## wpiea2020212-print-pdf

## Source details

**Canonical URL:** [wpiea2020212-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020212-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020212-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020212-print-pdf.pdf.json)

---

### Key findings
- On March 3, 2020, the Fed cut the benchmark rate by 50 basis points to 1-1.25 percent; on March 15, 2020, the Fed cut the rate to 0-0.25 percent.
- On June 10, 2020, the “dot plot” indicated that the rate would be kept in the range of 0-0.25 percent through 2022 and Chairman Powell announced that the purchases of MBS would continue, triggering the 30-year fixed mortgage rate to hit a record low of 2.97%.
- On July 16, the rate dropped again and marked the seventh record low this year.
- According to the Case-Shiller National Home Price Index, real house prices rose 45 percent from February 2012 through May 2020.
- The year-on-year growth rate of the median price per square foot in the recent four months (April-August 2020) has accelerated faster than any four-month period in the lead-up to the 2007-09 global financial crisis.
- The acceleration in median price per square foot in April-August 2020 is observed across urban, suburban, and rural areas.

### Mechanisms and interpretation
- Downward pressure on housing prices:
  - COVID-19 is an unprecedented negative shock to households’ incomes (both current and expected incomes due to massive layoffs), leading to significant downward pressure on housing prices via depressed demand.
- Upward pressure on housing prices:
  - (a) Higher housing demand induced by the central bank’s unprecedented monetary easing, part speculative housing demand (“fear of missing out”) and part demand from previously liquidity-constrained low-income households (who were borrowing up to the debt-service limits before the monetary easing).
  - (b) Lower housing supply due to stay-at-home orders and sellers’ reluctance to supply houses out of COVID infection fears.
- Net outcome remains unclear a priori because these forces operate in opposite directions.

### Key empirical findings (stylized facts)
- Fact #1 (aggregate price rebound)
  - After a temporary slow-down in March and April 2020, the median growth rates (year-on-year and month-on-month) of the median price per sqft bounced back and exceeded pre-COVID levels.
  - Year-on-year growth: 5.0 in June 2020, 5.7 in July, and 6.6 in August, all higher than the 4.8 in February 2020.
  - Some zip codes experienced year-on-year growth of more than 30 percent in each month since March 2020.
- Fact #2 (acceleration vs pre-GFC)
  - The year-on-year growth rates of median price per sqft accelerated by 2.6 percentage points from 4.0 percent in April 2020 to 6.6 percent in August 2020 (April–August 2020).
  - By comparison, the fastest four-month pre-GFC acceleration was 1.8 percentage points, from 8.1 percent year-on-year to 9.9 percent (February 2004 to June 2004).
  - After the interest rate cut in March 2020, the price-growth acceleration entered the steady phase in one month; after the November 2002 cut it took about eight months.
- Fact #3 (supply decline amplified)
  - Median year-on-year growth rate of active listings was negative since July 2019 and dropped sharply since March 2020.
  - Month-on-month growth rates after April 2020 became more negative, contributing to rapid price growth.
- Fact #4 (demand hotness surge)
  - Median views per property rose by 99 percent in July 2020 and 121 percent in August 2020 (year-on-year), record highs since data began in July 2017.
- Fact #5 (structural break in demand response to mortgage rate)
  - Zip code-level panel regressions show a structural break in March 2020 in the elasticity of housing demand to mortgage rates (Chow test rejects no-break).
  - From July 2017 to February 2020 the 30-year fixed mortgage rate was negatively associated with year-on-year growth of online views but statistically insignificant.
  - From March 2020 onwards: as the 30-year fixed mortgage rate drops by 1 percentage point, the associated year-on-year growth of online views per property increases by about 267 percentage points (Table 3 coefficient: -266.940*** in column (3); pooled column (6): -264.443***).
  - Falsification tests using March 2019 as breakpoint do not show this pattern.
- Fact #6 (price growth concentrated in lower-to-middle income zip codes)
  - Nonparametric estimates (March–August 2020 average per zip code) show year-on-year price growth decreases with median family income: about 15 percent in zip codes with median family income of $25,000 down to close to 0 percent in zip codes with $225,000, then increasing again for very rich zip codes.
- Fact #7 (U-shaped demand response across income)
  - Year-on-year growth of views per property displays a U-shaped relationship with median family income (March–August 2020): high at low incomes, declines through the middle, then rises at the top end.
  - Example: from around 58 percent in zip codes with median family income of $45,000 to about 37 percent in zip codes with $200,000, then rising to about 50 percent at the very top (full-sample description).
- Fact #8 (broad geographic reach)
  - Changes in housing price, demand, and supply since April 2020 are broad-based across metropolitan, micropolitan, small-town, and rural areas with similar magnitudes.
  - Examples:
    - Metropolitan and micropolitan: median housing price growth (y-o-y) accelerated from 4–4.5 percent in April 2020 to about 7 percent in August 2020 (acceleration of 3–3.5 percentage points).
    - Small towns: 3.5 percent to 5.3 percent (1.8 percentage point acceleration).
    - Rural areas: 3.5 percent to 6.0 percent (3.5 percentage point acceleration).
    - Demand (views): metropolitan/micropolitan/small towns accelerated from about 0 percent in April 2020 to about 125 percent in August 2020; rural accelerated to 178 percent in August.
    - Supply (active listings): metropolitan from -18 percent in April 2020 to -28 percent in August 2020; micropolitan and rural from about -10 percent to about -30 percent; small towns from -11 percent to -24 percent.

