## Appendix 5. Empirical Strategy and Specification for the Results on Earnings

## Source details

**Canonical URL:** [Appendix 5. Empirical Strategy and Specification for the Results on Earnings](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021201-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021201-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021201-print-pdf.pdf.json)

---

### I. Research framing and empirical focus
- Central research question:
  - What are the channels through which monetary policy may impact income or wealth inequality, and is there empirical evidence of economically significant effects?
- Scope and caveats:
  - Empirical focus on within-country inequality, mostly in advanced economies with high-quality micro data.
  - Paper does not cover net welfare effects in a unified framework, post-COVID extraordinary measures empirically (limited time/uncertainty), effects of the secular decline in r*, nor cross-border spillovers.
- Pre-COVID synthesis:
  - Properly identified monetary policy shocks often suggest monetary accommodation reduces income inequality mostly by reducing unemployment.
  - Net effects of transitory monetary easing on inequality are generally small and temporary; simulations and HANK experiments find effects dissipate after about 8 quarters.
  - Evidence on wealth inequality is less clear; country-specific results vary and unconventional policies may differ from conventional rate cuts.

### II. Stylized facts and structural context
- Key stylized facts:
  - Close to 95 million more people fell into extreme poverty in 2020.
  - Within-country income inequality trending up in advanced economies; wealth inequality rose (top 1 percent share in G20 increased).
  - Global real interest rates declined over the past 30 years; decline in the real neutral rate accelerated after the GFC.
- Structural drivers highlighted:
  - Trade and financial integration, automation, demographics implicated in rising inequality and lower interest rates.
  - Decline in real interest rates attributed to globalization, potential growth, and demographics.

### III. Channels through which monetary policy affects distribution
- Income channels:
  - Income composition channel: richer households derive larger share of income from business and capital income; monetary easing may increase inequality via capital-income responsiveness.
  - Earnings distribution channel: lower-income households’ wages and employment prospects are typically more sensitive to monetary policy; easing may reduce income/consumption inequality mainly by reducing unemployment.
- Wealth channels:
  - Balance sheet composition channel: differing asset/liability responses (real estate, bonds, stocks) make effects ambiguous across the wealth distribution.
  - Savings-redistribution channel: unexpected decreases in real interest rates favor borrowers and hurt savers; borrowers often have higher marginal propensity to consume.
- Additional channels and feedbacks:
  - Inflation channel, interest rate exposure channel (Auclert 2019), and general equilibrium feedbacks (consumption, investment, employment, current/financial account) can offset or amplify on-impact effects.

### IV. Household exposures and illustrative empirical evidence
- Data sources used for illustrations:
  - OECD income distribution database (IDD) 2014–17; SCF 2016 for the United States; HFCS 2016 for the European Union.
- Income and asset exposures:
  - In the United States, capital income is especially concentrated at the very top (top 1 and top 10 percent).
  - For most households (except top wealth groups), real estate is the largest asset share in both US and EU.
  - Household liabilities are dominated by borrowing against real estate, primarily long-term nominal mortgages.
- Unemployment exposure:
  - Young workers face much higher unemployment risk; monetary easing mitigates recessionary unemployment effects disproportionately benefiting younger, less-experienced, and lower-paid workers.
- Key numerical observations from portfolio/asset simulations:
  - Capital gains on equities largest at the top: US top gains 7.4 percent; EU top gains 3 percent.
  - Housing price increases: US bottom twenty percent gain 13.6 percent; EU bottom twenty percent gain 5.8 percent.
  - Wealthy in the US derive larger gains from equity holdings (7.4 percent); wealthy in the EU derive relatively larger gains from real estate assets (5.4 percent).
  - Caution: percentage gains can appear large for households with very low net wealth.

### V. Microsimulations: Unhedged Interest Rate Exposures (UREs) and asset-price scenarios
- URE metric and illustrative shock:
  - URE = Y − T − C + A − L (formula preserved exactly).
  - Illustrative shock: fall in the real interest rate by 100 bps.
- URE findings:
  - Poor and younger middle-class households with large debts and little financial assets tend to have larger negative UREs and thus benefit relatively more from a fall in interest rates.
  - UREs tend to increase with income and age; older, particularly rich households have the largest positive UREs.
  - Cross-country variation within the EU is substantial; UREs positive across all income groups on average for Germany and Austria, negative for substantial parts of distribution in Cyprus, Ireland, Netherlands, Portugal, and Spain.
  - Simulations do not include unrealized capital gains from prolonged easing that typically benefit the wealthy via equity price increases.
- Asset-price microsimulation scenario:
  - Illustrative scenario: 10 percent increase in asset prices (stocks, real estate, bonds).
  - Findings:
    - Higher equity prices favor wealthier households in both the US and the EU.
    - Real estate price increases favor mostly the bottom part of the distribution in the US and mostly the middle in the EU.
    - Bond price increases leave net wealth distribution largely unchanged.

### VI. DSGE / HANK model evidence (KMV and variations)
- KMV HANK model features:
  - Two assets: low-return short-dated asset and high-return long-dated asset subject to transaction costs.
  - Calibration targets: net wealth inequality with about 60 percent of households either in debt or holding close to zero cash; wealthiest decile holds 88 and 75 percent of long- and short-dated assets respectively; 32 percent of labor earnings accrue to the wealthiest decile.
- Benchmark shock and outcomes:
  - Shock: monetary policy loosening via an exogenous shock to the inflation-targeting Taylor rule of 100 bps (annualized).
  - Benchmark assumption: fiscal space generated by easing used to pay down debt.
  - Main dynamic outcomes (IRFs) under benchmark:
    - Higher income inequality as portfolio effects (higher total returns to assets for the rich) outweigh positive impacts on labor income.
    - Net wealth inequality roughly unchanged despite rising income inequality due to high responsiveness of wealth at the bottom and very high assets-to-income ratios at the top:
      - Average household holds assets worth 3 times their income.
      - At the top of the wealth distribution, the assets-to-income ratio is larger than 12; at the bottom it is essentially zero.
    - Lower consumption inequality: rising wages support consumption of the poorest while consumption of the rich barely moves.
- Fiscal policy variant:
  - Using fiscal space for transfers in combination with monetary loosening improves aggregate transmission and yields better distributional outcomes, particularly for consumption inequality.
- Sensitivity / Gini calibrations (Gini index, in percent):
  - Baseline: Liquid 0.98; Illiquid 0.81
  - Higher lending spread: Liquid 0.83; Illiquid 0.82
  - Higher cost of using illiquid account: Liquid 0.93; Illiquid 0.73
  - Higher neutral rate: Liquid 0.72; Illiquid 0.83
  - France (HFCS 2016): Liquid 0.77; Illiquid 0.70
  - Germany (HFCS 2016): Liquid 0.71; Illiquid 0.82
- Model caveats:
  - KMV’s strong portfolio channel arises from assumptions (flexible wages, sticky prices); alternative assumptions (e.g., wage rigidities, employed–unemployed margin) can reverse sign/magnitude of distributional responses in other HANK specifications.

