Seasonality in EPS Forecast Errors in the Finnish Stock Market – Seasonal Affective Disorder or Fundamental Driven?
Ruhanen, Vesa (2017)
Kuvaus
Opinnäytetyö kokotekstinä PDF-muodossa.
Tiivistelmä
EPS forecasts are one of the most important factors affecting stock prices. Previous literature has identified seasonality in analysts’ EPS forecast error. While traditional finance intends to explain the seasonality with firm fundamentals, behavioral finance emphasizes cognitive explanations. One such explanation is Seasonal Affective Disorder, according to which forecasts become more pessimistic as depression increases in the fall.
The main purpose of this thesis is to identify if SAD effect explains forecast error seasonality in EPS consensus estimates with a 1-year horizon, or if the error is explainable by traditional factors. Since previous literature combining analysts and SAD effect is based on data from the US, this thesis uses unique data from Finnish stock markets, where the effect should be theoretically stronger. Seasonality in EPS forecast error is studied with panel data consisting of 3665 common observations between 2010 and 2014. Along with SAD effect, independent variables control for firm specific fundamentals, changes in macroeconomic conditions, and analyst related factors. Regression analysis is performed for different subsamples according to firm size and forecast season. Validity of the results is verified by additional robustness tests that control for weaknesses in the methodology and possible inappropriate independent variables.
The results show general optimism, which is greater for small firms, and a seasonal pattern in forecast error. For full sample, the forecasts are 28.2% too optimistic on average. The error is the smallest in January (14.6%), peaks in March (38.2%), declines until September (21.8%), and stays flat for the year-end. Although there is weak pro SAD effect evidence, this is mitigated by robustness tests and inconsistent findings for small firms. Positive relationship is found between forecast error and previous quarter actual EPS suggesting that positive momentum decreases forecast error. Information asymmetry and analyst uncertainty about a firm’s earnings, on the other hand, are negatively related to forecast error and an increase in the variables causes higher optimism. Overall, the evidence suggests that in the setup fundamentals and analyst related factors drive forecast error instead of SAD effect. Although forecast error is not fully explainable by the chosen factors, this thesis presents a clear seasonal pattern, which is stronger for small firms, which helps investors to correct the bias in estimates.
The main purpose of this thesis is to identify if SAD effect explains forecast error seasonality in EPS consensus estimates with a 1-year horizon, or if the error is explainable by traditional factors. Since previous literature combining analysts and SAD effect is based on data from the US, this thesis uses unique data from Finnish stock markets, where the effect should be theoretically stronger. Seasonality in EPS forecast error is studied with panel data consisting of 3665 common observations between 2010 and 2014. Along with SAD effect, independent variables control for firm specific fundamentals, changes in macroeconomic conditions, and analyst related factors. Regression analysis is performed for different subsamples according to firm size and forecast season. Validity of the results is verified by additional robustness tests that control for weaknesses in the methodology and possible inappropriate independent variables.
The results show general optimism, which is greater for small firms, and a seasonal pattern in forecast error. For full sample, the forecasts are 28.2% too optimistic on average. The error is the smallest in January (14.6%), peaks in March (38.2%), declines until September (21.8%), and stays flat for the year-end. Although there is weak pro SAD effect evidence, this is mitigated by robustness tests and inconsistent findings for small firms. Positive relationship is found between forecast error and previous quarter actual EPS suggesting that positive momentum decreases forecast error. Information asymmetry and analyst uncertainty about a firm’s earnings, on the other hand, are negatively related to forecast error and an increase in the variables causes higher optimism. Overall, the evidence suggests that in the setup fundamentals and analyst related factors drive forecast error instead of SAD effect. Although forecast error is not fully explainable by the chosen factors, this thesis presents a clear seasonal pattern, which is stronger for small firms, which helps investors to correct the bias in estimates.