Test for different time trend for different groups?

[John Swiss]

I have panel data (`y`) at different time points (`t`) for individuals (`i`) in 2 different groups (`g`).

I am trying to test if the coefficient of `t` on `y` is larger for individuals in group 2. In other words, is the time trend more steep for a given group?

How would I go about doing that?

There is likely auto-correlation at the `i` level. Data looks like this

```
| y | t | i | g |
|---|---|---|---|
| 2 | 1 | 1 | 1 |
| 4 | 1 | 2 | 1 |
| 6 | 1 | 3 | 2 |

```

[Clyde Schechter]

Something like:

xtset i t
xtreg y i.g##c.t, fe vce(cluster i)
margins g, dydx(t)

The direct comparison of the time trend in group 2 vs group 1 will be found in the coefficient of 2.g#c.t.

The code itself does not depend on the number of groups, though the output will reflect that and in the presence of more than 2 groups the interpretation depends on whether the desired comparison is between group 2 and group 1 or some other contrast among groups.

[John Swiss]

Is this still valid if individuals are auto-correlated?

[Clyde Schechter]

Well, maybe you want Newey-West standard errors. See -help newey-.