The role of data limitations, seasonality and frequency in asset pricing models

► We measure how much a parameter estimate changes, on average, when the sample is increased by one observation. ► The results indicate that the estimates of the CAPM parameters significantly differ across samples, which are based on different days of the week (seasonality) and time intervals (changes in the distribution of returns over time). ► Parameter estimates based on a mixed sample (where seasonality and time variation is not accounted for) do converge to some values (but not necessarily to the true values). ► Therefore, if seasonality and time-varying moments are ignored in empirical analysis, statistical results may be unreliable, while this problem would not be noticeable.