Oil/gold, oil, and yields—and who invented the yield curve spread, anyway?

Building on our argument yesterday, that the gold/oil ratio can be used to predict 10-year treasury rates sixteen months in advance (although silver/oil is probably a much better tool), we would like to explore some of them implications of that conclusion.

For one, if everything we have said is more or less true, then the gold/oil ratio can be used to forecast treasury rates, while the oil/gold ratio can be used to forecast the Dow/gold ratio, which is shadowed by the P/E ratio, and the P/E ratio is just the inverse of EPS. In other words, gold/oil had been ‘pushing’ interest rates down throughout the ’80s, ’90s, and 2000s, up until the financial collapse, when the gold/oil ratio jumped to its third highest level in twenty years. At the same time, in the guise of the oil/gold ratio and its peculiar relationship with Dow/gold and P/E ratios, etc, it began pushing EPS higher beginning in 2000, for the first time in decades.

In other words, the oil/gold ratio was single-handedly trashing the “Fed model”, which says that EPS should generally track with treasury yields. As you can see from this chart below, from 2000, the yield spread between 10-year treasuries and the dividend yield fell.

since 2000, spread has been dropping

Moreover, EPS and treasury rates have taken on an inverse relationship since the Dow/gold ratio began to peak in the late 1990s.

P/E ratio and EPS reverses correlations w/ treasuries from turn of century

The only problem is that I still can’t explain why any of this happening. I am just collecting more and more patterns that indicate something more fundamental going on. George Cooper’s analogy between the financial system and ‘spatial flutter’ is constantly in my mind, and in the interplay between the yield curve spread and the oil/gold ratio, it always seems to be present but indefinable.

In the meantime, I might as well identify another odd pattern, and again it involves the gold/oil ratio. It is the kind of observation I generally do not put much stock into (if you’ll pardon the pun)—the notion of movements based on seemingly preordained time cycles, but I couldn’t help but notice this pattern while staring at the gold/oil chart below.

I'm trying to improve the quality of my charts!

It seems as if the gold/oil ratio is influenced by a five-year cycle. Specifically, it appears to achieve relative highs during years ending in ‘3’ or ‘8’, usually late in a given year and sometimes a month or two into the following year. It then hits relatively distinct lows in the third year (usually the latter half) or the fourth year (i.e., years ending in ‘5/6’ or ‘0/1’). In either instance of achieving highs or lows, it may move farther up or down, but the overall pattern seems to remain.

each vertical line marks a new year

achieves three year high in 1978; three year low in 1981

two-year high in 1983, followed by three-year lows in '85

two-year high in 1988; five-year low in 1990

five-year high in 1993; three-year lows in 1996

five-year high in 1998; decades low in 2000

unusual: two-year high in 2003 doesn't break 14.28 (oil/gold 0.07); all-time low in 2005

ten-year high in 2008 and trailing into 2009....

if the March 2011 low turns out to be the cyclical low, it would be the least impressive cyclical low ever, but it is now heading higher

As for our current situation, we should be hitting a fairly convincing new low in the gold/oil ratio sometime this year or next. In the first week of April, a month before silver skyrocketed to nearly $50, the gold/oil ratio challenged the 2009 lows, but has since turned back. It would seem, then, that the gold/oil ratio is likely to fall somewhere significantly below 12.7 this year, although it could be as late as 2012, and the question remains as to whether or not it will, as we have argued, break below the critical 8.3 level (i.e., above oil/gold at 0.12). Needless to say, it is hardly inconceivable that it will be so bold as to completely ignore precedent.

The other implication is the future of yields. If the gold/oil ratio leads by 16 months or so, and the gold/oil ratio runs along a five-year pattern like we have described, then yields should be susceptible to a similar pattern. With highs in yields coming in years ending with ‘4/5’ and ‘9/0’ and troughs in ‘6/7’ (maybe ‘8’) and ‘1/2’ (possibly ‘3’). And, that appears to be the case.

peaks in 1970; troughs in 1971; peaks in 1975; bottoms in 1976

historic highs in 1980; falls into 1982

peaks in 1984; bottoms in 1986

yields rising into 1989, then bottom in '92

yields peak in '94; bottom in '95-'96

yields peak in 2000; then fall...

yields peak in '05-'06; then fall...

yields bottom in 2008

10-year apparently peaked in Spring 2010 and probably has lower to go before pivoting upwards

The next question is, does that pattern hold true before gold and oil broke free of their price moorings after the ‘Nixon shock’? I won’t put up the charts, because there doesn’t appear to be any such pattern, but I recall that a 5-year rolling correlation between the Dow and the 10-year treasury could be used to time the stock market from as early as the 1930s up until the 1950s and ’60s. In any case, it would seem as if the oil/gold ratio sets the tone today.

