services Modernization Act, allowed bank holdings to conduct securities13
activities without limit in subsidiaries separate from their commercial banks. The timing of the de-regulation14
fits well with our finding of a possible structural break in the mid to late 1990s.515
Overall, our initial evidence shows that during the 1995 - 2007 period, the risk-taking channel of monetary16
policy operated strongly in a domestic context whereby permissive monetary policy is followed by a period of17
lower measured risks and expanded bank lending risk-capacity. In the next section, we will investigate the18
international dimension of the spillover effects of US monetary policy and how such a lower dollar funding19
cost results in higher cross-border bank flows through the internal capital markets of global banks.20
2.2.1. Variance Decompositions21
The above findings establish that monetary policy has a medium-run (two to three years) impact on22
broker leverage and VIX, and that broker dealer leverage has a statistically significant effect on the US23
dollar exchange rate. The effects are not only statistically significant, but also economically significant.24
Figure 5 shows what fraction of the structural variance of the four variables in the VAR is due to monetary25
policy shocks or BD leverage shocks. Monetary policy shocks account for almost 30% of the variance of26
VIX Index and between 10% and 20% of the variance of BD leverage at horizons longer than 10 quarters.27
In contrast, monetary policy shocks are less important drivers of the variance of US dollar exchange rate as28
given by REER.29
5We are grateful to the referee for this observation.
Capital flows and the risk-taking channel of monetary policy 10
[INSERT FIGURE 5 HERE]1
2
BD leverage shocks account for more than 20% of the variance of the exchange rate and for almost 40%3
of the variance of the Fed Funds rate at horizons longer than 10 quarters. BD leverage shocks also count for4
about 20% of the variance of VIX at horizons longer than 15 quarters. Our variance decomposition reveals5
a considerable degree of interactions between the variables in our model, and points to the importance of6
the leverage cycle of the global banks as a key determinant of the transmission of monetary policy shocks.7
2.2.2. Alternative Measures of Monetary Policy Shocks8
Figure 6 shows the impulse-response functions for alternative measures of monetary policy on REER,9
VIX and BD-leverage in the four-variable VAR with 2 lags and 1000 bootstrapped standard errors. The10
considered monetary policy shocks are residuals from a Taylor rule, M1 growth and nominal effective Fed11
Funds rate.12
[INSERT FIGURE 6 HERE]13
The first alternative measure of monetary policy stance is the difference between the nominal Fed Funds14
target rate and the Fed Funds rate implied by a backward looking Taylor rule. The Taylor rule rate assumes15
the natural real Fed funds and the target inflation rate to be 2%, while the output gap is computed as16
the percentage deviation of real GDP (from the IFS) from potential GDP (from the Congressional Budget17
Office). As shown by the top row of Figure 6, our qualitative conclusions when using the Fed Funds target18
rate as the monetary policy shock remain unchanged. A positive interest rate shock leads to an appreciation19
of the US dollar after a lag of 10 quarters, and the relation is consistent with a decline in banking sector20
leverage after around 7 quarters. In turn, the “risk-taking channel” is clearly evident in the middle cell of21
the top row, where a monetary policy shock is associated with greater measured risks after two quarters.22
We consider two further alternative measure of monetary policy shocks, shown in the second and third23
rows of Figure 6. One alternative measure is the growth rate of the US M1 money stock, where a positive24
shock to M1 corresponds to monetary policy loosening. In this specification, the qualitative conclusions25
persist in the impulse responses for the exchange rate and the banking sector leverage. The impact on the26
VIX dissipates more quickly than for the other monetary shock measures. One reason for the qualitative27
difference of the M1 variable may be the greater search for safe assets during periods when markets become28
turbulent, as investors seek out bank deposits rather than riskier claims.29
Capital flows and the risk-taking channel of monetary policy 11
Our third measure of monetary policy shock is the nominal effective Federal Funds rate, which measures1
actual transactions prices used in the Fed Funds market of interbank lending, rather than the Fed Funds2
target rate itself. Our earlier conclusions using the Fed Funds target rate are robust to this variation. To3
the extent that the difference between the Fed Funds target rate and the effective Fed Funds rate are small,4
high frequency deviations, our results are perhaps not surprising.5
2.2.3. Additional Robustness Checks6
A series of robustness exercises for our VARs are reported in the separate online appendix of the paper.7
The sensitivity of the recursive VAR to alternative orderings of the variables is a perennial theme in VAR8
analysis. The selection and ordering of the variables in our VAR follows the bank-to-bank mechanism9
outlined in Bruno and Shin (2014). Furthermore, the different degrees of inertia inherent in our selected10
variables give some basis for the specification of our VAR analysis (see Kilian (2011) for discussion of this11
point). The separate appendix to our paper displays various robustness exercises. Figure A.2 of the12
appendix illustrates the alternative ordering: (1) Fed Funds target rate (2) broker dealer leverage (3) REER13
and (4) VIX, where the two price variables REER and VIX are switched. Our key findings on the risk-taking14
channel remain unchanged to such an alternative order.15
Figure A.3 in the appendix also reports the impulse responses for the VAR when the Fed Funds rate is16
ordered last to investigate within-quarter policy responses of the Fed Funds rate to VIX or bank leverage.617
In the VAR with the ordering: (1) broker dealer leverage (2) VIX (3) REER and (4) Fed Funds target rate,18
our key results on the risk-taking channel are again qualitatively unchanged.19
Our results remain unchanged if the nominal effective exchange rate (NEER) replaces the real effective20
exchange rate (REER), as shown in Figure A.4 in the appendix.21
Our findings are also robust to the inclusion of the growth of US industrial production in the VAR,22
which examines the impact of macroeconomic conditions as a backdrop to monetary policy. Our results23
reported in Figure A.5 in the appendix indicate that including industrial production does not alter the main24
conclusions on the mechanism of the risk-taking channel through the leverage of the broker dealer sector.25
3. International Dimension26
This section focuses on the international dimension of the transmission mechanism of monetary policy.27
Taylor (2013) argued that the potential for monetary policy spillovers operating through divergent policy28
interest rates has led to an enforced coordination of interest rate policy among central banks who fear that29
6We thank Chris Sims for suggesting this alternative ordering for our robustness tests.
