@DTrim99 and I found that the CPS understates the number of children under five in DC by 30% compared to the more-reliable ACS (notebook), which is causing inaccurate forecasts of the District Child Tax Credit. Short of calibrating the nationwide CPS to each state (like what we're doing in the UK), we can start by adding state-level targets to the nationwide version. Currently the ECPS is even farther off for this, since there are no state targets. This may also improve other targets related to SALT and SNAP, for instance, or make them more resilient to out-of-sample targets that relate to state.
@DTrim99 and I found that the CPS understates the number of children under five in DC by 30% compared to the more-reliable ACS (notebook), which is causing inaccurate forecasts of the District Child Tax Credit. Short of calibrating the nationwide CPS to each state (like what we're doing in the UK), we can start by adding state-level targets to the nationwide version. Currently the ECPS is even farther off for this, since there are no state targets. This may also improve other targets related to SALT and SNAP, for instance, or make them more resilient to out-of-sample targets that relate to state.