OIL PREDICTS STOCK MARKET DIP
By Tom McClellan – McClellan Market Report
March 25, 2011
Just over a year ago, I looked at the 10-year leading indication that crude oil prices give for the stock market. It is time to take another look at that relationship, especially in light of the trouble that it suggests is coming for stock prices.
This week’s chart shows again how the price plot of crude oil prices has done a great job of giving us a macro view of what the trend should be 10 years later for the stock market. The periods when crude oil prices have moved sideways led to sideways periods for the stock market a decade later. And the periods when crude oil has trended upward were followed 10 years later by big bull markets in the stock market.
So the fact that crude oil prices have gone from a low of $11/barrel in 1998 to now above $100 is an indication that we should expect a persistent uptrend for stock prices in the decade ahead. But we should not expect it to be an unbroken uptrend.
When we zoom in closer, we see that oil’s price fluctuations can have important meaning for stock prices about 10 years later. The timing is not perfect, but the dance steps generally get repeated.
The one caveat to that principle is that oil price movements that are based on supply and demand forces tend to matter much more than oil price movements brought about by governmental or quasi-governmental forces. The Arab Oil Embargo in 1973 got the big oil price rise started, but stocks did not match the magnitude of that rise or the additional up leg caused by the Iranian revolution in 1979. And the oil price crash of 1986 that came about when Saudi Arabia abandoned the production quotas similarly did not bring stock prices down. The 1990 Iraq invasion of Kuwait caused oil prices to briefly double, but we did not see an exact echo of that spike in the stock market. When governments put a thumb on the scale and nudge oil prices away from where supply and demand factors would dictate, it does not show up as much 10 years later in the stock market.
Still, the background price pattern movements can clearly be seen as having been repeated in stock prices roughly 10 years afterward. And now we are into the 10-year echo point of the big oil price decline from Nov. 2000 to January 2002. So far, the Fed’s POMOs have kept the stock market going higher, so we have not yet seen the echo of that oil price decline being manifested in stock prices. But given the decades of correlation between stock prices and oil’s leading indication, it is hard to imagine that we will be exempted from seeing some kind of echo of that oil price drop. When the Fed stops doing POMOs in June, and when the stock market enters the part of the year when seasonality is much weaker, stock prices should finally be allowed to manifest an echo of that 2000-02 oil price decline.
The good news for long term investors is that later this decade we should see stocks echo the big rise in oil prices. The bad news is that the most likely way for this to happen is not from stocks being worth more, but rather that the dollars needed to buy stocks will be worth a lot less thanks to the Fed inflating the monetary base. So yes, in the late 2010s, your shares of stock will be worth more dollars. But those dollars won’t be worth as much.
Related Charts
| Dec 10, 2010
|
Oct 29, 2010
|
Feb 26, 2010
|
When you ignore the shocks of war, there is surprisingly high correlation. Nice graph, thanks.
This guy suggesting using 10 year old crude price action for todays stock market?
Oh boy…
http://www.astrologicalinvesting.com/
Correlation to oil prices doesn’t mean causation. USD destruction is the real driver.
I do not disagree that it is about the destruction of the dollar, and I made that very point in this article.
But the issue of correlation not being causation is irrelevant. It is not about causation, it is about revelation. What happens to oil prices now reveals what is more or less going to happen to stock prices 10 years later. The causative factor behind the relationship does not matter. It might make us feel better if we could identify a cause, and explain the “why”, but that is not essential.
Scientists are still working on explanations for the “why” of gravity, and they have had to invent an additional 10 dimensions just to get the math of string theory to work. But even without knowing the “why” of gravity, we can all accept that it does work; we have seen enough evidence. 110 years worth of seeing oil’s leading indication work ought to be sufficient.
Insisting that the relationship must be other than what is clearly evident is like the guy whose wife catches him in bed with another woman, who he insists is not actually there. “Who are you going to believe,” he asks her, “me or your own lying eyes?”
As an admirer of many of your previous articles, Mr McClellan, I find the justification you’ve written here more than shocking. Instead of providing a reasoned argument, you frankly admit that you have none and invoke the laws of physics (as if they equate) to insist that your chart, like the fact of gravity, should speak for itself. This is not only vainglorious, it is reprehensible.