We have already shown how it shapes the equity, gold, and now the bond markets. In addition, it seems to have a great deal to say about the oil market.

A lot of commentary about the oil/gold ratio has to do with using it trade oil vs gold. I think my ‘supersystem’ shows that this is not the best use the ratio can be put to, that it is far better—freakishly and uncannily better—to gauge the Dow/gold ratio. And, under rare circumstancesĀ (the mid-1970s and in 2009) that we catalogued in previous posts, it can be used to identify periods when oil will go parabolic.

Those instances may be less rare than I had previously thought. The gold/oil ratio can be used to identify those ‘Leeb shocks’ that we have incorporated into our system. On the chart below, I have used the oil/gold ratio and then flipped it upside down, so it is just another representation of the gold/oil ratio, but in this case I have moved it forward 12 months instead of 16. Every ‘Leeb shock’ (when oil increases by 80% a year or more) has been preceded by a massive fall in the oil/gold ratio (i.e., a massive rise in the gold/oil ratio), except for the shock caused by the Iraqi invasion of Kuwait in 1990. Usually, this involves a move from above 0.07 to below that level, but the shock that ’caused’ the financial crisis did not get near that low, having come from unprecedented highs down to the 0.083 oil/gold ratio (12 on the gold/oil for those keeping score at home). Moreover, the lag was 17 months during that period, rather than 12. So, don’t sell your first-born thinking you can use the money from trading oil to buy him back in a year’s time.

Even so, the correlation is uncanny.

oil/gold reversed 'forecasts' Leeb's Oil Indicator 12 months in advance

Obviously, any kind of major fall in the price of oil might reasonably be expected to be reflected on a jump in the year-on-year price of oil in twelve months time, but if it were that simple, then the concurrent oil/gold ratio would be an equally good reflection of the jump in oil. If you look at the chart below, there is a relationship, but it does not compare to the one above.


At this point, we have a royal mess on our hands. We have a gold/oil ratio that can be used to forecast movements in 10-year yields 16 months ahead of time, forecast major oil moves 12 months in advance, and somehow ‘determines’ the 3-month note in real time—or is it the other way around? Moreover, it appears to be under the influence of a rough five-year time cycle and have a number of critical levels, most importantly 0.05, 0.07, and 0.12, depending on how you intend to use the ratio, the best being to determine the Dow/gold ratio and, by extension, dividend yields. Is there anything the oil/gold ratio can’t do?

But, let’s stick with bonds and oil for now. If the oil/gold ratio can be used to forecast bonds 16 months down the road and oil 12 months down the road—although imperfectly—then shouldn’t there be a four month lag between bonds and year-on-year oil?

not an especially dramatic impact oil y/y has on bonds

In general, it is not especially clear that there is such a connection.

In fact, to muddy the waters a bit more—as if that was what we needed right now—it would appear that oil does not lead bonds four months in advance, but that the yield curve leads oil 16 months in advance. Specifically, when the yield curve inverts, then oil tends to jump significantly in the next 16 months. Or, to put it more precisely, of the seven ‘Leeb oil shocks’ we have had, each except the 1987 and 2009 shocks were preceded by the yield curve flattening to below 1.00. In 1987, it had only gone as low as 1.24. Of the other five instances, four had been preceded by inverted yield curves, with the 1999 shock having been preceded by a flattening in the curve to 0.20.

yield curve seems to forecast Leeb oil shocks

Alternatively, not every inversion of the yield curve has been followed by an oil spike in 16 months. The inversion of the curve in 2000 did not ‘bring about’ an oil spike, but rather a 50% drop in crude. But, the bottom in the yield curve spread coincided very closely with the oil/gold ratio breaking through the 0.12 level, and thereby ushering in a decade of falling Dow/gold.

Even so, since we have argued that both the yield curve spread and the oil/gold ratio have generally telegraphed ‘Leeb shocks’, one might expect that the yield spread and the oil/gold ratio would be positively correlated for the most part.

I'm not sure what to make of this

We mentioned before that the gold/oil ratio ‘predicted’ the 10-year treasury 16 months in advance (although we are still exceptionally curious about what the silver/oil ratio might say about yields), while the three month yield appeared to be a function of the current 10-year yield divided by the current gold/oil ratio. But, as far as the mathematical model for the three month is concerned, any one of those factors could be expressed as a function of the other two.

For example, that the current 10-year yield is a multiple of the gold/oil ratio and the three month yield, or that the gold/oil ratio is the yield curve, which is ironic, because the conclusion today’s argument had tended towards was that the oil/gold ratio generally and roughly matched the yield curve. Is it possible that both are true, because in one instance we are using division and in the other subtraction? That appears to be what is at issue in this instance. So, is it legitimate to use division to measure the yield curve spread? What is the justification behind using subtraction?


No comments yet

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: