Last year, I wrote about how medium-term movements (roughly five months to five years) in the long-term bond market seemed to be determined by two types of cyclicality. The bond rally on Friday has inspired me to revisit those elements to see what they might have to say about the bond market over the next two years. My sense is that we will almost definitely see significantly higher rates in 2014, but it is not clear if the rise in rates has already begun or will begin later in 2013.
That being the case, my general assessment is that although there is a possibility that we could see a bond (IEF) rally this year, the danger of being caught in the cyclical snapback is so severe that I cannot imagine taking anything but limited, tactical positions in bonds. Based on analysis elsewhere, I continue to believe that equities (DIA) exhibit a higher return/risk profile over the next year.
Some secular determinants
First, with respect to secular movements in interest rates, I think it is important to point out a few generalities.
1. Bond yields appear to have a natural tendency to conform to patterns set by equity yields, but this tendency is often diluted under the dollar standard.
2. Interest rates appear to be negatively correlated with debt/GDP ratios under the dollar standard.
The relationship between long-term bond yields and equity yields is complex. Since 1960, we can generally say that long-term interest rates are a product of the spread between the earnings and dividend yields plus the real short-term rate, such that:
10yr = EY - DY + 1yr - CPI%
But, without having very precise estimations of each of these elements, I am not sure how we can utilize this equation to predict future movements. I generally believe that each of these elements on the right side of the equation will experience primarily downward pressure for the next year (assuming that we can extrapolate that from the assumption that we are in an environment characterized by expanding P/E multiples), but without precision, that seems analytically useless. Research on the spread between earnings and dividend yields appears to be such a wasteland that I don't even know where to begin to estimate that, except to note that the equity yield spread tends to approximate the rate of inflation itself. But so does the dividend yield (since 1960), leaving us with the unhelpful observation that, under the dollar standard, the 10-year yield can be estimated according to minute interactions between inflation and equity yields. If and when ZIRP ends, things will only become more complicated.
That leaves us with the federal debt/GDP ratio. Debt continues to accumulate, so the question is whether or not GDP will grow faster than debt and thereby raise interest rates. GDP itself is composed of nominal GDP and the rate of inflation. Unfortunately, I am not aware of any especially reliable ways to predict GDP, nominal or real, never mind how its rate of growth will compare to the rate of growth of government debt over an extended period of time.
So, not being aware of any compelling ways to predict long-term trends in the bond market and having seen people a lot smarter than me call a premature end to the bond bull, I turn to cyclical trends in the market that have existed roughly since the demise of Bretton Woods I.
The first cyclical element is the five-year cycle. Roughly every five years, Treasury yields acquire intermediate lows, shortly after which they attain new intermediate highs in a relatively violent move upwards.
As something of an aside, I would like to point out that, except for 2008, this tendency was even stronger for short-term yields, and if one assumes that an inverted yield curve either signals or causes recessions (or even oil shocks), one should note how many of these inverted yield curves 'result' from highs in five-year cycles. In this article, I would like to stick with the tedious business of predicting yields themselves, and start to address the relationship between yields, the oil/gold ratio, and oil shocks in a follow-up article.
So, to return to the topic at hand, in my previous treatment of this cycle, I focused on the lows, but I wonder if what we should really be contemplating is the highs.
In the chart below, I have calculated the averages of the five-year cycle (counting from both 1955 and 1970) and set them beside the current (2010-2015) cycle and the previous one (2005-2009). Of course, no single cycle conforms to the average, but since 1980, the correlations have been broadly stronger.
(Source: St Louis Fed)
Looking at the two averages, the primary shift in this pattern that appears to be occurring is a tendency for the intermediate low to appear later, as it did in the last cycle (2005-2009), which got through the door just in the nick of time.
This time, the steep descent in 2012 suggested that the intermediate low arrived early (not unlike 1987), and although the rise in interest rates since then hasn't been especially powerful on an arithmetical basis, on a logarithmic basis, it has been of comparable magnitude to the typical rise after a low in the cycle. But, with the increasing tendency for these lows to show up later in the fourth year (counting from peak to peak in this article), I feel a distinct lack of confidence in declaring the absolute intermediate low to have been marked in the summer of last year.
