China weighs on copper prices

Copper is a fascinating and important asset. Traders have known for decades that “Dr. Copper” can foretell recessions, and indeed, we’ve found that including front-month copper futures in our models improves their ability to forecast nominal GDP.

Despite the general bullish turn in most asset prices, copper is down substantially from last year, trading around $2.60 per pound, from around $3.40 at the start of the year. I’d proposed in my last post on this subject that this was due to the increasingly weak Chinese economy. This thesis has held up upon deeper investigation.

Eyeball the plot above, comparing co-movements between copper and the Shanghai index, suggests there could be a link. In fact, the co-movement shows up at (at least) the frequency of daily changes and has strengthened in 2018, when the US-China trade war really began.

The plot below reports the correlation coefficient between the Shanghai Composite stock index and copper prices, based on daily log differences in both time series (a statistically sound way of measuring co-movement between two assets). The association was fairly weak in 2015, 2016 and 2017, with the correlation coefficient holding below 0.3. Given that we’re measuring daily log changes, a correlation coefficient of around 0.2 is still fairly high. The association, however, jumped markedly in the first half of 2018 and has soared in the second half, to 0.57 for the quarter which has just ended.

Chinese economic data are not useful for fine-grained detail. Official inflation and GDP series simply don’t exhibit the stochasticity that characterizes data from other countries, and tellingly they are harder to model than data from other countries.

Instead, Chinese figures appear to be sanitized. They’re plausible when averaged over a few years, but not always year-to-year and certainly not quarter-to-quarter. We simply don’t know quite what is going on in China’s economy today as a result. The markets at least, suggest that growth is slumping substantially, keeping copper (and likely other commodities) cheap.



  1. // Reply

    Hi … I think you might be missing the more salient point, which is that copper, along with many other things, is starting to improve a bit. Recall that China only started to put into action its new liquidity-increase push in July. That marked a bottom in the yuan, copper and certain EM series. Most of the rest of EM bottomed in August. So the issue is really whether we are in fact at some kind of turn here, or if it’s just a pause. Gold is a good barometer. So far it’s hanging in there at 1200. Conceptually, as long as the US long-bond yield is going up, the Fed is not too tight and none of the above series should get worse, no?

  2. Justin Iriving
    // Reply

    Copper and the SSE are both up a bit sure, but remain highly correlated which is what I find most interesting. That the Chinese were able to take some action to deal with the drop in growth expectations is not surprising, what is interesting is how little they’ve been able to do. I have not found gold data good for much. Copper and the SSE could get worse from here, even if the USA keeps growing briskly, if the US trader war on China ramps up further; say if US vassal states are brought into it for example.

  3. // Reply

    Well I am just saying there are really no leading indicators to say decisively that copper and China are heading back down. Might as well flip a coin. I mean … what would you propose as a data series that strongly leads copper?

  4. Justin Iriving
    // Reply

    I’d be surprised if anything leads copper reliably. The efficient markets hypothesis says that copper should be a random walk

  5. // Reply

    Wait … by the NGDP model shouldn’t copper be predictable by Fed policy, in the same way gold is? Gold is a pure reflection of Fed policy: Fed too tight = gold down, Fed too loose = gold up. Copper is one step removed from gold (both are raw commodities but copper suffers variable industrial demand unlike gold), copper should respond to Fed policy but with a longer lag.

  6. Justin Iriving
    // Reply

    The Market Monetarist approach says that all commodities and stocks have embedded within their price, a nearly information-efficient forecast of future NGDP growth. We use the markets to try and distill that estimate for a year-ahead NGDP outlook. If this unseen NGDP expectations path falls, then that will lower gold or copper or stock prices vis-a-vis the earlier, higher path. But economists can’t anticipate changes in the path of NGDP expectations, we can only try to read markets. If we could anticipate changes in NGDP expectations, we’d become extremely rich from trading, it’s really hard. The only way to beat the market reliably is to have inside information, which of course the Fed has at least conceivably, though the markets often seem to know what the Fed will do before the Fed knows.

  7. // Reply

    Fascinating post. Also, Golden Week box office was off 28% year-over-year.

Leave a Reply

Your email address will not be published. Required fields are marked *