ltc-paper-scripts 0.1.0__py3-none-any.whl

ltc_paper_scripts/latex_figures.py

@@ -0,0 +1,250 @@
+# --- Import necessary packages ---
+import math
+import os
+from pathlib import Path
+from reportlab.pdfgen import canvas
+from reportlab.lib.colors import black, HexColor
+from reportlab.pdfbase import pdfmetrics
+from reportlab.pdfbase.ttfonts import TTFont
+from figure_ticks import extended_breaks
+
+# Port of charter-schools/code/r/{constants,helpers}.R: same colors, physical
+# canvas sizes, discrete scales, cropped bar axes, centered CIs and gain brackets.
+# Prefer the manuscript font where installed. The open fallback ships with the
+# package; proprietary Office/Windows font files are read locally, not copied.
+BLUE = HexColor('#29B6A4')
+ORANGE = HexColor('#FAA523')
+GRAY = HexColor('#999999')
+FONT, BOLD = 'PaperSans', 'PaperSans-Bold'
+font_root=Path(__file__).parent/'fonts'
+office_fonts=Path('/Applications/Microsoft Word.app/Contents/Resources/DFonts')
+windows_fonts=Path(os.environ.get('WINDIR','C:/Windows'))/'Fonts'
+font_pairs=[
+    (font_root/'Lucida Sans.ttf',font_root/'Lucida Sans Demibold Roman.ttf'),
+    (office_fonts/'Lucida Sans.ttf',office_fonts/'Lucida Sans Demibold Roman.ttf'),
+    (windows_fonts/'LSANS.TTF',windows_fonts/'LSANSD.TTF'),
+    (font_root/'DejaVuSans.ttf',font_root/'DejaVuSans-Bold.ttf'),
+]
+regular,bold=next(pair for pair in font_pairs if all(p.is_file() for p in pair))
+for name,path in [(FONT,regular),(BOLD,bold)]:pdfmetrics.registerFont(TTFont(name,str(path)))
+
+FIGURE_EXHIBITS = {
+    'f1_bars':'earnings-early-adulthood', 'f1_observed':'earnings-early-adulthood',
+    'f1_surrogate':'earnings-early-adulthood', 'f2':'earnings-distribution',
+    'f3_attendance':'college-attendance-completion', 'f3_completion':'college-attendance-completion',
+    'a2':'appendix-college-by-age-placeholder', 'a3_networks':'appendix-cmo-leave-one-out',
+    'a3_baseline':'appendix-cmo-leave-one-out', 'a4_observed':'appendix-earnings-by-age-firm-chars',
+    'a4_surrogate':'appendix-earnings-by-age-firm-chars',
+}
+
+
+def label(c, x, y, text, size=22, align='center', bold=False, color=black):
+    c.setFillColor(color); c.setFont(BOLD if bold else FONT,size)
+    method = {'center':c.drawCentredString,'left':c.drawString,'right':c.drawRightString}[align]
+    method(x,y,text)
+
+
+def vertical_title(c, x, y, text):
+    c.saveState(); c.translate(x,y); c.rotate(90)
+    lines=text.split('\n')
+    for i,line in enumerate(lines):label(c,0,(len(lines)-1)*14-i*28,line,size=26)
+    c.restoreState()
+
+
+def axes(c, left, bottom, right, top, y, ticks, formatter):
+    c.setStrokeColor(black); c.setLineWidth(1.067); c.setDash([])
+    c.line(left,bottom,right,bottom); c.line(left,bottom,left,top)
+    for tick in ticks:
+        c.line(left-8,y(tick),left,y(tick))
+        label(c,left-13,y(tick)-8,formatter(tick),align='right')
+
+
+def make_figures(data, figures, folder):
+    folder.mkdir(parents=True, exist_ok=True)
+    audit = {}
+    for name, series, filename, ylabel in [
+        ('f1', [('f1_observed', BLUE, 'Observed Earnings'), ('f1_surrogate', ORANGE, 'Surrogate Index Prediction')], 'earnings_by_age.pdf', 'Local Average Treatment Effect on\nEarnings'),
+        ('a2', [('a2', BLUE, None)], 'college_attendance_by_age.pdf', 'Local Average Treatment Effect on\nCollege Attendance'),
+        ('a4', [('a4_observed', BLUE, 'Observed Earnings'), ('a4_surrogate', ORANGE, 'Firm-Characteristics Surrogate')], 'firm_surrogate_by_age.pdf', 'Local Average Treatment Effect on\nEarnings'),
+    ]:
+        if not all(key in figures for key, _, _ in series):continue
+        prepared=[]
+        for key,color,legend in series:
+            points=[]
+            for b in figures[key]:
+                coef=data.value(dict(b,field='coef'));se=data.value(dict(b,field='se'))
+                if coef is None or se is None:
+                    raise ValueError(f'{key}, age {b["age"]}: supply the coefficient and standard error in the mapped workbook.')
+                points.append((b['age'],coef,se))
+            if not points or len({p[0] for p in points}) != len(points):
+                raise ValueError(f'{key}: expected a nonempty series with unique ages in the figure mapping')
+            points.sort()
+            audit[key]=[{'age':age,'coef':e,'se':se,**data.used[b['record']]}
+                        for (age,e,se),b in zip(points,sorted(figures[key],key=lambda b:b['age']))]
+            prepared.append((points,color,legend))
+        line_figure(prepared,folder/filename,ylabel,dollars=name!='a2')
+    for key,filename,dollars in [
+        ('f1_bars','earnings_age25.pdf',True),
+        ('f2','earnings_distribution.pdf',False),
+        ('f3_attendance','college_attendance.pdf',False),
+        ('f3_completion','college_completion.pdf',False),
+    ]:
+        if key in figures:audit[key]=paired_bars(data,figures[key],folder/filename,dollars)
+    if 'a3_networks' in figures:
+        audit['a3']=leave_one_out(data,figures['a3_networks'],figures['a3_baseline'][0],folder/'network_leave_one_out.pdf')
+    return audit
+
+
+def line_figure(prepared,path,ylabel,dollars=True):
+    ages=sorted({a for points,_,_ in prepared for a,_,_ in points})
+    low=min(0,*(e-1.96*se for points,_,_ in prepared for _,e,se in points))
+    high=max(0,*(e+1.96*se for points,_,_ in prepared for _,e,se in points))
+    span=max(high-low,1);lo,hi=low-.05*span,high+.05*span
+    width,height=1296,648
+    left,bottom,right,top=159,74,1285,637
+    x=lambda age:left+(ages.index(age)+.6)*(right-left)/(len(ages)-1+1.2)
+    y=lambda val:bottom+(val-lo)*(top-bottom)/(hi-lo)
+    c=canvas.Canvas(str(path),pagesize=(width,height),invariant=1)
+    axes(c,left,bottom,right,top,y,extended_breaks(lo,hi),
+         (lambda v:f'${v/1000:.1f}K') if dollars else (lambda v:f'{v:.1f}%'))
+    for age in ages:
+        c.line(x(age),bottom-8,x(age),bottom);label(c,x(age),bottom-29,str(age))
+    label(c,(left+right)/2,16,'Age')
+    vertical_title(c,44,(bottom+top)/2,ylabel)
+    c.setStrokeColor(HexColor('#808080'));c.setLineWidth(1.067);c.setDash([4.27,4.27]);c.line(left,y(0),right,y(0));c.setDash([])
+    for points,color,_ in prepared:
+        c.setStrokeColor(color);c.setLineCap(0)
+        for sign in [-1,1,0]:
+            # R's dotted linetype uses one unit on / three off, scaled by width.
+            c.setLineWidth(3.201 if sign else 2.134)
+            c.setDash([3.201,9.603] if sign else [])
+            p=c.beginPath()
+            for i,(age,e,se) in enumerate(points):
+                (p.moveTo if i==0 else p.lineTo)(x(age),y(e+sign*1.96*se))
+            c.drawPath(p)
+        c.setFillColor(color)
+        for age,e,se in points:c.circle(x(age),y(e),6,stroke=0,fill=1)
+    legends=[(color,text) for _,color,text in prepared if text]
+    if legends:
+        legend_width=max(pdfmetrics.stringWidth(text,FONT,22) for _,text in legends)+30
+        lx=right-.02*(right-left)-legend_width
+        for i,(color,text) in enumerate(legends):
+            ly=bottom+.02*(top-bottom)+10+(len(legends)-1-i)*29