What I feel more confident in saying is that we are likely to see significantly higher rates in 2014 than we will this year, and the outstanding question is whether we will see the market take us there this year or next. So, when I compare the possibility that we have already seen the lows in interest rates (i.e., in 2012) with the probability that we will see significant highs in 2014, the prospect that we could achieve new lows this year seems less enticing, although not at all impossible.
From my own perspective, where I feel somewhat more comfortable in predicting equity market bullishness and commodity bearishness, it is much easier for me to recommend equities, but obviously different people will weigh these factors differently, according to their own insights and risk appetite.
Right. So, based on five-year cycles, we have a vague notion of what the topography of the market will look like over the next couple years (although it hardly need be said, I hope, that there are no guarantees), so let's look at the other type of bond market cyclicality.
This is the 16-month cyclicality (69 weeks or roughly five quarters) that appears to show up in other contexts (for example, the time between an inverted yield curve and a recession or a flattened yield curve and an oil shock). Moreover, this 16-month cycle seems to be composed of three legs, which would seem to point to a series of 23-week cycles. I should point out that, unlike the five-year cycle, these 69- and 23-week cycles are not trough-to-trough but rather trough-to-peak. In other words, a relative low in yields will often be followed by a relative high in roughly 23 weeks' or 69 weeks' time.
(Source: St Louis Fed)
I don't want to overplay this hand. Sometimes, a period of rising interest rates is only followed by a decline in momentum rather than a bona fide change in direction. And, rarely are the periods precisely 23 or 69 weeks long. If you look at the current 5-year cycle we are in (in the chart prior to the one above), for example, you can see that the high in the winter of 2011 (the fourteenth month on the chart) was followed seventeen months later by the 2012 summer low (the thirty-first month). Within that crash in interest rates, there was one leg sharply down, another leg that held ground, and then a final leg sharply down again.
At least up until last week, the ride back up was not wholly contemptible. So, forgetting the 5-year cycles for the moment, what do yields prophesy of themselves now?
Well, we can posit that, eight months in, it was probably time for some sort of "correction." But, before continuing on in that vein, I should also point out that the oil/gold ratio has a similar sort of cyclicality to interest rates and that we can use it nearly as well as the latter in predicting future rate moves five to sixteen months in advance.
Cyclicality in gold/oil and bonds
In the chart below, where I've detrended the gold/oil ratio and the 10-year yield, you can see this tendency (albeit imperfect) of the gold/oil ratio and yields to "forecast" future changes in bond momentum (as well as demonstrate the five-year cycles mentioned above). Again, I don't want to overplay these relationships, because they can be off by months or be dead wrong at times, and they can be especially wrong when it comes to precise timing of five-years low.
(Source: St Louis Fed)
Finally, it also should be said that although the oil/gold ratio tends to correlate with Treasury yields in the short-term, it is much more volatile than the latter and doesn't conform over the long-term. That means giant moves in the gold/oil ratio do not always manifest themselves in interest rate moves, but they can help highlight the cyclicality.
Interest rate outlook
Now that those caveats are all in place, let's look at the gold/oil ratio as it stands today.
A naïve interpretation would tell us that interest rates should have peaked in December or January and then sharply declined into April, after which they might reverse or perhaps stabilize and then reverse in the summer. Since there tends to be a shift in momentum every 23 weeks (roughly five months), if we could pick out the top that follows, we could count backwards to better estimate a bottom, but the gold/oil ratio has been quite tame since the spike in precious metals in 2011 and it's hard to pick out a clear top.
Moreover, we know that we cannot interpret these patterns naively, so the chart could be deceiving us about not only the direction but the force of future moves. But, combining the possibility that a five-year low has not been made, I think there may be some reason to expect a modicum of bond strength through the summer. It will be interesting to see how the competition between the oil/gold ratio and interest rates resolves itself in the coming weeks, and this might provide us with a clearer picture fairly soon.
In my opinion, both the interior cyclicality of interest rates, as well as growing tensions in the broader picture, not only with respect to cyclicality in the oil/gold ratio, but in the total intermarket environment, suggest to me that equities remain the least risky asset class over the course of the next twelve months.
Additional disclosure: I am long June Dow futures.