### VII. Empirical specification and identification for earnings results
- Data and method:
  - Local projection method of Jordà (2005) augmented with smooth transition regression (Granger and Teravistra 1993).
  - US macro and household data at quarterly frequency over 1980–2016 to compare pre- and post-2008 periods.
  - Individual-level earnings and wages from the Current Population Survey (CPS); CPS data are top-coded.
- Estimated specification (demeaned four-quarter earnings change as dependent variable):
  - Left-hand side: 푤푤_{i,t+4} − 푤푤_{i,t} (demeaned earnings change over four quarters).
  - Controls vector 푿_{i,t} includes:
    - a second-order polynomial in age;
    - state fixed effects;
    - occupation fixed effects;
    - industry fixed effects;
    - education fixed effects (10 categories);
    - seasonal fixed effects (12 categories);
    - gender;
    - race;
    - individual relationship to household head (4 categories).
  - 푤푤_{i,t} is demeaned earnings: log wages_{i,t} − 1/N ∑ log wages_{i′,t}.
- Identification of monetary policy shocks:
  - Use high-frequency variations in interest rate futures within a narrow window around FOMC announcements (following Gertler and Karadi (2015)).
  - Futures surprises around FOMC announcements used as instruments: first stage regresses changes in the FFR on futures surprises; second stage uses instrumented FFR changes to estimate earnings effects.

### VIII. Key empirical findings on labor-market responses
- Estimated dynamic effects (reported in text):
  - A 100 bps increase in the FFR decreases year-on-year average wages by 0.6 percent in the first quarter after the monetary policy shock.
  - The same 100 bps increase decreases year-on-year average wages by 1.1 percent in the fourth quarter after the shock.
- Heterogeneity and margins:
  - Unemployment (extensive margin): illustrative 100 bps monetary loosening yields a cumulative 4 percentage-point drop in unemployment after four quarters in the illustrative specification; benefits disproportionately accrue to younger, less experienced, and lower-paid workers.
  - Wages (intensive margin): for those remaining employed, wages at the top rise disproportionately more than at the bottom following monetary loosening, partly offsetting the unemployment effect.
  - Net effect via the earnings distribution channel: small and inequality-reducing because the unemployment (extensive) effect dominates the wage (intensive) effect.
- Robustness:
  - Introducing interactions for initial wage levels confirms significant heterogeneity.
  - Interacting the change in the FFR with a dummy for wages larger than the year-specific average yields similar conclusions.

### IX. Synthesized empirical/quantitative takeaways
- Net pre-COVID effects of transitory monetary policy easing on inequality:
  - Generally small and temporary when properly identified; dissipate after about 8 quarters in simulations and HANK experiments.
  - Heterogeneity across countries and studies; unconventional policies may have different distributional impacts than conventional rate cuts.
- Literature synthesis (three key takeaways):
  - (i) Magnitude of net distributional effect is small compared to trend evolution in inequality within countries.
  - (ii) Monetary policy easing may reduce income and consumption inequality in the medium term via labor-market channels.
  - (iii) In the short term, wealth and perhaps income inequality may rise in response to monetary policy easing, driven by asset-price effects.

### X. Policy implications and communications
- Core policy stance:
  - Monetary policy should remain focused on macroeconomic stabilization and price stability as a primary goal; adding distributional objectives to central bank mandates risks reducing effectiveness and undermining independence.
- Recommendations for central banks:
  - Model and analyze heterogeneity (income and wealth distributions) within existing frameworks to better understand transmission.
  - Communicate distributional effects and counterfactuals to address public concerns and explain how countercyclical policy enhances welfare.
  - Consider coordination with fiscal policy: monetary loosening can create fiscal space that, if used for targeted transfers, improves both macro and distributional outcomes.
- Communication priorities:
  1. Clearly state objectives and mandate and explain how policy conduct achieves them.
  2. Emphasize monetary policy is a blunt tool for group-specific concerns.
  3. Recognize and discuss public concerns using central bank data and research expertise.

*Source: wpiea2021201-print-pdf — Appendix 5. Empirical Strategy and Specification for the Results on Earnings*

### References..............................................................................................................

### wpiea2021201-print-pdf - References..............................................................................................................

### Listed components in this content unit
- References....................................................................................................................... 35
- Appendix 1. Definition of Variables In Household Surveys ............................................... 41
- Appendix 2. Benchmarking “Unhedged Interest Rate Exposures”...................................... 43
- Appendix 3. Benchmarking Distributional Implications of  Asset Price Increases................ 44
- Appendix 4. The Heterogeneous Agent New Keynesian model of KMV............................ 45

*Source: wpiea2021201-print-pdf - References..............................................................................................................*