+            c.setStrokeColor(color);c.setFillColor(color);c.setDash([]);c.setLineWidth(2.134)
+            c.line(lx,ly+7,lx+17,ly+7);c.circle(lx+8.5,ly+7,6,stroke=0,fill=1)
+            label(c,lx+23,ly,text,align='left')
+    c.save()
+
+
+def paired_bars(data,bindings,path,dollars=False):
+    """Match create_barplot_panel: move the mean label, never the CI center."""
+    prepared=[]
+    for b in bindings:
+        row={field:data.value(dict(b,field=field)) for field in ['coef','se','ccm']}
+        if any(v is None for v in row.values()):
+            raise ValueError(f'{b["record"]}: the bar figure requires coef, se, and ccm.')
+        row.update(label=b['label'],treated=row['ccm']+row['coef'],
+                   percent_gain=100*row['coef']/row['ccm'] if row['ccm'] else None,
+                   record=b['record'],source=data.used[b['record']])
+        prepared.append(row)
+    count=len(prepared);width,height=(1296/2.1,648) if dollars else (1296,648)
+    gap=.025 if count==4 else .1
+    panel_width=width/(count+(count-1)*gap)
+    c=canvas.Canvas(str(path),pagesize=(width,height),invariant=1)
+    for i,row in enumerate(prepared):
+        origin=i*panel_width*(1+gap)
+        left=origin+11+61.2+(72 if dollars else 0)
+        right=origin+panel_width-11;bottom=86.6;top=637 if dollars else 613.2
+        control,treated,se=row['ccm'],row['treated'],row['se']
+        ci_lo,ci_hi=treated-1.96*se,treated+1.96*se
+        padding=.02*max(abs(v) for v in [control,treated,ci_lo,ci_hi])
+        lo=min(.75*min(control,treated),ci_lo-padding)
+        upper=max(control,ci_hi)
+        # Same fractional headroom and expansion as the R source. No zero floor.
+        span=max((upper-lo)/(1-.05-.12),1e-6)
+        hi=lo+1.05*span
+        y=lambda v:bottom+(v-lo)*(top-bottom)/(hi-lo)
+        x=lambda v:left+(v-.4)*(right-left)/(2.5 if count==4 else 2.2)
+        # The R labels round ticks to whole $K / whole percent, even when the
+        # extended-break locations include halves. This also preserves its fixed
+        # y-axis label width (long decimal labels would extend off the canvas).
+        fmt=(lambda v:f'${v/1000:,.0f}K') if dollars else (lambda v:f'{v:.0f}%')
+        axes(c,left,bottom,right,top,y,extended_breaks(lo,hi),fmt)
+        # coord_cartesian crops bars at the lower axis, including negative values.
+        c.saveState();clip=c.beginPath();clip.rect(left,bottom,right-left,top-bottom);c.clipPath(clip,stroke=0)
+        for pos,value,color in [(1,control,ORANGE),(2,treated,BLUE)]:
+            c.setFillColor(color)
+            c.rect(x(pos-.35),min(y(0),y(value)),x(pos+.35)-x(pos-.35),abs(y(value)-y(0)),stroke=0,fill=1)
+        c.restoreState()
+        c.setStrokeColor(black);c.setLineWidth(1.067);c.line(left,bottom,right,bottom)
+        c.setStrokeColor(black);c.setLineWidth(2.134)
+        c.line(x(2),y(ci_lo),x(2),y(ci_hi))
+        for end in [ci_lo,ci_hi]:c.line(x(1.95),y(end),x(2.05),y(end))
+        mean=lambda v:f'${v/1000:.1f}K' if dollars else f'{v:.1f}%'
+        label(c,x(1),y(control)+5,mean(control))
+        label(c,x(2.05),y(treated)+5,mean(treated),align='left')
+        bracket=upper+.05*span
+        # ggpubr tip.length is a fraction of its data range, before coord zoom.
+        tip=.03*(bracket-min(ci_lo,control))
+        c.setLineWidth(2.134)
+        c.lines([(x(1),y(bracket-tip),x(1),y(bracket)),
+                 (x(1),y(bracket),x(2),y(bracket)),
+                 (x(2),y(bracket),x(2),y(bracket-tip))])
+        if dollars:
+            raw=f'{"+" if row["coef"]>=0 else "−"}${abs(row["coef"])/1000:.1f}K (s.e.=${se/1000:.1f}K)'
+            label(c,x(1.5),y(bracket)+7,raw)
+            if row['percent_gain'] is not None:
+                label(c,x(1.5),y(bracket+.055*span)+5,f'{row["percent_gain"]:+.1f}%',size=26)
+            else:label(c,x(1.5),y(bracket+.055*span)+5,'—',size=26)
+            vertical_title(c,29,(bottom+top)/2,'Mean Wage Earnings at Age 25')
+            names=['Control\nComplier Mean','Treated Mean\n(CCM + LATE)']
+        else:
+            label(c,x(1.5),y(bracket)+36,f'{row["coef"]:+.1f}pp')
+            label(c,x(1.5),y(bracket)+7,f'(s.e.={se:.1f}pp)')
+            label(c,(left+right)/2,height-27,row['label'],bold=True)
+            names=['Control\nComplier\nMean','Treated\nMean']
+        c.setStrokeColor(black);c.setLineWidth(1.067)
+        for pos,text in enumerate(names,1):
+            c.line(x(pos),bottom-8,x(pos),bottom)
+            for j,line in enumerate(text.split('\n')):label(c,x(pos),bottom-29-j*23.8,line)
+        row['layout']={'control_x':x(1),'treated_x':x(2),'ci_x':x(2),
+                       'bracket_left':x(1),'bracket_right':x(2),'axis_min':lo,'axis_max':hi}
+    c.save()
+    return prepared
+
+
+def leave_one_out(data,bindings,baseline,path):
+    if len(bindings)<2:raise ValueError('Figure A3 requires at least two networks in input_mapping().')
+    rows=[]
+    for b in bindings:
+        coef=data.value(dict(b,field='coef'));se=data.value(dict(b,field='se'))
+        if coef is None or se is None or se<=0:
+            raise ValueError(f'{b["record"]}: Figure A3 requires a coefficient and positive standard error.')
+        rows.append(dict(label=b['label'],coef=coef,se=se,source=data.used[b['record']]))
+    if len({tuple(sorted(row['source']['where'].items())) for row in rows}) != len(rows):
+        raise ValueError('Figure A3 network samples must identify distinct rows.')
+    pooled=data.value(dict(baseline,field='coef'))
+    if pooled is None:raise ValueError('Figure A3 requires the pooled baseline earnings coefficient.')
+    estimates=[]
+    for i,row in enumerate(rows):
+        others=[r for j,r in enumerate(rows) if j!=i]
+        precision=sum(1/r['se']**2 for r in others)
+        estimates.append({'label':row['label'],'coef':sum(r['coef']/r['se']**2 for r in others)/precision,
+                          'se':math.sqrt(1/precision)})
+    # The paper sorts the bars and suppresses network names, CIs and reference line.
+    ordered=sorted(estimates,key=lambda r:r['coef'])
+    low=min(0,*(r['coef'] for r in estimates));high=max(4.5,*(1.15*r['coef'] for r in estimates))
+    span=high-low;lo,hi=low-.05*span,high+.05*span
+    left,bottom,right,top=159,89,1285,637
+    y=lambda v:bottom+(v-lo)*(top-bottom)/(hi-lo)
+    x=lambda pos:left+(pos-.4)*(right-left)/(len(ordered)+.2)
+    c=canvas.Canvas(str(path),pagesize=(1296,648),invariant=1)
+    axes(c,left,bottom,right,top,y,extended_breaks(lo,hi),lambda v:f'${v:.1f}K')
+    for i,row in enumerate(ordered,1):
+        c.setFillColor(GRAY);c.rect(x(i-.35),min(y(0),y(row['coef'])),x(i+.35)-x(i-.35),abs(y(row['coef'])-y(0)),stroke=0,fill=1)
+    label(c,right-20,top-32,f'Baseline Pooled Estimate: ${pooled:.1f}K',align='right',color=BLUE)
+    label(c,(left+right)/2,23,'Each Bar Plots LATE Leaving Out One Charter Network')
+    vertical_title(c,44,(bottom+top)/2,'Treatment Effect on Mean Wage Earnings\nat Age 25')
+    c.save()
+    return {'networks':rows,'leave_one_out':estimates,'baseline':pooled,
+            'baseline_source':data.used[baseline['record']]}

