When would you use 2sls instead of ils?

Two-stage least squares (2SLS) is used instead of ordinary least squares (OLS) when there is concern about endogeneity in the regression model, such as when an independent variable is correlated with the error term. This typically arises in the presence of omitted variable bias, measurement error, or simultaneous causality. 2SLS helps to provide consistent estimators by using instrumental variables that are correlated with the endogenous explanatory variables but uncorrelated with the error term. In contrast, OLS is appropriate when all variables are exogenous and there are no such concerns.