With monetary authorities around the world preparing for their exit, there are fears in some circles that a new Armageddon is in sight. Volatility could shoot up, it is argued, as investors try to figure out the impact of a synchronous global tightening on their respective asset classes—let alone the difference between, say, the effective fed funds rate and the interest on excess reserves!
The fears are not unjustified, so I thought of going back to see whether history, can inform us about the chances of an impending “Vola-geddon”!
I’ll focus on the behavior of volatility during three “exit” precedents: (a) The Bank of Japan (BoJ)’s exit in 2006, which is the only available precedent of a central bank’s exit from quantitative easing (QE); (b) the Fed’s exit in 2004, which followed a stated commitment to a “low for long” rates policy and therefore bears similarities to the current situation; and (c) the Fed’s exit in 1994, a year that saw volatility in capital markets go up.
Earlier exit episodes are not as relevant, mainly because central bank communications were much more opaque than today—e.g. prior to February 1994, the Fed did not even announce its target policy rate, while changes in the target rate were often made outside scheduled FOMC meetings, leaving markets guessing.
So what does experience tell us?
First “lesson” is that, under normal circumstances (and I’ll define “abnormal” below), central banks will begin their exit only after the recovery seems to be well entrenched. This moment is usually associated by a preceding period of steadily declining volatility/risk aversion.
For example, by the time the Fed hiked in Feb 1994, the VIX (volatility implied from options on the S&P500) had been on a declining path for more than two years, reflecting the improving risk environment. Ditto for the Fed’s 2004 exit and the BoJ’s in March 2006. In other words, by the time the hikes begin, markets are pretty well equipped to withstand a rise in volatility, were this to occur.
Now, does volatility rise during exits? Clearly my sample is minute but there is a useful qualitative comparison between the 2004 and 2006 episodes (when the VIX did not rise); and the 1994 one (when, on the day of the Feb 1994 announcement, the VIX jumped 50% and hovered around those levels throughout the year).
In the case of the former two, the exit was largely anticipated by the market. You can see that by looking at the 3-month Libor expected three months forward (3m/3m). In the US, this had already begun to move up in early April, even though the first hike actually occurred end-June. In Japan, it moved as early as October 2005, around the time when the BoJ published its economic outlook, which hinted that the exit from QE was near.
In contrast, in 1994, 3m/3m rates suggest that the Fed’s February hike was largely unanticipated. News items at the time also point to ongoing market uncertainty about the impact of successive rate increases on the growth outlook (Mexico’s tequila crisis also contributed to higher vol later that year).
These examples provide support to the hypothesis that the improvement in the market’s understanding of central banks’ modus operandi—itself the result of enhanced central bank transparency over the years—has helped reduce policy-related uncertainty and financial market volatility. (Empirical research in the academic literature is consistent with this view).
Against this backdrop, both the Fed and the ECB are going at great lengths to explain the sequencing of their exit to the markets. This tells me that investors should take their words at face value when forming rates forecasts, rather than bracing for a nasty surprise.
Turning specifically to bond market volatility… Monetary policy can help contain it in (at least) two ways: First, with a credible commitment to low inflation, which anchors expectations and reduces the volatility of inflation forecasts; and second, by reducing uncertainty about the path of monetary policy itself.
The former calls for a strong emphasis on the path of inflation expectations as a guide to policy (an emphasis the Fed has explicitly affirmed in its FOMC statements); the latter calls (once again) for clear communication on the rationale and sequencing of the exit process.
So let’s what happened during the three exit episodes. In what follows, I’ll proxy uncertainty about the monetary policy path by the volatility in money markets, in line with a 1996 BIS paper by Borio and McCauley (volatility measured as the standard deviation in the daily change of the 3m/3m rate, over a 3-month period). The “test” then is to see whether money market volatility was “passed-through” to volatility in bonds of long-term maturities.
Turns out Japan is not a very informative case: Volatility in money markets was elevated during 2006 but remained within the bounds seen in late ’04 and throughout ’05. Importantly, the pass-through from money markets to long-term bond volatility was limited, especially for longer tenors (e.g. 10-year). I should mention that the BoJ did a good job explaining its exit plan ahead of time, in late-2005/early 2006. It’s just that the data do not seem to provide a conclusive link between monetary policy uncertainty (so defined) and bond market volatility.
In the US, volatility increased in both tightening episodes, though in 2004 the rise was shorter-lived—a brief interruption to a downward path in the context of a benign global risk environment. What differs from the Japanese case is that there is a clearer pass-through from money market to bond market volatility. In other words, *assuming* money market volatility can be interpreted as uncertainty about monetary policy, central bank communication may have more influence on bond market volatility in the US than in Japan.
So when could things go really wrong? I mentioned earlier I would define “abnormal” so here we go—a few “tail” events with Vola-geddon potential:
One has to do with the impact on long-term rates of the Fed’s downsizing of its balance sheet. Part of the reason is uncertainty about (a) how much impact the Fed’s MBS/Treasury purchases have had on long-term rates in the first place; and (b) whether it was the stock or the flow of Fed purchases that mattered (the Fed’s current thinking seems to be leaning towards the stock theory).
I personally expect a limited impact here, simply because the Fed is taking a very cautious approach—indicating it will opt for a passive downsizing in the short run (ie allowing its assets to mature), with active purchases only much later in the exit process. Note that the ECB did not really enter the asset purchase game (except for covered bonds) so that’s not an issue for them (but it’s a big issue for the Brits).
Another “tail” risk is a sudden shift in investor perceptions of sovereigns’ ability to service their debt (including the US). Unfortunately, triggers for such shifts are often “exogenous”: Sovereigns can be vulnerable for years (see Greece) before markets decide it’s time to pull the plug. So I have little to contribute in trying to predict the timing of this.
Finally, you could have a scenario where goldbugs and monetarist nutheads gain vogue, prompting a spike in inflation expectations that would force the Fed to act earlier and more forcefully on the rates front. Given the chatter in some circles, I can’t say that’s impossible, but in light of the dynamics in the real economy I’ll say just this: The probability I assign to this risk is exactly equal to the probability of Ben topping the charts with a song titled “I’m a printing machine”… you pick the number!
Originally published at Models & Agents and reproduced here with the author’s permission.