## wpiea2025090-print-pdf

## Source details

**Canonical URL:** [wpiea2025090-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025090-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025090-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025090-print-pdf.pdf.json)

---

### Introduction and motivation
- Post-pandemic monetary tightening coincided with historically high mortgage rates and soaring property prices; the rate at which US renters transitioned into home-ownership plummeted by 25 percent on average, between 2021 and 2023.
- Rental prices surged, particularly in cities where the decline in first-time buyers was more severe, intensifying affordability pressures.
- Gap in literature: sparse research on how monetary policy affects rental prices and its distributional effects.
- Objective: Trace how rising mortgage rates interacted with mortgage payment-to-income (MTI) underwriting thresholds to affect transitions from renting to first-time home-ownership and the consequent effects on local rental markets.

### Data and methodological approach
- Data:
  - American Housing Survey (AHS), core national (urban) sample; waves 2019 through 2023; placebo waves 1978 and 1980 used.
  - Urban sample size: 23,000–28,000 observations per wave.
  - Owner-occupied share: about 62–64 percent of units.
  - Reports on 15 MSAs; rich unit-level data on physical characteristics, housing costs, financing, occupant demographics, neighborhood attributes.
- Identification and measurement:
  - First-time home-buyers (FTHB) identified as households switching from renting to owning between survey periods.
  - MTI for FTHB computed as annual mortgage payments divided by annual household income.
  - Imputed MTIs for renters (if buying current unit) via hedonic imputation of house price, imputed mortgage amount, imputed interest rate, assumed mortgage maturity and amortization (details in Appendix B).
  - First-year payment formula preserved exactly: Payment_imt = [Mortgage_imt / 30] + [Mortgage_imt * Rate_imt]; MTI_Rimt = Payment_imt / Annual Income_Rimt.
- Empirical strategies:
  - Regression discontinuity design around MTI = 31 (FHA front-end underwriting threshold).
  - Non-parametric counterfactual: re-price 2021 first-time buyers at 2023 mortgage rates to simulate MTI distribution changes (baseline: uniform 4.2 percentage points increase in 30-year FRM between Q1 2021 and Q1 2023).
  - Demand-adjusted alternative: reduce loan amounts by 2 percent for each percentage point increase in mortgage rates and rescale loan amounts to MSA-level income and house-price changes.
  - Placebo test using 1978–1980 AHS waves (mortgage rates rose by 4.7 percentage points in that episode).

### Principal empirical findings — transitions and underwriting discontinuity
- Regression-discontinuity evidence:
  - Significant discontinuity in probability of switching from renting to buying around MTI = 31.
  - Moving from just below to just above MTI = 31 reduces the probability of becoming a first-time buyer by about 5 percent (Threshold coefficients around -0.052*** to -0.045*** across specifications).
  - Table 2 highlights (AHS 2019–2023):
    - Column (1) sample MTI 0-100: Threshold = -0.052*** (0.007); MTI = 0.006*** (0.000); Threshold*MTI = -0.005*** (0.000); Observations = 23,181; R-squared = 0.107.
    - Column (2) sample MTI 0-50: Threshold = -0.045*** (0.010); MTI = 0.005*** (0.001); Threshold*MTI = -0.006*** (0.001); Observations = 18,268; R-squared = 0.102.
    - Column (3) sample MTI 25-37 (optimal bandwidth): Threshold = -0.052*** (0.017); MTI = 0.011*** (0.003); Observations = 4,768; R-squared = 0.136.
  - Placebo thresholds and other observables remain continuous around MTI = 31; determinants like leverage and interest rates are smooth.

### Counterfactual distributions and volume impacts (2021 re-priced to 2023 rates)
- Re-pricing assumptions and baseline:
  - Uniform increase in interest rates applied: 4.2 percentage points (average change in newly originated 30-year FRM between Q1 2021 and Q1 2023).
  - Hold house value, loan amount, amortization schedule, household income constant in baseline.
- Average effects of re-pricing 2021 buyers at 2023 rates:
  - Average increase in mortgage payments: about 780$ per household/month.
  - MTIs increase by 9 percentage points on average.
- Volume and distributional findings (relative to 2021 volumes):
  - Aggregate decline in first-time buyers between waves: 25 percent.
  - Below MTI = 31:
    - Volume increases modestly by 4 percent relative to the 2021 counterfactual baseline.
    - Demand-adjusted robustness: increase is 5 percent below the limit.
  - Above MTI = 31:
    - Cumulative drop is 29 percent of 2021 levels.
    - Majority of the difference (20 percent) concentrated between MTIs of 31 and 50 (FHA overall DTI upper limit).
    - Loan volumes decline by 9 percent above 50 percent MTI.
  - Between MTIs of 25 and 31: volumes decline modestly by 5 percent (no bunching just below threshold).
- Interpretation:
  - Because MTI underwriting limits remained constant, the reduction in first-time buyers is concentrated above MTI = 31 (29 percent decline above vs. 4 percent increase below), pointing to binding payment-to-income constraints as mortgage rates rose.
- Robustness and placebo:
  - Demand-adjusted counterfactual yields quantitatively similar results.
  - Placebo 1978–1980: mortgage rates rose by 4.7 percentage points; first-time buyers declined from about 3.2 percent to roughly 2.4 percent of the sample, but no sharp drop at MTI = 31 — consistent with less strict MTI enforcement then and supporting the 2021–2023 interpretation.

