AUD/USD has rallied in the last couple of weeks as “risk on” has dominated the market. But it seems that China really is very weak and the RBA may well cut interest rates further (it sounded doveish after the last meeting). AUD/USD is quite strongly correlated with the 2-year interest-rate spread between the US and Australia, and the market is presently towards the top of the expected range given the current spread. If Australian rates are likely to be cut further and to stay lower for longer (on account of Chinese weakness) then it would be sensible to be short AUD/USD when it is at the extreme of the expected range. A short-term rally in the markets could well take it there, and I will keep watching for such a development. On the same argument, it would have been reasonable to short it a week ago, but I am wary of catching falling knives after my experiences this year (when I have been too early to go long gold and to short USD/JPY).
What is the Value of Econometrics?
I have been thinking a lot recently about the idea of systematic fundamental macro. Bridgewater has become very successful with this kind of strategy. I do not have access to the fund’s track record, but if it has consistently made money then it must be because its models incorporate knowledge about how the economy works that I do not have.
How to work out these models for oneself? One could posit and test potential relationships using econometrics; or one could codify the ad-hoc models in one’s head about the relationships between macroeconomic and financial variables and back-test the result. I expect that a combination of the two is the right approach. (I don’t think the idea of using academic macroeconomists’ models, which might seem an obvious third approach, would have any merit.)
That has led me to wondering whether I know enough about econometrics (i.e. statistics applied to economic data). I am sceptical about whether advanced statistics are really necessary for teasing out the important relationships (and even simple multiple regressions may be more than is needed — volatility-weighted composite indices do not do a bad job at capturing the effects of a range of factors on some variable in my experience) but I fear that there is a chance that that scepticism might be based more on ignorance that on knowledge. So I have ordered some books on the subject.
- US durable goods orders 1.1% b.e. May. Core 0.4% d.e.
- US pending home sales 5.9% b.e. May.
- German employment data
- UK current account
- UK final GDP
- European Council meeting
- Initial claims
- US final GDP