Capital flows and the risk-taking channel of monetary policy 12
failure to follow suit in lowering rates would undermine other macro objectives. The role of global banks1
that channel wholesale funding across borders is perhaps the most important channel for such transmission2
of financial conditions. For instance, Cetorelli and Goldberg (2012) have found that global banks respond3
to changes in US monetary policy by reallocating funds between the head office and its foreign offices, thus4
contributing to the international propagation of domestic liquidity shocks.5
[INSERT FIGURE 7 HERE]6
Figure 7 uses data from the BIS locational banking statistics to plot the foreign currency assets and7
liabilities of BIS-reporting banks, classified according to currency. The top plot represents the US dollar-8
denominated assets of BIS-reporting banks in foreign currency, and hence gives the US dollar assets of banks9
outside the United States. The bottom plot in Figure 7 gives the corresponding US dollar-denominated10
liabilities of banks outside the United States. It is clear from Figure 7 that the US dollar plays a much more11
prominent role in cross-border banking than does the euro, sterling or yen.12
[INSERT FIGURE 8 HERE]13
To gain some perspective on the size of the US dollar assets in Figure 7, compare the total assets series14
next to the aggregate commercial banking sector in the United States, which is given in Figure 8. US15
dollar assets of banks outside the US exceeded $10 trillion in 2008Q1, briefly overtaking the US chartered16
commercial banking sector in terms of total assets. So, the sums are substantial. It is as if an offshore17
banking sector of comparable size to the US commercial banking sector is intermediating US dollar claims18
and obligations. Shin (2012) shows that the European global banks account for a large fraction of the US19
dollar intermediation activity that takes place outside the United States.720
As shown earlier by Figure 1, capital flows through the international banking system was a very sub-21
stantial proportion of total cross-border debt flows. Such banking flows have also played a major role in22
the expansion of domestic lending. At the peak in 2007, for instance, bank-to-bank cross-border liabilities23
accounted for 20% of total private credit and for over 30% of the percentage of global GDP. The large weight24
of the banking sector prior to 2008 lends weight to the bank-to-bank mechanism in Bruno and Shin (2014).25
Rey (2013) also documents the rapid increase in credit flows relative to FDI and portfolio equity flows.26
7A BIS (2010) study describes how the branches and subsidiaries of foreign banks in the United States borrow from moneymarket funds and then channel the funds to their headquarters. See also Baba, McCauley and Ramaswamy (2009), McGuireand von Peter (2009), IMF (2011) and Shin (2012), who note that in the run-up to the crisis, roughly 50% of the assets of U.S.prime money market funds were obligations of European banks.
Capital flows and the risk-taking channel of monetary policy 13
3.1. Structural VAR Analysis of Cross-Border Banking Flows1
We now turn our attention to the cross-border dimension. In order to address the international dimension2
of monetary policy spillovers, the existing list of VAR variables expands to include a measure of cross-border3
banking sector flows into our existing VAR analysis that is consistent with the model in Bruno and Shin4
(2014). In particular, the analyisis focuses on the cross-border lending by banks to other banks, as given5
by the BIS banking statistics.6
The choice of our capital flow variable conforms to the mechanism in Bruno and Shin (2014) which builds7
on the institutional features underpinning the international banking system. In particular, global and local8
banks operate in a “double-decker” model of banking where regional banks borrow in US dollars from global9
banks in order to lend to local corporate borrowers. In turn, the global banks finance cross-border lending10
to regional banks by tapping US dollar money market funds in financial centers.11
In this setting, when the US dollar risk-free rate interest rate falls, the spread between the local lending12
rate and the U.S. dollar funding rate increases. The resulting lower dollar funding costs leads to an accel-13
eration of bank capital flows and more permissive credit conditions in recipient economies. An implication14
of the model is that the loosening of US monetary policy will result in greater cross-border liabilities, with15
spillover effects on global financial conditions through the bank leverage channel.16
We measure international banking flows as the growth (log difference) in cross-border loans of BIS-17
reporting banks on banking sector counterparties, as measured by the difference between Table 7A (all18
borrowers) and Table 7B (non-bank borrowers) from the BIS Locational Bank Statistics. Global banks19
account for most of the international exposures. Since European banks have a pivotal role in the transmission20
of global liquidity (see Shin (2012)) and the US dollar is the currency underpinning the global banking system,21
this variable is a good proxy for banking claims of global banks that use US dollar wholesale funding. This22
measure fits our objective to capture the activities of the internationally active banks that were instrumental23
in channeling dollar funding globally.24
The model uses the following Cholesky ordering: (1) Fed Funds target rate (2) broker dealer leverage (3)25
BIS banking flows (4) VIX and (5) US dollar REER. Capital flows reflect the speed of balance adjustment26
of the intermediaries so they order between the Fed Funds rate and the market variables, but after the27
broker dealer leverage. Figure 9 presents the impulse responses together with bootstrapped bias-corrected28
90% confidence bands for the model with two lags.8 As before, Figure 9 is organized so that the rows of the29
matrix indicate the shocked variable and the columns of the matrix indicate the response variable. Each30
8Results are robust to 95% bootstrapped confidence intervals.