### Appendix 5. Empirical Strategy and Specif ication for the Results on Earnings ................... 46

### Appendix 5. Empirical Strategy and Specification for the Results on Earnings

### I. Introduction — framing and research question
- Rising income and wealth inequality in advanced economies since the 1980s coincided with prolonged monetary easing since the Global Financial Crisis (GFC).  
- Central research question: What are the channels through which monetary policy may impact income or wealth inequality, and is there empirical evidence of economically significant effects?  
- Key contextual facts reported:
  - The increase in inequality and the decrease in interest rates have both been long-term phenomena, likely driven by structural factors (Dabla-Norris et al. 2015).  
  - Since the GFC, many advanced economies experienced substantial monetary easing, including unconventional measures.  
- Empirical focus: within-country inequality, mostly in advanced economies with high-quality micro data.  
- Summary of pre-COVID empirical evidence:
  - Properly identified monetary policy shocks suggest monetary accommodation reduces income inequality mostly by reducing unemployment (e.g., Coibion et al. 2017; Furceri et al. 2018).  
  - Evidence on wealth inequality is less clear due to shorter time series.  
  - Many studies find mixed and often economically negligible net effects of transitory monetary policy easing on income inequality, with variation across countries and policy tools.  
  - Quantitative models with heterogeneity and many hand-to-mouth consumers often find beneficial effects of monetary easing on inequality via general equilibrium effects (Ampudia et al. 2018; Kaplan et al. 2018).  
  - Micro-simulations and a HANK model simulation for this paper confirm pre-COVID net effects of monetary easing on inequality are small and temporary (they dissipate after about 8 quarters).  
  - Contrasting evidence: Danish administrative data show large asset price increases from easing accrue disproportionately to wealthier households, raising inequality (Andersen et al. 2020).  
- Policy stance regarding central banks:
  - Central banks should communicate distributional effects to address public concerns and explain how countercyclical policy enhances welfare.  
  - Central banks should model and analyze distributions of income and wealth within existing frameworks because heterogeneity affects transmission.  
  - Monetary policy should remain focused on macroeconomic stability; adding objectives risks reducing effectiveness and may undermine central bank independence.  
- Caveats and boundaries of the paper:
  - Does not cover net welfare effects of monetary policy in a unified framework (output, inflation, unemployment, and distributional effects).  
  - Does not empirically cover distributional effects of post-COVID extraordinary measures due to limited time and high uncertainty.  
  - Does not cover effects of the secular decline in the natural rate of interest on inequality, nor cross-border monetary policy spillovers.

### II. Stylized facts documented
- Inequality intensified post-GFC and was further aggravated by COVID-19:
  - Pandemics tend to increase inequality (Figure 2; Furceri et al. 2020).  
  - Close to 95 million more people fell into extreme poverty in 2020 (IMF 2021).  
  - COVID-19 had particularly adverse consequences for young workers, women, and lower-income households (Schnabel 2020; Dossche et al. 2020).  
- Monetary policy response patterns:
  - Major central banks implemented aggressive easing during the GFC and again during COVID-19; unconventional measures rose when policy rates hit the zero-lower bound.  
  - Heavy reliance on accommodation increased public interest in central banks after the GFC.  
- Long-term coexistence of trends:
  - Within-country income inequality trending up in advanced economies; emerging markets show a mixed picture but higher initial levels.  
  - Wealth inequality rose; in the G20, the share of wealth of the top 1 percent increased over the past two decades (Figure 4).  
  - Global real interest rates have declined over the past 30 years; the decline in the real neutral rate accelerated after the GFC (Figure 5).  
- Structural drivers highlighted:
  - Trade and financial integration, automation, and demographics implicated in rising inequality and lower interest rates.  
  - Dabla-Norris et al. (2015) (sample ~100 AEs and EMDEs, period 1980-2012) found greater financial openness and technological progress and easing of labor market regulations associated with rising within-country income inequality; trade openness had a negative relationship with inequality.  
  - Decline in real interest rates attributed to globalization (Natal and Stoffels 2019), potential growth (Pescatori and Turunen 2015), and demographics (Han 2019).  
  - COVID-19 may affect these structural factors and put added pressure on interest rates via persistent changes in consumption and savings preferences (Jordà et al. 2020).

### III. Channels for distributional effects of monetary policy (stylized framework)
- Two broad sets of channels: income channels and wealth channels. Heterogeneity across households (income, wealth, age, skill, sector, marginal propensity to consume, access to finance) is the building block.
- Income channels:
  - Income composition channel:
    - Richer households derive a larger share of income from business and financial (capital) income, typically more responsive to monetary policy than labor income (Gornemann et al. 2016).  
    - Monetary easing may benefit richer households more and increase income/consumption inequality under this channel.  
  - Earnings distribution channel:
    - Lower-income households’ wages and employment prospects are typically more sensitive to monetary policy and business cycles (Heathcote et al. 2010).  
    - Monetary easing may reduce income and consumption inequality by stimulating activity and reducing unemployment, disproportionately helping less-skilled and younger workers.  
- Wealth channels:
  - Balance sheet composition channel:
    - Assets (real estate, bonds, stocks) and liabilities differ in their response to interest rate changes; effects on wealth are ambiguous and depend on type and duration.  
    - Monetary easing may decrease or increase wealth and consumption inequality.  
  - Savings redistribution channel:
    - Unexpected decreases in real interest rates favor borrowers and hurt savers; borrowers often have higher marginal propensity to consume.  
    - Monetary easing may reduce wealth and consumption inequality via redistribution from savers to borrowers.  
- Additional channels (treated as wealth channels in framework):
  - Inflation channel: heterogeneity in cash holdings and nominally fixed debt affects redistribution (Erosa and Ventura 2002; Doepke and Schneider 2006).  
  - Interest rate exposure channel: exposure measured by difference between maturing assets and liabilities at a point in time (Auclert 2019).  
- General equilibrium feedbacks:
  - Consumption, investment, employment and labor income, and current/financial account movements feed back over longer horizons and can offset on-impact effects; ultimate direction depends on heterogeneities and rigidities (prices/wages).

### IV. Household exposures and illustrative evidence
- Data sources and scope used to illustrate exposures: OECD income distribution database (IDD) for OECD countries over 2014–17; Survey of Consumer Finances (SCF) 2016 for the United States; Household Finance and Consumption Survey (HFCS) 2016 for the European Union; other cited datasets and studies.
- Income composition exposures:
  - On average for OECD countries, the share of labor income and transfers varies significantly across income quintiles (Figure 7).  
  - In the United States, capital income is especially concentrated at the very top (top 1 and top 10 percent).  
  - In the euro area, richer households hold more financial assets and receive a larger share of capital income; bottom households benefit relatively more from transfers.  
- Unemployment risk exposure:
  - Young workers face much higher unemployment risk and volatility than older workers; youth unemployment is more sensitive to economic cycles (2000–19 volatility comparisons).  
  - Monetary easing mitigates recessionary unemployment effects, disproportionately benefiting younger, less experienced, and lower-paid workers and thereby reducing labor earnings disparities. This is highlighted as the most powerful inequality-reducing channel of monetary easing in several instances (Carney 2016; Draghi 2016; Amberg et al. 2021).  
  - Offsetting forces: skill-premium channel—high-skilled labor may benefit relatively more due to capital-skill complementarity (Dolado et al. 2021).  
- Balance sheet composition exposures:
  - In the US and EU (SCF 2016 and HFCS 2016):
    - The wealthiest households hold a larger share of financial assets (US: wealthiest 10 percent’s assets are ~46 percent stocks and 16 percent bonds).  
    - For most households (except the top wealth groups), real estate is the largest asset share in both the US and EU.  
    - Monetary easing may have more equitable effects via house prices than via capital income because real estate is more broadly held.  
  - Liabilities:
    - Household balance sheets are dominated by borrowing against real estate, primarily long-term nominal mortgages.  
    - Lower-income households’ mortgage liabilities are more likely to be of subprime quality; terms on these loans may be more sensitive to tightening (adjustable-rate mortgages, marginal borrowers cut first).  
    - Because bottom households have higher marginal propensity to consume, reductions in debt service from easing (especially on ARMs) can increase consumption and other loan types (e.g., car loans) (Di Maggio et al. 2017).