ltc_paper_scripts/latex_requirements.txt

@@ -0,0 +1,4 @@
+openpyxl>=3.1
+reportlab>=4,<5; python_version < '3.10'
+reportlab>=4; python_version >= '3.10'
+pypdf>=4

ltc_paper_scripts/latex_source/outputs/tables/output/age25_earnings_table.tex

@@ -0,0 +1,38 @@
+\setlength{\tabcolsep}{5pt}
+\begin{tabular}{L{3.5cm} *{5}{C{2cm}}}
+  \toprule
+  & \shortstack{First\\Stage}
+  & \shortstack{Earnings\\ITT}
+  & \shortstack{Earnings\\LATE}
+  & \shortstack{Earnings\\LATE with\\Controls}
+  & AGI \\
+  & (1) & (2) & (3) & (4) & (5) \\
+  \midrule
+  Offered vs. Control
+    & @@age25_earnings_table.0.0@@
+    & @@age25_earnings_table.2.0@@
+    & & & \\
+    & @@age25_earnings_table.1.0@@
+    & @@age25_earnings_table.3.0@@
+    & & & \\
+  Control Mean
+    & @@age25_earnings_table.0.1@@
+    & @@age25_earnings_table.2.1@@
+    & & & \\
+  \addlinespace
+  Enrolled vs. Control
+    & &
+    & @@age25_earnings_table.4.0@@
+    & @@age25_earnings_table.6.0@@
+    & @@age25_earnings_table.8.0@@ \\
+    & &
+    & @@age25_earnings_table.5.0@@
+    & @@age25_earnings_table.7.0@@
+    & @@age25_earnings_table.9.0@@ \\
+  Control Complier Mean
+    & &
+    & @@age25_earnings_table.4.1@@
+    & @@age25_earnings_table.6.1@@
+    & @@age25_earnings_table.8.1@@ \\
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/balance_table.tex

