Accessible Version
Monetary Policy and the Distribution of Income: Evidence from U.S. Metropolitan Areas, Accessible Data
Figure 1. Top 10 Percent Income Share in the United States, 1998–2019
Line chart, by percent, 1998 to 2019. Data are yearly. There are three series, labeled “our measure”, “Piketty, Saez, and Zucman (2018)”, and “Auten and Splinter (2024)”. The “our measure” series starts at approximately 46 percent, falls to about 45 percent by 2003, rises to about 46 percent by 2007, falls to about 45 percent by 2009, and rises steadily to about 47 percent by 2019. The series, “Piketty, Saez, and Zucman (2018)” starts at about 42 percent, rises to about 44 percent by 2006, falls to 43 percent by 2009, rises to 46 percent by 2012, and remains relatively flat until 2019. The series, “Auten and Splinter (2024)” starts at about 36 percent in 1998, rises to about 39 percent by 2006, falls to about 37 percent by 2009, rises to about 40 percent by 2012, and remains relatively flat until 2019.
Note: The black line depicts the share of adjusted gross income earned by the top 10 percent of zip codes in the SOI data. The dotted red line and the dashed green line show the shares of pre-tax national income earned by the top 10 percent of households using administrative tax returns as computed by Piketty et al. (2018) and Auten and Splinter (2024), respectively.
Source: Authors’ calculations using data from the IRS Statistics of Income.
Figure 2. Income Distribution across U.S. Metro Areas and Time
Four charts (top-left is panel A, top-right is panel B, bottom-left is panel C, bottom-right is panel D)
Panel A: Avg. P90/P10 ratio: Adjusted gross income
Map of the United States (excluding Hawaii and Alaska) showing the log p90/p10 ratio of adjusted gross income for zip codes within core-based statistical areas (CBSAs) across the US. The scale is a sliding scale, where CBSAs with the lowest ratios (closer to 1) are shaded yellow. As the log ratio increases, the shading turns from yellow to orange (around 3.3) to red (around 5.5) to dark red (around 9). CBSAs that do not have enough zip codes for this ratio calculation are left unshaded. The west and east coasts (particularly the northeast) have more red and dark red CBSAs. The central US is more sparsely shaded than the coasts, and the CBSAs that are shaded tend to have yellow or orange shading. The only state that does not have any shaded CBSAs on this map is Wyoming.
Panel B: Avg. P90/P10 ratio: Salaries & wages
Map of the United States (excluding Hawaii and Alaska) showing the log p90/p10 ratio of salaries and wages for zip codes within core-based statistical areas (CBSAs) across the US. The scale is a sliding scale, where CBSAs with the lowest ratios (closer to 1) are shaded yellow. As the log ratio increases, the shading turns from yellow to orange (around 2) to red (around 3) to dark red (around 5). CBSAs that do not have enough zip codes for this ratio calculation are left unshaded. The west and east coasts (particularly the northeast) have more red and dark red CBSAs. The central US is more sparsely shaded than the coasts, and the CBSAs that are shaded tend have yellow or orange shading. The only state that does not have any shaded CBSAs on this map is Wyoming.
Panel C: Adjusted gross income across metro areas
Line chart. The x-axis shows percentile, starting from 1 and ending at 99. They y-axis shows adjusted gross income across metro areas per household (in 2017 dollars), starting at 0 and ending at $300,000. There are two series, “1998” and “2019”. The 1998 series starts at around $30,000 at the 1st percentile and steadily increases to around $70,000 by the 80th percentile. From here, the slope begins to sharply increase, until the line is nearly vertical and reaches around $220,000 by the 99th percentile. The 2019 series starts at around $30,000 at the 1st percentile and steadily increases to around $80,000 by the 80th percentile. From here, the slope begins to sharply increase, until the line is nearly vertical and reaches over $300,000 by the 99th percentile. The 2019 series remains always above the 1998 series.