### V. Empirical/quantitative takeaways and specification implications (as framed)
- Net pre-COVID effects of transitory monetary policy easing on inequality:
  - Generally small and temporary when properly identified; dissipate after about 8 quarters in simulations and HANK model experiments reported.  
  - Heterogeneity in signs and magnitudes across studies and countries; unconventional policies may have different distributional impacts than conventional rate cuts.  
- Importance of identification and decomposing channels:
  - Empirical strategies should isolate monetary policy shocks and quantify contributions of income-composition, earnings-distribution, balance-sheet composition, and savings-redistribution channels.  
  - Microdata (income surveys, SCF, HFCS, administrative data) are crucial to measure exposures across the income and wealth distributions.  
- Policy implications emphasized:
  - Central banks should explain counterfactuals showing how monetary policy improves outcomes relative to alternatives.  
  - Central banks should incorporate heterogeneity in models and analyses to better understand transmission and distributional implications.  
  - Despite distributional concerns, maintaining focus on macroeconomic stability is essential to preserve policy effectiveness and central bank independence.

*Italic: Source — wpiea2021201-print-pdf: Appendix 5. Empirical Strategy and Specification for the Results on Earnings*

### Appendix 1 for variable definitions.

### Appendix 1 for variable definitions.

### Household exposures and composition of liabilities/assets
- Household liabilities are dominated by borrowing against real estate, except for the poor in the US.
- EU group comprises Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Lithuania, Latvia, Malta, Netherlands, Poland, Portugal, the Slovak Republic, Slovenia, and Spain.
- Asset/debt distribution cohort definitions used in figures and simulations: Bottom 20, Middle 70, Top 10; age cohorts: Prime age [20–55] and Older [56–85].

### Savings-Redistribution Channel: microsimulations
- Welfare metric: "unhedged interest rate exposures" (UREs).
- Illustrative shock: fall in the real interest rate by 100 bps.
- Key findings:
  - Poor and younger middle-class households with large debts and little financial assets tend to have larger negative UREs and therefore benefit relatively more from a fall in interest rates.
  - UREs tend to increase with income and age; older, particularly rich households have the largest positive UREs.
  - Cross-country variation within the EU is substantial: UREs are positive across all income groups on average for Germany and Austria, and negative for a substantial part of the income and net wealth distribution in Cyprus, Ireland, Netherlands, Portugal, and Spain (Tzamourani 2021, based on 2010 HFCS data).
  - Simulations do not include unrealized capital gains from prolonged monetary easing that typically benefit the wealthy via equity price increases.

### Asset Price Channel: microsimulations
- Illustrative scenario: 10 percent increase in asset prices (stocks, real estate, bonds).
- Key findings:
  - Higher equity prices favor wealthier households in both the US and the EU.
  - Increases in real estate prices:
    - Favor mostly the bottom part of the distribution in the US.
    - Favor mostly the middle of the wealth distribution in the EU.
  - Bond price increases leave the net wealth distribution largely unchanged.
  - Quantitative examples reported:
    - Capital gains on equities largest at the top: US top gains 7.4 percent; EU top gains 3 percent.
    - US bottom twenty percent gain 13.6 percent from housing price increases; EU bottom twenty percent gain 5.8 percent from housing price increases.
    - Wealthy in the US derive larger gains from equity holdings (7.4 percent); wealthy in the EU derive relatively larger gains from real estate assets (5.4 percent).
- Caution: percentage gains can look large for households with very low net wealth; the result that the US bottom 20 percent benefit most from housing price increases should be taken with caution.

### DSGE model for the United States (KMV)
- Shock specification: monetary policy loosening via an exogenous shock to the inflation-targeting Taylor rule of 100 bps (annualized).
- Benchmark calibration assumes fiscal space generated by monetary easing is used to pay down debt.
- Main dynamic outcomes (IRFs) under the benchmark:
  - Higher income inequality as portfolio effects (higher total returns to assets for the rich) outweigh positive impacts on labor income.
  - Net wealth inequality roughly unchanged despite rising income inequality, due to high responsiveness of wealth at the bottom to increases in income and very high assets-to-income ratios at the top.
    - Average household holds assets worth 3 times their income.
    - At the top of the wealth distribution, the assets-to-income ratio is larger than 12; at the bottom it is essentially zero.
  - Lower consumption inequality: rising wages support consumption of the poorest while consumption of the rich barely moves.
- Fiscal policy variant: using increased fiscal space for transfers in combination with monetary loosening improves aggregate transmission and yields better distributional outcomes, particularly for consumption inequality.
- Alternative parametrizations and key numeric calibrations (Gini index, in percent) for initial inequality in holdings of liquid and illiquid assets:
  - Table 1 (Gini index, in percent)
    - Baseline: Liquid 0.98; Illiquid 0.81
    - Higher lending spread: Liquid 0.83; Illiquid 0.82
    - Higher cost of using illiquid account: Liquid 0.93; Illiquid 0.73
    - Higher neutral rate: Liquid 0.72; Illiquid 0.83
    - France (HFCS 2016): Liquid 0.77; Illiquid 0.70
    - Germany (HFCS 2016): Liquid 0.71; Illiquid 0.82
- Alternative parametrization details and effects:
  - First alternative: increase wedge between lending rates and policy rates to 8 percent (compared to 4 percent in baseline) → reduces fraction of households with negative net wealth; IRFs for Gini indices similar to baseline.
  - Second alternative: higher cost of using the illiquid account → slightly larger aggregate effects; slightly less net wealth inequality; reduces initial illiquid net wealth inequality closer to France.
  - Third alternative: raise equilibrium long-run real interest rate r* to 4 percent → monetary policy less effective (lower MPCs), produces larger net wealth inequality driven by direct interest rate channel; calibration close to Germany.