A statistician would have a field day with your assertion that ’110 years worth of seeing oil’s leading indication work ought to be sufficient.’ No sir, strictly speaking it is neither sufficient nor statistically significant: even according to your own chart and claims, you’re actually dealing with just six major long-term trends in oil prices, bull, bear and sideways, over that period. Six data points does not amount to a statistical hill of beans.
Aside from which, you have blithely overlooked a blazingly obvious explanation for the relationship, in favour of one for which you can find no logical reason. Appalling! Oil prices are, quite clearly, an inverse coincident indicator. They are a headwind to stocks when rising and a tailwind when falling or steady – which is why when you push oil prices forward ten years on your chart you appear to have (voila!) a ‘leading’ indicator. Shift oil prices back ten years instead and see for yourself. Why on earth – and I speak as a stock market contrarian – would you want to complicate something as straightforward as this?
There’s a horrible temptation as a technician to see patterns and correlations where there are none – God knows I’ve been caught in this hall of mirrors myself. But I am truly staggered to see a respected analyst write that ‘the causative factor…does not matter’, and that ‘It is not about causation, it is about revelation.’
Something is being revealed here, but it sure isn’t about oil prices.
It is important for anyone proposing a new theory to be able to withstand even the most trivial criticisms. So bravo to you, Mr. Money, for serving up that opportunity so neatly for me.
Your conjuring of what “a statistician” would do with this relationship is reminiscent of what longtime Los Angeles Dodgers broadcaster said about statistics resembling the relationship that drunks have with lampposts, in that they rely upon them more for support than for illumination.
If we were to restore the relationship between oil prices and the DJIA (both log-scaled) to a coincident one as you propose, we would see a general correlation owing to the fact that both are trending upward. On the chart, this is seen as both plots moving from lower left to upper right.
But such trends can lead to erroneous statistical conclusions because of the statistical principle known as “autocorrelation”. That word has multiple meanings in the mathematical realm, so please read the article at http://www.mcoscillator.com/learning_center/weekly_chart/correlations_may_not_be_what_they_seem/. It discusses the use of correlation coefficients in uptrending data, and how they can lead to incorrect conclusions.
Another way to examine the data to get at the correlative (or not) relationship is to show data-point pairs on a scatterplot (X-Y) chart. If we were to examine the long term history of oil prices and the DJIA on such a coincident scatterplot, both logarithmically scaled, it would look like this:
http://mcoscillator.com/data/charts/weekly/DJIA_Oil_Scatterplot.gif
The operative idea here is that the closer such a relationship is to linear, the stronger is the correlation. If the dots are scattered wildly on the chart, then it shows a lack of correlation, and it suggests that we have to view the relationship with doubt.
But if we look at the same data and incorporate the 10-year offset which I propose, we see this relationship:
http://mcoscillator.com/data/charts/weekly/DJIA_Oil_Scatterplot_offset.gif
It might not seem like there is much of a difference, but that owes to the multi-decade uptrend inducing an autocorrelation. But when we compare each chart and evaluate which one shows greater linearity, we find that the 10-year offset causes all of the dots to tighten up into a much more linear relationship, especially when we accept that the exogenous examples were caused by OPEC putting its thumb on the scale.
So kudos to you, Mr. Money (may I be familiar and call you “Onthe”?) for offering me this chance to spread my statistical wings in order to help make the logical point which should have been evident to anyone willing to actually look at the chart without prejudice.
And did you notice that I chose to make my argument without once using the term “sphincter”. That must somehow signal a rhetorical weakness on my part.
The condescension in your response is absolutely breathtaking.
Do you really think you can wave away a causal relationship as clearly established as the one between oil prices and stocks with a scatter plot? Why on earth are you attempting to create a (10-year!) leading indicator out of one which works entirely logically as a coincident one?
Not only is your theory wrong-headed, it is glaringly inconsistent. You are happy to use the OPEC price hikes of the 1970s to support your thesis (lo! stocks rise in the 1980s), then simply dismiss the flip side collapse (and Saudi abandonment of quotas) in the 1980s when stocks continue to rise in the 90s.