### Regional heterogeneity and effects on rent price inflation
- Hypothesis: Rising mortgage rates excluded potential buyers, increasing short-run rental demand and rent inflation; rental supply adjusts slowly, generating higher rents.
- Identification: cross-MSA variation in ShareC_m,19−21 (share of constrained home-buyers in an MSA measured ex-ante — 2019 or 2021 buyers with observed MTI < 31 and counterfactual MTI > 31).
- Main rent-price regression (Equation 2):
  - ∆Rent_im,23 = α + β1 ShareC_m,19−21 + γ_N X_N,i,23 + θ_N ∆λ_N,m,23 + ε_im,23
  - ∆Rent_im,23 is log difference in rent for unit i in city m between 2021 and 2023; X_N,i,23 are unit and tenant controls; ∆λ_N,m,23 are city-level time-varying controls including ∆ house prices, ∆ income, ∆ population, share rent-controlled units, share voucher recipients.
- Main quantitative results:
  - ShareC_m,19−21 coefficients (Table 3 Columns (4)-(6)): 0.231***, 0.218***, 0.226***.
  - Change in share of first-time home-buyers (∆FTHB_m,23) coefficients (Columns (1)-(3)): -0.078***, -0.079***, -0.075***.
  - Controls with significant effects include ∆HP_m,23 (0.229*** to 0.275***), ∆Income_m,23 (0.363** to 0.679***), Population_m,23 (0.051** to 0.095***).
  - Observations = 5,911; R-squared range = 0.008 to 0.020.
  - Magnitude interpretation: one standard deviation increase in share of constrained buyers (5 percentage points) increases rent prices on the average unit by about 1 percent more (or 4% of a standard deviation of the dependent variable).
- Supply-side and mechanism evidence:
  - No significant conversion of owner-occupied units into rentals as ShareC increases (ShareC coefficient on Converted Units = 0.040, SE = 0.035).
  - New construction increases (ShareC coefficient on New Units = 0.073**, SE = 0.035).
  - Rental vacancies decline significantly (ShareC coefficient on Rental Vacancies = -0.048***, SE = 0.016), indicating tighter rental markets.
- Heterogeneity across tenants and segments:
  - Lower-income tenants (bottom four deciles) experience sharper rent increases; richer renters (top three deciles) relatively shielded.
  - Effect not driven by voucher prevalence or rent-control incidence among poorer renters.
  - Smaller units (up to 750 square feet) face greater rent growth; ex-ante vacancy rates in size buckets mediate the effect.
  - Conditioning on ex-ante vacancy rates removes differential rent-price effects across income distribution: the mechanism operates through tighter segments where lower-income renters concentrate.
- Pre-trend and robustness:
  - Pre-trend tests (2019–2021) produce small and statistically insignificant coefficients on ShareC, supporting parallel trends.
  - Results robust to alternative controls and using ex-ante (2019) measures of ShareC.

### Key statistics and descriptive summaries
- Table 1 (AHS 2019–2023 descriptive statistics):
  - First-time buyers (percent): 3.6, 3.7, 2.7 for 2019, 2021, 2023 respectively.
  - Share renters (percent): 37, 37, 37 for 2019, 2021, 2023 respectively.
  - Annual Income (2019USD): 107,043; 107,942; 109,507 for 2019, 2021, 2023 respectively.
  - Age head: 51.6; 51.8; 52.4 for 2019, 2021, 2023 respectively.
  - MTI (FTHB, percent): 23; 22; 25 for 2019, 2021, 2023 respectively.
  - Change in rent (percent): 7.6; 6.4; 11.4 for 2019, 2021, 2023 respectively.
  - Observations: 28,147; 27,157; 23,942 for 2019, 2021, 2023 respectively.
- Appendix A / sample-level statistics:
  - Table A1 regression observations: 23,181; 18,268; 4,768 by column.
  - Table A2 weighted observations by year: 1,239,685 (2019); 1,330,602 (2021); 1,031,170 (2023).
  - Table A2 means (first-time buyers): HH income (2019USD) = 135,323; 134,821; 135,178 for 2019, 2021, 2023 respectively; Age = 40; 40; 41; Graduate degree (Percent) = 54; 60; 60; LTV (Percent) = 86; 84; 76; MTI (Percent) = 23; 22; 25.
  - Table A3 constrained vs not constrained: White population (Percent) = 67 (Constrained) vs 73 (Not constrained); HH Gross Income (2019USD) = 108,130 (Constrained) vs 163,001 (Not constrained); Percent of total = 25 (Constrained) vs 75 (Not constrained).
  - Table A6 rental-supply regressions observations and ShareC coefficients: Converted Units Observations = 15,096 (ShareC = 0.040, SE = 0.035); New Units Observations = 8,901 (ShareC = 0.073**, SE = 0.035); Rental Vacancies Observations = 11,778 (ShareC = -0.048***, SE = 0.016).