### Labor Earnings Channel: empirical assessment
- Decomposition: extensive margin (employment status/unemployment) and intensive margin (wages).
- Unemployment effects:
  - Illustrative shock: monetary loosening of 100 bps.
  - After four quarters, cumulative impulse response: 4 percentage-point drop in unemployment.
  - Benefits disproportionately accrue to younger, less experienced, and lower-paid workers.
- Wage effects:
  - For those who stay employed, wages at the top rise disproportionately more than those at the bottom following monetary loosening, partly offsetting the unemployment effect.
  - Earnings at the top affected mainly by changes in hourly wages; earnings at the bottom affected mainly by changes in hours worked and employment status.
  - No evidence of an effect on hours worked (results not shown).
- Net effect via the earnings distribution channel: small and inequality-reducing because the unemployment (extensive margin) effect dominates the wage (intensive margin) effect.
- Policy implication: assess both extensive and intensive margins to understand distributional impacts and interpret increased earnings inequality carefully (could reflect a more inclusive labor force).

### Results from other studies and synthesis
- Three main approaches in literature: micro-simulations, macroeconomic modeling, and empirical analyses.
- Common challenges: measurement of inequality, data availability/timeliness, and identification of exogenous monetary policy shocks.
- Representative findings across approaches:
  - Micro-simulations: often small effects; asset-class-specific impacts (equity and bond price increases can raise wealth inequality; house price increases tend to reduce wealth inequality).
  - Model-based analyses: income inequality increases short run and declines medium run; wealth inequality may increase short-lived; consumption inequality decreases short and medium run.
  - Empirical analyses: many studies report expansionary monetary policy reduces inequality (e.g., euro area, Italy, United Kingdom, United States); some studies report increases in inequality for specific countries or with short time series.
- Three key takeaways from the literature:
  - (i) Magnitude of net distributional effect is small compared to trend evolution in inequality within countries.
  - (ii) Monetary policy easing may reduce income and consumption inequality in the medium term.
  - (iii) In the short term, wealth and perhaps income inequality may rise in response to monetary policy easing.
- Additional notes:
  - Some studies that find increases in income inequality associated with reductions in policy rates may conflate exogenous policy shocks with systematic policy responses to worsening economic conditions.
  - Microsimulations are partial equilibrium exercises focusing on on-impact distributional responses and rely on identification of exogenous shocks and assumptions on household behavior.
  - Country-specific structural factors (labor market flexibility, mobility constraints, mortgage/credit market structure) matter for heterogeneity in distributional effects.

### Policy implications — central bank mandate and coordination with fiscal policy
- Primary recommendation: maintain a coherent and transparent monetary policy framework with a clear primary mandate (often price stability) to preserve credibility and effectiveness.
- Drawbacks of adding distributional objectives to central bank mandates:
  - May reduce overall effectiveness of monetary policy.
  - Can undermine credibility and complicate communications.
  - Risks to central bank independence.
- Nonetheless, central banks should:
  - Factor heterogeneity among economic agents into analysis (model income and wealth distributions) and disseminate results, as heterogeneity affects transmission (e.g., collateral effects of real estate wealth, differing marginal propensities to consume).
  - Consider distributional implications when choosing policy targets (e.g., differences between inflation targeting and price-level targeting).
- Role of fiscal policy:
  - Countercyclical fiscal policy is an important complementary tool, especially near the effective lower bound.
  - Monetary loosening can create fiscal space which, if used for targeted transfers, can improve both macro and distributional outcomes (illustrated in KMV simulations combining transfers and monetary loosening).
  - Fiscal tools such as automatic stabilizers provide social insurance and should be used alongside monetary policy when possible.

*Source: wpiea2021201-print-pdf - Appendix 1 for variable definitions.*

### Box 1. Can Central Banks Handle Multiple Policy Objectives?

### Box 1. Can Central Banks Handle Multiple Policy Objectives?

### Multiple objectives, mandates, and institutional implications
- A single monetary policy instrument cannot achieve multiple independent policy goals (Tinbergen 1952).
- Adding goals to the central bank’s mandate can reduce welfare in the context of multiple policymakers and multiple frictions because making the central bank the residual claimant of macroeconomic problems may discourage action by other policymakers or induce non–welfare-enhancing behavior (Davig and Gürkaynak 2015).1/
- Multiple objectives may threaten central bank independence:
  - Independence is a way to solve time inconsistency in monetary policy (Kydland and Prescott 1977; Barro and Gordon 1983; Rogoff 1985).
  - Empirical evidence credibly links central bank independence (with accountability) to lower inflation (Klomp and De Haan 2010).
  - Multiple objectives increase susceptibility to political or special interest pressure.
- Multiple objectives complicate communication, reduce transparency, and weaken accountability:
  - A single quantitative objective—e.g., medium-term inflation—facilitates monitoring and benchmarking of central bank performance.
  - Multiple goals increase the risk that announcements foster confusion and hinder private-sector decision-making (Eusepi and Preston 2010; Reis 2013).
- Complementarities and tradeoffs:
  - Objectives can be complementary; central banks that target inflation also care about growth, financial stability, and financial inclusion.
  - Short-run tradeoffs between inflation and unemployment can justify pursuing complementary real economic goals (absence of “divine coincidence” per Blanchard and Gali 2007; Reis 2013).
  - Conflicting objectives create difficult tradeoffs.
- Preconditions for adding monetary policy targets (Reis 2013):
  - The target must be measurable in an unambiguous way.
  - Monetary policy must be useful in attaining it.
  - Achieving one objective should not guarantee achieving the other; the target must create a genuine tradeoff requiring a change in policy.
- Measurement and efficacy concerns for distributional objectives:
  - Inequality has many dimensions (consumption, income, wealth), making unambiguous measurement difficult.
  - Monetary policy effects on inequality are argued to be weak.

