Cash Flow

Abstract

Quick EDA’s to explore cash flow

Baseline Cash Flow

Income Sources

```{python}
#| label: fig-income-sources-pct
#| fig-cap: Average percentage of cash income from each source at baseline

inflow_df = (pd.DataFrame({'Category': [inflow_name_map.get(c, c) for c in inflow_pct_cols],
                           'Mean': [baseline[c].mean() for c in inflow_pct_cols]})
             .sort_values('Mean', ascending=True))
inflow_df['Category'] = pd.Categorical(inflow_df['Category'], categories=inflow_df['Category'], ordered=True)

(ggplot(inflow_df, aes(x='Category', y='Mean'))
 + geom_col(fill='steelblue')
 + geom_text(aes(label='Mean'), format_string='{:.1f}%', ha='left', nudge_y=0.5, size=8)
 + coord_flip()
 + labs(x='', y='Mean % of Total Cash Inflow', title='Income Sources at Baseline')
 + theme(figure_size=(7, 5), axis_text_y=element_text(angle=45, ha='right')))
```

Figure 1: Average percentage of cash income from each source at baseline

```{python}
#| label: fig-income-sources-amount
#| fig-cap: Average amount of cash income from each source at baseline (% × total inflow)

bl_pos = baseline[baseline['cash_inflow'] > 0]
amount_df = (pd.DataFrame({
    'Category': [inflow_name_map.get(c, c) for c in inflow_pct_cols],
    'Amount': [(bl_pos[c] / 100 * bl_pos['cash_inflow_99']).mean() for c in inflow_pct_cols],
    'N': [(bl_pos[c] > 0).sum() for c in inflow_pct_cols]
}).sort_values('Amount', ascending=True))
amount_df['Category'] = pd.Categorical(amount_df['Category'], categories=amount_df['Category'], ordered=True)
amount_df['label'] = amount_df.apply(lambda r: f"{r['Amount']:,.0f} (n={r['N']})", axis=1)

(ggplot(amount_df, aes(x='Category', y='Amount'))
 + geom_col(fill='steelblue')
 + geom_text(aes(label='label'), ha='left', nudge_y=50, size=7)
 + coord_flip()
 + labs(x='', y='Mean Amount', title='Income Sources at Baseline: Amount')
 + theme(figure_size=(7, 5), axis_text_y=element_text(angle=45, ha='right')))
```

Figure 2: Average amount of cash income from each source at baseline (% × total inflow)

Caution

915 housholds (40%) report zero cash inflow at baseline

```{python}
#| label: fig-dominant-income-source
#| fig-cap: Number of households with concentrated income (≥70% from single source)

bl_pos = baseline[baseline['cash_inflow'] > 0].copy()
bl_pos['dominant_pct'] = bl_pos[inflow_pct_cols].max(axis=1)
bl_pos['dominant_src'] = bl_pos[inflow_pct_cols].idxmax(axis=1).map(inflow_name_map)
concentrated = bl_pos[bl_pos['dominant_pct'] >= 70]
n_conc, n_tot = len(concentrated), len(bl_pos)

conc_counts = concentrated['dominant_src'].value_counts().reset_index()
conc_counts.columns = ['Source', 'N']
conc_counts = conc_counts.sort_values('N', ascending=True)
conc_counts['Source'] = pd.Categorical(conc_counts['Source'], categories=conc_counts['Source'], ordered=True)

(ggplot(conc_counts, aes(x='Source', y='N'))
 + geom_col(fill='steelblue')
 + geom_text(aes(label='N'), ha='left', nudge_y=1, size=8)
 + coord_flip()
 + labs(x='', y='Number of Households',
        title=f'Concentrated Income (≥70%): {n_conc}/ 2272 HH ({(n_conc/2272)*100:.1f}%)')
 + theme(figure_size=(7, 5), axis_text_y=element_text(angle=45, ha='right')))
```

Figure 3: Number of households with concentrated income (≥70% from single source)