### Interpretations, policy-relevant insights, and conclusions
- Interpretation of core mechanism:
  - Rising mortgage rates between 2021 and 2023 (average 30-year FRM increase of 4.2 percentage points between Q1 2021 and Q1 2023) raised mortgage payments and MTIs substantially (about 780$ per household/month and 9 percentage points on MTI, on average).
  - The 25 percent aggregate decline in first-time buyer volumes is concentrated above MTI = 31 (29 percent decline above vs. 4 percent increase below), with 20 percent of the difference concentrated between MTIs of 31 and 50 — consistent with binding payment-to-income underwriting constraints.
- Distributional and macro implications:
  - Reduction in first-time buyers increased rental demand in constrained MSAs, lowered vacancies, and raised hedonic rent prices, with the largest increases for smaller units and renters in the bottom four income deciles.
  - Because rents and owner-equivalent rents account for about 33% of the US CPI basket, these channels likely contributed to persistence of elevated headline inflation during the post-pandemic tightening phase.
- Broader contribution:
  - Novel focus on rent-to-own transitions and property-level analysis shows that pre-existing mortgage underwriting standards mediate how monetary policy affects tenure decisions, rental equilibria, and distributional outcomes across renter groups.
- Policy-relevant takeaway:
  - Rising interest rates interacting with MTI underwriting limits can generate second-round distributional effects via rental markets, disproportionately affecting lower-income renters for whom shelter is a larger consumption share.

*Source: wpiea2025090-print-pdf — Introduction, Sections 3–4, and Appendix materials (American Housing Survey based analyses; Working Paper No. WP/2025/090).*

### 1.    Introduction

### wpiea2025090-print-pdf - 1.    Introduction

### Key motivation and overview
- Post-pandemic monetary tightening coincided with historically high mortgage rates and soaring property prices; the rate at which US renters transitioned into home-ownership plummeted by 25 percent on average, between 2021 and 2023.
- Rental prices surged, particularly in cities where the decline in first-time buyers was more severe, intensifying affordability pressures.
- Gap in literature: sparse research on how monetary policy affects rental prices and its distributional effects.
- Objective: Trace how rising mortgage rates interacted with mortgage payment-to-income (MTI) underwriting thresholds to affect transitions from renting to first-time home-ownership and the consequent effects on local rental markets.

### Methodological approach
- Data: Property-level, longitudinal information from the American Housing Survey (AHS), using the core national (urban) sample and five waves: 2019 through 2023 and two waves in 1978 and 1980 for placebo analysis.
- Identification:
  - First-time home-buyers (FTHB) identified as households switching from renting to owning between two survey periods.
  - Household-specific MTI ratio for FTHB calculated as annual mortgage payments divided by annual household income.
  - Imputed MTI ratios for renters if they were to purchase the same property they live in (imputation details referenced in Appendix B).
- Empirical strategies:
  - Regression discontinuity design exploiting a discontinuity in switching probability around an MTI ratio of 31 (the FHA front-end underwriting threshold).
  - Non-parametric counterfactual exercise simulating how the MTI distribution for 2021 first-time buyers would look if borrowers faced 2023 interest rates (holding loan demand and housing choices constant).
  - Placebo test using late 1970s data when mortgage rates rose similarly but underwriting limits were less binding.

### Principal empirical findings
- Transition and underwriting discontinuity:
  - There is a significant discontinuity in the probability of switching from renting to buying around an MTI ratio of 31.
  - The probability of switching from renting to buying is 5 percent lower just above an MTI ratio of 31 than just below, for households with similar demographics living in the same city and year.
- Counterfactual and volume impacts:
  - Differences between observed and counterfactual MTI distributions are negligible below the underwriting MTI limit of 31.
  - Above the MTI threshold of 31, the observed MTI distribution drops by 29 percent relative to the 2021 counterfactual, accounting for the majority of the drop in volumes of first-time buyers between the two periods.
  - Using 2019 as a counterfactual year and adjusting for intensive-margin mortgage demand changes yield qualitatively similar results.
  - Placebo test (late 1970s): supports interpretation that the 2021–2023 mortgage rate rise pushed many renters above underwriting payment-to-income limits, reducing ability to enter home-ownership.
- Rental market effects:
  - The relative prevalence of constrained first-time buyers in a metro area strongly predicts local rental market tightness between 2021 and 2023, including:
    - Declines in vacancies.
    - Significant increases in hedonic rent prices.
  - Housing units with similar characteristics exhibit significantly higher rent price growth the higher the share of ex-ante constrained home-buyers in the city. Ex-ante constrained buyers defined as 2019–2021 first-time buyers with observed MTI below 31 and counterfactual MTI above 31.
  - Effects are robust to controls for city-level time-varying developments (population growth, average incomes, house prices).
  - Price dynamics are particularly pronounced for smaller units occupied by renters in the bottom four deciles of the income distribution, indicating stronger impacts on lower-income renters.

### Relation to existing literature and contributions
- Connects to literature on interaction between monetary policy, debt-servicing limits, and housing tenure decisions (references within source).
- Confirms and extends findings that tighter borrowing conditions reduce home-ownership rates among younger and lower-income buyers.
- Novel contributions:
  - Focus on transitions from renting to first-time home-ownership.
  - Explicit analysis of how tenure decision fluctuations affect rental market equilibria using property-level data.
  - Evidence that pre-existing mortgage underwriting standards are key to understanding monetary policy effects on rental price growth and on distributional outcomes across renter groups.
- Policy-relevant insight: Rising interest rates interacting with MTI underwriting limits can generate second-round distributional effects via rental markets, disproportionately affecting lower-income renters for whom shelter is a larger consumption share.