Note on instruments and loss functions (footnote 1/ in source):
- Even if more instruments cannot be added, assigning a loss function to the central bank that mimics the social loss function (including a distributional term) yields optimal monetary policy conditional on the toolkit and state variables; but “optimal monetary policy” may not equal “optimal policy” (Davig and Gürkaynak).

### Tools outside monetary policy to address distributional impacts
- Distributional impacts can be addressed with targeted policies specifically designed for redistribution and group-specific cyclical impacts:
  - Well-targeted redistributive fiscal policies across households and regions can reduce inequality of disposable incomes and best address group-specific concerns.
  - Training, education reforms, and labor market policies can reduce inequality by addressing skill gaps (IMF 2014 and IMF 2019).23
- Central banks and financial regulators have non-monetary tools with more targeted distributional effects:
  - Policies fostering financial development and financial inclusion can have beneficial distributional consequences (Carstens 2021).
  - Reducing costs of participating in capital markets can decrease wealth inequality (Favilukis 2013).
  - Increased access to payments, savings, and insurance tends to reduce inequality; access to credit can either increase or reduce inequality depending on regulation, supervision, and channel predominance (Cihak et al. 2020).

### Central Bank Communication
- Clear communication helps fill perception gaps and allows monetary policy to remain focused on macroeconomic stabilization.
- Persistently low interest rates can create perceptions of increased inequality because counterfactuals are unobservable, while high asset prices or low returns on savings are visible.
- Bank of England’s Haldane: “When it comes to evidence on the distributional impact of monetary policy, there are wide gaps in understanding and even wider gaps in perception.” (Haldane 2018)
- Three communication priorities for central banks:
  1. Clearly state objectives and mandate, and explain how monetary policy conduct helps achieve those objectives. Clarifying what central banks can and cannot do fosters accountability and trust, reducing political pressure to take on additional objectives.
  2. Emphasize that monetary policy is a blunt tool to address group-specific concerns; spikes in food and energy prices or sector-specific shocks can disproportionately affect certain groups.
  3. Recognize and discuss public concerns about distributional effects; central banks should use their data and research expertise to measure and clarify distributional effects on income and wealth (see Honohan 2019).
- Trends in central bank communications:
  - Mentions of distributional effects or inequality were largely non-existent in ECB speeches prior to 2006 but have increased in recent years (Figure 16; examples include Constâncio 2017, Lane 2019, Schnabel 2020).
  - The ECB has attempted to gauge how inequality shapes the transmission of monetary policy to consumption.24
- Communication strategies to address public concerns:
  - Highlight counterfactuals: explain that inequality would likely have been worse without monetary accommodation, and argue that monetary policy actions are welfare-increasing despite possible distributional effects.25
  - Emphasize the role of structural factors behind secular increases in inequality and long-run declines in interest rates.

### Box 2 (summarized): Central Bank Communication on Distributional Issues — Chile and Sweden
- Chile:
  - Historically high income inequality; monetary policy rate at 0.5 percent in December 2020.
  - October 2019 protests focused on inequality and structural reforms; central bank policies were not a target, possibly due to the success of BCdC’s inflation targeting in stabilizing inflation.
  - BCdC provided liquidity, foreign exchange, and monetary stimulus within its mandate and communicated tightly on price stability.
  - BCdC emphasized that high inflation and unemployment hurt the poor and increase inequality; Governor Mario Marcel (December 5, 2019) provided a counterfactual of high inflation (6 percent) and unemployment (10 percent) leading to inequality rising to levels not seen in 27 years, mostly through rising unemployment.
  - BCdC has reported on labor-market heterogeneities and published a special report in December 2018 using disaggregated data on the Chilean labor market.
- Sweden:
  - One of the lowest income-inequality levels among advanced economies but higher and rising wealth inequality.
  - Riksbank cut its main policy rate to negative territory in February 2015 (where it remained until December 2019) and introduced an asset purchase program.
  - Rising asset prices during persistently negative policy rates and strong preference for equality generated a robust policy debate on distributional impacts.
  - Riksdag Committee on Finance (terms of reference set in 2016) assessed whether monetary policy should take distributional considerations into account and concluded distributional consequences since the GFC have been small and have likely ameliorated income differences through lower unemployment.

### Conclusions and research gaps
- Monetary policy has not been a driver of the rise in inequality in many advanced economies since the 1980s; structural factors driving inequality have also pushed the neutral real interest rate (r*) to very low levels, prompting policy-rate cuts.
- With COVID-19, monetary policy space has narrowed further, including in EMDEs; low r* made conventional countercyclical policy more difficult, and unconventional tools post-GFC have been controversial in part due to debates about their association with increased wealth inequality.
- Literature mostly finds small net effects of monetary policy on inequality:
  - Monetary easing mostly reduces income inequality through the labor-income channel by lowering unemployment and increasing participation among unskilled and younger workers.
  - Effects on the wealth distribution are less clear and depend on household balance-sheet composition.
  - Size of distributional effects depends on relative importance of different channels and country characteristics; simulations in the paper align broadly with these findings but require further calibration across jurisdictions.
- Policy implications:
  - Distributional considerations do not warrant changing central bank mandates; monetary policy should remain focused on macroeconomic stabilization and price stability as a primary goal.
  - An additional mandate to reduce inequality could adversely affect central bank accountability and independence and reduce effectiveness in achieving price stability.
  - Other policies (e.g., fiscal policy) are better suited to address distributional issues and group-specific concerns.
  - Central banks should better model and analyze heterogeneities in income and wealth within existing frameworks to understand transmission differences.
  - Central banks’ key role in the inequality debate is clear communication—via speeches, reports, and outreach—addressing public concerns about distributional effects.
- Areas requiring further study:
  - Distributional effects of unconventional monetary policy, including asset purchases, are less understood than conventional tools.
  - The unprecedented global monetary response to COVID-19 may have different or novel distributional effects.
  - Distributional consequences in EMDEs are largely unknown, given weaker social safety nets and stronger exchange-rate channels.
  - Ongoing reviews of major central banks’ frameworks could have important implications for how central banks address inequality; these are fertile grounds for further research.

