How does weather influence market returns and trading activity in Belgium ?

This study⁽¹⁾ explores the influence of weather conditions on market returns and trading activity. It is assumed that meteorological parameters influence mood, which will in turn affect individuals when making investment decisions. Unlike the few behavioural finance studies on the subject, this work relies not only on market data but also on a sample of retail trading accounts to better capture heterogeneity across investors. The results reveal a significant influence of sunshine duration, precipitation quantity and temperature on financial markets.

1. Introduction

Weather is a key factor that shapes people’s daily lives. Significant changes in meteorological conditions can affect them both psychologically and physically. Researchers in psychology have established links between meteorological conditions and mood (e.g., Cunningham, 1979). Past research has also shown that individuals’ decisions may partially depend on their mood (Cao and Wei, 2005). Focusing on financial decisions, weather is expected to influence investor behavior through its effect on mood.

Adding to this stream of research, this study investigates the influence of weather conditions on market returns and retail investor behavior. Unlike the few studies on the subject, this work relies not only on market data but also on a sample of retail trading accounts. Using data at the individual level allows heterogeneity across investors to be better taken into account. Spanning a period of approximately 20 years, this study focuses on three weather variables: sunshine duration, temperature, and precipitation quantity.

The remainder of this paper is organized as follows. Section 2 summarizes the literature review. The hypotheses and data are presented respectively in Sections 3 and 4. The results of the empirical work are reported in Section 5. Section 6 concludes.

2. What do we know so far?

The psychology literature highlights that individuals’ mood can fluctuate significantly over short periods of time due to external factors such as weather conditions (Kööts et al., 2011). Past research has shown that meteorological conditions can indeed influence individuals’ affect as their senses respond to various climatic factors (Lu and Chou, 2012). In particular, sunshine is usually associated with improved mood, whereas the opposite is commonly observed with cloud cover, humidity, or rain (Cunningham, 1979; Sanders and Brizzolara, 1982; Schwarz and Clore, 1983; Parrott and Sabini, 1990). Findings related to the effects of temperature are more mixed. Some papers document a positive relationship between individuals’ affect and this weather parameter (Cunningham, 1979; Howarth and Hoffman, 1984), while others highlight a negative correlation between these variables (Bell and Baron, 1976).

Another well-established finding in psychology is that mood and emotions can influence individuals’ decision-making processes. For instance, Johnson and Tversky (1983) support the “mood as information” hypothesis, which posits that mood variations can influence people’s decisions, even when the source of these fluctuations is unrelated to the choices to be made (Lu and Chou, 2012). From this perspective, affect can function as a form of information when making decisions (Slovic et al., 2002). Two main factors are said to determine the extent to which mood influences the decision-making process: the level of uncertainty and the complexity of the decision (Forgas, 1995). It is therefore reasonable to consider that emotions can play a significant role in financial decision-making, given that financial choices are generally characterized by a high degree of uncertainty and complexity (Frühwirth and Sögner, 2015).

Based on these theoretical foundations, researchers in behavioral finance have examined whether weather conditions influence financial markets, through their impact on investors’ affect. For instance, Dong and Tremblay (2018) assumed that “pleasant weather”, such as sunshine, should improve investors’ optimism, thereby increasing their propensity to purchase assets. Most of the studies in this field focus on market returns and report that sunshine is positively correlated with returns, unlike cloud cover, temperature and rainfall (Hirshleifer and Shumway, 2003; Cao and Wei, 2005). However, these relationships may differ depending on the location and the season, except for sunshine that consistently remains positively associated with market re-turns (Dong and Tremblay, 2018).

Researchers have also addressed the relationships between weather conditions and retail investors’ trading activity, but to a lesser extent. Moreover, most of them have used aggregate market data, such as market depth or volatility, as proxies for trading activity. According to the existing literature, only two studies really focus on retail investors. The first study is from Goetzmann and Zhu (2005), who find no significant influence of cloud cover on trading activity of a sample of individual investors. The second paper is from Schmittmann et al. (2015), who report that cloud cover and precipitation are positively associated with retail trading activity, while temperature has the opposite effect. In addition, their findings support a positive influence of temperature on investors’ tendency to purchase assets, whereas the opposite is observed with precipitation or cloud cover.

