There was interesting movement in yesterday morning's markets after the (mostly expected) disappointing 1Q GDP report. Equities fell slightly as trading began, but bond yields jumped. Considering the current context of a tentative recovery in economic production and interest rates, we might have expected interest rates to decline as a result of a disappointing economic report.

I have previously speculated that the non-normal distribution of interest rate expectations resulting from the zero lower bound may be creating odd behaviors in the forward rate curve, and I think this may be an example.

This first chart describes a context where there is a negative skew in forward rate expectations that is truncated by the zero lower bound. So, the mode expected rates are higher than the median expected rates. Normally, the mean expected rate, which we might suspect to be near the market rate, would be even lower, but the zero lower bound causes a distortion in the distribution of expectations which pushes the mean higher.

As uncertainty declines, which we would expect over time as we approach a given date, the variance in the distribution would decline, and the mode, mean, and median would converge. This is one factor that I think is in play in the current forward rate market.

Now, if one were to take a naïve position on forward rates to take advantage of this adjustment over time, this would be basically a position that earns profits by taking on long-tail negative risks. If some very negative economic development comes to pass, rates will decline sharply. But, as long as the economy progresses somewhat reasonably, these profits will accrue. It's the proverbial nickel in front of the steamroller.

I am using growth in the mortgage market as a signal of economic expansion, so I am taking this position with the idea that the mortgage signal gives me insight into the likelihood of failure. The hope is to capture the gains from the trend but to exit the position if the mortgage signal weakens.

This next graph is a very simple example of the influences we might see on rates. The orange line is a hypothetical forward yield curve, based on rates rising in September and climbing at a somewhat conservative rate of 35 basis points per quarter. The blue line is a bimodal distribution, one set of investors expecting the orange line and one set of investors expecting us to remain at the zero lower bound. Here, I have set the expectation of persistent ZLB at 50%. I'm sure that is higher than the market expectation, but for now I'm not too concerned with getting the numbers exactly right.

In the next graph, the blue line remains, and it compares to today's opening forward rates (the purple line). The red line is the high mark in the forward rate market after the GDP report today. The dark blue line is a hypothetical yield curve with adjusted expectations. The expected date of the first rate hike didn't seem to move today. But, the GDP report would, counterintuitively, have had two countervailing effects on forward rate expectations.

First, the disappointing numbers would have reduced the expected slope of rate hikes. But, second, the fact that the report wasn't catastrophic would have reduced the variance in expected outcomes. In this simple model, that would mean that the percentage of investors with the ZLB expectation would fall. If the change in rates after the report reflects no change in the expected slope, then a reduction in the ZLB percentage of 3% would create today's rate changes. If the change in rates after the report reflects a decline of 2 basis points per quarter in the slope, then a decline of 8% in the proportion of investors expecting a persistent ZLB would pull us up to the later market rates.

So, my point is that as time passes and we avoid a complete meltdown, rates should move naturally up from the blue line to some version of the orange line. I also think there is another factor at work here, which is that the lack of nominal real estate investments is creating an oversupply of savings in the long term bond market, and this is dragging down forward rates in general, especially at the long end of the curve. So, right now, the orange line probably reflects a lower slope (and, there is also probably a smaller probability of remaining at the ZLB). But, in addition to seeing the median and mean expected rates move up toward the orange line as the ZLB expectations decline, we should also see the orange line move up, too, as savings finds new outlets in the real estate sector. This could happen through new building or mortgage growth, and it is clearly a convergence that will happen in a functional recovery from this point. Returns to real estate are far outside the historical norm compared to bonds.

Since each forward rate is the product of its own set of forward expectations, there is a chance that news of persistent growth in new home building or mortgage expansion could cause fairly large sudden shifts in this direction. If recovery persists and mortgages expand, an eventual rise of around 2% or so at the long end of the yield curve seems like a reasonable expectation.

The US is not Japan or Europe. But we may see rates go down not up. After all, what is the track record of the last 30 years?

ReplyDeleteThe Primary Dealer survey from the Fed shows an expected slope of about 30bp per quarter and a 15% or so chance of pulling back to the ZLB. But, the same survey shows expected short term rates that are 1% or more higher than the current forward market rates. Strange how these surveys can be so far from the market prices.

Deletehttp://www.newyorkfed.org/markets/survey/2015/March-result.pdf