In previous articles, I have written about how, in the absence of a way to ascertain reversals in the secular disposition of treasuries, we can use a couple of intermediate-term tricks to help us along the way. And, I have surmised that, to a certain extent, we can leverage these techniques in such a way as to give us an idea as to when the bond bull has died. In this essay, I would like to talk about a third such trick that is a kind of hybrid of the previous two, and why this continues to make me bearish about bonds for the rest of the year and, somewhat more tentatively, for the next two years.
First, I have noted how there appears to be a strange five-year cyclicality to 10-year treasury rates. Since the demise of Bretton Woods, rates have tended to form a deep trough in years ending with '3' and '8,' which are then followed by sharp upward spikes into years ending with '5' and '0'.
When bonds are in a secular bear market, such as in the 1970s, those spikes have broken through the pre-trough high. When they are in a secular bull market, as they have been since the early 1980s, the spikes fail to breach the pre-trough high.
Second, the gold/oil ratio often gives us a sixteen-month preview of changes in interest rates. Although I believe that this relationship is somewhat incidental, insofar as I suspect that the behavior of the gold/oil ratio is tied to equity yields, nevertheless, there appears to be a deeper cyclicality to yields of all kinds that is reflected in different markets and ratios.
Both the five-year cyclicality and the relationship between the gold/oil ratio and treasury yields is reflected in the chart below.
As you can see, neither correlation is perfect. The link between the gold/oil ratio and treasury yields was especially weak during the 2000s, which is also when equity yields and treasury rates parted ways.
That brings us to the phenomenon I would like to draw out now, namely the internal cyclicality of rates at 23- and 69-week intervals. In other words, I would like to argue that treasury yields undergo bearish and bullish phases that alternate in sixteen-month (i.e., 69-week) phases, and that each phase can also be divided into three 23-week portions, in which the middle phase is counter-cyclical with reference to the larger 69-week phase.
Like the other relationships I have tried to posit, this one is not perfect, but I believe that when used in conjunction with other techniques, it provides important clues as to future bond market moves.
I should also say that the phrase "69- and 23-week intervals" suggests greater precision than I intend. Calling them 16-month and 5/6-month cycles gives a slightly more honest impression of their roughness, and it may turn out that it is 70 and 23.3 weeks or some other combination that gives a better result, but for now I just want to identify the general pattern.
Finally, one last caveat: this cyclicality is embedded in broader decades-long secular trends and can be obscured or overridden by other market forces, so I confess that it takes a little imagination to see it and that it cannot be the only consideration when it comes to guessing what rates will do next, but I will do my best to tease it out below and the reader can decide for him- or herself.
In the chart above, I am trying to point out the idealized 23-week cyclicality of treasury yields over the last three years. I believe that the bottom in July marked the end of a 20-week collapse and the beginning of a new uptrend or at least a change in momentum.
Although I do not really want to discuss the gold/oil ratio here, I believe that that ratio exhibits similar behavior and can be used to forecast treasury yield momentum.
Over the last decade, the regularity of the treasury yield rhythm has been imperfect but impressive.
In the chart above, I am comparing the rise and fall of interest rates over a 23-week period with the rise and fall of the subsequent 23 weeks.
During the period when the bond market was shifting from a secular bear to a secular bull -- the early 1980s -- it did not exhibit such neat patterns. Although I believe it still exhibits some of that 23-week cyclicality, it is certainly more chaotic, especially when the five-year cyclicality is imposing itself as in 1983 and 1987.
In the chart below, I am using the same technique to look at to what extent 69-week changes can be used to predict the following 69 weeks.
Even during the period where this appears to break down most, from 1978-1982, I believe that a closer look shows that this 69-week cyclicality still makes itself felt.
Here is another look at 69-week changes, this time next to nominal 10-year yields.
A change in the neighborhood of plus or minus 200 basis points often marks an extreme in yields, although there is certainly precedent for much higher moves.
Now, if a 69-week high should be followed by something like a 69-week low, then that means we should see relative highs (or lows) every 138 weeks.
As you can see, however, there are rather extreme deviations from that cyclicality. The blue line marks the current rate and the pink the forward rate. Interestingly, the greatest deviations appear to be related to the five-year cyclicality I mentioned previously.
Somewhat counterintuitively, I believe we can use this breakdown in the 16-month cyclicality to ultimately bolster our argument.
What I mean to say is that the 138-week change gives us somewhat less noise than the 69-week change and it gives us the sense that the plus or minus 2 level is rather meaningful. Since 1987, a -2 change has indicated that a five-year trough is imminent. During much of the 1970s, a +2 change indicated a five-year peak had been reached. In both instances, a '0' change often marked top or bottom, respectively.
The 1980s make it quite clear that these levels are not sacred, but I think they do give us a hint, especially when used in conjunction with other techniques.
Changes over 32 months (138 weeks) also bear a degree of similarity to changes in the yield curve.
I believe that, taken together, the behavior of treasury yields suggests that we have probably passed or perhaps will soon achieve a five-year trough in yields and that the remainder of the year and the bulk of 2013 will see sharply rising yields.
Assuming that is true, the high we see in roughly 16 months' time will either set a new lower level of resistance for yields, or if it breaches the previous resistance, may indicate the end of the bond bull.
In this final chart, a scatter chart of treasury rates compared to cyclical changes in rates over the last 50 years, I think it gives us a fairly decent notion of the extreme levels we are at. It doesn't mean that we could not go yet lower in the coming years, but it does something to illustrate the risks for those still going long on treasuries.
Additional disclosure: I am short AUDCHF and Sept Dow futures.