This work contributes to the aforementioned literature by exploring the influence of weather conditions on market returns and retail investor behavior in Belgium.

3. Hypotheses

Building on the literature, empirically testable hypotheses specific to the influence of sunshine, temperature, and rainfall on market returns and retail trading activity were formulated.

Regarding market returns, researchers generally found a positive relationship with sunshine and a negative relationship with rainfall or temperature. The first three hypotheses are consistent with similar effects expected in Belgium.

Regarding investor behavior, past research reports evidence of a positive relationship between sunshine and mood. This led to the assumption that both trading activity and purchasing behavior are positively associated with sunshine duration. Since temperature and precipitation are known to negatively influence mood (Bell and Baron, 1976; Schwarz and Clore, 1983), it is assumed that they are both negatively associated with investors’ trading activity and tendency to purchase assets.

4. Weather, market return and trading account data

The data used in this study come from three sources and span approximately two decades from January 2000 to August 2021.

The meteorological data were provided by the Belgian Royal Meteorological Institute (RMI). Monthly records of sunshine duration, average temperature, and precipitation quantity were used in the analyses. The detailed information available for each province was aggregated to obtain data at the country level. Following Hirshleifer and Shumway (2003), Goetzmann and Zhu (2005), and Cao and Wei (2005), the weather parameters were de-seasonalized to ensure that the observed effects could be attributed to the variables under scrutiny⁽²⁾.

Concerning market returns, data relating to the Bel-20 and Brussels All-Share (Bel-all) indices were downloaded on Bloomberg terminals. Focusing on monthly closing prices from January 2000, monthly returns (starting in February 2000) were computed for both indices.

The sample of retail trading account data was made available by the supervisor of this master’s thesis (against signing a strict confidentiality agreement). This sample comes from a larger anonymized database provided by a Belgian online brokerage firm. It includes 5064 investors, for whom the number of monthly transactions and the monthly monetary trading volumes (in euros) are available over the entire period. Furthermore, these data also contain individual characteristics, including gender, birthdate, education level, as well as subjective measures of financial literacy and risk tolerance. The mean age of investors in 2021 is 60.86, and approximately 93% of them are male. 72.39% of these investors hold a university degree. Under the MiFID regulations and the related “Know Your Customer” process, investors were asked to self-assess their financial literacy on a 5-level scale. The average subjective financial literacy in the sample is 3.56 out of 5. Investors were also asked to self-report their risk tolerance on a 5-level scale. The sample’s mean risk tolerance is just under level 4 (out of 5).

5. Empirical findings

5.1. Sunshine, rainfall and market returns

To examine the relationships between mrket returns and meteorological conditions, several time-series OLS regressions were performed. Significant results were obtained for sunshine and precipitation. All things being equal, an increase in monthly sunshine duration is associ-ated with a rise in both Bel-all and Bel-20 monthly returns. The opposite result is found for precipitation, as expected. By contrast, there is no significant link between temperature and market returns.

5.2. Sunshine and aggregate trading activity

To explore the relationships between climatic factors and trading activity at the aggregate level, several time-series OLS regressions were estimated. Before running these regressions, two approaches were adopted to build the aggregate monthly measures of trading activity. The first approach only included active investors for each month. They are defined as individuals who executed at least one trade in that month. This approach measures trading activity only among investors who effectively decide to trade. The second approach, in addition to active investors, also integrates individuals who held a stock portfolio in that month and could have traded stocks, even if they did not. Put differently, this second approach refers to potential investors and is broader as it considers both regular and occasional investors. Both approaches were used to compute monthly averages per investor⁽³⁾ for the number of trades and trading volumes (in euros).