@@ -0,0 +1,54 @@
+% Workbook-driven table; source variable mappings are in build_latex.py.
+% Edit labels, row order, panels, and column structure here.
+
+
+\begin{tabular}{L{6.4cm} *{4}{C{2.2cm}}}
+  \toprule
+  & \multicolumn{2}{c}{Age 20+ Sample}
+  & \multicolumn{2}{c}{Age 25+ Sample} \\
+  \cmidrule(lr){2-3}\cmidrule(lr){4-5}
+  & Control Group Mean & Offered vs. Control & Control Group Mean & Offered vs. Control
+  \\
+  & (1) & (2) & (3) & (4) \\
+  \midrule
+
+
+  \addlinespace
+  \textit{A. Tax-Data Linkage} & & & & \\
+  \hspace{0.1cm}Linked to tax data (\%)  & @@balance_table.14.0@@  & @@balance_table.14.1@@  & @@balance_table.14.2@@  & @@balance_table.14.3@@ \\
+     &   & @@balance_table.15.1@@  &   & @@balance_table.15.3@@ \\
+
+  \addlinespace
+
+  \textit{B. Demographic Characteristics} & & & & \\
+  \hspace{0.1cm}Black (\%)
+     & @@balance_table.0.0@@  & @@balance_table.0.1@@  & @@balance_table.0.2@@  & @@balance_table.0.3@@ \\
+     &   & @@balance_table.1.1@@  &   & @@balance_table.1.3@@ \\
+  \hspace{0.1cm}Hispanic (\%)
+     & @@balance_table.2.0@@  & @@balance_table.2.1@@  & @@balance_table.2.2@@  & @@balance_table.2.3@@ \\
+     &   & @@balance_table.3.1@@  &   & @@balance_table.3.3@@ \\
+  \hspace{0.1cm}White (\%)
+     & @@balance_table.4.0@@  & @@balance_table.4.1@@  & @@balance_table.4.2@@  & @@balance_table.4.3@@ \\
+     &   & @@balance_table.5.1@@  &   & @@balance_table.5.3@@ \\
+  \hspace{0.1cm}Female (\%)
+     & @@balance_table.6.0@@  & @@balance_table.6.1@@  & @@balance_table.6.2@@  & @@balance_table.6.3@@ \\
+     &   & @@balance_table.7.1@@  &   & @@balance_table.7.3@@ \\
+
+  \addlinespace
+  \textit{C. Family and Neighborhood Characteristics} & & & & \\
+  \hspace{0.1cm}Parent income rank (percentile)
+     & @@balance_table.8.0@@  & @@balance_table.8.1@@  & @@balance_table.8.2@@  & @@balance_table.8.3@@ \\
+     &   & @@balance_table.9.1@@  &   & @@balance_table.9.3@@ \\
+  \hspace{0.1cm}Median tract household income (\$)
+     & @@balance_table.10.0@@  & @@balance_table.10.1@@  & @@balance_table.10.2@@  & @@balance_table.10.3@@ \\
+     &   & @@balance_table.11.1@@  &   & @@balance_table.11.3@@ \\
+  \hspace{0.1cm}Predicted age 25 earnings (\$)
+     & @@balance_table.12.0@@  & @@balance_table.12.1@@  & @@balance_table.12.2@@  & @@balance_table.12.3@@ \\
+     &   & @@balance_table.13.1@@  &   & @@balance_table.13.3@@ \\
+
+  \addlinespace[4pt]
+  Number of applicants
+     &   & @@balance_table.16.1@@  &   & @@balance_table.16.3@@ \\
+
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/college_data_comparison_table.tex

@@ -0,0 +1,20 @@
+% Synthetic review overlay; restore originals from data/codex_synthetic/review_source.
+% The Form 1098-T benchmark is shared by both NSC comparisons.
+
+
+\setlength{\tabcolsep}{8pt}
+\begin{tabular}{L{4cm} *{3}{C{3cm}}}
+  \toprule
+  & \shortstack{Form 1098-T\\Benchmark}
+  & \shortstack{NSC\\Original Sample}
+  & \shortstack{NSC\\Aligned Sample} \\
+  & (1) & (2) & (3) \\
+  \midrule
+  Enrolled vs. Control
+     & @@college_data_comparison_table.0.0@@  & @@college_data_comparison_table.0.1@@  & @@college_data_comparison_table.0.2@@ \\
+     & @@college_data_comparison_table.1.0@@  & @@college_data_comparison_table.1.1@@  & @@college_data_comparison_table.1.2@@ \\
+  \addlinespace
+  Control Complier Mean
+     & @@college_data_comparison_table.2.0@@  & @@college_data_comparison_table.2.1@@  & @@college_data_comparison_table.2.2@@ \\
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/college_enrollment_completion_table.tex

