ریسک، عدم اطمینان و سیاست پولی

The VIX, the stock market option-based implied volatility, strongly co-moves with measures of the monetary policy stance. When decomposing the VIX into two components, a proxy for risk aversion and expected stock market volatility (“uncertainty”), we find that a lax monetary policy decreases both risk aversion and uncertainty, with the former effect being stronger. The result holds in a structural vector autoregressive framework, controlling for business cycle movements and using a variety of identification schemes for the vector autoregression in general and monetary policy shocks in particular. The effect of monetary policy on risk aversion is also apparent in regressions using high frequency data.

A popular indicator of risk aversion in financial markets, the VIX index, shows strong co-movements with measures of the monetary policy stance. Fig. 1 considers the cross-correlogram between the real interest rate (the Fed funds rate minus inflation), a measure of the monetary policy stance, and the logarithm of end-of-month readings of the VIX index. The VIX index essentially measures the “risk-neutral” expected stock market variance for the US S&P500 index. The correlogram reveals a very strong positive correlation between real interest rates and future VIX levels. While the current VIX is positively associated with future real rates, the relationship turns negative and significant after 13 months: high VIX readings are correlated with expansionary monetary policy in the medium-run future.The strong interaction between a “fear index” (Whaley, 2000) in the asset markets and monetary policy indicators may have important implications for a number of literatures. First, the recent crisis has rekindled the idea that lax monetary policy can be conducive to financial instability. The Federal Reserve's pattern of providing liquidity to financial markets following market tensions, which became known as the “Greenspan put,” has been cited as one of the contributing factors to the build-up of a speculative bubble prior to the 2007–09 financial crisis.1 Whereas some rather informal stories have linked monetary policy to risk-taking in financial markets (Rajan, 2006, Adrian and Shin, 2008 and Borio and Zhu, 2008), it is fair to say that no extant research establishes a firm empirical link between monetary policy and risk aversion in asset markets.2 Second, Bloom (2009) and Bloom et al. (2009) show that heightened “economic uncertainty” decreases employment and output. It is therefore conceivable that the monetary authority responds to uncertainty shocks, in order to affect economic outcomes. However, the VIX index, used by Bloom (2009) to measure uncertainty, can be decomposed into a component that reflects actual expected stock market volatility (uncertainty) and a residual, the so-called variance premium (see, for example, Carr and Wu, 2009), that reflects risk aversion and other non-linear pricing effects, perhaps even Knightian uncertainty. Establishing which component drives the strong co-movements between the monetary policy stance and the VIX is therefore particularly important. Third, analyzing the relationship between monetary policy and the VIX and its components may help clarify the relationship between monetary policy and the stock market, explored in a large number of empirical papers (Thorbecke, 1997, Rigobon and Sack, 2004 and Bernanke and Kuttner, 2005). The extant studies all find that expansionary (contractionary) monetary policy affects the stock market positively (negatively). Interestingly, Bernanke and Kuttner (2005) ascribe the bulk of the effect to easier monetary policy lowering risk premiums, reflecting both a reduction in economic and financial volatility and an increase in the capacity of financial investors to bear risk. By using the VIX and its two components, we test the effect of monetary policy on stock market risk, but also provide more precise information on the exact channel. This article characterizes the dynamic links between risk aversion, uncertainty and monetary policy in a simple vector-autoregressive (VAR) system. Such analysis faces a number of difficulties. First, because risk aversion and the stance of monetary policy are jointly endogenous variables and display strong contemporaneous correlation (see Fig. 1), a structural interpretation of the dynamic effects requires identifying restrictions. Monetary policy may indeed affect asset prices through its effect on risk aversion, as suggested by the literature on monetary policy news and the stock market, but monetary policy makers may also react to a nervous and uncertain market place by loosening monetary policy. In fact, Rigobon and Sack (2003) find that the Federal Reserve does systematically respond to stock prices.3 Second, the relationship between risk aversion and monetary policy may also reflect the joint response to an omitted variable, with business cycle variation being a prime candidate. Recessions may be associated with high risk aversion (see Campbell and Cochrane, 1999 for a model generating counter-cyclical risk aversion) and at the same time lead to lax monetary policy. Our VARs always include a business cycle indicator. Third, measuring the monetary policy stance is the subject of a large literature (see, for example, Bernanke and Mihov, 1998a); and measuring policy shocks correctly is difficult. Models featuring time-varying risk aversion and/or uncertainty, such as Bekaert et al. (2009), imply an equilibrium contemporaneous link between interest rates and risk aversion and uncertainty, through precautionary savings effects for example. Such relation should not be associated with a policy shock. However, our results are robust to alternative measures of the monetary policy stance and of monetary policy shocks. In particular, the results are robust to identifying monetary policy shocks using a standard structural VAR, using high frequency Fed funds futures changes following Gürkaynak et al. (2005), and using the approach in Bernanke and Kuttner (2005), based on the unexpected change in the Fed Funds rate on a monthly basis. The remainder of the paper is organized as follows. Section 2 details the measurement of the key variables in the VAR, including monetary policy indicators, monetary policy shocks and business cycle indicators. First and foremost, we provide intuition on how the VIX is related to the actual expected variance of stock returns and to risk preferences. While the literature has proposed a number of risk appetite measures (see Baker and Wurgler, 2007 and Coudert and Gex, 2008), our measure is increasing in risk aversion in a variety of realistic economic settings. This motivates our empirical strategy in which the VIX is split into a pure volatility component (“uncertainty”) and a residual, which should be more closely associated with risk aversion. Section 3 analyzes the dynamic relationship between monetary policy and risk aversion and uncertainty in standard structural VARs. The results are remarkably robust to a long list of robustness checks with respect to VAR specification, variable definitions and alternative identification methods. Section 4 employs two alternative methods to identify monetary policy shocks relying on Fed futures data.4 Our main findings are as follows. A lax monetary policy decreases risk aversion in the stock market after about nine months. This effect is persistent, lasting for more than two years. Moreover, monetary policy shocks account for a significant proportion of the variance of the risk aversion proxy. Monetary policy shocks have a significant impact on risk aversion also in regressions using high frequency data. The effects of monetary policy on uncertainty are similar but somewhat weaker. On the other hand, periods of both high uncertainty and high risk aversion are followed by a looser monetary policy stance but these results are less robust and weaker statistically. Finally, it is the uncertainty component of the VIX that has the statistically stronger effect on the business cycle, not the risk aversion component.