The results are mixed and suggest behavioral differences between active and potential investors as in Schmittmann et al. (2015). Potential investors tend to trade less (in number of trades) during sunny periods, which could be explained by the opportunity cost effect. This means that the opportunity cost of spending time trading is perceived as higher on sunny days. When the sun shines brighter than usual, potential investors might trade to a lesser extent because they prefer to spend time enjoying the sunshine.

When focusing on active investors, the average number of transactions does not vary with weather conditions. Nevertheless, they trade higher monetary volumes when sunshine duration rises, which is consistent with an increased optimism induced by sunshine.

This study also investigates whether investors’ buying behavior is influenced by weather conditions. The aggregate monthly “Buy-Sell Imbalance” (Lee, 2024) was thus measured to deter-mine if investors, taken all together, are net buyers (BSI > 0) or sellers (BSI < 0). No significant results were found.

5.3. Good vs. bad weather and individual trading activity

A further analysis of the relationships between weather variables and investor behavior was conducted using panel data models. The goal is to better account for heterogeneity across investors. Specifically, a balanced panel was used, which is consistent with some investors deliberately choosing not to trade at certain points in time. The dependent variables of these models (defined per month and per investor) are the number of trades, the monetary trading volume, and the BSI, respectively. Several control variables were included in these models. Retail investors’ personal characteristics are used as individual-varying variables, while dummies accounting for crisis periods serve as time-varying variables. Two individual- and time-varying variables were also added, namely each investor’s portfolio value and size (i.e., number of distinct stocks in the portfolio).

The results show that temperature is negatively correlated with both trading activity and pur-chasing behavior, which is consistent with this study’s hypotheses. By contrast, the findings regarding sunshine and precipitation do not support the hypotheses. Investors trade less and sell rather than buy when sunshine duration rises, while both the number of trades and trading volumes are higher with increased precipitation. Intuitively, Belgian investors might prefer to spend more time outdoors during sunnier or warmer periods (i.e., enjoying what is called a “good” or “nice” weather in Belgium), which results in reduced trading activity. In the oppo-site, during periods of heavier rainfall, people might be more inclined to stay indoors, thereby providing them with more opportunities to trade. As mentioned earlier, these findings align with the opportunity cost hypothesis proposed by Schmittmann et al. (2015).

Furthermore, investors may choose to sell rather than buy when temperature or sunshine increases, anticipating that they will spend more time outdoors and pay less attention to financial markets during that period. Such results suggest that the influence of weather on investor behavior may not entirely stem from its impact on mood, but also from changes in people’s lifestyles and daily activities caused by meteorological conditions.

6. Conclusion

This study addressed the influence of weather conditions on both market returns and retail trading activity in Belgium. Past research has established links between financial markets and climatic variables, but most studies have used aggregated market data as proxies for trading activity. Relying on a sample of retail trading account data spanning almost 20 years, this work adds to the literature with a direct focus on retail investor behavior.

Key findings indicate that market returns tend to increase (decrease) when the monthly sunshine duration (precipitation quantity) rises. At the aggregate level, the results suggest differences between active and potential investors. When it is sunnier than usual, potential investors execute fewer trades while active investors trade higher monetary volumes. Next, the findings based on panel data models show a positive relationship between precipitation and retail trading activity (in both number of trades and monetary volumes), while the opposite is found for higher temperatures or increased sunshine duration. The empirical work also reveals a higher propensity to sell when sunshine or temperature is higher than usual. Within the Belgian temperate maritime climate context, the results are consistent with changes in people’s lifestyle and day-to-day activities caused by meteorological conditions, without ruling out potential weather-induced mood effects.

This study’s findings have several implications. First, they deliver relevant insights into retail investor behavior, which are useful for brokers and financial advisors keen on providing better advice to their clients, any practitioners in the finance industry, and regulators whose mandate is to ensure investor protection. The results could also be helpful for retail investors seeking to better understand their own behavior. Next, some of the findings might help assess the usefulness of constructing any weather-based investment strategies for portfolio management (Hirshleifer and Shumway, 2003).