@@ -0,0 +1,31 @@
+% Synthetic review overlay; restore originals from data/codex_synthetic/review_source.
+% Table layout backed by outputs/tables/input/college_enrollment_completion_table_input.csv.
+
+
+\setlength{\tabcolsep}{2.5pt}
+\begin{tabular}{L{3.2cm} *{8}{C{1.45cm}}}
+  \toprule
+  & \multicolumn{4}{c}{College Enrollment by Age 20}
+  & \multicolumn{4}{c}{College Completion by Age 24} \\
+  \cmidrule(lr){2-5}\cmidrule(lr){6-9}
+  & Any
+  & \shortstack{Two-\\ Year}
+  & \shortstack{Four-\\ Year}
+  & \shortstack{Highly\\ Selective}
+  & Any
+  & \shortstack{Associate's\\ Degree}
+  & \shortstack{Bachelor's\\ Degree}
+  & \shortstack{Highly\\ Selective} \\
+  & (1) & (2) & (3) & (4) & (5) & (6) & (7) & (8) \\
+  \midrule
+
+  Enrolled vs. Control
+     & @@college_enrollment_completion_table.0.0@@  & @@college_enrollment_completion_table.0.1@@  & @@college_enrollment_completion_table.0.2@@  & @@college_enrollment_completion_table.0.3@@  & @@college_enrollment_completion_table.0.4@@  & @@college_enrollment_completion_table.0.5@@  & @@college_enrollment_completion_table.0.6@@  & @@college_enrollment_completion_table.0.7@@ \\
+     & @@college_enrollment_completion_table.1.0@@  & @@college_enrollment_completion_table.1.1@@  & @@college_enrollment_completion_table.1.2@@  & @@college_enrollment_completion_table.1.3@@  & @@college_enrollment_completion_table.1.4@@  & @@college_enrollment_completion_table.1.5@@  & @@college_enrollment_completion_table.1.6@@  & @@college_enrollment_completion_table.1.7@@ \\
+
+  \addlinespace
+  Control Complier Mean
+     & @@college_enrollment_completion_table.2.0@@  & @@college_enrollment_completion_table.2.1@@  & @@college_enrollment_completion_table.2.2@@  & @@college_enrollment_completion_table.2.3@@  & @@college_enrollment_completion_table.2.4@@  & @@college_enrollment_completion_table.2.5@@  & @@college_enrollment_completion_table.2.6@@  & @@college_enrollment_completion_table.2.7@@ \\
+
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/college_heterogeneity_table.tex

@@ -0,0 +1,36 @@
+% Synthetic review overlay; restore originals from data/codex_synthetic/review_source.
+% Table layout backed by outputs/tables/input/college_heterogeneity_table_input.csv.
+
+
+\setlength{\tabcolsep}{2pt}
+\begin{tabular}{L{3cm} *{10}{C{1.15cm}}}
+  \toprule
+  & \multicolumn{2}{c}{Parent Income}
+  & \multicolumn{2}{c}{Neighborhood Income}
+  & \multicolumn{2}{c}{Sex}
+  & \multicolumn{2}{c}{Race/Ethnicity}
+  & \multicolumn{2}{c}{Entry Grade} \\
+  \cmidrule(lr){2-3}\cmidrule(lr){4-5}\cmidrule(lr){6-7}\cmidrule(lr){8-9}\cmidrule(lr){10-11}
+  & Low & High
+  & Low & High
+  & Boys & Girls
+  & Black & Hispanic
+  & \shortstack{High\\ School} & \shortstack{Middle\\ School} \\
+  & (1) & (2) & (3) & (4) & (5) & (6) & (7) & (8) & (9) & (10) \\
+  \midrule
+
+  \multicolumn{11}{l}{\textit{Panel A. College Attendance by Age 20}} \\
+  \addlinespace[2pt]
+  Enrolled vs. Control
+     & @@college_heterogeneity_table.0.0@@  & @@college_heterogeneity_table.0.1@@  & @@college_heterogeneity_table.0.2@@  & @@college_heterogeneity_table.0.3@@  & @@college_heterogeneity_table.0.4@@  & @@college_heterogeneity_table.0.5@@  & @@college_heterogeneity_table.0.6@@  & @@college_heterogeneity_table.0.7@@  & @@college_heterogeneity_table.0.8@@  & @@college_heterogeneity_table.0.9@@ \\
+     & @@college_heterogeneity_table.1.0@@  & @@college_heterogeneity_table.1.1@@  & @@college_heterogeneity_table.1.2@@  & @@college_heterogeneity_table.1.3@@  & @@college_heterogeneity_table.1.4@@  & @@college_heterogeneity_table.1.5@@  & @@college_heterogeneity_table.1.6@@  & @@college_heterogeneity_table.1.7@@  & @@college_heterogeneity_table.1.8@@  & @@college_heterogeneity_table.1.9@@ \\
+
+  \addlinespace
+  \multicolumn{11}{l}{\textit{Panel B. Four-Year Degree Completion}} \\
+  \addlinespace[2pt]
+  Enrolled vs. Control
+     & @@college_heterogeneity_table.2.0@@  & @@college_heterogeneity_table.2.1@@  & @@college_heterogeneity_table.2.2@@  & @@college_heterogeneity_table.2.3@@  & @@college_heterogeneity_table.2.4@@  & @@college_heterogeneity_table.2.5@@  & @@college_heterogeneity_table.2.6@@  & @@college_heterogeneity_table.2.7@@  & @@college_heterogeneity_table.2.8@@  & @@college_heterogeneity_table.2.9@@ \\
+     & @@college_heterogeneity_table.3.0@@  & @@college_heterogeneity_table.3.1@@  & @@college_heterogeneity_table.3.2@@  & @@college_heterogeneity_table.3.3@@  & @@college_heterogeneity_table.3.4@@  & @@college_heterogeneity_table.3.5@@  & @@college_heterogeneity_table.3.6@@  & @@college_heterogeneity_table.3.7@@  & @@college_heterogeneity_table.3.8@@  & @@college_heterogeneity_table.3.9@@ \\
+
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/college_mediation_additional_specs_table.tex