Your explanation of the latter (but, notably, not the former) is that prices set by supply and demand ‘tend to matter much more than oil price movements brought about by governmental or quasi-governmental forces’. Why? If you have a rigorous theory, let’s hear it. But you cannot dismiss flaws in your argument by insisting that causation ‘does not matter’. That’s just risible.
Your scatter plots tell you that the coincident relationship of oil and stocks is ‘in doubt’. Hmm, let’s take a look.
1. 1890 – 1920 Oil volatile but trending upwards: Stocks make no progress.
2. 1920 – 1929 Oil drops back sharply and stabilizes. Stocks soar.
3. 1929 – 1933 Oil falls sharply during the Depression, with stocks.
4. 1933 – 1948
Oil rises very modestly then remains essentially flat, rising sharply from 1946 into 1948. Stocks recover then oscillate, dropping sharply from 1946 – 1948.
5. 1948 – 1973 Oil flatlines. Stocks enjoy a major secular bull market.
6. 1973 – 1981 Oil spikes. Stocks endure a major secular bear market.
7. 1981 – 1999 Oil plunges. Stocks enjoy a major secular bull market.
(Oil enters a flat to modestly falling trend from 1987 onwards, with a brief spike in 1990. Stocks and the economy push relentlessly higher with one recession in 1990.)
8. 1999 – 2008
Oil surges into 2000, then moves higher and spikes into 2008. Stocks make no progress, with two recessions following each oil price surge.
9. 2008 – Oil falls sharply in the recession, with stocks (as in the early ’30s) paving the way for a market rebound. However emerging market demand, geopolitical concerns and speculative flows are now conspiring to create another oil price surge. A doubling of prices Y.O.Y. has preceded recession on each occasion since trading began in the 1970s. Let’s see if this time is different.
http://pragcap.com/oil-prices-say-100-chance-of-a-new-recession
I invite the reader to decide whether the coincident correlation is ‘in doubt’.
Analysis which lives in a parallel charting universe, ignoring the blindingly obvious and moving lines forward and backward to create an illusion of ‘revelation’, is not analysis but self-deception. It is hocus-pocus, which fools only the magician.
By all means persuade us with reasoned argument (after all, to use your own analogy, even the string theorists can produce that). But if you can’t support a technical argument with fundamental logic – brazenly defying reason in the process – don’t expect it to be taken seriously. As Truzzi says, extraordinary claims require extraordinary evidence.
A perfect example of a technical analyst disappearing so far up his own fundament he can’t see the light through his sphincter.
He says in his linked article “I don’t have a good hypothesis to explain why it should work this way, which is a little bit troubling…” Duh. The reason he can’t figure it out is that the correlation is ass-backwards.
For some reason he wants to shift the oil price forwards 10 years and make it a leading indicator. But if he just looked at the movement of oil prices here and now, he would see that the coincident correlation works entirely logically. Shift oil prices back ten years so they run concurrently with the Dow and take another look at his chart.
When prices peak and fall, or even just stay stable at a moderate level – as they did between 1920-1929, 1950-70, 1981-87 & 1990-98 – the conditions are ripe for corporations to thrive and for a stock market advance. When prices surge (1973-81, 1987-90, 1998-2008) there are huge headwinds.
If anything, rapidly rising oil prices are a leading – and logical – indicator of recessions, as PragCap and others have highlighted.
http://pragcap.com/oil-prices-say-100-chance-of-a-new-recession
Agood one, On the Money!
InvestorX
Doesn’t make any sense. It is hard to see something with a 10-year lag.
I’d say this was a classic example of false cross-correlation as with “Storks and Babies”. With enough data points you can get 95%+ R Sq. but it doesn’t mean it’s useful info, just that the two series happen to have the same shape.
I agree with Cy.
Higher levels of production and consumption should always lead to higher stock and oil prices.
Consequently , oil pricing can be a lagging indicator.
The correlation is fantastic. How does the percentage gain look? If oil moves from $10 to $100, a 10x gain, what percentage gain is the stock market looking at by 2020? Great job Tom.
I tend to agree with OntheMoney. I think this is a spurious correlation. Why would it take 10 years for a change in oil prices to affect the stock market? And, why would the correlation be the opposite of what one would expect?
Assuming that all increases in oil prices are driven by increased demand, then wouldn’t we expect to see spikes following peaks in the stock market?