Capital flows and the risk-taking channel of monetary policy 14
cell of the tables gives the impulse responses over 20 quarters to a one-standard-deviation variable shock1
identified in the first column.2
[INSERT FIGURE 9 HERE]3
Figure 9 reveals how capital flows through the banking sector are an important element of the narrative of4
the risk-taking channel. Figure 10 gathers six of the panels for a more succinct summary of the relationships.5
The top two left panels of Figure 10 show the impact of lower cost of borrowing (Fed Funds rate) and Value-6
at-Risk measures (VIX) on banking sector leverage, as already documented in Figure 3.7
[INSERT FIGURE 10 HERE]8
9
The other panels in Figure 10 show the mechanism of how the risk taking channel of monetary policy10
impact capital flows and the US dollar exchange rate through the banking sector.11
The top right panel in Figure 10 shows that an increase in broker dealer leverage leads to an immediate12
marked increase in cross-border bank flows and a long term increase after 3 quarters, persisting over the entire13
20 quarters. Such a vehement increase in cross-border flows is associated with US currency depreciation14
that starts at quarter 3 and lasts until quarter 11 and starts again after quarter 17 (bottom left panel). The15
last two panels together show that monetary policy eventually leaves its mark on the US dollar exchange16
rate and the capital flows funded by the US dollar. Figure A.6 presented in the appendix confirms the our17
results are robust to an alternative ordering of the variables.18
Importantly, the bank leverage effect explored here seems to be tied to the special role of the US dollar. In19
fact, when replicating our results by using other currencies, the leverage channel does not have amplification20
effects for the exchange rate of currencies other than the US dollar. This provides additional illustration21
of the role of the US dollar as the currency underpinning the global banking system, based on US dollar22
wholesale funding (see Bruno and Shin (2014))23
Overall, the empirical regularities uncovered in our VAR results show the risk-taking channel of monetary24
policy and its impact on financial and real variables through bank leverage. Bank leverage is thus the25
linchpin that translates lower measures of risk (lower cost of borrowing and Value-at-Risk) into greater26
cross-border banking flows and local currency appreciation. Such empirical features corroborate the finding27
in Eichenbaum and Evans (1995) that the US dollar tends to depreciate over a protracted period when the28
US dollar funding cost declines. Our complementary evidence shows that the impact of monetary policy29
works through the bank leverage channel.30
Capital flows and the risk-taking channel of monetary policy 15
Our results are also consistent with the findings in Gourinchas and Obstfeld, who conduct an empirical1
study using data from 1973 to 2010 and find that two factors emerge consistently as the most robust and2
significant predictors of financial crises, namely a rapid increase in leverage and a sharp real appreciation of3
the currency. Shularick and Taylor (2012) similarly highlight the role of leverage in financial vulnerability,4
especially that associated with the banking sector.5
4. Concluding Remarks6
The main contribution of our paper relative to earlier studies is in highlighting the role of the banking7
sector in the cross-border transmission of monetary policy. Our findings underline the role of banks as8
intermediaries whose financing costs are closely tied to the policy rate chosen by the central bank. If9
funding costs affect decisions on how much exposures to take on, monetary policy will then affect the10
economy through greater risk-taking by the banking sector. Given the pre-eminent role of the U.S. dollar11
as the currency that underpins the global banking system, our findings suggest that the impact of the policy12
rate chosen by the Federal Reserve has an international dimension, as well as a domestic one.13
The role played by the U.S. dollar in the global banking system suggests that the value of the U.S. dollar14
may thus be a bellwether for global financial conditions, as recently suggested by Lustig, Roussanov and15
Verdelhan (2012) and Maggiori (2010).16
More broadly, the role of the US dollar in the global banking system opens up important questions on17
the transmission of financial conditions across borders. In a financial system with interlocking claims and18
obligations, one party’s obligation is another party’s asset. When global banks apply more lenient conditions19
on local banks, the more lenient credit conditions are transmitted to the recipient economy. In this way,20
more permissive liquidity conditions in the sense of greater availability of credit will be transmitted across21
borders through the interactions of global and local banks. Calvo, Leiderman and Reinhart (1993, 1996)22
famously distinguished the global “push” factors for capital flows from the country-specific “pull” factors,23
and emphasized the importance of external push factors in explaining capital flows to emerging economies24
in the 1990s. Eickmeier, Gambacorta and Hofmann (2013) and Chen et al. (2012) are two papers in a recent25
literature that has attempted to elucidate the concept of “global liquidity” that was first formally studied by26
the official sector in the BIS study on global liquidity (BIS (2011)). Conversely, during times of crises, the27
deleveraging of the global banks is associated with “dollar shortages” as documented by Baba, McCauley28
and Ramaswamy (2009) and McGuire and von Peter (2009).29
The results in our paper suggest that further research on the impact of the risk-taking channel of monetary30
Capital flows and the risk-taking channel of monetary policy 16
policy may yield insights into the transmission of global liquidity conditions across borders. One key question1
is to what extent future episodes of cross-border financial spillovers will resemble the banking sector-led model2
examined in this paper. The fact that banking sector leverage has been subdued since the crisis suggests3
that the future channels of transmission will not be bank-driven, but instead involve alternative mechanisms4
- perhaps through the market for debt securities (Shin (2013)). One contribution of our paper has been to5
establish a benchmark for comparison during a period when banking sector activity was particularly strong;6
alternative channels can be gauged relative to such a benchmark.7
Capital flows and the risk-taking channel of monetary policy 17
References1
Adrian, T., Shin, H.S., 2010. Liquidity and Leverage. Journal of Financial Intermediation 19, 418-437.2
Adrian, T., Shin, H.S., 2014. Procyclical Leverage and Value-at-Risk. Review of Financial Studies 27 (2),3
373-403.4
Baba, N., Robert, N.M., Ramaswamy, S., 2009. US Dollar Money Market Funds and Non-US Banks. BIS5
Quarterly Review March 2009, 65-81.6
Bank for International Settlements, 2011. Global liquidity – concept, measurement and policy implications.7
CGFS Papers 45, Committee on the Global Financial System (http://www.bis.org/publ/cgfs45.pdf).8
Bekaert, G., Hoerova, M., Lo Duca, M., 2013. Risk, Uncertainty and Monetary Policy. Journal of Monetary9
Economics 60 (7), 771-788.10
Borio, C., Zhu, H., 2012. Capital regulation, risk-taking and monetary policy: a missing link in the trans-11
mission mechanism? Journal of Financial Stability 8 (4), 236-251.12
Bruno, V., Shin, H.S., 2014. Cross-Border Banking and Global Liquidity. Forthcoming in the Review of13
Economic Studies.14
Calvo, G. A., Leiderman, L., Reinhart, C., 1993. Capital Inflows and Real Exchange Rate Appreciation in15
Latin America: The Role of External Factors. IMF Staff Papers 40 (1), 108-151.16
Calvo, G. A., Leiderman, L., Reinhart, C., 1996. Capital Flows to Developing Countries in the 1990s: Causes17
and Effects. Journal of Economic Perspectives 10, Spring 1996, 123-139.18
Cetorelli, N., Goldberg, L. S., 2012. Banking Globalization and Monetary Transmission. Journal of Finance19
67(5), 1811-1843.20
Chen, S., Liu, P., Maechler, A., Marsh, C., Saksonovs, S., and Shin, H. S., 2012. Exploring the dynamics of21
global liquidity. IMF Working Paper, WP/12/246.22
David, A., Veronesi, P., 2014. Investors’ and Central Bank’s Uncertainty Embedded in Index Options.23
Review of Financial Studies, 27 (6), 1661-1716.24
Dell’Ariccia, G., Laeven, L., and Suarez, G., 2013. Bank Leverage and Monetary Policy’s Risk-Taking25
Channel: Evidence from the United States. IMF Working Paper, WP/13/143.26
Eichenbaum, M., Evans, C. L., 1995. Some Empirical Evidence on the Effects of Shocks to Monetary Policy27
on Exchange Rates. Quarterly Journal of Economics 110 (4), 975-1009.28
Eickmeier, S., Gambacorta, L., Hofmann, B., 2014. Understanding Global Liquidity. European Economic29
Review 68, 1-18.30
Forbes, K. J., Warnock, F. E., 2012. Capital Flow Waves: Surges, Stops, Flight and Retrenchment. Journal31
of International Economics 88 (2), 235-251.32
Capital flows and the risk-taking channel of monetary policy 18
Froot, K., Ramadori, T., 2005. Currency Returns, Intrinsic Value, and Institutional-Investor Flows. Journal1
of Finance 60 (3), 1535 - 1566.2
Gabaix, X., Maggiori, M., 2014. International Liquidity and Exchange Rate Dynamics. NBER Working3
Paper, No. w19854.4
Gourinchas, P. O., Maurice, O., 2012. Stories of the Twentieth Century for the Twenty-First. American5
Economic Journal: Macroeconomics 4 (1), 226-65.6
Jimenez, G., Ongena, S., Peydro, J.-L., Saurina, J., 2014. Hazardous Times for Monetary Policy: What7
Do Twenty-Three Million Bank Loans Say about the Effects of Monetary Policy on Credit Risk-Taking?8
Econometrica 82 (2), 463-505.9
Kilian, L., 2011. Structural Vector Autoregressions. University of Michigan, mimeo.10
Liu, Z., Waggoner, D., Zha, T., 2011. Sources of Macroeconomic Fluctuations: A Regime-Switching DSGE11
Approach. Quantitative Economics 2 (2), 251-301.12
Lustig, H. N., Nikolai, L. R., Adrien, V., 2014. Countercyclical Currency Risk Premia. Journal of Financial13
Economics, 111 (3), 527-553.14
Lund-Jensen, K., 2012. Monitoring Systemic Risk Based on Dynamic Thresholds. IMF Working Paper No.15
12/159.16
Maddaloni A., Peydro, J.-L., 2011. Bank Risk-Taking, Securitization, Supervision, and Low Interest Rates:17
Evidence from Euro-Area and US Lending Standards. Review of Financial Studies 24, 2121-2165.18
Maggiori, M., 2010. The U.S. Dollar Safety Premium. Working paper, NYU.19
McGuire, P., Goetz, V. P., 2009. The US Dollar Shortage in Global Banking. BIS Quarterly Review, March20
2009, Bank for International Settlements.21
Rey, H., 2013. Dilemma not Trilemma: The global financial cycle and monetary policy independence.22
Jackson Hole Economic Symposium.23
Schularick, M., Taylor, A. M., 2012. Credit Booms Gone Bust: Monetary Policy, Leverage Cycles, and24
Financial Crises, 1870-2008. American Economic Review 102, 1029-61.25
Shin, H. S., 2012. Global Banking Glut and Loan Risk Premium. Mundell-Fleming Lecture, IMF Economic26
Review 60 (2), 155-192.27
Shin, H. S., 2013. The Second Phase of Global Liquidity and Its Impact on Emerging Economies. Remarks28
at 2013 Federal Reserve Bank of San Francisco Asia Economic Policy Conference.29
Sims, C. A., 1980. Macroeconomic and Reality. Econometrica 48, 1-48.30
Stock, J. H., Watson, M. W., 2001. Vector Autoregressions. Journal of Economic Perspectives 15, 101-115.31
Taylor, J. B., 2013. International Monetary Coordination and the Great Deviation. Journal of Policy32
Modeling 35 (3), 463-472.33
Capital flows and the risk-taking channel of monetary policy 19
Capital flows and the risk-taking channel of monetary policy 19
Categories of Cross-border Liabilities by Counterparty (All countries)
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
Tri
llio
n U
S d
olla
rs
Non-bank to Non-bank
Non-bank to Bank
Bank to Non-bank
Bank to Bank
Bank to Bank Liabilities as Percentage of Private Credit
0%
5%
10%
15%
20%
25%
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Per
cent
of P
rivat
e C
redi
t
All
Developed
Developing
Figure 1: Cross-border liabilities by type of counterparty. Left panel shows cross-border debt liabilities by pairwiseclassification of borrower and lender. “Bank to bank” refers to cross-border claims of banks on other banks (BIS bankingstatistics table 7A minus 7B). “Bank to non-bank” refers to cross-border claims of banks on non-banks (BIS table 7B). Claimsof non-banks are from BIS international debt security statistics, tables 11A and 11B). The right panel shows cross-borderbank-to-bank debt liabilities as percentage of total private credit in recipient economy. GDP and private credit data are fromthe World Bank.
1
Figure 1: Cross-border liabilities by type of counterparty.
Left panel shows cross-border debt liabilities by pairwise classification of borrower and lender. “Bank tobank” refers to cross-border claims of banks on other banks (BIS banking statistics table 7A minus 7B).“Bank to non-bank” refers to cross-border claims of banks on non-banks (BIS table 7B). Claims of non-banksare from BIS international debt security statistics, tables 11A and 11B). The right panel shows cross-borderbank-to-bank debt liabilities as percentage of total private credit in recipient economy. GDP and privatecredit data are from the World Bank.
Capital flows and the risk-taking channel of monetary policy 20
2007Q2
5.0
10.0
15.0
20.0
25.0
30.0
35.0
1995Q
4
1996Q
4
1997Q4
1998Q
4
1999Q
4
2000Q4
2001Q
4
2002Q4
2003Q
4
2004Q
4
2005Q4
2006Q
4
2007Q
4
2008Q
4
2009Q
4
2010Q4
2011Q
4
10.0
15.0
20.0
25.0
30.0
35.0
2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0
log_vix(-1)
BD
lev
era
ge
Figure 2: The left panel plots the leverage of the US broker dealer sector from the Federal Reserve’s Flow of Funds series(1995Q4 - 2012Q2). Leverage is defined as (equity + total liabilities)/equity. The right panel plots the scatter chart of USbroker dealer leverage against the log VIX index lagged one quarter. The dark shaded squares are the post-crisis observationsafter 2007Q4 (Source: Federal Reserve and CBOE)
Recursive
VAR
Ordering
12 3 4
Impact of (↓) On Fed Funds On BD Leverage On VIX On US dollar REER
Fed Funds-.5
0
.5
1
0 5 10 15 20-1
-.5
0
.5
1
0 5 10 15 20
-.05
0
.05
0 5 10 15 20
-.005
0
.005
0 5 10 15 20
BD Leverage-1
-.5
0
.5
0 5 10 15 20
0
.5
1
1.5
0 5 10 15 20
-.05
0
.05
0 5 10 15 20
-.005
0
.005
.01
0 5 10 15 20
VIX-.4
-.2
0
.2
.4
0 5 10 15 20
-1
-.5
0
.5
0 5 10 15 20
0
.05
.1
.15
0 5 10 15 20
-.01
-.005
0
.005
0 5 10 15 20
US dollar
REER-.4
-.2
0
.2
0 5 10 15 20
-.5
0
.5
1
0 5 10 15 20
-.04
-.02
0
.02
.04
0 5 10 15 20
-.01
0
.01
.02
0 5 10 15 20
Figure 3: Impulse response functions in recursive VAR. This figure presents estimated impulse-response functions forthe four variable recursive VAR (Fed Funds, BD leverage, VIX and REER) and 90 percent bootstrapped confidence intervalsfor the model with two lags, based on 1000 replications.