Panel D: Salaries & wages across metro areas
Line chart. The x-axis shows percentile, starting from 1 and ending at 99. They y-axis shows salaries and wages across metro areas per household (in 2017 dollars), starting at 0 and ending at $300,000. There are two series, “1998” and “2019”. The 1998 series starts at around $25,000 at the 1st percentile and steadily increases to around $50,000 by the 80th percentile. From here, the slope begins to sharply increase, until the line is nearly vertical and reaches around $120,000 by the 99th percentile. The 2019 series starts at around $25,000 at the 1st percentile and steadily increases to around $60,000 by the 80th percentile. From here, the slope begins to sharply increase, until the line is nearly vertical and reaches around $150,000 by the 99th percentile. The 2019 series remains always above the 1998 series.
Note: Panels A and B depict the distributions of the CBSA-specific time-series averages of the P90/P10 ratio of total income and the CBSA-specific time-series averages of the P90/P10 ratio of labor income, respectively, from 1998 through 2019. Panels C and D depict the 1998 (blue dashed line) and 2019 (solid black line) distributions of (real) total income per household and (real) labor income per household, respectively.
Source: Authors’ calculations using data from the IRS Statistics of Income.
Figure 3. Response of Income Inequality to a Monetary Policy Tightening
Four charts (top-left is panel A, top-right is panel B, bottom-left is panel C, bottom-right is panel D)
Panel A: P90/10 ratio: Adjusted gross income
Scatter plot with whiskers on estimates. The x-axis shows years after the monetary policy surprise from 1 to 4. The y-axis shows the impulse response of a monetary policy shock on adjusted gross income, in percent, from -2 to 2. There is one series with points: “Estimate”. Additionally, there is one line series, “90% CI”, that shows the 90 percent confidence interval around each estimate. The estimate for year one is around 0.75 percent, with the confidence interval going from about -0.1 to about 1.75. In year 2, the estimate is around 1.1, with the confidence interval going from about 0.4 to about 1.8. In year 3, the estimate is around 0.5, with the confidence interval going from slightly below 0 to slightly above 1. In year 4, the estimate is around 0.25, with the confidence interval going from around -1 to around 1.8.
Panel B: P90 and P10: Adjusted gross income
Scatter plot with whiskers on estimates. The x-axis shows years after the monetary policy surprise from 1 to 4. The y-axis shows the impulse response of a monetary policy shock on adjusted gross income, in percent, from -4 to 2. There are two series with points: “Estimate (P90)” and “Estimate (P10)”. Additionally, there is one line series, “90% CI”, that shows the 90 percent confidence interval around each estimate for both series. The P90 estimate for year one is around 0.5 percent, with the confidence interval going from about -0.5 to about 1.5. In year 2, the P90 estimate is around 0.5, with the confidence interval going from about -0.5 to about 1.3. In year 3, the P90 estimate is slightly below -1, with the confidence interval going from around -1.9 to around -0.2. In year 4, the P90 estimate is around -2.25, with the confidence interval going from around -2.9 to around -1.2. The P10 estimates are shown slightly to the right of each P90 estimate for all years. The P10 estimate for year one is around -0.25 percent, with the confidence interval going from about -1.2 to about 0.5. In year 2, the P10 estimate is around -0.5, with the confidence interval going from about -1.5 to about 0.2. In year 3, the P10 estimate is around -1.5, with the confidence interval going from around -2.1 to around -0.8. In year 4, the P10 estimate is around -2.5, with the confidence interval going from around -3.8 to around -1.2.
Panel C: P90/10 ratio: Salaries & wages
Scatter plot with whiskers on estimates. The x-axis shows years after the monetary policy surprise from 1 to 4. The y-axis shows the impulse response of a monetary policy shock on salaries and wages, in percent, from -2 to 2. There is one series with points: “Estimate”. Additionally, there is one line series, “90% CI”, that shows the 90 percent confidence interval around each estimate. The estimate for year one is around 0.5 percent, with the confidence interval going from about -0.2 to about 1.2. In year 2, the estimate is around 0.75, with the confidence interval going from about 0.2 to about 1.2. In year 3, the estimate is slightly below 1, with the confidence interval going from around 0.5 to around 1.5. In year 4, the estimate is around 1.5, with the confidence interval going from around 0.5 to around 2.25.