*Source: Box 1, “Can Central Banks Handle Multiple Policy Objectives?” (extracted from the supplied IMF content).*

### references highlighted the negative impact

### wpiea2021201-print-pdf - references highlighted the negative impact

### Riksbank communications and distributional effects
- The Sveriges Riksbank Act of 1999 grounds the price stability mandate on the beneficial effects of low inflation in terms of reducing arbitrary redistributions of income between savers and borrowers.
- Between the global financial crisis and 2016, the Swedish central bank did not communicate about the distributional effects of monetary policy.
- Starting in 2016, the Riksbank has included the distributional effects of monetary policy in its communications.
- Since 2017, at the urging of the Riksdag Committee on Finance, the central bank has included a section on the distributional effects of monetary policy in its annual Accounts of Monetary Policy submitted to the Committee.
- The November 2020 Monetary Policy Report included an article on the distributional effects of the Riksbank’s measures.
- The Riksbank’s communications emphasize:
  - Expansionary monetary policy contributes to higher asset prices in the short term, benefitting certain groups more than others.
  - Expansionary monetary policy also leads to higher employment and stronger growth, which equalize incomes and wealth.
  - High and volatile inflation has distributional effects that a monetary policy focused on an inflation target counteracts.
  - The need to reform housing and tax policies to counteract the steep rise in housing prices and the redistribution of wealth between owners and rentals and existing homeowners and first-time buyers.

### Figure 2.2 notes and data coverage
- Figure title: Riksbank Speeches Mentioning Distributional Issues (percent of total).
- Sources: Sveriges Riksbank and IMF staff calculations.
- Notes:
  - The analysis was performed on speeches available at https://www.riksbank.se/en-gb/press-and-published/speeches-and-presentations/, as of 31 Jan 2020.
  - The chart plots the share of speeches by Riksbank officials per year that mention (case-insensitive) any of: inequality, distributional effects, distribution of income, income dispersion or income distribution.

### Selected methodological descriptions and data facts (household surveys)
- Survey of Consumer Finances (SCF):
  - Triennial cross-sectional survey of US families conducted by the Federal Reserve Board.
  - Includes information on families’ balance sheets, pensions, income, and demographic characteristics.
  - The 2016 SCF covers more than 6,500 households.
- Household Finance and Consumption Survey (HFCS):
  - Coordinated by the European Central Bank.
  - Based on 84,000 interviews in 18-euro area countries (Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Malta, Netherlands, Portugal, Slovak Republic, Slovenia, and Spain) as well as Poland and Hungary.
- Asset and liability categorizations used to illustrate decomposition:
  - Assets: real estate primary, real estate secondary, bonds, stocks, liquid, and others.
  - Liabilities: mortgages secured by primary residence, mortgages secured by other residence, and loans.
- Simplifying assumptions for harmonization of SCF and HFCS:
  - For the US, split “other managed assets” equally into bonds and assets.
  - Assume that 79 percent of “quasi-liquid retirement accounts” is liquid assets.
  - For the EU, assume that 80 percent of mutual funds are held in stocks and 20 percent in bonds.
  - For “managed account,” 50 percent allocated to assets while 15 percent to bonds.
- Households ranked and grouped by net wealth distribution into:
  - Bottom 20 percent (the poor).
  - Middle 70 percent (the middle class).
  - Top 10 percent (the rich).

### Appendix 2 — Unhedged Interest Rate Exposures (UREs) benchmark
- Purpose: Microsimulations based on a welfare metric, “Unhedged Interest Rate Exposures” (UREs), capturing households’ exposures to changes in real interest rates; proposed by Auclert (2019) and applied for the United States and Italy. Tzamourani et al. (2019) follow the same approach for euro area households.
- Intuition: A fall in interest rate can redistribute away from households with large positive UREs (e.g., large share of short-term fixed-income investments such as certificates of deposits and/or fixed-rate mortgages) towards households with large negative UREs (e.g., large share of long-term bond investments and/or adjustable-rate mortgage liabilities).
- URE definition (formula preserved exactly):
  - URE = Y − T − C + A − L
  - where Y−T is net annual disposable income, C is a consumption measure that includes durable and non-durable goods as well as interest and principal payments. A and L are the remaining assets and liabilities maturing in a year, respectively.
- Data used:
  - Micro data surveys for the US and the EU: the 2016 SCF for the US and the 2016 HFCS for the EU.
- Caveats explicitly stated:
  - Simulations are based on several simplifying assumptions related to the type and maturity of various assets and liabilities, as in Auclert (2019).
  - For the US, simulations also use the 2016 Survey of Consumer Expenditures to benchmark household consumption in the 2016 SCF.
  - For the EU, household consumption data are directly observable in the HFCS while household disposable income is proxied by gross income due to data availability.
  - Simulations do not include the effect of a prolonged monetary policy easing on asset prices resulting from unrealized capital gains from higher assets prices, which typically benefit the wealthy but might not be immediately reflected in households’ income.
  - Simulations do not account for the effect of higher real estate prices on debt levels and home affordability, particularly at the bottom of the distribution.
  - Results are based on the 2016 survey data (latest available) and potential changes in household income and finances since then can alter the results.
  - Results are subject to measurement error, a common caveat with survey data.
  - Given these caveats, the results are for illustration only, as different assumptions may lead to different results.
- Consistency note: Broad findings align with Auclert’s results for the US and Tzamourani’s results for the EU.

*Content based on wpiea2021201-print-pdf - references highlighted the negative impact*

### APPENDIX 3. BENCHMARKING DISTRIBUTIONAL IMPLICATIONS OF ASSET PRICE

### APPENDIX 3. BENCHMARKING DISTRIBUTIONAL IMPLICATIONS OF ASSET PRICE INCREASES

### Microsimulation methodology
- Microsimulations use the 2016 Survey of Consumer Finances (SCF) for the US and the 2016 Household Finance and Consumption Survey (HFCS) for the EU.
- Household net wealth is computed as the difference between household assets and liabilities using portfolio information from the 2016 SCF (US) and 2016 HFCS (EU).
- Household holdings of bonds, real estate, and equities are scaled by net wealth position.
- Household capital gains on each asset are computed by multiplying the asset-to-net wealth ratio by a hypothetical 10 percent price increase.

