I'm trying to test the relationship (only in cross-section - not with time series data) between national consumption and the fragile state index and my project advisor suggested I include inflation and interest rate variables in my OLS regression model. I'm having trouble immediately seeing the theoretical reason for this given that both inflation and interest rates are theoretically consequential from previous time period conditions, and interest rates are determined exogenously by government policy based on the other variables for the purpose of influencing future values of those variables. Am I wrong in this and there is another way of viewing this that is consistent with my advisors recommendation?


1 Answer 1


Consumption via the Euler equation should depend on the real interest rate, which is a function of nominal interest rates and expected inflation. Current inflation may be a good proxy for expected inflation. Hence, consistent with your adviser's recommendation, these two variables may indeed be relevant regressors for consumption. The big question is whether that will give rise to an omitted variable bias. For that you need to consider whether the fragile state index may be correlated with these variables. Often it is easiest to simply include the two additional control variable to see whether the main result is robust.

  • $\begingroup$ Thanks. For general reference, see here for checking for OVB. You can check for covariance by regressing variables on each other. In STATA there is a "vif" command to check for collinearity in a model. Then scatter plot or regress the residuals on the other independent variables to see if there are any correlations, indicating OVB. statisticsbyjim.com/regression/confounding-variables-bias/…. $\endgroup$
    – kleinerde
    Mar 11, 2022 at 14:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.