July 2017 – what a weather ride! First, temperatures up to 37 degrees Celsius in Stuttgart. Then, downpours for days on end. How did this affect you – apart from having to change plans for outings at short notice, or going to outdoor parties in wellingtons and raincoats? As finance researchers, we like to question whether certain phenomena also affect financial markets.

The 2005 article “Rain or Shine: Where is the Weather Effect?” published in European Financial Management by Goetzmann and Zhu addresses exactly the relation between weather and market makers’ behavior!

Adjusting weather for seasonal effects

Goetzmann and Zhu consider investors from five major urban areas in the United States (New York, Los Angeles, San Francisco, Chicago and Philadelphia) and analyze the relation between local weather on the stock market and on trading activity. They measure “weather” via cloudiness (daily sky cover/SKC), relative to a seasonal average. Hence, the results will not be affected by generally lower trading in summer, or widespread phenomena as the winter blues. The stock market’s evolution is measured via NYSE/AMEX daily index returns.

There is a weather effect, …

In line with earlier studies, Goetzmann and Zhu find that cloudiness negatively correlates with stock returns. Table 1 shows that the statistical significance is high with a 5% significance level.

NYSE/AMEX return *100 Net buy in shares
SKC -0.038** -51, 642

Table 1: Correlation between daily sky cover SKC and NYSE/AMEX return, net buy in shares. Source: Table 2 and 4 of Goetzmann/Zhu (2005). ** indicates signifcance at the 5% level.

But where does this effect come from? To zoom in on the mechanism, the authors check whether investors are more likely to sell stocks on cloudy days. If that were the case, the selling pressure would drive stock prices down, leading to negative returns. But Goetzmann and Zhu find no significant impact of cloudiness on individual investors’ propensity to sell (or buy) stocks. In summary, there is no relation between cloudiness and trading volume in any urban area!

…but who is weather-sensitive?

So, how does the relation between returns and weather arise? Goetzmann and Zhu propose an explanation that is related to a research area at our department: empirical market microstructure. In equity markets, a specific group of market participants called “market makers” stand ready to buy stocks if you want to sell, and sell stocks if you want to buy. These market makers affect stock prices directly through the price they charge for providing immediacy. When the bid-ask spread is large, an investor must buy at a high price, and sell at a low price. This reduces the returns that investors can earn.

Weather and market makers’ behavior

When the authors run a regression of the (change in the) bid-ask spread on cloudiness, they find a positive and significant relationship. Table 2 shows that spreads decrease on sunny days, and increase on cloudy days.

Change in bid-ask spread*100
SKC 0.0599***
SKCAM 0.904***

Table 2: Regression of SKC and SKCAM on change in bid-ask spread. Source: Table 13 of Goetzmann/Zhu (2005). *** indicates significance at the 1% level.

This result is robust to a number of specifications: including the index return, changes in trading volume, dummies for rain and snow, and interaction terms leaves the result mostly unchanged. One intuitive observation: morning weather (SKCAM) has a stronger impact on market makers – who are usually working in windowless rooms, and thus most affected by the morning weather.

So, do market makers cause the return effect?

What remains of the weather effect once the authors adjust for bid-ask spreads? Table 3 shows that once Goetzmann and Zhu include liquidity changes, and an interaction between cloudiness and liquidity changes, the effect of cloudiness on returns disappears!

NYSE/AMEX return*100
SKC -0.00818
Liquidity changes -3.97***
SKC*Liquidity changes -0.682***

Table 3: Regression of NYSE/AMEX index return on SKC, liquidity changes, and interaction term between SKC and liquidity changes. Source: Table 14 of Goetzmann/Zhu (2005). *** indicates significance at the 1% level.

Open question: Why do market makers react to the weather?

The results imply that market makers react to the weather, and spread this effect to the entire market. One remaining question: why do market makers react to the weather at all? One explanation is that, unlike individual investors, market makers work at the exchange physically. So the local weather during the commute may influence their behavior. Additionally, market makers do work in an environment that is not fully competitive. This makes it harder for other market participants to arbitrage away profits from market makers’ irrational behavior (such as a weather effect).

Author: Linda Klingler