This study acknowledges limitations such as the small sample size and the specific characteristics of the climate in Belgium. Belgian retail investors might not react to meteorological conditions in the same way as investors living in countries with another climate type.

This study paves the way for further research on the influence of extreme weather conditions on retail investor behavior. This research direction would be particularly relevant, as climate change is leading to more frequent and severe weather events.

Endnotes

⁽¹⁾ This master’s thesis was achieved under the supervision of Prof. Catherine D’Hondt at the Louvain School of Management, UCLouvain.  Antoine Trotin and Catherine D’Hondt are grateful to the online brokerage house for providing the retail trading data.  They also thank the Belgian Royal Meteorological Institute for providing them with meteorological data.  The master’s thesis is available upon request (antoine.trotin@uclouvain.be), or via this link.

⁽²⁾ Since the empirical work is based on monthly data, an average of each weather factor for each month across all years was first computed. Subsequently, the deseasonalized variable for each month was obtained by subtracting the cross-sectional average of a given month from the observations of the corresponding month.

⁽³⁾ Working with measures per investor is necessary because all the investors did not start trading on the same date. In the sample, the number of retail investors is slightly increasing over the entire period.

References

• Bell, P. A., & Baron, R. A. (1976). Aggression and Heat: The Mediating Role of Nega-tive Affect 1. Journal of Applied Social Psychology, 6(1), 18-30.

• Cao, M., & Wei, J. (2005). Stock market returns: A note on temperature anomaly. Journal of Banking & Finance, 29(6), 1559-1573.

• Cunningham, M. R. (1979). Weather, mood, and helping behavior: Quasi experiments with the sunshine Samaritan. Journal of personality and social psychology, 37(11), 1947.

• Dong, M., & Tremblay, A. (2018). Does the weather influence global stock returns?. Critical Finance Review.

• Forgas, J. P. (1995). Mood and judgment: the affect infusion model (AIM). Psychological bulletin, 117(1), 39.

• Frühwirth, M., & Sögner, L. (2015). Weather and SAD related mood effects on the financial market. The Quarterly Review of Economics and Finance, 57, 11-31.

• Goetzmann, W. N., & Zhu, N. (2005). Rain or shine: where is the weather effect?. European Financial Management, 11(5), 559-578.

• Hirshleifer, D., & Shumway, T. (2003). Good day sunshine: Stock returns and the weather. The journal of Finance, 58(3), 1009-1032.
• Howarth, E., & Hoffman, M. S. (1984). A multidimensional approach to the relationship between mood and weather. British journal of psychology, 75(1), 15-23.
• Johnson, E. J., & Tversky, A. (1983). Affect, generalization, and the perception of risk. Journal of personality and social psychology, 45(1), 20.

• Kööts, L., Realo, A., & Allik, J. (2011). The influence of the weather on affective experience. Journal of individual differences.

• Lee, A. (2024). Buy-Sell Imbalances on and around Round Numbers and High-Frequency Trading. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3331198

• Lu, J., & Chou, R. K. (2012). Does the weather have impacts on returns and trading activities in order-driven stock markets? Evidence from China. Journal of Empirical Finance, 19(1), 79-93.

• Parrott, W. G., & Sabini, J. (1990). Mood and memory under natural conditions: Evi-dence for mood incongruent recall. Journal of personality and Social Psychology, 59(2), 321.

• Sanders, J. L., & Brizzolara, M. S. (1982). Relationships between weather and mood. The Journal of general psychology, 107(1), 155-156.

• Schmittmann, J. M., Pirschel, J., Meyer, S., & Hackethal, A. (2015). The impact of weather on German retail investors. Review of Finance, 19(3), 1143-1183.

• Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgments of well-being: Informative and directive functions of affective states. Journal of personality and social psychology, 45(3), 513.

• Slovic, P., Finucane, M., Peters, E., & MacGregor, D. G. (2002). Rational actors or rational fools: Implications of the affect heuristic for behavioral economics. The journal of socio-economics, 31(4), 329-342.