### Data description and key statistics
- AHS longitudinal (urban) sample features:
  - Urban sample size: 23,000–28,000 observations per wave.
  - Owner-occupied share: about 62–64 percent of units.
  - Reports on 15 MSAs (list of MSAs contained in source).
  - Rich unit-level data: physical characteristics, housing costs (rent, mortgage, utilities, maintenance), financing details (mortgage size, interest rate), occupant demographics, neighborhood attributes.
- Table 1: Descriptive statistics by survey year (American Housing Survey, 2019–2023)
  - First-time buyers (percent): 3.6, 3.7, 2.7 for 2019, 2021, 2023 respectively.
  - Share renters (percent): 37, 37, 37 for 2019, 2021, 2023 respectively.
  - Annual Income (2019USD): 107,043; 107,942; 109,507 for 2019, 2021, 2023 respectively.
  - Age head: 51.6; 51.8; 52.4 for 2019, 2021, 2023 respectively.
  - MTI (FTHB, percent): 23; 22; 25 for 2019, 2021, 2023 respectively.
  - Change in rent (percent): 7.6; 6.4; 11.4 for 2019, 2021, 2023 respectively.
  - Observations: 28,147; 27,157; 23,942 for 2019, 2021, 2023 respectively.
  - Notes: Change in rent is average percent change relative to the previous wave (2 years before). Household income expressed in 2019USD. Observations weighted by frequency weights.
- Imputation approach for renters’ potential MTIs:
  - Predict sale price of rented property using hedonic pricing model based on similar houses in same city and year.
  - Impute mortgage amount using loan amounts of first-time home-buyers with similar demographics and property prices in same city and year.
  - Predict interest rate faced by a renter based on mortgage rates paid by first-time buyers purchasing similar properties and borrowing similar amounts in same city and year.
  - Assume loan maturity structure as described in Appendix B (AHS does not contain mortgage maturity information).
  - Result: imputed potential annual mortgage payments for renters, yielding imputed MTIs when rescaled by annual income.
  - These imputed MTIs are used to estimate the discontinuity in propensity to switch from renting to owning around underwriting thresholds.

*Source: wpiea2025090-print-pdf - 1.    Introduction — https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025090-print-pdf.pdf*

### 3.    Interest rates and first-time home buyers

### 3.    Interest rates and first-time home buyers

### Interaction of mortgage rates and underwriting standards (overview)
- The section studies how changes in mortgage rates interact with underwriting standards to determine the share of renters who can become home-owners.
- Key premise: the share of mortgage payments to income, or front-end ratio (MTI), is a key parameter in determining the probability of becoming a first-time home-buyer because it determines eligibility for a mortgage.
- Underwriting limits remained unchanged after 2021 and, when combined with rising interest rates, excluded a significant portion of potential buyers from the owner-occupied housing market in 2023.

### First-time buyers and mortgage underwriting standards: regression discontinuity
- FHA underwriting parameters cited:
  - Overall DTI ratios typically not to exceed 43-50 percent of the borrowers’ monthly income.
  - Front-end ratio (MTI) for FHA loans is not supposed to exceed 31 percent of the borrowers’ combined annual income.
  - FHA mortgages with front-end ratios exceeding this threshold generally require manual underwriting and potentially compensating factors (higher collateral, accumulated savings, or previous credit history).
- Empirical approach:
  - A discontinuity in the distribution of observed MTIs for first-time home-buyers is evident around the 31 threshold, with bunching just below (Figure2).
  - MTIs for renters are imputed (the MTI they would face if purchasing the same home they currently live in) and compared with observed MTIs for first-time buyers to estimate changes in the probability of switching from renting to buying in a close bandwidth around the 31 MTI threshold, holding observable characteristics constant.
  - The regression-discontinuity design is formalized in Equation (1), where FTHB is a dummy (1 if unit is occupied by a first-time buyer, 0 if occupied by a renter), MTI is mortgage-to-income ratio (observed for buyers, imputed for renters), Ximt is a vector of household and unit-specific controls, and θmt are MSA-by-year fixed effects.
- Key empirical findings:
  - The probability of becoming a first-time buyer shows a clear discontinuity around the underwriting MTI threshold of 31 (Figure3), while other observed characteristics remain continuous around the threshold.
  - Placebo tests show no relevant discontinuities around other thresholds.
  - Determinants of MTI ratios such as leverage ratios and interest rates remain fairly smooth around the threshold.
- Table 2: Probability of becoming a first-time buyer as a function of MTI limits (AHS 2019-2023)
  - Column (1): sample MTI 0-100
    - Threshold: -0.052*** (0.007)
    - MTI: 0.006*** (0.000)
    - Threshold*MTI: -0.005*** (0.000)
    - Observations: 23,181
    - R-squared: 0.107
    - MSA by Year FE: Yes
    - HH and Unit controls: Yes
  - Column (2): sample MTI 0-50
    - Threshold: -0.045*** (0.010)
    - MTI: 0.005*** (0.001)
    - Threshold*MTI: -0.006*** (0.001)
    - Observations: 18,268
    - R-squared: 0.102
    - MSA by Year FE: Yes
    - HH and Unit controls: Yes
  - Column (3): sample MTI 25-37 (optimal bandwidth)
    - Threshold: -0.052*** (0.017)
    - MTI: 0.011*** (0.003)
    - Threshold*MTI: -0.000 (0.005)
    - Observations: 4,768
    - R-squared: 0.136
    - MSA by Year FE: Yes
    - HH and Unit controls: Yes
  - Interpretation:
    - Moving from just below to just above the MTI limit of 31 reduces the probability of finding a first-time home buyer in the sample by about 5 percent.
    - Results are robust to alternative specifications (including quadratic polynomial) and robust bias-corrected confidence intervals.

