![cluster standard errors stata cluster standard errors stata](https://www.stata.com/new-in-stata/difference-in-differences-DID-DDD/img/trendplot.png)
![cluster standard errors stata cluster standard errors stata](https://i.ytimg.com/vi/YP2uziGL_bo/hqdefault.jpg)
#CLUSTER STANDARD ERRORS STATA SERIAL#
In general, when working with time-series data, it is usually safe to assume temporal serial correlation in the error terms within your groups. You can account for firm-level fixed effects, but there still may be some unexplained variation in your dependent variable that is correlated across time. (independently and identically distributed).Ī classic example is if you have many observations for a panel of firms across time. In these cases, it is usually a good idea to use a fixed-effects model.Ĭlustered standard errors are for accounting for situations where observations WITHIN each group are not i.i.d. Unless your X variables have been randomly assigned (which will always be the case with observation data), it is usually fairly easy to make the argument for omitted variables bias. If your dependent variable is affected by unobservable variables that systematically vary across groups in your panel, then the coefficient on any variable that is correlated with this variation will be biased. Which approach you use should be dictated by the structure of your data and how they were gathered.įixed effects are for removing unobserved heterogeneity BETWEEN different groups in your data. It is perfectly acceptable to use fixed effects and clustered errors at the same time or independently from each other. It’s important to realize that these methods are neither mutually exclusive nor mutually reinforcing. Then I’ll use an explicit example to provide some context of when you might use one vs. I’ll describe the high-level distinction between the two strategies by first explaining what it is they seek to accomplish. I found myself writing a long-winded answer to a question on StatsExchange about the difference between using fixed effects and clustered errors when running linear regressions on panel data.