Hi,
I am trying to run the following model using country-year level panel data: y= lag(x) + neighbor
Y is a continuous variable between 0 and 1 measuring a country's level of compliance with environmental regulation, x is whether or not it was sanctioned for non-compliance in t-1. I want to create an additional control (neighbor) that is a binary indicator of whether a country that is "similar" to the country was sanctioned in t-1. I want to define similarity based on three variables: a level of Y in t-2, a geographical region categorical variable, and a GDP variable. I'm not sure how to think about creating these "similarity" groups. Is a matching algorithm appropriate here, and does this make sense if one of the variables is time invariant (geography) and the others I want to measure for t-2.
I am trying to run the following model using country-year level panel data: y= lag(x) + neighbor
Y is a continuous variable between 0 and 1 measuring a country's level of compliance with environmental regulation, x is whether or not it was sanctioned for non-compliance in t-1. I want to create an additional control (neighbor) that is a binary indicator of whether a country that is "similar" to the country was sanctioned in t-1. I want to define similarity based on three variables: a level of Y in t-2, a geographical region categorical variable, and a GDP variable. I'm not sure how to think about creating these "similarity" groups. Is a matching algorithm appropriate here, and does this make sense if one of the variables is time invariant (geography) and the others I want to measure for t-2.