### Measuring the effect of interest rates on purchasing behavior: counterfactual distributions
- Sample and strategy:
  - Restrict sample to first-time home-buyers observed between 2019 and 2023.
  - Use older waves to compare volumes in a low-rate environment (2019-2021) to volumes in a high-rate environment (2023).
  - Build counterfactual distributions simulating what would have happened to volumes of home-buyers in the absence of credit supply constraints, following Bosshardt et al. (2024).
- Observed volume changes:
  - The number of first-time buyers declined significantly between 2021 and 2023: the frequency-weighted drop is about 300,000 units, roughly corresponding to the 25% decline reported in Table 1.
  - The decline is broadly uniform over the MTI distribution, with the only significant spike in volumes compared to 2021 concentrated at 50 percent (the upper bound of the underwriting comprehensive DTI limit) (Figure4).
- Use of counterfactual:
  - The 2021 and 2023 MTI distributions appear similar aside from volume differences and both exhibit a sharp drop in volumes around the front-end ratio of 31 percent.
  - The 2021 distribution of first-time buyers can be used to build a counterfactual distribution for 2023 to help uncover what proportion of the missing mass of home buyers can be attributed to rising mortgage rates pushing potential buyers above binding payment-to-income limits.

*Italic: Source: wpiea2025090-print-pdf - 3.    Interest rates and first-time home buyers*

### Appendix Table A2shows that on average, first time home-buyers in2021and2023are very similar

### Appendix: Counterfactual MTI Distributions and First-Time Home-Buyers (2021–2023)

### Data and methodological setup
- Data source: American Housing Survey (AHS), waves for 2021 and 2023 (also placebo AHS waves 1978 and 1980).
- Construction of counterfactual:
  - Re-calculate individual MTI ratios of 2021 first-time buyers assuming mortgage interest rates equal those prevalent in 2023.
  - Apply a uniform increase in interest rates corresponding to the average change in rates on newly originated 30-year FRM between the first quarter of 2021 and the first quarter of 2023: a 4.2 percentage points increase.
  - Apply this average increase to the borrower-specific mortgage rate as reported in the survey; where individual interest-rate data are missing, apply the average percent increase in MTI ratios between observed and counterfactual across first-time buyers in a given wave.
  - Hold other parameters constant: house value, loan amount, amortization schedule, household income (baseline extensive-margin approach).
- Alternative (demand-adjusted) specification:
  - Reduce loan amounts by 2 percent for each percentage point increase in mortgage rates (elasticities from DeFusco and Paciorek (2017)).
  - Further adjust mortgage amounts to changes in average incomes and house prices recorded in each MSA between waves (regressions of log loan amounts on log income and log house value, conditional on MSA and year fixed effects; average elasticities multiplied by average change in incomes and house prices; individual loan amounts rescaled and divided by household income).

### Main counterfactual results (2021 re-priced to 2023 mortgage rates)
- Average changes from observed 2021 to counterfactual:
  - Average increase in mortgage payments: about 780$ per household/month.
  - MTIs increase by 9 percentage points on average.
- Distributional findings:
  - The counterfactual 2021 distribution (2021 buyers priced at 2023 rates) is smoother around MTI thresholds by construction.
  - The observed 2023 distribution compared to the counterfactual reveals where first-time buyer volumes declined between 2021 and 2023.
- Key magnitudes (relative to 2021 volumes):
  - Aggregate decline in first-time buyers between the two waves: 25 percent (as presented in Table 1).
  - Below the MTI limit of 31:
    - Volume of first-time buyers increases modestly by 4 percent relative to 2021 levels (counterfactual baseline).
    - In demand-adjusted robustness: increase is 5 percent below the limit.
  - Above the MTI limit of 31:
    - Cumulative drop is 29 percent of 2021 levels.
    - The majority of the difference (20 percent) is concentrated between MTIs of 31 and 50 (the FHA upper limit for overall DTI ratios).
    - Loan volumes decline by 9 percent above 50 percent MTI.
  - Between MTIs of 25 and 31:
    - Volumes decline modestly by 5 percent, indicating no bunching just below the threshold to avoid the MTI limit.
- Interpretation:
  - Because MTI underwriting limits remain constant over this period, and the reduction in the number of first-time buyers relative to 2021 occurs almost entirely above the MTI threshold of 31 (29 percent reduction above, compared to 4 percent increase below), the drop in volumes can largely be ascribed to binding payment-to-income constraints as mortgage rates rose.
  - The counterfactual does not alter the overall volume of 2021 buyers but changes their placement in the MTI distribution; the integral of differences across the MTI distribution equals the overall 25 percent decline.