Panel D: P90 and P10: Salaries & wages
Scatter plot with whiskers on estimates. The x-axis shows years after the monetary policy surprise from 1 to 4. The y-axis shows the impulse response of a monetary policy shock on salaries and wages, in percent, from -4 to 2. There are two series with points: “Estimate (P90)” and “Estimate (P10)”. Additionally, there is one line series, “90% CI”, that shows the 90 percent confidence interval around each estimate for both series. The P90 estimate for year one is around –0.5 percent, with the confidence interval going from about -1 to about 0. In year 2, the P90 estimate is around -0.6, with the confidence interval going from slightly below -1 to about -0.25. In year 3, the P90 estimate is slightly above -1, with the confidence interval going from around -1.2 to around -0.6. In year 4, the P90 estimate is around -1.75, with the confidence interval going from around -2.7 to slightly above -1. The P10 estimates are shown slightly to the right of each P90 estimate for all years. The P10 estimate for year one is slightly above -1, with the confidence interval going from slightly above -2 to about 0. In year 2, the P10 estimate is around -1.25, with the confidence interval going from slightly below -2 to about -0.5. In year 3, the P10 estimate is around -1.8, with the confidence interval going from around -2.5 to slightly below -1. In year 4, the P10 estimate is around -3.1, with the confidence interval going from around –4.9 to around -1.5.
Note: Panels A and C depict the impulse response of an unexpected 25 basis point tightening in monetary policy on total income inequality and labor income inequality, respectively, over a horizon of one to four years. Point estimates are shown using black circular dots. Panels B and D decompose these impulse responses into changes in the (real) P90 income level (square red dots) and (real) P10 income level (circular blue dots) for total income and labor income, respectively, over the same four-year horizon. The whiskers in all panels depict the 90 percent confidence interval for a given point estimate.
Figure 4. Response of Labor Income Inequality to a Monetary Policy Tightening (In Weak Local Labor Markets)
Panel A: P90/10 ratio: Salaries & wages
Scatter plot with whiskers on estimates. The x-axis shows years after the monetary policy surprise from 1 to 4. The y-axis shows the impulse response of a monetary policy shock on salaries & wages, in percent, from -1 to 2. There is one series with points: “Estimate”. Additionally, there is one line series, “90% CI”, that shows the 90 percent confidence interval around each estimate. The estimate for year one is around -0.1 percent, with the confidence interval going from about -0.4 to about 0.2. In year 2, the estimate is around 1.3, with the confidence interval going from about 0 to about 0.8. In year 3, the estimate is around 1.1, with the confidence interval going from around 0.5 to around 1.6. In year 4, the estimate is around 1.1, with the confidence interval going from around 0.7 to around 1.5.
Panel B: P90 and P10: Salaries & wages
Scatter plot with whiskers on estimates. The x-axis shows years after the monetary policy surprise from 1 to 4. The y-axis shows the impulse response of a monetary policy shock on salaries and wages, in percent, from -4 to 1. There are two series with points: “Estimate (P90)” and “Estimate (P10)”. Additionally, there is one line series, “90% CI”, that shows the 90 percent confidence interval around each estimate for both series. The P90 estimate for year one is slightly below 0, with the confidence interval going from about -0.5 to about 0.5. In year 2, the P90 estimate is around -0.2, with the confidence interval going from about -0.8 to about 0.3. In year 3, the P90 estimate is around -0.1, with the confidence interval going from around -0.75 to around 0.4. In year 4, the P90 estimate is slightly below -1, with the confidence interval going from around -1.7 to around -0.4. The P10 estimates are shown slightly to the right of each P90 estimate for all years. The P10 estimate for year one is slightly above 0, with the confidence interval going from about -0.7 to about 0.9. In year 2, the P10 estimate is around -0.6, with the confidence interval going from about -1.2 to slightly above 0. In year 3, the P10 estimate is around -1.2, with the confidence interval going from around -1.9 to around 0.4. In year 4, the P10 estimate is around -2.1, with the confidence interval going from around -2.9 to around -1.3.
Note: Panel A depicts the impulse response of an unexpected 25 basis point tightening in monetary policy on labor income inequality when local labor markets are weak over a horizon of one to four years. Point estimates are shown using black circular dots. Panel B decomposes this impulse response when local labor markets are weak into changes in the (real) P90 income level (square red dots) and (real) P10 income level (circular blue dots), respectively, over the same four-year horizon. The whiskers in both panels depict the 90 percent confidence interval for a given point estimate.