I’m running a regression to test the stock market in Sweden. In essence, I have made a categorization of two types of buys and two types of Sells. Then I have divided the sample containing all trades over time into small, medium, and large size firms and run the same regression on the different samples.
When I regress in the small data sample and the large one, I get nothing unusual. However, running the regression on medium sized firms I get an unusally high intercept, which is statistically significant. This goes for all of the following scenarios:
- Only running the regression on the two buy dummy variables.
- Only running the regression on the two sell dummy variables.
- And running it on all dummy variables but excluding the largest one (which will then be characterized by the intercept).
What might cause the coefficient of the intercept to be so high? I have also conducted the regression on all three scenarios in the large data sample, where I have not distinguished between firm size, and there I do not get anything out of the ordinary..
Hope you can help, and that you understand the issue I’m dealing with.