### Robustness checks and placebo tests
- Demand-adjusted counterfactual (accounts for intensive margin responses and MSA-level income/price changes):
  - Results remain quantitatively similar to the baseline counterfactual.
  - Below the MTI limit, 2023 volumes increase by 5 percent relative to the 2021 demand-adjusted counterfactual; they decline by 29 percent above the limit.
- Placebo test: 1978–1980 AHS waves
  - Context:
    - Average mortgage rates increased from 9.3 percent in the first semester of 1978 to 14 percent in early 1980, a 4.7 percentage points increase—similar in magnitude to the 2021–2023 increase.
    - Total number of first-time home-buyers declined from about 3.2 percent of the sample to roughly 2.4 percent (a decline similar to that observed between 2021 and 2023 in percentage terms).
  - Observed pattern in 1978–1980:
    - Distribution of MTIs for first-time buyers shifts decisively to the right in 1980 relative to 1978.
    - No sharp drop in volumes at the MTI threshold of 31 in either year, consistent with less strict enforcement of MTI limits at the time (manual underwriting predominated; FHA and Agencies applied limits less strictly).
  - Counterfactual exercise for 1978 re-priced to 1980 rates:
    - The counterfactual 1978 distribution shifts right and closely resembles the observed 1980 distribution.
    - The relative decline in loan volumes between the two waves occurs mostly below the MTI threshold of 31 (contrast with 2021–2023 results).
  - Interpretation of placebo:
    - The contrast between the 1978–1980 placebo and the 2021–2023 counterfactual supports the view that the 2021–2023 decline in first-time buyers reflects the interaction between rising mortgage rates and binding MTI underwriting limits, rather than a mechanical artifact of the simulation.
    - In 1978–1980, the more uniform drop in volumes is consistent with reductions in loan demand due to rising rates and/or credit supply constraints binding on other margins than payment-to-income limits.

### Summary conclusion from the analysis
- Rising mortgage rates between 2021 and 2023 (average 30-year FRM increase of 4.2 percentage points between Q1 2021 and Q1 2023) raised mortgage payments and MTIs substantially (about 780$ per household/month and 9 percentage points on MTI, on average).
- The decline in first-time buyer volumes between 2021 and 2023 (aggregate 25 percent) is concentrated above the MTI underwriting threshold of 31 (29 percent decline above vs. 4 percent increase below), with 20 percent of the difference concentrated between MTIs of 31 and 50.
- These distributional patterns, robustness to demand-adjusted specifications, and the contrasting placebo results from 1978–1980 point to binding payment-to-income constraints as a primary driver of the reduction in first-time buyer activity in 2023.

*Source: Appendix material from wpiea2025090-print-pdf (AHS-based counterfactual analyses, 2021–2023 and 1978–1980).*

### 4.    Regional heterogeneity and effects on rent price inflation

### 4.    Regional heterogeneity and effects on rent price inflation

### Hypothesis and mechanism
- Hypothesis: Rising mortgage rates excluded potential homebuyers from the owner-occupied market, causing these households to rent for longer and increasing short-run rental demand and rent inflation.
- Mechanism emphasized: Rental supply is unlikely to adjust quickly to relative increases in demand, producing higher rent prices for all tenants.

### Empirical strategy
- Identification: Exploit regional variation in the proportion of potential buyers facing binding borrowing constraints, following approaches similar to Acharya et al., 2022 and Bosshardt et al., 2024.
- Key variable: ShareC_m,19−21 — the share of constrained home-buyers in a given MSA, defined as 2019 or 2021 buyers who would have been pushed above the underwriting MTI threshold by the increase in mortgage rates observed between their survey year and 2023 (observed MTI below 31 and counterfactual MTI above 31).
- Main regression (Equation 2): 
  - ∆Rent_im,23 = α + β1 ShareC_m,19−21 + γ_N X_N,i,23 + θ_N ∆λ_N,m,23 + ε_im,23
  - ∆Rent_im,23 is the log difference in rent price for unit i in city m between 2021 and 2023.
  - X_N,i,23: unit and tenant characteristics (size, n.bedrooms, n.bathrooms, type of build, building age; age and income decile of occupants; number of people in unit).
  - ∆λ_N,m,23: city-level time-varying controls (log change in house prices, income levels and population between 2021 and 2023; average incomes and population levels in 2023), plus city-level housing policy indicators (share of rent-controlled units, share of housing voucher recipients).
- Design interpreted as difference-in-differences comparing changes in rental prices between 2021 and 2023 for similar units across MSAs.

### Main quantitative findings
- Cross-sectional correlation: The change in the share of first-time buyers across cities is negatively correlated with rent price growth between 2021 and 2023.
- Table 3 results summary:
  - Change in the share of first-time home-buyers (∆FT HB_m,23) coefficients in Columns (1)-(3): -0.078***, -0.079***, -0.075*** (standard errors reported in table).
  - ShareC_m,19−21 coefficients in Columns (4)-(6): 0.231***, 0.218***, 0.226*** (standard errors reported in table).
  - Other controls with significant coefficients include ∆HP_m,23 (0.229*** to 0.275***) and ∆Income_m,23 (0.363** to 0.679***), Population_m,23 (0.051** to 0.095***).
  - Observations: 5,911; R-squared range: 0.008 to 0.020 across specifications.
- Magnitude: For one standard deviation increase in the share of constrained buyers (5percentage points), rent prices on the average unit increase by about 1percent more (or 4% of a standard deviation of the dependent variable).
- Robustness:
  - Results are robust to progressively more comprehensive sets of controls and to including the history of unit-level rent price developments (Table A4).
  - Pre-trend test (re-estimating Equation 2 on 2019–2021 change in rents): coefficients on share of constrained buyers are small and statistically insignificant (Table A5), supporting parallel trends.
  - Using the share measured ex-ante (2019) yields similar robustness.

### Supply-side dynamics and mechanism evidence
- No significant conversion of owner-occupied units into rental units as the share of constrained buyers increases (Table A6, column 1) — consistent with segmentation between renter and owner-occupied markets.
- New construction increases in response to stronger demand pressures (Table A6, column 2).
- Rental vacancies decline significantly (Table A6, column 3), indicating rental market tightness.
- Interpretation: Rent price inflation likely originates from increasing scarcity of rental units in the short run as tighter credit conditions raise rental demand and supply cannot fully adjust.