### Asset definitions and portfolio mapping
- EU:
  - Housing wealth includes households’ real estate and holdings of mutual funds that predominately invest in real estate.
  - Bond holdings = direct bond holdings + mutual funds predominantly investing in bonds + 79 percent of private pension holdings.
  - Equity holdings = stocks + business wealth + equity mutual funds + 21 percent of private pension holdings.
- US:
  - Real estate = primary residence + other residential property + 16 percent of other assets.
  - Bond holdings = savings bonds + bonds + 20 percent of pooled investment fund + 79 percent of retirement accounts + 15 percent of other assets.
  - Equity holdings = stocks + business wealth + 80 percent of pooled investment funds + 21 percent of other assets.

### Scaling principle and interpretive note
- Capital gains are scaled by net wealth in line with the principle of scale invariance (inequality measure invariant to equi-proportional changes in initial net wealth).
- Illustrative example from text:
  - If net wealth is 1 unit for one bracket and 20 units for another, multiplying both by 100 yields 100 and 2000, implying unchanged inequality.
  - Under a total new wealth of 2079, the top bracket would receive more than 95 percent of total new wealth: top bracket increases by 1980 units, bottom bracket by 99 units.

---

### APPENDIX 4. THE HETEROGENEOUS AGENT NEW KEYNESIAN MODEL OF KMV

### Main features of the KMV HANK model
- The model of Kaplan, Moll and Violante (2018) combines cross-sectional household heterogeneity under incomplete markets with New Keynesian features: price rigidities, monopolistic competition, and a Taylor rule.
- Agents can save in two assets:
  - a low-return short-dated asset; and
  - a high-return long-dated asset subject to transaction costs.
- Calibration targets to US micro data include:
  - net wealth inequality: about 60 percent of households either in debt or holding close to zero cash;
  - inequality in asset holdings: the wealthiest decile holds 88 and 75 percent of all long- and short-dated assets respectively;
  - inequality in labor earnings: 32 percent of labor earnings accrue to the wealthiest decile.

### Mechanism and macro implications
- The model generates a large endogenous aggregate marginal propensity to consume (MPC), making consumption highly responsive to monetary policy shocks through general equilibrium ("indirect") effects.
- Monetary policy loosening:
  - incentivizes immediate consumption increases, especially for agents facing a borrowing constraint;
  - raises demand for goods, increasing labor demand and wages, which reinforce consumption effects.

### Caveats and model limitations
- The portfolio channel in KMV is very strong and typically dominates direct and other channels affecting income inequality.
- This strong portfolio channel arises because wages are assumed flexible but prices are sticky; profits contract after monetary expansion, lowering the price of equity and raising subsequent returns to equity.
- If wages were rigid:
  - households would not see significant labor income rises following monetary expansion, suppressing income and consumption of the poorest the most;
  - profits could become procyclical, so asset returns would not rise as much as under flexible wages;
  - wealth of the rich would be supported but their income suppressed; consumption of the rich would be roughly unaffected due to stable permanent income.
- KMV does not capture all relevant channels; for example:
  - Gornemann et al. (2016) build a HANK model with an employed-unemployed margin (not modeled in KMV) and find looser monetary policy reduces consumption and income inequality—the opposite of KMV’s response.
  - Gornemann et al. (2016) abstract from the liquid/illiquid assets dichotomy emphasized by KMV.

---

### APPENDIX 5. EMPIRICAL STRATEGY AND SPECIFICATION FOR THE RESULTS ON EARNINGS

### Data, sample, and overall approach
- Analysis follows Albrizio et al. (2021) using:
  - the local projection method of Jordà (2005), augmented with the smooth transition regression approach of Granger and Teravistra (1993) for state-dependent responses;
  - US macro and household data at quarterly frequency over 1980–2016 to compare pre- and post-2008 periods.
- Individual-level earnings and wages come from the Current Population Survey (CPS); note that CPS data are top-coded, preventing observation of dynamics at the upper end of the distribution.

### Estimated specification for labor market outcomes
- Estimated equation (paraphrased from text):
  - The left-hand side is 푤푤_{i,t+4} − 푤푤_{i,t} (demeaned earnings change over four quarters).
  - Right-hand side includes:
    - 훼 times control vector 푿_{i,t};
    - summed terms of interest capturing responses to changes in the Federal Fund Rate (FFR) over horizons q;
    - individual fixed effects and an error term.
- 푤푤_{i,t} is demeaned earnings for individual i in quarter t; 퐹𝐹𝐹𝐹𝑈_{t} is the Federal Fund Rate.
- Controls vector 푿_{i,t} includes:
  - a second-order polynomial in age;
  - state fixed effects;
  - occupation fixed effects;
  - industry fixed effects;
  - education fixed effects (10 categories);
  - seasonal fixed effects (12 categories);
  - gender;
  - race;
  - individual relationship to household head (4 categories).
- Parameters of interest are 휸_q; large and significant 휸_q indicate heterogeneous impacts of monetary shocks on labor outcomes.
- Demeaned earnings are defined as log wages_{i,t} − 1/N ∑ log wages_{i′,t}.

### Identification of monetary policy shocks
- Monetary policy shocks (changes in the FFR) are identified using high-frequency variations in interest rate futures within a narrow window around FOMC announcements (following Gertler and Karadi (2015)).
- Futures response to FOMC announcements proxies investors’ surprises; surprises that are proportional to shifts in future prices are used as instruments.
- First stage: regress changes in the FFR on surprises in futures around FOMC announcements.
- Second stage: changes in the FFR instrumented by surprises in futures are used to estimate effects on earnings.

### Key empirical results (reported in text)
- Column (1) of the second-stage regressions reports dynamic effect of an exogenous quarterly change in the FFR on year-on-year (yoy) change in wages:
  - A 100 bps increase in the FFR decreases yoy average wages by 0.6 percent in the first quarter after the monetary policy shock.
  - The same 100 bps increase decreases yoy average wages by 1.1 percent in the fourth quarter after the shock.
- Column (2) introduces interaction terms for the initial level of wages (one year ago) and confirms significant heterogeneity in labor market outcomes in response to a monetary policy shock.
- Because individual-level wages in controls are demeaned using quarter-specific wage averages, the point estimate for coefficients on the change in the FFR is unchanged by construction.
- Robustness note: interacting the change in the FFR with a dummy for people with a wage larger than the (year-specific) average yields similar conclusions.

*Source: wpiea2021201-print-pdf - APPENDIX 3. BENCHMARKING DISTRIBUTIONAL IMPLICATIONS OF ASSET PRICE*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021201-print-pdf.pdf_