A number of recent studies point at a potential link between loose monetary policy and excessive risk-taking in financial markets. Rajan (2006) conjectures that in times of ample liquidity supplied by the central bank, investment managers have a tendency to engage in risky, correlated investments. To earn excess returns in a low interest rate environment, their investment strategies may entail risky, tail-risk sensitive and illiquid securities (“search for yield”). Moreover, a tendency for herding behavior emerges due to the particular structure of managerial compensation contracts. Managers are evaluated vis-à-vis their peers and by pursuing strategies similar to others, they can ensure that they do not under perform. This “behavioral” channel of monetary policy transmission can lead to the formation of asset prices bubbles and can threaten financial stability. Yet, there is no empirical evidence on the links between risk aversion in financial markets and monetary policy. This article has attempted to provide a first characterization of the dynamic links between risk, uncertainty and monetary policy, using a simple vector-autoregressive framework. Implied volatility is decomposed into two components, risk aversion and uncertainty, and the interactions between each of the components and monetary policy are studied under a variety of identification schemes for monetary policy shocks. It is consistently found that lax monetary policy increases risk appetite (decreases risk aversion) in the future, with the effect lasting for more than two years and starting to be significant after about nine months. The effect on uncertainty is similar but the immediate response of uncertainty to monetary policy shocks in high frequency regressions is weaker than that of risk aversion. Conversely, high uncertainty and high risk aversion lead to laxer monetary policy in the near-term future but these effects are not always statistically significant. These results are robust to controlling for business cycle movements. Consequently, our VAR analysis provides a clean interpretation of the stylized facts regarding the dynamic relations between the VIX and the monetary policy stance depicted in Fig. 1. The primary component driving the co-movement between past monetary policy stance and current VIX levels (first column of Fig. 1) is risk aversion but uncertainty also reacts to monetary policy. Both components of the VIX lie behind the negative relation in the opposite direction (second column of Fig. 1) but statistical confidence in this structural link is smaller. We hope that our analysis will inspire further empirical work and research on the exact theoretical links between monetary policy and risk-taking behavior in asset markets. A recent literature, mostly focusing on the origins of the financial crisis, has considered a few channels that deserve further scrutiny. Adrian and Shin (2008) stress the balance sheets of financial intermediaries and repo growth; Adalid and Detken (2007) and Alessi and Detken (2011) stress the buildup of liquidity through money growth; and Borio and Lowe (2002) emphasize rapid credit expansion.20 Recent work in the consumption-based asset pricing literature attempts to understand the structural sources of the VIX dynamics (see Bekaert and Engstrom, 2013, Bollerslev et al., 2009 and Drechsler and Yaron, 2011). Yet, none of these models incorporates monetary policy equations. In macroeconomics, a number of articles have embedded term structure dynamics into the standard New-Keynesian workhorse model (Bekaert et al., 2010 and Rudebusch and Wu, 2008), but no models accommodate the dynamic interactions between monetary policy, risk aversion and uncertainty, uncovered in this article. The policy implications of our work are also potentially important. Because monetary policy significantly affects risk aversion and uncertainty and these financial variables may affect the business cycle, we seem to have uncovered a monetary policy transmission mechanism missing in extant macroeconomic models. Fed chairman Bernanke (see Bernanke, 2002) interprets his work on the effect of monetary policy on the stock market (Bernanke and Kuttner, 2005) as suggesting that monetary policy would not have a sufficiently strong effect on asset markets to pop a “bubble” (see also Bernanke and Gertler, 2001, Gilchrist and Leahy, 2002 and Greenspan, 2002). However, if monetary policy significantly affects risk appetite in asset markets, this conclusion may not hold. If one channel is that lax monetary policy induces excess leverage as in Adrian and Shin (2008), perhaps monetary policy is potent enough to weed out financial excess. Conversely, in times of crisis and heightened risk aversion, monetary policy can influence risk aversion and uncertainty in the market place, and therefore affect real outcomes.