### Heterogeneity across tenants and market segments
- Income-based heterogeneity:
  - Tenants in the bottom four deciles of the city-year income distribution experience sharper rent increases in response to an increase in the share of constrained first-time buyers.
  - Richer renters, particularly those in the top three deciles, are relatively shielded (Table A7, column 1).
  - The effect is not driven by higher prevalence of voucher recipients among poorer renters (column 2) or incidence of rent-controlled units in this segment (column 3).
  - Excluding newly leased units or short-tenancy units does not materially alter results (columns 4 and 5).
- Segment explanation:
  - Lower-income tenants predominantly rent smaller units (up to 750 square feet) which exhibited significantly lower vacancy rates than average already in 2021 (Figure A13).
  - Table A8: Rent prices grow more for smaller units in response to the same prevalence of constrained home-buyers (column 1); other unit characteristics correlated with size matter less (columns 2 and 3).
  - Column 4 of Table A8: Ex-ante vacancy rates in a given size bucket are key — greater slack reduces the rent response to constrained buyers.
  - Conditioning on ex-ante vacancy rates in the size segment removes the differential rent-price effects across the income distribution (column 5).
- Interpretation: Poorer tenants faced higher exposure to rent increases because they concentrate in tighter rental segments; this implies second-round distributional consequences of monetary policy via the rental market, amplified by shelter typically accounting for a larger share of consumption among poorer tenants.

### Conclusions (key takeaways)
- First takeaway: The large decline in first-time home-buyers between 2021 and 2023 can in great part be attributed to rising mortgage rates, which pushed many potential buyers above pre-existing mortgage-to-income underwriting thresholds.
- Second takeaway: The reduction in first-time home-buyers increased pressure on some segments of the rental market, driving a steep increase in average rent prices; rent price inflation was more pronounced for units occupied by lower-income renters, highlighting potential unintended distributional consequences of monetary policy.
- Broader implication: Given that rents and owner-equivalent rents currently account for about 33% of the US Consumer Price Index (CPI) basket, the channels identified here likely contributed to the persistence of elevated headline inflation during the post-pandemic tightening phase.

*Source: IMF Working Paper — chapter "4.    Regional heterogeneity and effects on rent price inflation"*

### References

### References

### Key cited literature
- Extensive citations on monetary policy, housing, and distributional effects, including works by Abramson et al. (2025); Acharya et al. (2022); Acolin et al. (2016); Adams et al. (2024); Albuquerque et al. (2025); Ampudia et al. (2018); Andersen et al. (2023); Bartscher et al. (2022); Bhutta and Ringo (2021); Bolhuis et al. (2022); Bosshardt et al. (2024); Calonico et al. (2014); Castellanos et al. (2024); Cloyne et al. (2019); Coibion et al. (2017); De Stefani and Mano (2025); Defusco et al. (2020); DeFusco and Paciorek (2017); Dennis (1981); Dias and Duarte (2019, 2022); Gete and Reher (2018); Greenwald (2018); Groiss and Syrichas (2025); Jaravel (2024); Mabille (2023); Molloy (2024); Ringo (2024).

### Empirical and methodological anchors (from cited and appendix content)
- American Housing Survey waves used: 2019-2023; specific empirical analyses cite AHS 2019-2023 and AHS 2021-2023 samples.
- Regression discontinuity specifications include quadratic polynomials and MSA-by-year fixed effects.
- MTI thresholds and cutoffs referenced: MTI rounded to integer buckets; underwriting threshold highlighted at MTI = 31.
- Counterfactual scenarios assume a 3.5 percentage points increase in mortgage rates (average increase between the first semester of 2019 and the first semester of 2023).
- Mortgage assumptions for imputed renter MTIs:
  - Mortgage maturity: 30-year (assumed).
  - Constant amortization schedule assumed for simplicity in first-year MTI calculations.
  - Imputed down-payment assumption in Figure B2: 20 percent.
- Imputation approach mirrors Owner Equivalent Rent hedonic methods: hedonic regression of HousePrice_imt on large vector X_imt, year interactions (λ_t * X_imt), and MSA-by-year fixed effects φ_mt.
- Mortgage amount imputed via regression including imputed HousePrice_imt and unit controls.
- Interest rate imputed via regression on imputed HousePrice_imt, Mortgage_imt, unit controls, interactions, and φ_mt.
- First-year annual mortgage payment formula preserved exactly:
  - Payment_imt = [Mortgage_imt / 30] + [Mortgage_imt * Rate_imt]
  - MTI_Rimt = Payment_imt / Annual Income_Rimt

