Publication - Market volatility, monetary policy and the term premium - by Sushanta K Mallick, Madhusudan Mohanty and Fabrizio Zampolli
This paper provides an empirical assessment of the role that measures of expected financial market volatility and monetary policy play in the determination of the government bond term premium. Three main considerations motivate such a focus.
First, in financially sophisticated economies such as the United States the long-term interest rate is an important determinant of broad financial conditions and potentially of real activity. Second, changes in the long-term interest rate are not driven only by changes in the expected path of future short-term interest rates, as assumed by mainstream macroeconomic theory,1 but also by changes in the term premium. Indeed, the possibility of influencing the term premium directly is one of the motivations behind the large-scale purchases of bonds by the Federal Reserve postcrisis. Third, risk-taking and its interaction with monetary policy and bond dealers’ leverage are in principle important, but relatively unexplored, mechanisms in the determination of the term premium.
Risk-neutral measures of expected volatility, such as the Chicago Board Options Exchange Volatility Index (VIX) or the Merrill Lynch Option Volatility Expectations (MOVE) index, are a proxy not only for investors’ uncertainty about future equity and bond prices, respectively, but also for investors’ risk aversion.2 When uncertainty or risk aversion increases, investors may demand higher compensation for holding risky assets, including long-duration government securities. Furthermore, investors that actively manage their balance sheet (such as bond dealers) may face tighter funding or capital constraints and therefore be forced to reduce their exposures. Thus, market-making capacity and liquidity in the bond market may decline, leading to wider swings in bond yields (eg Gromb and Vayanos (2002), Brunnermeier and Pedersen (2005), He and Krishnamurthy (2008) and Adrian and Shin (2010)). Conversely, a decline in expected volatility may induce investors to take on more risk as well as improve market liquidity, thus leading to a narrowing of term premia.3 To the extent that monetary policy affects investors’ perception of uncertainty and risks, risk-taking in bond markets may therefore constitute an additional channel of monetary transmission.
The interaction between the term premium, expected volatility and monetary policy has not received much attention in the empirical literature. Recent empirical work supports the view that a time-varying term premium is an important channel in the monetary transmission mechanism (Gertler and Karadi (2015)), a finding that is sharply at odds with the standard New Keynesian model (Woodford (2003)). But this work does not examine how the term premium is affected by changes in expected volatility or other measures of risk aversion or uncertainty. By contrast, other empirical studies have documented how expected stock market volatility, as captured by the VIX, has relevant macroeconomic effects. Bekaert et al (2013) find that looser monetary policy reduces the risk aversion component of the VIX and increases output, while Bruno and Shin (2015) shows that a decline in US dollar bank funding costs leads to an increase in bank leverage through the mitigation of volatility risk.4 Yet none of these studies pay attention to expected bond market volatility and the term premium, despite the bond market having gained greater relevance post-crisis.
A casual inspection of Graph 1 suggests that, post-crisis, spells of higher expected bond market volatility, as captured by the MOVE index, have been accompanied by upward movements in the 10-year term premium – for example, during the 2013 taper tantrum episode (Graph 1).5 However, other drivers, including asset purchases, may be relevant. An interesting question is whether the VIX also counts. While it may be tempting to assume that only a measure of expected volatility matters for the term premium, the VIX may also be empirically relevant. For one, investors may hold diversified portfolios of bonds and equities. Second, the VIX has been shown to correlate well not only with US equity prices, but also with the prices of a range of other risky assets, including foreign assets. Third, while the MOVE and the VIX tend to co-move, their correlation changes over time. Post-crisis, the two measures have occasionally diverged (Graph 1, left-hand panel). All of this suggests that both volatility indicators, rather than the MOVE alone, may be useful in characterising the overall attitude towards risk and the degree of uncertainty faced by bond investors.
Switzerland BISBCHBB
Website:www.bis.org
