I am studying the effects of stock splits on the liquidity of a firm. In my sample I have c. 150 firms that have split their firm's share in the period 2000-2020 as well as 150 control firms. For each firm, I have 4 observations before the split, and four observations (each observation represents one quarter) after the split. In total, I have 8 observations per company (4 before and 4 after). The timing for each split varies per firm, so firm A might have split their shares in 2011Q1 and firm B might have split their shares in quarter 2010Q4. As with all DiD analysis, I have a treatment dummy that takes the value of one for all observations of a firm, only if the firm has actually split their shares. Additionally, a post-split dummy is added which takes the value of one in the four observations after the stock split.
For each firm, I have gathered a control firm which is observed in the same quarters (based on size/book-to-market) but has not split its shares. This would normally be named the control group. My dataset thus starts with 8 observations from Firm A (which has split its shares), followed by 8 observations from Firm C (which serves as a control group). The post-split dummy for Firm C behaves the same as the post-split dummy for firm A. Please see below for a stylized example:
A | 2010Q1 | 0.5 | 0 | 1 |
A | 2010Q2 | 0.5 | 0 | 1 |
A | 2010Q3 | 0.5 | 0 | 1 |
A | 2010Q4 | 0.6 | 0 | 1 |
A | 2011Q1 | 0.7 | 1 | 1 |
A | 2011Q2 | 0.7 | 1 | 1 |
A | 2011Q3 | 0.8 | 1 | 1 |
A | 2011Q4 | 0.9 | 1 | 1 |
C | 2010Q1 | 0.5 | 0 | 0 |
C | 2010Q2 | 0.5 | 0 | 0 |
C | 2010Q3 | 0.8 | 0 | 0 |
C | 2010Q4 | 0.8 | 0 | 0 |
C | 2011Q1 | 0.8 | 1 | 0 |
C | 2011Q2 | 0.9 | 1 | 0 |
C | 2011Q3 | 0.8 | 1 | 0 |
C | 2011Q4 | 0.9 | 1 | 0 |
etc. | ... | .... | ... | ... |
Any help would be greatly appreciated, many thanks!
FYI: Cross-posted at Stata reddit.
0 Response to How to do a DiD-analysis in stata with multiple pre- and post-observations per firm and different treatment times per firm?
Post a Comment