@@ -0,0 +1,30 @@
+% Synthetic review overlay; restore originals from data/codex_synthetic/review_source.
+% Placeholder layout: all estimates await disclosure.
+\setlength{\tabcolsep}{3pt}
+\begin{tabular}{L{3cm} *{8}{C{1.45cm}}}
+  \toprule
+  & \multicolumn{2}{c}{Earnings Baseline Spec.}
+  & \multicolumn{3}{c}{Actual Completion: 25+ Sample}
+  & \multicolumn{3}{c}{Predicted Completion: 25+ Sample} \\
+  \cmidrule(lr){2-3}\cmidrule(lr){4-6}\cmidrule(lr){7-9}
+  & \shortstack{Earnings\\Ever\\College}
+  & \shortstack{Earnings\\Never\\College}
+  & \shortstack{Any College\\Completion}
+  & \shortstack{BA\\Completion}
+  & \shortstack{Two-Year\\Completion}
+  & \shortstack{Any College\\Completion}
+  & \shortstack{Four-Year\\Completion}
+  & \shortstack{Two-Year\\Completion} \\
+  & (1) & (2) & (3) & (4) & (5) & (6) & (7) & (8) \\
+  \midrule
+
+  Enrolled vs. Control
+     & @@college_mediation_additional_specs_table.0.0@@  & @@college_mediation_additional_specs_table.0.1@@  & @@college_mediation_additional_specs_table.0.2@@  & @@college_mediation_additional_specs_table.0.3@@  & @@college_mediation_additional_specs_table.0.4@@  & @@college_mediation_additional_specs_table.0.5@@  & @@college_mediation_additional_specs_table.0.6@@  & @@college_mediation_additional_specs_table.0.7@@ \\
+     & @@college_mediation_additional_specs_table.1.0@@  & @@college_mediation_additional_specs_table.1.1@@  & @@college_mediation_additional_specs_table.1.2@@  & @@college_mediation_additional_specs_table.1.3@@  & @@college_mediation_additional_specs_table.1.4@@  & @@college_mediation_additional_specs_table.1.5@@  & @@college_mediation_additional_specs_table.1.6@@  & @@college_mediation_additional_specs_table.1.7@@ \\
+
+  \addlinespace
+  Control Complier Mean
+     & @@college_mediation_additional_specs_table.2.0@@  & @@college_mediation_additional_specs_table.2.1@@  & @@college_mediation_additional_specs_table.2.2@@  & @@college_mediation_additional_specs_table.2.3@@  & @@college_mediation_additional_specs_table.2.4@@  & @@college_mediation_additional_specs_table.2.5@@  & @@college_mediation_additional_specs_table.2.6@@  & @@college_mediation_additional_specs_table.2.7@@ \\
+
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/college_test_score_mediation_table.tex

@@ -0,0 +1,18 @@
+\setlength{\tabcolsep}{4pt}
+\begin{tabular}{L{4.7cm} C{1.8cm} C{2.5cm} C{2.5cm} C{2.7cm}}
+  \toprule
+  \textit{A. College} & Effect (pp) & Earnings Return (\% per pp) & Implied Earnings Gain (\%) & Observed Gain Explained (\%) \\
+  \midrule
+  BA completion  & @@college_test_score_mediation_table.0.0@@ & @@college_test_score_mediation_table.0.1@@ & @@college_test_score_mediation_table.0.2@@ & @@college_test_score_mediation_table.0.3@@ \\ \addlinespace
+  Selective BA completion  & @@college_test_score_mediation_table.1.0@@ & @@college_test_score_mediation_table.1.1@@ & @@college_test_score_mediation_table.1.2@@ & @@college_test_score_mediation_table.1.3@@ \\ \addlinespace
+  Highly selective BA completion  & @@college_test_score_mediation_table.2.0@@ & @@college_test_score_mediation_table.2.1@@ & @@college_test_score_mediation_table.2.2@@ & @@college_test_score_mediation_table.2.3@@ \\ \addlinespace
+  \cmidrule(lr){4-5}
+  Total college contribution  &  &  & @@college_test_score_mediation_table.3.2@@ & @@college_test_score_mediation_table.3.3@@ \\
+
+  \addlinespace[10pt]
+  \textit{B. Test Scores} & Effect (SD/year) & Earnings Return (\% per SD) & Implied Earnings Gain (\%) & Observed Gain Explained (\%) \\
+  \midrule
+  \hspace{0.1cm}Annual math gain
+     & @@college_test_score_mediation_table.4.0@@ & @@college_test_score_mediation_table.4.1@@ & @@college_test_score_mediation_table.4.2@@ & @@college_test_score_mediation_table.4.3@@ \\
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/earnings_heterogeneity_table.tex

