(This post takes as its jumping off point the market monetarist point of view that monetary policy has been too tight since 2007, and that, because of this, stock market returns and interest rates have been positively correlated during that time, as periods of looser monetary policy have resulted in higher inflation expectations and real economic growth expectations.)
I think what may have been missing from simple models of the nominal interest rate is an inflation uncertainty factor. When we compare TIPS and nominal bonds, we can get a basic idea of real rates & inflation expectations. But, because of the uncertainty around the Fed’s expanded balance sheet and duration risk, even if expected inflation is typical, there is probably a much larger range of possible inflation outcomes than usual. The effect of this uncertainty should be threefold:
1) TIPS bonds would fetch a premium above the level produced by average inflation expectations alone. This would create the illusion of lower real rates and higher expected inflation when comparing TIPS to nominal rates.
2) Because this is a risk premium, the premium would be shared by both TIPS and nominal bonds. In other words, as TIPS are bid up, some investors will choose to purchase nominal bonds because they will be willing to take on inflation uncertainty in order to capture some of the premium from #1.
3) The net effect will be to lower yields in general because concerns about future Fed errors related to the management of the balance sheet lower average expectations about future economic growth. In summary, the inflation yield will look higher and the real yield will look lower, in equal amounts, plus there will be an additional downward effect on the actual real yield due to the uncertainty itself.
So, how does this help to understand recent market movements?
First, what exactly happened in the yield curve? I have a model that uses Eurodollar futures to derive the expected date of the first rate increase, the quarterly rate of subsequent rate increases, and the level of uncertainty around the timing of the first increase. This model shows two distinct periods of activity:
1) From the end of April to mid-June, rising forward rates suggest a movement in the first rate increase from late 2015 to late 2014. This presumably was the product of improving expectations about the economy, which could be from a mixture of expectations of Fed accommodation and empirically improving macro data.
2) After mid-June, the expected date of the rate increase remained around the end of 2014. What changed was an increase in the slope of the forward yield curve, and a decrease in the uncertainty around the timing of the first rate increase (first around June 13, but then especially around the Fed meeting on June 18-19). Without getting into the nuts & bolts of the model, in the graph below, you can see how the second period of change comes from a sharper convexity around the expected rate increase and a steeper slope for the later dates, whereas the June 11 curve is basically the May 1 curve moved back 1 year in time.
My speculation is that the May & early June rate increases are easily explained by improved economic expectations. And the late June interest rate movement did not come from Fed tightening. In fact, the stationary expected date of the rate change suggests that the market did not see any net change in the Fed’s monetary stance. I think the largest effect was the clarification of the winding down of QE3, which decreased the inflation uncertainty factor discussed above. The market’s marginal expectations about inflation and Fed OMO have not changed, but the market is much more confident about it, and is less worried about a premature or late exit from QE3. What effects would we expect this change to have?
1) Apparent real rates in the TIPS market would go up.
2) Apparent inflation expectations would go down.
3) Actual real rates would go up, as confidence about managed economic expansion increases, raising nominal rates.
4) And we can add one further expectation: Stock performance has been positively correlated with rising rates since 2008, which is a sign that monetary policy has been too tight. Why, then, did the S&P 500 dip on June 19 at the same time that interest rates jumped? If we take this uncertainty factor into account, we would expect that there would be two factors affecting the stock market. (1) a bullish effect from increased certainty about the end of QE3 and (2) a bearish effect from the decrease in the inflation uncertainty premium, since stocks can serve as an inflation hedge. The net effect would be indeterminate, but could be slightly bearish, which is what we saw.
I don’t have a systematic way of measuring this uncertainty effect, but for the purpose of visualizing it, I have manually created an estimate of the effect based on the quantity of Fed holdings, the increase in duration during Operation Twist, and gradual improvement as positive economic news came in over 2013, with a one time positive shock at the Fed announcement on June 19. This is admittedly ad hoc and unscientific.
The dotted lines below represent the typical Real Rate & Inflation component derived from TIPS and nominal rates. The solid lines include my imprecise uncertainty effect and the inflation and real components once they have been adjusted for the distortions of this effect on TIPS yields. Whether or not the scale of the effect is accurate, as graphed, the basic idea remains that the actual effect of the QEs on inflation expectations has probably been decreasing over time, relative to the effect suggested by the TIPS spread; the effect of the QEs on the real rate has been stronger than suspected, but has had to fight the increasing uncertainty premium until recent economic optimism led to a bottoming of this effect; and the recent reactions of interest rates and stock prices to Fed clarifications setting more precise limits on QE3 could have resulted from a relatively small pulse change in this effect, not from changes in the real rate or inflation expectations.