Spending Categories

```{python}
#| label: fig-outflow-sources-pct
#| fig-cap: Average percentage of cash outflow to each category at baseline

outflow_df = (pd.DataFrame({'Category': [outflow_name_map.get(c, c) for c in outflow_pct_cols],
                            'Mean': [baseline[c].mean() for c in outflow_pct_cols]})
              .sort_values('Mean', ascending=True))
outflow_df['Category'] = pd.Categorical(outflow_df['Category'], categories=outflow_df['Category'], ordered=True)

(ggplot(outflow_df, aes(x='Category', y='Mean'))
 + geom_col(fill='coral')
 + geom_text(aes(label='Mean'), format_string='{:.1f}%', ha='left', nudge_y=0.5, size=8)
 + coord_flip()
 + labs(x='', y='Mean % of Total Cash Outflow', title='Spending Categories at Baseline')
 + theme(figure_size=(7, 5), axis_text_y=element_text(angle=45, ha='right')))
```

Figure 4: Average percentage of cash outflow to each category at baseline

```{python}
#| label: fig-outflow-sources-amount
#| fig-cap: Average amount of cash outflow to each category at baseline (% × total outflow)

bl_out_pos = baseline[baseline['cash_outflow'] > 0]
out_amount_df = (pd.DataFrame({
    'Category': [outflow_name_map.get(c, c) for c in outflow_pct_cols],
    'Amount': [(bl_out_pos[c] / 100 * bl_out_pos['cash_outflow_99']).mean() for c in outflow_pct_cols],
    'N': [(bl_out_pos[c] > 0).sum() for c in outflow_pct_cols]
}).sort_values('Amount', ascending=True))
out_amount_df['Category'] = pd.Categorical(out_amount_df['Category'], categories=out_amount_df['Category'], ordered=True)
out_amount_df['label'] = out_amount_df.apply(lambda r: f"{r['Amount']:,.0f} (n={r['N']})", axis=1)

(ggplot(out_amount_df, aes(x='Category', y='Amount'))
 + geom_col(fill='coral')
 + geom_text(aes(label='label'), ha='left', nudge_y=50, size=7)
 + coord_flip()
 + labs(x='', y='Mean Amount', title='Spending Categories at Baseline: Amount')
 + theme(figure_size=(7, 5), axis_text_y=element_text(angle=45, ha='right')))
```

Figure 5: Average amount of cash outflow to each category at baseline (% × total outflow)

```{python}
#| label: fig-dominant-outflow-source
#| fig-cap: Number of households with concentrated spending (≥70% to single category)

bl_out_pos = baseline[baseline['cash_outflow'] > 0].copy()
bl_out_pos['dominant_pct'] = bl_out_pos[outflow_pct_cols].max(axis=1)
bl_out_pos['dominant_src'] = bl_out_pos[outflow_pct_cols].idxmax(axis=1).map(outflow_name_map)
concentrated_out = bl_out_pos[bl_out_pos['dominant_pct'] >= 70]
n_conc_out, n_tot_out = len(concentrated_out), len(bl_out_pos)

out_counts = concentrated_out['dominant_src'].value_counts().reset_index()
out_counts.columns = ['Source', 'N']
out_counts = out_counts.sort_values('N', ascending=True)
out_counts['Source'] = pd.Categorical(out_counts['Source'], categories=out_counts['Source'], ordered=True)

(ggplot(out_counts, aes(x='Source', y='N'))
 + geom_col(fill='coral')
 + geom_text(aes(label='N'), ha='left', nudge_y=1, size=8)
 + coord_flip()
 + labs(x='', y='Number of Households',
        title=f'Concentrated Spending (≥70%): {n_conc_out}/{n_tot_out} HH ({n_conc_out/n_tot_out*100:.1f}%)')
 + theme(figure_size=(7, 5), axis_text_y=element_text(angle=45, ha='right')))
```

Figure 6: Number of households with concentrated spending (≥70% to single category)

Time Trends

Income Amount by Source Over Time

```{python}
#| label: fig-income-amount-time
#| fig-cap: Average income amount from each source over time

# Calculate cash_inflow amount for each category by period
df_pos = df[df['cash_inflow'] > 0].copy()
trends = []
for period in sorted(df_pos['period'].dropna().unique()):
    p_df = df_pos[df_pos['period'] == period]
    for c in inflow_pct_cols:
        amt = (p_df[c] / 100 * p_df['cash_inflow_99']).mean()
        trends.append({'Period': int(period), 'Category': inflow_name_map.get(c, c), 'Amount': amt})
trends_df = pd.DataFrame(trends)


(ggplot(trends_df, aes(x='Period', y='Amount', color='Category'))
 + geom_line(size=1)
 + geom_point(size=2)
 + labs(x='Period', y='Mean Amount', title='Income by Source Over Time')
 + theme(figure_size=(8, 5), legend_position='right'))
```

