Working Paper: Carry Trades, Risk Aversion, and Negative Betas

[Update Added Below: Is the link broken? Evidence suggests that this may be the case, at least temporarily. Once again it highligts the need to investigate the spuriousness of the link described below.]

[Update 2: I have had to remove the paper since the (few I imagine) downloads ate up my entire bandwith; email me if you want a copy]

Sorry for the silence this week, but I have not exactly been sitting idle. In this way I have been scribbling on one of those papers I said that I would finish and what is more ... I managed to finish the damn thing too. The paper is a working paper I have been writing on for the last couple of weeks and when I handed in my final project at school I figured that I might as well punch through. The paper is written in the context of my work along side the analysts at the Global Economy Matters blog.

What is it about then?

Well, the story begins back when I wrote a piece on the USD/JPY and how I related daily movements in the currency pair with movements in equities. More specifically, the piece homed in on the idea of carry trading currencies as so-called risk sentiment gauges in financial markets; especially since the credit turmoil began. The idea here is very simple in the sense that more than notional evidence suggests how the JPY (and indeed carry trade crosses in general) exhibit negative correlations and betas with risky assets.

Over the course of the past couple of months this story has also brewed elsewhere. In particular, the point about how the JPY is negatively correlated to equities was taken a step further when Greg Mankiw suggested (here too) that the Yen in fact is a negative beta asset. As per usual when Mankiw massages a topic it prompted many responses (e.g. here) but more importantly it alerted me to a more general inquiry in to what it actually means that this 'carry trade' behavior is embedded in the market. In short, this is the topic of my working paper entitled: Of Low Yielders and Carry Trading – the JPY and CHF as Market Risk Sentiment Gauges. (see above for access options)

Here is the abstract and conclusion:

Ever since the credit turmoil took hold in the summer 2007 financial markets have been on the brink. On this background, strong evidence is provided for the idea of carry trading currencies as risk sentiment gauges in the market. Using daily returns from 2006 to May 2008 this paper shows how traditional carry trading currency crosses (mainly JPY and CHF crosses) exhibit strong negative correlation and beta values with three key equity markets (SP500, Nikkei 225 and DAX 30). This paper also shows how this relationship, in relation to specific currency pairs, has been particularly strong since the advent of the credit crisis. This paper also homes in on the idea of carry trading currencies as means of hedging equity returns and fluctuations on a daily basis. At an initial glance such relationships are however bound to be highly spurious. As such, this paper also attempts to qualify its findings in a more general and solid empirical context. 


The principles of carry trading and how to bet against the patchy theory of uncovered interest rate parity are well known. However, carry trading and the effect of investors pursuing it have almost turned in to an urban legend on financial markets where many derivative effects of ‘carry trading behavior’ are cited. This paper has attempted to scrutinize one of the most widespread of these. It has consequently been shown how the JPY and CHF, often cited as the traditional funding currencies in carry trades, exhibit strong negative correlations and beta values to equities. This lends evidence to the idea of the CHF and JPY as risk sentiment gauges in the market. A conclusion important to this argument was also that the negative beta values and their explanatory power increase in the context of an exogenous event which brings volatility and uncertainty to the market.

Regarding the potential for investors to use carry trade crosses to hedge equity positions the evidence appears strong. An initial glance will thus reveal that most currency pairs not only exhibit negative correlations with equities but also how they posses significant, in statistical terms, negative beta values. This suggests that investors can indeed cover equity positions by buying low yielding currencies in times of volatility. Yet, these conclusions need to be taken with a pinch of salt. First of all, the study is based on daily returns which means that only investors of a certain pedigree would be able to benefit from these correlations (i.e. investors trading on a daily basis). Moreover, it is not certain that such correlations can be assumed to persist. In this regard it is important to watch the currency pairs with significant negative beta values and relatively high coefficients of determinations across the two periods (GBP/JPY and GBP/CHF to the DAX 30 and EUR/CHF and AUD/JPY to the Nikkei 225).

Further studies on this topic should attempt to widen the time span of the sample to gauge the general validity of the results as well as attempt to make forecasts of daily exchange rate returns based on the relationships cited above.

Now, the conclusions are operationalized through regression analysis and correlations and I would not consider the paper heavy in quantitative terms. However, these things are always subjective since academic economists will probably find it too light on the math (I should have used GARCH, tested for unequal variance in the context of the Chow Test etc, I know :)) whereas others will shun it because of the statistical operationalization of the argument. More interesting is the validity of the results. I mean, it was only a few days ago when Macro Man smugly noted how bad hedges were falling apart for good people citing the exact same relationship (or more aptly its spuriousness) which I attempt to formalize in my paper. So it could seem as if I am too late with my findings in that the horse has already bolted across the meadows; or perhaps, to stay in Macro Man territory, it was always a pink flamingo? I will let my astute readers figure that one out for themselves.

Lastly, and specifically in the context of the JPY the paper also has implications for the conceptualization of those famous fundamentals driving the JPY. On the one hand we have the steady and relentless decline in Japanese investors' home bias and subsequent outflow of funds (remember to read Stephen Jen here too). Just this week JPMorgan's chief currency strategist noted how the Yen could decline to 114 against the USD on outflows alone. On the other hand there is the effect I latch onto in my paper by which risk sentiment and risk aversion drives the movements of funding carry trade currencies. When the first effect stops and the other takes over (and vice versa) is difficult to say but for investors it is important know them both.

All tables and results are available in the paper which can be downloaded above. If you want the actual spreadsheet(s) drop me a line and I will ship them over. Remember also that this is a first draft which basically means "not finished" and lots of typos. I will probably go over it in a month or so and refine it. Also, my impression is that a similar study has to exist within the vaults of some of the IB's or hedge funds' research edifices or is this just too common knowledge to jot down?