Crude Approximations

Today the Boy King is worried about oil.

This is somewhat puzzling, as we have no position in oil, either long or short.

Unfortunately, what we do have is a position in government bonds. And the bond market is a fickle beast. Some days the bond market seems to care only about abstract macroeconomic news: growth and output data, employment and wage data, housing market data, cost-of-living indices, the prices of raw materials, and so on. On other days, bonds dance in lockstep with their friends across the aisle, stocks. There are periods of weeks on end when bond prices seem to be determined entirely by (say) the dollar-yen exchange rate; every other indicator is ignored. Then, in the blink of an eye, the correlation with exchange rates vanishes, and a new day-to-day correlation develops, with (say) emerging markets. This lasts a few days, before the next fashion takes hold. Lather, rinse, repeat.

The current determinant of choice is oil. Bond yields have tracked oil prices almost tick for tick these last few weeks. Unfortunately for us, oil has been in freefall, driving yields lower, and we’d been betting that yields will go higher. So, every morning for the last few weeks, I’ve been greeted by the sight of Jimmy’s worried face as he informs me what happened in the oil market overnight. “Oil dropped 30 cents yesterday!” “Oil is up 5 cents today!” He’s becoming an oil junkie. Every penny of fluctuation in the oil price is met with theatrical moans and sighs. It’s beginning to get me down.

Personally I think the latest move is overdone. Both my medium-term “fundamental analysis” of growth and inflation, and my short-term “trader’s intuition” tell me that bond yields have fallen too far; they’re ripe for a rebound.

I go to Jimmy and tell him I think yields are too low and they’re heading back up. A somewhat surreal conversation ensues.

Me: I think yields are going up.

Jimmy: So you think oil is going to rebound?

Me: No, I didn’t say that. I have no real opinion about oil. I’m not even trading oil. I’m trading bonds. And as a bond trader I think that yields have dropped too far, too fast.

Jimmy: But bond yields are completely correlated with oil.

Me: Umm, not quite.

At this point I go into lecture mode. Usually if I spout enough jargon I can confuse him into agreeing with me; let’s see if the trick will work today.

Me: I agree with you that bond yields have been highly correlated with oil prices over the last few weeks. But correlation does not imply causation: maybe both markets are being driven by some external factor that we’re not aware of. In any case the signals are simultaneous; there’s no predictive power in oil price information. And who knows, the correlation could change – after all, it didn’t exist two months ago, and I’m willing to bet it won’t exist two months hence. Oh, and your sample size is too small to be robust – three weeks of data don’t prove anything.

Unfortunately, the strategy backfires. My explanation confuses him all right, but instead of backing off and agreeing to let me do what I want, he wants a simpler explanation. He wants a graph.

I’m not sure what the graph will prove, and I tell him so. He thinks I’m contradicting him and says in what he imagines are firm, decisive tones, “I insist on seeing a graph”.

I ignore the request. Creating a graph of oil against bonds, while trivial to do, would be of absolutely no use. Graphs are horribly non-quantitative – about the only thing they’re good for is reinforcing your existing prejudices. In this particular case, a graph would merely confirm what we already know: that bond yields have tracked oil prices over the last few weeks. Big deal.

But I underestimate Jimmy’s doggedness. He digs into our research archive and pulls out a piece of analysis from a large investment bank showing exactly the graph he wanted me to generate.

He waves this graph triumphantly at me: See! There’s a correlation!

Me: (sighing) I never said there wasn’t a correlation. But it may not be robust, it doesn’t imply causation, and it certainly says nothing about future behavior.

Jimmy: But there’s a correlation!

Me: I don’t care. I’m not buying oil, I’m betting that bond yields will go up.

Jimmy: But betting on yields is just like buying oil, because of the correlation!

Me: I just explained why the correlation is meaningless.

Jimmy: So does that mean you don’t think oil will rise?

Me: It may rise. It may fall. I don’t have an opinion either way.

Jimmy: But then why are you betting that bond yields will go up?

I realize that this is a battle I can’t win. Beating my head against a wall is more fun than trying to change Jimmy’s mind once it’s made up. There’s only one thing for it: make a graceful retreat.

I tell him I think bond yields will go up because oil is poised for a major move higher.

Update, a week later: Bond yields have risen, while the price of oil hasn’t changed. The Boy King comes up to me and says, “It’s very interesting, isn’t it? The correlation has broken down, just as we thought it would”.