### Key empirical sample sizes and summary statistics (as reported in tables)
- Table A1 regression observations: 23,181; 18,268; 4,768 (by column).
- Table A2 observations (weighted) by year: 1,239,685 (2019); 1,330,602 (2021); 1,031,170 (2023).
- Table A2 means (first-time buyers by survey year):
  - HH income (2019USD): 135,323 (2019), 134,821 (2021), 135,178 (2023).
  - Age: 40 (2019), 40 (2021), 41 (2023).
  - Graduate degree (Percent): 54 (2019), 60 (2021), 60 (2023).
  - White population (Percent): 70 (2019), 69 (2021), 69 (2023).
  - LTV (Percent): 86 (2019), 84 (2021), 76 (2023).
  - MTI (Percent): 23 (2019), 22 (2021), 25 (2023).
- Table A3 demographic comparison (2019-2021):
  - White population (Percent): 67 (Constrained) vs 73 (Not constrained).
  - HH Gross Income (2019USD): 108,130 (Constrained) vs 163,001 (Not constrained).
  - Head graduate (Percent): 55 (Constrained) vs 65 (Not constrained).
  - LTV (Percent): 88 (Constrained) vs 78 (Not constrained).
  - Percent of total: 25 (Constrained) vs 75 (Not constrained).
- Table A4 regression sample: Observations = 3,975; R-squared values: 0.119, 0.124, 0.132.
  - Coefficient on ShareC_m,19−21: 0.217***; standard errors (0.070), (0.069), (0.069) across specifications.
  - Coefficient on ∆Rent_i,21: -0.327*** with SEs (0.019), (0.020), (0.019).
- Table A5 pre-trends sample: Observations = 6,614; R-squared range: 0.006–0.013.
- Table A6 rental-supply regressions:
  - Observations: Converted Units 15,096; New Units 8,901; Rental Vacancies 11,778.
  - ShareC_m,19−21 coefficients: 0.040 (Converted Units), 0.073** (New Units), -0.048*** (Rental Vacancies) with SEs (0.035), (0.035), (0.016).
- Heterogeneity tables:
  - Table A7 observations range: 5,911; 4,760; 5,613; 5,386; 3,778. R-squared around 0.021–0.026.
  - Table A8 observations: 5,911; R-squared around 0.020–0.022.
- Figure and appendix numeric markers and labels preserved: figures labeled A1–A13, B1–B2; appendix pages indicate page numbers referenced (e.g., pages 28–21 in appendix sequence).

### Appendix A: Figures — concise content elements
- Figure A1: Housing affordability over time—series rescaled to 2010q1 = 100; Mortgage rates are averages on newly originated 30-year FRMs; series plotted for Mortgage Rate, Median Income, Rent Index, House Price Index, Affordability across quarters 2010q1 to 2025q1.
- Figure A2: Change in average rents by city, 2021-2023; left panel change in average rent computed as frequency-weighted log difference between 2021 and 2023; right panel correlation between city-level change in rent prices and change in share of first-time home-buyers between 2021 and 2023.
- Figures A3–A13: Present volumes by unit size, placebo tests by MTI, interest rates and LTVs by MTI, observed vs counterfactual MTI distributions, demand-adjusted counterfactuals (methodology following Bosshardt et al. (2024) assumptions), city-level shares of constrained home-buyers, correlations with city demographics and outcomes, income deciles and vacancy rates by unit size. Specific methodological notes:
  - MTIs are rounded up to the closest integer in figures.
  - Frequency or observation weights applied across figures unless otherwise noted.

### Appendix B: Imputing Renter’s MTI — methodological summary
- Goal: Impute renters’ MTI by mirroring Owner Equivalent Rent imputation; use hedonic regressions estimated on owner-occupied sample to impute HousePrice_imt for renter units with same characteristics X_imt in same MSA and year.
- Hedonic house price regression specification:
  - HousePrice_imt = α + Σ_n γ_n X_imt + Σ_n θ_n (λ_t * X_imt) + φ_mt + ε_imt
  - X_imt includes: building type (apartment/house); age and age^2; square feet; fixed effects for number of bedrooms and bathrooms; indicator for significant neighborhood crime.
  - Year interactions (λ_t * X_imt) allow time-varying characteristics effects; φ_mt are MSA-by-year fixed effects.
- Mortgage amount imputation:
  - Mortgage_imt = α + β1 HousePrice_imt + Σ_n γ_n X_imt + Σ_n θ_n (λ_t * X_imt) + φ_mt + ε_imt
- Interest rate imputation:
  - Rate_imt = α + β1 HousePrice_imt + β2 Mortgage_imt + Σ_n γ_n X_imt + Σ_n θ_n (λ_t * X_imt) + φ_mt + ε_imt
- Payment and MTI calculations (first year):
  - Payment_imt = [Mortgage_imt / 30] + [Mortgage_imt * Rate_imt]
  - MTI_Rimt = Payment_imt / Annual Income_Rimt
- Empirical patterns reported:
  - Renters’ imputed MTIs are overall higher than first-time buyers’ observed MTIs; renters’ MTI distribution shifted to the right in 2023 relative to 2021, consistent with higher interest rates and rising house prices.
  - Visible spike in renters with imputed MTI at or just above the front-end threshold in 2023.
  - Figure B2 compares imputed mortgage-to-income vs rent-to-income by city for waves 2021 and 2023; noted city examples: DC, Miami (buying often cheaper in 2021), New York (costs comparable in 2021), San Francisco (buying typically more expensive in 2021). In 2023, higher mortgage rates and price changes increased buying costs across many cities.
- Assumptions and simplifications explicitly stated:
  - Mortgage maturity assumed 30-year.
  - Constant amortization schedule assumed for first-year MTI calculations (simplification justified as most mortgages issued with fixed nominal payments at least for the first year).
  - Imputed down-payment assumption in Figure B2: 20 percent.

*Missing Home-Buyers and Rent Inflation: The Role of Interest Rates and Mortgage Underwriting Standards — Working Paper No. WP/2025/090*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025090-print-pdf.pdf_