Figure 2: The left panel plots the leverage of the US broker dealer sector from the Federal Reserve’s Flow ofFunds series (1995Q4 - 2012Q2). Leverage is defined as (equity + total liabilities)/equity. The right panelplots the scatter chart of US broker dealer leverage against the log VIX index lagged one quarter. The darkshaded squares are the post-crisis observations after 2007Q4 (Source: Federal Reserve and CBOE)
Capital flows and the risk-taking channel of monetary policy 20
2007Q2
5.0
10.0
15.0
20.0
25.0
30.0
35.0
1995Q
4
1996Q
4
1997Q4
1998Q
4
1999Q
4
2000Q4
2001Q
4
2002Q4
2003Q
4
2004Q
4
2005Q4
2006Q
4
2007Q
4
2008Q
4
2009Q
4
2010Q4
2011Q
4
10.0
15.0
20.0
25.0
30.0
35.0
2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0
log_vix(-1)
BD
lev
era
ge
Figure 2: The left panel plots the leverage of the US broker dealer sector from the Federal Reserve’s Flow of Funds series(1995Q4 - 2012Q2). Leverage is defined as (equity + total liabilities)/equity. The right panel plots the scatter chart of USbroker dealer leverage against the log VIX index lagged one quarter. The dark shaded squares are the post-crisis observationsafter 2007Q4 (Source: Federal Reserve and CBOE)
Recursive
VAR
Ordering
12 3 4
Impact of (↓) On Fed Funds On BD Leverage On VIX On US dollar REER
Fed Funds-.5
0
.5
1
0 5 10 15 20-1
-.5
0
.5
1
0 5 10 15 20
-.05
0
.05
0 5 10 15 20
-.005
0
.005
0 5 10 15 20
BD Leverage-1
-.5
0
.5
0 5 10 15 20
0
.5
1
1.5
0 5 10 15 20
-.05
0
.05
0 5 10 15 20
-.005
0
.005
.01
0 5 10 15 20
VIX-.4
-.2
0
.2
.4
0 5 10 15 20
-1
-.5
0
.5
0 5 10 15 20
0
.05
.1
.15
0 5 10 15 20
-.01
-.005
0
.005
0 5 10 15 20
US dollar
REER-.4
-.2
0
.2
0 5 10 15 20
-.5
0
.5
1
0 5 10 15 20
-.04
-.02
0
.02
.04
0 5 10 15 20
-.01
0
.01
.02
0 5 10 15 20
Figure 3: Impulse response functions in recursive VAR. This figure presents estimated impulse-response functions forthe four variable recursive VAR (Fed Funds, BD leverage, VIX and REER) and 90 percent bootstrapped confidence intervalsfor the model with two lags, based on 1000 replications.
Figure 3: Impulse response functions in recursive VAR. This figure presents estimated impulse-responsefunctions for the four variable recursive VAR (Fed Funds, BD leverage, VIX and REER) and 90 percentbootstrapped confidence intervals for the model with two lags, based on 1000 replications.
Capital flows and the risk-taking channel of monetary policy 20
Capital flows and the risk-taking channel of monetary policy 21
Impact of Fed Funds
on VIX
Impact of VIX
on BD Leverage
Impact of Fed Funds
on BD Leverage
-.05
0
.05
0 5 10 15 20
-1
-.5
0
.5
0 5 10 15 20-1
-.5
0
.5
1
0 5 10 15 20
Figure 4: Impulse response functions in recursive VAR. This figure presents three panels from the impulse responsefunctions of the four variable VAR (Fed Funds, BD leverage, VIX and REER) illustrating the impact of a Fed Funds targetrate shock on the leverage of the US broker dealer sector. A positive Fed Funds target rate shock leads to a decline in brokerdealer leverage, via the fall in the VIX index. The panels show 90 percent bootstrapped confidence intervals for the model withtwo lags, based on 1000 replications.
Variance decomposition: impact of Fed Funds shocks
On US dollar REER On VIX On BD Leverage On Fed Funds
0
.07
1
0 5 10 15 200
.1
.2
.3
.5
1
0 5 10 15 20
0
.1
.2
1
0 5 10 15 200
.5
1
0 5 10 15 20
Variance decomposition: impact of BD Leverage shocks
On US dollar REER On VIX On BD Leverage On Fed Funds
0
.1
.2
.3
1
0 5 10 15 20
0
.1
.2
1
0 5 10 15 20
0
.5
1
0 5 10 15 20
0
.2
.4
1
0 5 10 15 20
Figure 5: Variance Decomposition. This figure presents variance decompositions from the four variable VAR giving thefractions of the structural variance due to Fed Fund or Leverage shocks for the four variables REER, VIX, BD Leverage andFed Fund (model with 2 lags).
Figure 4: Impulse response functions in recursive VAR. This figure presents estimated impulse-response functions for the four variable recursive VAR (Fed Funds, BD leverage, VIX and REER) and 90percent bootstrapped confidence intervals for the model with two lags, based on 1000 replications.
