Hi,

I am attempting to estimate the impact of hosting the World Cup on the growth rate of countries. I am considering the use of two samples. The first is the full sample of countries for which data is available (160+). I believe the treatment effects estimated in this sample would ATE.

The second is a sample of countries who have hosted or unsuccessfully bid to host the World Cup. These countries self-select into the sample by submitting a World Cup but do not get to choose whether they are treated or not. I have been told that the treatment effects in this sample would be equivalent to ATT (average treatment effects on the treated), although I am not sure this is strictly true as not all countries in this sample are treated. The unsuccessful hosts serve as the comparison group.

Could anyone offer advice as to whether estimation in the second sample is equivalent to estimating ATT? If not, is there a way to refer to treatment effects such as these?

Thanks,

Nic