Okay: I’ve checked my door locks, made sure my kids are safe, and braced myself for the inevitable incendiary incoming comments. So, I feel secure in pointing this out:

Gold’s real return for the last 10 years has been a blistering 1.07% per year. And worse, that’s higher than you ought to expect for the next 10 years.

Here’s the math. Gold on July 19, 2008, was at $955. Today it is at $1223, for a gain of 28.1%. But the overall price level (CPI) was at 218.815 in June 2008, and at 251.989 in June 2018 (we won’t get July figures for another month so this is the best we can do at the moment), for a 15.2% rise in the overall price level.

1.07% = [(1+28.1%) / (1 + 15.2%)] ^ 0.1 – 1

It might be even worse than that. Gold bugs are fond of telling me how the CPI is manipulated and there’s really so much more inflation than that; if that’s so, then the real return is obviously much worse than the calculation above implies.

Now, this shouldn’t be terribly surprising. You start with a pile of real stuff, which doesn’t grow or shrink for ten years…your real return is, at least in units of that real stuff, precisely 0%. And that’s what we should expect, in the very long run, from the holding of any non-productive real asset like a hard commodity. (If you hold gold via futures, then you also earn a collateral return of course. And if you hold warehouse receipts for physical gold, in principle you can earn lease income. But the metal itself has an a priori expected real return of zero). Indeed, some people argue that gold should be the measuring stick, in which case it isn’t gold which is changing price but rather the dollar. In that case, it’s really obvious that the real return is zero because the price of gold (in units of gold) is always 1.0.

So, while everyone has been obsessing recently about the surprisingly poor performance of gold, the reality is that over the longer time horizon, it has done about what it is supposed to do.

That’s actually a little bit of a coincidence, deriving from the fact that at $955 ten years ago, gold was reasonably near the fair price. Since then, gold prices soared and became very expensive, and now are sagging and getting cheaper. However, on my model gold prices are still too high to expect positive expected returns over the next decade (see chart, source Enduring Intellectual Properties).

The ‘expected return’ here is derived from a (nonlinear) regression of historical real prices against subsequent real returns. To be sure, because this is a market that is subject to immense speculative pressure both in the bull phases and in the bear phases, gold moves around with a lot more volatility than the price level does; consequently, it swings over time from being very undervalued (1998-2001) to wildly overvalued (2011-2013). I wouldn’t ever use this model to day-trade gold! However, it’s a useful model when deciding whether gold should have a small, middling, or large position in your portfolio. And currently, despite the selloff, the model suggests a small position: gold is much more likely to rise by less than the price level over the next decade, and possibly significantly less as in the 1980-1990 period (although I’d say probably not that bad).