Alternative Viewpoints: Raining on the weather/return correlation parade

In our monthly column featuring the thoughts of a member of the Chartered Alternative Investment Analyst (CAIA) Association, we feature an active publisher in highly rated journals who has recently written an article on weather variables and their impact on financial markets. Wessel Marquering, Ph.D., CAIA, is quantitative researcher at the Talergroup.  Marquering and fellow researcher Ben Jacobson wrote an interesting paper on weather and financial markets for the Journal of Banking & Finance.  What follows are Marquering’s thoughts on the promise and peril of trying to extract alpha from the weather.

Alpha in the Weather: Alternative Viewpoints, powered by CAIA

Special to by: Dr. Wessel Marquering, CAIA, Talergroup

As readers of this website are no doubt aware, weather derivatives trading is taking off – with trading volumes going through the roof and more hedge funds venturing into this space. Basically, a weather derivative is a financial product in which two parties agree to exchange cash flows determined by reference to a weather index. The reference indices include temperature, rainfall, wind speed, humidity, snowfall, to name a few, but the most heavily traded contracts are based on temperature indices.

On the one hand, weather derivatives can be used to manage risk, by insuring for example farmers against a bad crop, as an insurance against bad weather on holidays, by decreasing the exposure to temperature-related risk factors, etc. On the other hand, they have become a relatively new way to generate alpha.

These alpha opportunities arise because weather derivatives are difficult to price. And since weather patterns are not random, the Black-Scholes option model is not entirely appropriate. Some hedge funds actually hire meteorologists and run highly quantitative models to forecast the weather in an attempt to identify bargain contracts.  Since the weather is uncorrelated to, for example, sub-prime, Iraq war, etc., they are a great addition to investors’ portfolios.

Having said that, the weather derivative market is largely dominated by some very large participants. So any inefficiencies can still be arbitraged away quickly. Moreover, the market for weather derivatives requires managers specialization and can be highly volatile – aspects ignored by some (now defunct) hedge funds.
Nevertheless, there are some interesting opportunities for hedge funds in the weather derivatives market. Cross market trading the weather risk in emission rights, for example, offers some interesting opportunities to create alpha. The figure below shows the evolution of the temperature anomaly (temperature relative to the mean temperature between 1961 and 1990) in Europe over the period 1500-2007. As you might expect, it shows a clear increase in the last decades.

Source: Luterbacher et al.( 2007): Exceptional European Warmth and Phenology, Geophysical Research Letters, 34. pp. 1-6. (related presentation)

As a result, there has been a relatively new market of CO2 emission rights. But global warming has led to an increase in weather products in general, and there are many more sources of alphas, in which most cases the temperature is the driving factor. But unfortunately, not much attention has been paid on this topic in academics.

A recent study by Cao and Wei (Stock Market Returns: A Note on Temperature Anomaly), however, claims to have found interesting relationships between temperate and financial markets. Since weather conditions affect not only agricultural, retail, tourism, and energy sectors, but they also influence the mood of many people (and thus the mood of investors) and therefore stock prices themselves.  Cao and Wei cite psychological studies that show the relationship between temperature and human behaviour to support their argument. The authors find that lower temperatures are associated with higher stock market returns (due to aggressive risk-taking) and higher temperatures can lead to higher or lower stock returns, depending on which mood, aggression (risk-taking) or apathy (risk-avoidance) dominates.

In a study I co-authored for the Journal of Banking and Finance, we find that temperature is highly correlated with some other seasonal market indicators.  Taken together, these indicators all reflect an intra-annual seasonal pattern in stock market returns.  This pattern, which suggests that stock returns are higher during winter months (November through April) than summer months (May–October period), is commonly known as the Sell-in-May effect.

Source: Bouman, Sven and Ben Jacobsen (2002), The Halloween Indicator, Sell in May and Go Away: Another Puzzle, American Economic Review, 92 (5), pp. 1618-1635.

The above figure summarizes the average returns in winter and summer periods for 19 countries. For all the countries there is a clear seasonal pattern of higher average returns in winter periods. Simple Sell-in-May trading strategies (in at May 1st and out at November 1st) generates high and statistically significant estimated alphas in most countries – returns that cannot be explained by data-mining, the January effect, or alternative risk factors. While this strategy may sound overly simplistic, I have actually come across several mutual and hedge funds managers that employ this strategy as one of their tools to generate alpha.

In our study we re-examine several weather related anomalies for 48 countries. Among other econometric methods, we apply the J-test of Davidson and Mackinnon, which is a powerful tool to take into account possible multicollinearity.  We find that the Sell-in-May variable (which is naturally very highly correlated to temperature) remains a more likely candidate to explain the seasonality than the temperature variables themselves.  Thus, we conclude that it is simply not enough to link temperature directly to stock returns on the assumption that these variables affect mood and therefore affect stock returns.

Further, we show that other variables with a strong seasonal pattern do at least as well at explaining stock market returns.  For example, variables based on seasonal affective disorder and even the sales of ice cream, which surely can’t influence stock returns, can still be linked to the performance of stock returns.

The bottom line is that the correlation between weather-induced mood shifts and stock market returns does not imply causation. Without any further support this means that the suggested relation between temperature and stock market returns could just be data-driven inference based on spurious correlations.

Nevertheless, the possible causal relationship between the weather and stock market returns will surely continue to capture the imagination of investors and fund managers. Perhaps someday, researchers will uncover what philosophers have espoused for eons – that the world and its people are part of one interconnected system. Until then, don’t be so quick to sell on a rainy day.

Be Sociable, Share!

Leave A Reply

← Is an "integrated" 130/30 portfolio always better than a "combined" one? A "small-cap bias" in hedge funds themselves? →