remove unsubstantiated info from claude

Author: MaxGhenis

paper/sections/abstract.tex

@@ -1,3 +1,3 @@
 \section*{Abstract}
 
-We combine the demographic detail of the Current Population Survey (CPS) with the tax precision of the IRS Public Use File (PUF) to create an enhanced microsimulation dataset. Our method uses quantile regression forests to transfer income and tax variables from the PUF to demographically-similar CPS households, followed by a dropout-regularized gradient descent procedure that reweights households to match administrative targets. The enhanced dataset reduces discrepancies in key tax components by 40\% compared to the baseline CPS while preserving demographic relationships and program participation patterns. Validation against IRS Statistics of Income shows the enhanced data captures capital gains within 12\% of administrative totals (vs. 45\% baseline error), business income within 8\% (vs. 38\%), and dividend income within 7\% (vs. 32\%). The dataset matches state-level EITC claims within 5\% for 45 states and maintains the CPS's high accuracy for poverty estimation and program participation analysis. We release both the enhanced dataset and our open-source enhancement procedure to support transparent policy analysis.
+We combine the demographic detail of the Current Population Survey (CPS) with the tax precision of the IRS Public Use File (PUF) to create an enhanced microsimulation dataset. Our method uses quantile regression forests to transfer income and tax variables from the PUF to demographically-similar CPS households. We create a synthetic CPS-structured dataset using PUF tax information, stack it alongside the original CPS records, then use dropout-regularized gradient descent to reweight households toward administrative targets from IRS Statistics of Income, Census population estimates, and program participation data. This preserves the CPS's granular demographic and geographic information while leveraging the PUF's tax reporting accuracy. The enhanced dataset provides a foundation for analyzing federal tax policy, state tax systems, and benefit programs. We release both the enhanced dataset and our open-source enhancement procedure to support transparent policy analysis.

paper/sections/results.tex

@@ -1,46 +1,70 @@
 \section{Results}
 
-We validate our enhanced dataset against the administrative targets used in our loss matrix construction.
+We evaluate our enhanced dataset against administrative targets by constructing a loss matrix (defined in utils/loss.py) measuring relative deviations from:
 
-\subsection{Target Categories}
+\subsection{IRS Statistics of Income Targets}
 
-Our loss matrix tracks matches against:
-
-\subsubsection{IRS Statistics of Income}
-By AGI bracket and filing status:
+By AGI bracket and filing status, we track:
 \begin{itemize}
-    \item Adjusted gross income
-    \item Employment income
-    \item Business net profits/losses
-    \item Capital gains
+    \item Adjusted gross income totals
+    \item Return counts
+    \item Wages, salaries, and tips
+    \item Business net profits and losses (separately)
+    \item Capital gains (gross amounts and distributions)
     \item Ordinary dividends
-    \item Partnership and S-corporation income
+    \item Partnership and S-corporation income and losses
     \item Qualified dividends
     \item Taxable interest income
     \item Pension income
     \item Social Security benefits
+    \item Estate income and losses
+    \item Tax-exempt interest
+    \item IRA distributions
+    \item Rent and royalty income and losses
+    \item Taxable pension income
+    \item Taxable Social Security
+    \item Unemployment compensation
+\end{itemize}
+
+\subsection{Census Population Targets}
+
+From Census projections:
+\begin{itemize}
+    \item Population counts for each single year of age from 0 to 85
 \end{itemize}
 
-\subsubsection{Census Population Projections}
-Single-year age populations from ages 0-85.
+\subsection{CBO Program Totals}
 
-\subsubsection{CBO Projections}
-Annual totals for:
+From Congressional Budget Office projections:
 \begin{itemize}
-    \item Income tax
-    \item SNAP
-    \item Social Security
-    \item SSI
+    \item Income tax revenue
+    \item SNAP benefit payments
+    \item Social Security benefit payments
+    \item SSI payments
     \item Unemployment compensation
 \end{itemize}
 
-\subsubsection{Treasury EITC Statistics}
+\subsection{EITC Statistics}
+
+From Treasury data:
 \begin{itemize}
-    \item EITC claims and amounts by number of qualifying children
+    \item Number of returns claiming EITC by number of qualifying children
+    \item Total EITC amounts by number of qualifying children
 \end{itemize}
 
-% TODO: Add specific quantitative results from running the enhanced dataset against these targets
+\subsection{Other Targets}
+
+From various government sources:
+\begin{itemize}
+    \item Healthcare spending by age group and type
+    \item Child support payments
+    \item Housing costs and subsidies
+    \item Market income losses
+\end{itemize}
 
-\subsection{Computational Performance}
+The reweighting procedure minimizes the relative squared error between weighted sums of these variables and their administrative targets. 
 
-% TODO: Add measured runtime statistics from production enhancement runs
+% TODO: Add specific quantitative results showing:
+% - Initial deviations from targets in base CPS
+% - Final deviations after enhancement
+% - Distribution of weights between original and synthetic records