@@ -0,0 +1,54 @@
+% Workbook-driven table; source variable mappings are in build_latex.py.
+\setlength{\tabcolsep}{2pt}
+\begin{tabular}{L{3cm} *{10}{C{1.15cm}}}
+  \toprule
+  & \multicolumn{2}{c}{Parent Income}
+  & \multicolumn{2}{c}{Neighborhood Income}
+  & \multicolumn{2}{c}{Sex}
+  & \multicolumn{2}{c}{Race/Ethnicity}
+  & \multicolumn{2}{c}{Entry Grade} \\
+  \cmidrule(lr){2-3}\cmidrule(lr){4-5}\cmidrule(lr){6-7}\cmidrule(lr){8-9}\cmidrule(lr){10-11}
+  & Low & High
+  & Low & High
+  & Boys & Girls
+  & Black & Hispanic
+  & \shortstack{High\\ School} & \shortstack{Middle\\ School} \\
+  & (1) & (2) & (3) & (4) & (5) & (6) & (7) & (8) & (9) & (10) \\
+  \midrule
+  \multicolumn{11}{l}{\textit{Panel A. Earnings at Age 25}} \\
+  \addlinespace[2pt]
+
+  Enrolled vs. Control
+     & @@earnings_heterogeneity_table.0.0@@  & @@earnings_heterogeneity_table.0.1@@  & @@earnings_heterogeneity_table.0.2@@  & @@earnings_heterogeneity_table.0.3@@  & @@earnings_heterogeneity_table.0.4@@  & @@earnings_heterogeneity_table.0.5@@  & @@earnings_heterogeneity_table.0.6@@  & @@earnings_heterogeneity_table.0.7@@  & @@earnings_heterogeneity_table.0.8@@  & @@earnings_heterogeneity_table.0.9@@ \\
+     & @@earnings_heterogeneity_table.1.0@@  & @@earnings_heterogeneity_table.1.1@@  & @@earnings_heterogeneity_table.1.2@@  & @@earnings_heterogeneity_table.1.3@@  & @@earnings_heterogeneity_table.1.4@@  & @@earnings_heterogeneity_table.1.5@@  & @@earnings_heterogeneity_table.1.6@@  & @@earnings_heterogeneity_table.1.7@@  & @@earnings_heterogeneity_table.1.8@@  & @@earnings_heterogeneity_table.1.9@@ \\
+
+Control Complier Mean
+     & @@earnings_heterogeneity_table.2.0@@  & @@earnings_heterogeneity_table.2.1@@  & @@earnings_heterogeneity_table.2.2@@  & @@earnings_heterogeneity_table.2.3@@  & @@earnings_heterogeneity_table.2.4@@  & @@earnings_heterogeneity_table.2.5@@  & @@earnings_heterogeneity_table.2.6@@  & @@earnings_heterogeneity_table.2.7@@  & @@earnings_heterogeneity_table.2.8@@  & @@earnings_heterogeneity_table.2.9@@ \\
+
+  \addlinespace[8pt]
+  \multicolumn{11}{l}{\textit{Panel B. Surrogate Earnings at Age 31}} \\
+  \addlinespace[2pt]
+  Enrolled vs. Control & @@earnings_heterogeneity_table.3.0@@ & @@earnings_heterogeneity_table.3.1@@ & @@earnings_heterogeneity_table.3.2@@ & @@earnings_heterogeneity_table.3.3@@ & @@earnings_heterogeneity_table.3.4@@ & @@earnings_heterogeneity_table.3.5@@ & @@earnings_heterogeneity_table.3.6@@ & @@earnings_heterogeneity_table.3.7@@ & @@earnings_heterogeneity_table.3.8@@ & @@earnings_heterogeneity_table.3.9@@ \\
+   & @@earnings_heterogeneity_table.4.0@@ & @@earnings_heterogeneity_table.4.1@@ & @@earnings_heterogeneity_table.4.2@@ & @@earnings_heterogeneity_table.4.3@@ & @@earnings_heterogeneity_table.4.4@@ & @@earnings_heterogeneity_table.4.5@@ & @@earnings_heterogeneity_table.4.6@@ & @@earnings_heterogeneity_table.4.7@@ & @@earnings_heterogeneity_table.4.8@@ & @@earnings_heterogeneity_table.4.9@@ \\
+  Control Complier Mean & @@earnings_heterogeneity_table.5.0@@ & @@earnings_heterogeneity_table.5.1@@ & @@earnings_heterogeneity_table.5.2@@ & @@earnings_heterogeneity_table.5.3@@ & @@earnings_heterogeneity_table.5.4@@ & @@earnings_heterogeneity_table.5.5@@ & @@earnings_heterogeneity_table.5.6@@ & @@earnings_heterogeneity_table.5.7@@ & @@earnings_heterogeneity_table.5.8@@ & @@earnings_heterogeneity_table.5.9@@ \\
+  \addlinespace
+  \multicolumn{11}{l}{\textit{Panel C. College Attendance by Age 20}} \\
+  \addlinespace[2pt]
+  Enrolled vs. Control
+     & @@earnings_heterogeneity_table.6.0@@  & @@earnings_heterogeneity_table.6.1@@  & @@earnings_heterogeneity_table.6.2@@  & @@earnings_heterogeneity_table.6.3@@  & @@earnings_heterogeneity_table.6.4@@  & @@earnings_heterogeneity_table.6.5@@  & @@earnings_heterogeneity_table.6.6@@  & @@earnings_heterogeneity_table.6.7@@  & @@earnings_heterogeneity_table.6.8@@  & @@earnings_heterogeneity_table.6.9@@ \\
+     & @@earnings_heterogeneity_table.7.0@@  & @@earnings_heterogeneity_table.7.1@@  & @@earnings_heterogeneity_table.7.2@@  & @@earnings_heterogeneity_table.7.3@@  & @@earnings_heterogeneity_table.7.4@@  & @@earnings_heterogeneity_table.7.5@@  & @@earnings_heterogeneity_table.7.6@@  & @@earnings_heterogeneity_table.7.7@@  & @@earnings_heterogeneity_table.7.8@@  & @@earnings_heterogeneity_table.7.9@@ \\
+
+
+  \addlinespace[2pt]
+  Control Complier Mean
+     & @@earnings_heterogeneity_table.8.0@@  & @@earnings_heterogeneity_table.8.1@@  & @@earnings_heterogeneity_table.8.2@@  & @@earnings_heterogeneity_table.8.3@@  & @@earnings_heterogeneity_table.8.4@@  & @@earnings_heterogeneity_table.8.5@@  & @@earnings_heterogeneity_table.8.6@@  & @@earnings_heterogeneity_table.8.7@@  & @@earnings_heterogeneity_table.8.8@@  & @@earnings_heterogeneity_table.8.9@@ \\
+
+  \addlinespace[8pt]
+  \multicolumn{11}{l}{\textit{Panel D. Four-Year Degree Completion by Age 24}} \\
+  \addlinespace[2pt]
+  Enrolled vs. Control & @@earnings_heterogeneity_table.9.0@@ & @@earnings_heterogeneity_table.9.1@@ & @@earnings_heterogeneity_table.9.2@@ & @@earnings_heterogeneity_table.9.3@@ & @@earnings_heterogeneity_table.9.4@@ & @@earnings_heterogeneity_table.9.5@@ & @@earnings_heterogeneity_table.9.6@@ & @@earnings_heterogeneity_table.9.7@@ & @@earnings_heterogeneity_table.9.8@@ & @@earnings_heterogeneity_table.9.9@@ \\
+   & @@earnings_heterogeneity_table.10.0@@ & @@earnings_heterogeneity_table.10.1@@ & @@earnings_heterogeneity_table.10.2@@ & @@earnings_heterogeneity_table.10.3@@ & @@earnings_heterogeneity_table.10.4@@ & @@earnings_heterogeneity_table.10.5@@ & @@earnings_heterogeneity_table.10.6@@ & @@earnings_heterogeneity_table.10.7@@ & @@earnings_heterogeneity_table.10.8@@ & @@earnings_heterogeneity_table.10.9@@ \\
+  Control Complier Mean & @@earnings_heterogeneity_table.11.0@@ & @@earnings_heterogeneity_table.11.1@@ & @@earnings_heterogeneity_table.11.2@@ & @@earnings_heterogeneity_table.11.3@@ & @@earnings_heterogeneity_table.11.4@@ & @@earnings_heterogeneity_table.11.5@@ & @@earnings_heterogeneity_table.11.6@@ & @@earnings_heterogeneity_table.11.7@@ & @@earnings_heterogeneity_table.11.8@@ & @@earnings_heterogeneity_table.11.9@@ \\
+
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/earnings_robustness_table.tex