Capital flows and the risk-taking channel of monetary policy 21
Impact of Fed Funds
on VIX
Impact of VIX
on BD Leverage
Impact of Fed Funds
on BD Leverage
-.05
0
.05
0 5 10 15 20
-1
-.5
0
.5
0 5 10 15 20-1
-.5
0
.5
1
0 5 10 15 20
Figure 4: Impulse response functions in recursive VAR. This figure presents three panels from the impulse responsefunctions of the four variable VAR (Fed Funds, BD leverage, VIX and REER) illustrating the impact of a Fed Funds targetrate shock on the leverage of the US broker dealer sector. A positive Fed Funds target rate shock leads to a decline in brokerdealer leverage, via the fall in the VIX index. The panels show 90 percent bootstrapped confidence intervals for the model withtwo lags, based on 1000 replications.
Variance decomposition: impact of Fed Funds shocks
On US dollar REER On VIX On BD Leverage On Fed Funds
0
.07
1
0 5 10 15 200
.1
.2
.3
.5
1
0 5 10 15 20
0
.1
.2
1
0 5 10 15 200
.5
1
0 5 10 15 20
Variance decomposition: impact of BD Leverage shocks
On US dollar REER On VIX On BD Leverage On Fed Funds
0
.1
.2
.3
1
0 5 10 15 20
0
.1
.2
1
0 5 10 15 20
0
.5
1
0 5 10 15 20
0
.2
.4
1
0 5 10 15 20
Figure 5: Variance Decomposition. This figure presents variance decompositions from the four variable VAR giving thefractions of the structural variance due to Fed Fund or Leverage shocks for the four variables REER, VIX, BD Leverage andFed Fund (model with 2 lags).Figure 5: Variance Decomposition. This figure presents variance decompositions from the four variable
VAR giving the fractions of the structural variance due to Fed Fund or Leverage shocks for the four variablesREER, VIX, BD Leverage and Fed Fund (model with 2 lags).
Capital flows and the risk-taking channel of monetary policy 22
Impact of monetary policy shocks
On Exchange Rate On VIX On BD Leverage
Taylor Rule residual as monetary policy shock
-.005
0
.005
.01
0 5 10 15 20-.05
0
.05
0 5 10 15 20
-1
-.5
0
.5
1
0 5 10 15 20
M1 growth as monetary policy shock
-.01
-.005
0
.005
0 5 10 15 20
-.1
-.05
0
.05
0 5 10 15 20-.5
0
.5
1
0 5 10 15 20
Nominal effective Fed Funds rate as monetary policy shock
-.002
0
.002
.004
.006
0 5 10 15 20
-.05
0
.05
0 5 10 15 20-1
-.5
0
.5
0 5 10 15 20
Figure 6: Alternative definitions of monetary policy shocks. This figure shows the impulse-response functions and 90percent confidence bands for alternative monetary policy shocks on REER, VIX and BD-leverage in the four-variable modelwith two lags and 1000 bootstrapped standard errors. Monetary policy shocks considered are residuals from a Taylor rule, M1growth and nominal effective Fed Funds rate.
-12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
Mar.1999
Dec.1999
Sep.2000
Jun.2001
Mar.2002
Dec.2002
Sep.2003
Jun.2004
Mar.2005
Dec.2005
Sep.2006
Jun.2007
Mar.2008
Dec.2008
Sep.2009
Jun.2010
Tril
lion
Dol
lars U.S. dollar assets of
banks outside US
Euro assets of banksoutside eurozone
Sterling assets ofbanks outside UK
Yen assets of banksoutside Japan
Yen liabilities of banksoutside Japan
Sterling liabilities ofbanks outside UK
Euro liabilities of banksoutside eurozone
U.S. dollar liabilities ofbanks outside US
Figure 7: Foreign currency assets and liabilities of BIS reporting banks by currency (Source: BIS locational banking statistics,Table 5A)
Figure 6: Alternative definitions of monetary policy shocks. This figure shows the impulse-responsefunctions and 90 percent confidence bands for alternative monetary policy shocks on REER, VIX and BD-leverage in the four-variable model with two lags and 1000 bootstrapped standard errors. Monetary policyshocks considered are residuals from a Taylor rule, M1 growth and nominal effective Fed Funds rate.
Capital flows and the risk-taking channel of monetary policy 21
Capital flows and the risk-taking channel of monetary policy 22
Impact of monetary policy shocks
On Exchange Rate On VIX On BD Leverage
Taylor Rule residual as monetary policy shock
-.005
0
.005
.01
0 5 10 15 20-.05
0
.05
0 5 10 15 20
-1
-.5
0
.5
1
0 5 10 15 20
M1 growth as monetary policy shock
-.01
-.005
0
.005
0 5 10 15 20
-.1
-.05
0
.05
0 5 10 15 20-.5
0
.5
1
0 5 10 15 20
Nominal effective Fed Funds rate as monetary policy shock
-.002
0
.002
.004
.006
0 5 10 15 20
-.05
0
.05
0 5 10 15 20-1
-.5
0
.5
0 5 10 15 20
Figure 6: Alternative definitions of monetary policy shocks. This figure shows the impulse-response functions and 90percent confidence bands for alternative monetary policy shocks on REER, VIX and BD-leverage in the four-variable modelwith two lags and 1000 bootstrapped standard errors. Monetary policy shocks considered are residuals from a Taylor rule, M1growth and nominal effective Fed Funds rate.
