\FloatBarrier

\begin{table}[H]
  \caption{Policy responses}\label{REGPOLICYRESPONSE}
\begingroup
\centering
\begin{tabular}{lccc}
   \tabularnewline \midrule \midrule
   Dependent Variable: & \multicolumn{3}{c}{AF}\\
                           & \textbf{2011 Policy Model}  & \textbf{Direct Effect}  & \textbf{Neighbor Response} \\    
   Model:                  & (1)                         & (2)                      & (3)\\  
   \midrule
   \emph{Variables}\\
   Sbd.1-Post 2011         & -30.87$^{***}$              &                          &   \\   
                           & (8.477)                     &                          &   \\   
   In CREP-Post 2011       & -31.17$^{***}$              &                          &   \\   
                           & (7.072)                     &                          &   \\   
   Near to CREP-Post 2011  & -5.407$^{**}$               &                          &   \\   
                           & (2.218)                     &                          &   \\   
   CREP Effect (ATT)  &   & -38.70$^{***}$ & -2.788$^{**}$\\  
		      &                           & (2.905)        & (1.021)\\
  \midrule
   \emph{Fixed effects}\\
   Near CREP-Post CREP     & $\checkmark$                &                          & \\  
   In Fallow Program       & $\checkmark$                &                          & \\  
   Well                    & $\checkmark$                & $\checkmark$             & $\checkmark$\\   
   Year                    & $\checkmark$                &                          & \\  
   Subdistrict-Year        & $\checkmark$                & $\checkmark$             & $\checkmark$\\   
   Year-Ditch              & $\checkmark$                & $\checkmark$             & $\checkmark$\\   
   In Fallow Program       &                             & $\checkmark$             & $\checkmark$\\   
   Subdistrict-Year        &                             & $\checkmark$             & $\checkmark$\\   
   In CREP After Treatment &                             &                          & $\checkmark$\\   
   In CREP-Year            &                             &                          & $\checkmark$\\   
   \midrule
   \emph{Fit statistics}\\
   Observations            & 48,563                      & 49,439                   & 49,439\\  
   R$^2$                   & 0.74535                     & 0.74748                  & 0.74705\\  
   Within R$^2$            & 0.00271                     & 0.01114                  & 0.00174\\  
   \midrule \midrule
   \multicolumn{4}{l}{\emph{Clustered (Well \& Year) standard errors in parentheses}}\\
   \multicolumn{4}{l}{\emph{Signif. Codes: ***: 0.01, **: 0.05, *: 0.1}}\\
\end{tabular}
 
\par \raggedright 
In model 1, all active CREP wells are removed. So a well that enters CREP in 2014 has data from 2009-2013.\\
A well is considered a neighbor to a CREP or Fallow Program well if they are within one-half mile.\\
Each of the 10 largest ditches are included in the Ditch-Year fixed effect.\\
Subdistricts One through Six included in the Subdistrict-Year fixed effect.\\
CREP wells are removed from the dataset when estimating neighbor responses.
\par\endgroup
\end{table}