@@ -0,0 +1,29 @@
+\setlength{\tabcolsep}{3pt}
+\begin{tabular}{L{3.3cm} *{7}{C{1.55cm}}}
+  \toprule
+  & \multicolumn{4}{C{6.83cm}}{Alternative Specifications for Earnings at 25}
+  & \multicolumn{3}{C{5.07cm}}{Baseline Specification for Other Outcomes at 25} \\
+  \cmidrule(lr){2-5}\cmidrule(lr){6-8}
+  & \shortstack{Initial Offer\\Only}
+  & Initial and Waitlist Offers
+  & Pre-Analysis Plan
+  & \shortstack{OLS with\\Tract FEs}
+  & AGI
+  & \shortstack{Spouse\\Earnings}
+  & Marriage \\
+  & (1) & (2) & (3) & (4) & (5) & (6) & (7) \\
+  \midrule
+  Enrollment effect
+    & @@earnings_robustness_table.0.0@@ & @@earnings_robustness_table.2.0@@ & @@earnings_robustness_table.4.0@@
+    & @@earnings_robustness_table.12.0@@
+    & @@earnings_robustness_table.6.0@@ & @@earnings_robustness_table.8.0@@ & @@earnings_robustness_table.10.0@@ \\
+    & @@earnings_robustness_table.1.0@@ & @@earnings_robustness_table.3.0@@ & @@earnings_robustness_table.5.0@@
+    & @@earnings_robustness_table.13.0@@
+    & @@earnings_robustness_table.7.0@@ & @@earnings_robustness_table.9.0@@ & @@earnings_robustness_table.11.0@@ \\
+  \addlinespace
+  Control Complier Mean
+    & @@earnings_robustness_table.0.1@@ & @@earnings_robustness_table.2.1@@ & @@earnings_robustness_table.4.1@@ & & @@earnings_robustness_table.6.1@@ & @@earnings_robustness_table.8.1@@ & @@earnings_robustness_table.10.1@@ \\
+  Control Mean
+    & & & & @@earnings_robustness_table.12.1@@ & & & \\
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/earnings_threshold_table.tex

@@ -0,0 +1,27 @@
+% Synthetic review overlay; restore originals from data/codex_synthetic/review_source.
+% Table layout backed by outputs/tables/input/earnings_threshold_table_input.csv.
+
+
+\setlength{\tabcolsep}{3pt}
+\begin{tabular}{L{3cm} *{7}{C{1.65cm}}}
+  \toprule
+  & Employed
+  & \multicolumn{2}{c}{Earning $>$\$20K}
+  & \multicolumn{2}{c}{Earning $>$\$50K}
+  & \multicolumn{2}{c}{Earning $>$\$100K} \\
+  \cmidrule(lr){2-2}\cmidrule(lr){3-4}\cmidrule(lr){5-6}\cmidrule(lr){7-8}
+  & Actual & Actual & Predicted
+  & Actual & Predicted & Actual & Predicted \\
+  & (1) & (2) & (3) & (4) & (5) & (6) & (7) \\
+  \midrule
+
+  Enrolled vs. Control
+     & @@earnings_threshold_table.0.0@@  & @@earnings_threshold_table.0.1@@  & @@earnings_threshold_table.0.2@@  & @@earnings_threshold_table.0.3@@  & @@earnings_threshold_table.0.4@@  & @@earnings_threshold_table.0.5@@  & @@earnings_threshold_table.0.6@@ \\
+     & @@earnings_threshold_table.1.0@@  & @@earnings_threshold_table.1.1@@  &   & @@earnings_threshold_table.1.3@@  &   & @@earnings_threshold_table.1.5@@  &  \\
+
+  \addlinespace
+  Control Complier Mean
+     & @@earnings_threshold_table.2.0@@  & @@earnings_threshold_table.2.1@@  &   & @@earnings_threshold_table.2.3@@  &   & @@earnings_threshold_table.2.5@@  &  \\
+
+  \bottomrule
+\end{tabular}

ltc_paper_scripts/latex_source/outputs/tables/output/other_mediators_table.tex

@@ -0,0 +1,31 @@
+\setlength{\tabcolsep}{3pt}
+\begin{tabular}{L{3.5cm} *{5}{C{2.1cm}}}
+  \toprule
+  & \shortstack{Peer Parent\\Income\\Percentile}
+  & \shortstack{Teen Birth\\Rate for\\Girls}
+  & \shortstack{Selective\\College BA\\Rate at\\Age-19 Firm}
+  & \shortstack{Worker Wages\\Ages 30--40\\at Age-19 Firm}
+  & \shortstack{Retention Rate\\at Age-19 Firm} \\
+  & (1) & (2) & (3) & (4) & (5) \\
+  \midrule
+
+  Enrolled vs. Control
+    & @@other_mediators_table.0.0@@
+    & @@other_mediators_table.0.1@@
+    & @@other_mediators_table.0.3@@ & @@other_mediators_table.0.4@@ & @@other_mediators_table.0.5@@ \\
+    &
+    & @@other_mediators_table.1.1@@
+    & @@other_mediators_table.1.3@@ & @@other_mediators_table.1.4@@ & @@other_mediators_table.1.5@@ \\
+
+  \addlinespace
+  Comparison Mean
+    & @@other_mediators_table.2.0@@
+    & & & & \\
+
+  Control Complier Mean
+    &
+    & @@other_mediators_table.3.1@@
+    & @@other_mediators_table.3.3@@ & @@other_mediators_table.3.4@@ & @@other_mediators_table.3.5@@ \\
+
+  \bottomrule
+\end{tabular}