Figure 7: Average income amount from each source over time

Employment

```{python}
#| label: fig-income-trend-employment
#| fig-cap: Employment income over time by treatment arm

cat_df = trends_trt_df[trends_trt_df['Category'] == 'Employment']
(ggplot(cat_df, aes(x='Period', y='Amount', color='Treatment'))
 + geom_line(size=1) + geom_point(size=2)
 + labs(x='Period', y='Mean Amount', title='Employment Income by Treatment Arm')
 + theme(figure_size=(7, 4), legend_position='right'))
```

Figure 8: Employment income over time by treatment arm

Crop Sales

```{python}
#| label: fig-income-trend-cropsales
#| fig-cap: Crop sales income over time by treatment arm

cat_df = trends_trt_df[trends_trt_df['Category'] == 'Crop Sales']
(ggplot(cat_df, aes(x='Period', y='Amount', color='Treatment'))
 + geom_line(size=1) + geom_point(size=2)
 + labs(x='Period', y='Mean Amount', title='Crop Sales Income by Treatment Arm')
 + theme(figure_size=(7, 4), legend_position='right'))
```

Figure 9: Crop sales income over time by treatment arm

Other Agri. Income

```{python}
#| label: fig-income-trend-agincome
#| fig-cap: Other agricultural income over time by treatment arm

cat_df = trends_trt_df[trends_trt_df['Category'] == 'Other Agri. Income']
(ggplot(cat_df, aes(x='Period', y='Amount', color='Treatment'))
 + geom_line(size=1) + geom_point(size=2)
 + labs(x='Period', y='Mean Amount', title='Other Agri. Income by Treatment Arm')
 + theme(figure_size=(7, 4), legend_position='right'))
```

Figure 10: Other agricultural income over time by treatment arm

New Loans

```{python}
#| label: fig-income-trend-newloans
#| fig-cap: New loans income over time by treatment arm

cat_df = trends_trt_df[trends_trt_df['Category'] == 'New Loans']
(ggplot(cat_df, aes(x='Period', y='Amount', color='Treatment'))
 + geom_line(size=1) + geom_point(size=2)
 + labs(x='Period', y='Mean Amount', title='New Loans by Treatment Arm')
 + theme(figure_size=(7, 4), legend_position='right'))
```

Figure 11: New loans income over time by treatment arm

Other Income

```{python}
#| label: fig-income-trend-otherincome
#| fig-cap: Other income over time by treatment arm

cat_df = trends_trt_df[trends_trt_df['Category'] == 'Other Income']
(ggplot(cat_df, aes(x='Period', y='Amount', color='Treatment'))
 + geom_line(size=1) + geom_point(size=2)
 + labs(x='Period', y='Mean Amount', title='Other Income by Treatment Arm')
 + theme(figure_size=(7, 4), legend_position='right'))
```

Figure 12: Other income over time by treatment arm

Income Concentration Over Time

```{python}
#| label: fig-income-concentration-time
#| fig-cap: Number of households with ≥70% income from single source over time

conc_trends = []
for period in sorted(df['period'].dropna().unique()):
    p_df = df[(df['period'] == period) & (df['cash_inflow'] > 0)].copy()
    p_df['dominant_pct'] = p_df[inflow_pct_cols].max(axis=1)
    p_df['dominant_src'] = p_df[inflow_pct_cols].idxmax(axis=1).map(inflow_name_map)
    conc = p_df[p_df['dominant_pct'] >= 70]
    for src in conc['dominant_src'].unique():
        conc_trends.append({'Period': int(period), 'Source': src, 'N': (conc['dominant_src'] == src).sum()})
conc_df = pd.DataFrame(conc_trends)

top_sources = conc_df.groupby('Source')['N'].sum().nlargest(5).index.tolist()
conc_top = conc_df[conc_df['Source'].isin(top_sources)]

(ggplot(conc_top, aes(x='Period', y='N', color='Source'))
 + geom_line(size=1)
 + geom_point(size=2)
 + labs(x='Period', y='N Households (≥70% concentrated)', title='Income Concentration Over Time')
 + theme(figure_size=(8, 5), legend_position='right'))
```

Figure 13: Number of households with ≥70% income from single source over time