-12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
Mar.1999
Dec.1999
Sep.2000
Jun.2001
Mar.2002
Dec.2002
Sep.2003
Jun.2004
Mar.2005
Dec.2005
Sep.2006
Jun.2007
Mar.2008
Dec.2008
Sep.2009
Jun.2010
Tril
lion
Dol
lars U.S. dollar assets of
banks outside US
Euro assets of banksoutside eurozone
Sterling assets ofbanks outside UK
Yen assets of banksoutside Japan
Yen liabilities of banksoutside Japan
Sterling liabilities ofbanks outside UK
Euro liabilities of banksoutside eurozone
U.S. dollar liabilities ofbanks outside US
Figure 7: Foreign currency assets and liabilities of BIS reporting banks by currency (Source: BIS locational banking statistics,Table 5A)
Figure 7: Foreign currency assets and liabilities of BIS reporting banks by currency (Source: BIS locationalbanking statistics, Table 5A)
Capital flows and the risk-taking channel of monetary policy 23
2008Q1
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
11.0
1999Q1
2000Q2
2001Q3
2002Q4
2004Q1
2005Q2
2006Q3
2007Q4
2009Q1
2010Q2
TrillionDollars
US charteredcommercialbanks' totalfinancialassets
US dollarassets ofbanks outsideUS
Figure 8: US dollar foreign currency claims of BIS-reporting banks and US commercial bank total assets (Source: Flow ofFunds, Federal Reserve and BIS locational banking statistics, Table 5A)Figure 8: US dollar foreign currency claims of BIS-reporting banks and US commercial bank total assets
(Source: Flow of Funds, Federal Reserve and BIS locational banking statistics, Table 5A)
Capital flows and the risk-taking channel of monetary policy 22Capital flows and the risk-taking channel of monetary policy 24
Recursive
VAR
Ordering
12 3 4 5
Impact of (↓) On Fed Funds On BD Leverage On BIS bank flows On VIX On USD REER
Fed Funds-.5
0
.5
1
0 5 10 15 20
-.5
0
.5
0 5 10 15 20
-.01
-.005
0
.005
0 5 10 15 20-.05
0
.05
0 5 10 15 20
-.005
0
.005
0 5 10 15 20
BD
Leverage-1
-.5
0
0 5 10 15 20
-.5
0
.5
1
1.5
0 5 10 15 20
-.01
0
.01
0 5 10 15 20
-.05
0
.05
0 5 10 15 20
-.005
0
.005
0 5 10 15 20
BIS
Bank
flows-.2
0
.2
.4
0 5 10 15 20
-.5
0
.5
1
0 5 10 15 20-.02
0
.02
.04
0 5 10 15 20
-.1
-.05
0
.05
0 5 10 15 20
-.005
0
.005
.01
0 5 10 15 20
VIX-.4
-.2
0
.2
.4
0 5 10 15 20
-1
-.5
0
.5
0 5 10 15 20
-.01
0
.01
0 5 10 15 20
0
.05
.1
0 5 10 15 20
-.005
0
.005
0 5 10 15 20
USD
REER-.4
-.2
0
.2
.4
0 5 10 15 20
0
.5
0 5 10 15 20
-.01
-.005
0
.005
.01
0 5 10 15 20
-.04
-.02
0
.02
0 5 10 15 20
-.01
0
.01
.02
0 5 10 15 20
Figure 9: Impulse response functions in recursive VAR. This figure presents estimated impulse-response functions forthe five variable structural VAR (Fed Funds, BD leverage, BIS bank flows, VIX and REER) and 90 percent bootstrappedconfidence intervals for the model with two lags, based on 1000 replications.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––—
Figure 9: Impulse response functions in recursive VAR. This figure presents estimated impulse-response functions for the five variable structural VAR (Fed Funds, BD leverage, BIS bank flows, VIXand REER) and 90 percent bootstrapped confidence intervals for the model with two lags, based on 1000replications.
Capital flows and the risk-taking channel of monetary policy 25
Impact of Fed Funds
on BD Leverage
Impact of VIX
on BD Leverage
Impact of BD Leverage
on BIS bank flows
-.5
0
.5
0 5 10 15 20
-1
-.5
0
.5
0 5 10 15 20
-.01
0
.01
0 5 10 15 20
Impact of BD Leverage
on REER
Impact of Fed Funds
on BIS bank flows
Impact of Fed Funds
on REER
-.005
0
.005
0 5 10 15 20
-.01
-.005
0
.005
0 5 10 15 20-.005
0
.005
0 5 10 15 20
Figure 10: Impulse response functions in recursive VAR. This figure presents six panels from the impulse responsefunctions of the five variable VAR (Fed Funds, BD leverage, BIS bank flows, VIX and REER) illustrating the impact of a FedFunds target rate shock on BIS bank capital flows and REER. A positive Fed Funds target rate shock leads to decline in bankcapital flows, via the fall in the leverage of the banking sector. The panels show 90 percent bootstrapped confidence intervalsfor the model with two lags, based on 1000 replications.
Figure 10: Impulse response functions in recursive VAR.
This figure presents six panels from the impulse response functions of the five variable VAR (Fed Funds,BD leverage, BIS bank flows, VIX and REER) illustrating the impact of a Fed Funds target rate shock onBIS bank capital flows and REER. A positive Fed Funds target rate shock leads to decline in bank capitalflows, via the fall in the leverage of the banking sector. The panels show 90 percent bootstrapped confidenceintervals for the model with two lags, based on 1000 replications.