### Robustness and additional controls
- Composition control
  - Results hold after restricting to zip codes with year-on-year median growth rates of median sqft between -5 and 5 percent (similar house sizes sample).
  - Example: With similar house sizes, the four-month acceleration after Fed easing was 2.4 percentage points (from 3.9 percent in April 2020 to 6.3 percent in August 2020).
- Mobility controls (Google mobility data)
  - Merged county-level Google mobility (workplace, residential, overall) with zip code housing data (Feb–Aug 2020).
  - Workplace mobility positively correlated with online views; residential mobility negatively correlated.
  - In fixed-effect regressions (Table 8), FRM_30Yr coefficients remain large and negative (e.g., -264.645*** in column (1)).
  - After adding normalized PMI year-on-year growth (Table 9), mortgage-rate coefficients become even larger in magnitude (e.g., FRM_30Yr -313.395*** to -340.559*** across specifications), while workplace mobility remains positively associated and residential mobility negatively associated with views.
  - Interpretation: mobility patterns during stay-at-home orders tended to reduce online views, reinforcing that the observed demand surge is linked to low mortgage rates rather than only to increased online activity during lockdown.
- Omitted variable considerations
  - Local COVID case growth, unemployment, and economic disruption variables (if controlled) would likely require an even stronger interest-rate response to explain the observed increase in views (direction of omitted variable bias uncertain for other omitted factors like subjective valuation of owning a house).

### Behavioral scenarios consistent with findings
- Scenario (a): Fear of Missing Out (FOMO)
  - Potential homebuyers responded strongly to historically low mortgage rates, rushing to buy to avoid missing low financing costs and rising prices.
- Scenario (b): COVID-induced fundamental preference/behavior changes
  - Stay-at-home orders and work-from-home expectations raised the value of owning larger homes, increased forced saving and down-payment ability, and shifted search behavior to online channels.
- These scenarios have different implications for sustainability and financial stability; further empirical work is needed to disentangle them.

### Key regression coefficients and summary statistics (exact values preserved)
- Table 3 (panel regressions, full sample): FRM_30Yr coefficients
  - FE All: -22.927* (p-value 0.056)
  - FE Until Feb: -1.007 (p-value 0.807)
  - FE Since March: -266.940*** (p-value 0.006)
  - Pool All: -22.235* (p-value 0.059)
  - Pool Until Feb: -0.703 (p-value 0.863)
  - Pool Since March: -264.443*** (p-value 0.004)
- Table 6 (panel regressions, similar house sizes): FRM_30Yr coefficients
  - FE All: -23.799** (p-value 0.042)
  - FE Until Feb: -2.232 (p-value 0.561)
  - FE Since March: -268.867*** (p-value 0.004)
  - Pool All: -22.271* (p-value 0.058)
  - Pool Until Feb: -1.292 (p-value 0.754)
  - Pool Since March: -269.286*** (p-value 0.004)
- Table 8 (mobility controls, similar house sizes, Feb–Aug 2020): FRM_30Yr and mobility
  - FRM_30Yr: -264.645*** (FE, column 1), -223.346*** (FE, col 2), -201.779*** (FE, col 3)
  - Workplace: 2.531*** (FE, col 1)
  - Residential: -5.236*** (FE, col 2)
  - Overall: 1.990*** (FE, col 3)
- Table 9 (mobility + PMI): FRM_30Yr and PMI_yy
  - FRM_30Yr: -313.395*** (FE, col 1), -261.913*** (FE, col 2), -312.828*** (FE, col 3); pooled: -334.613***, -311.265***, -340.559*** respectively
  - PMI_yy: 0.315 (FE, col 1), 0.208 (FE, col 2), 0.554** (FE, col 3); pooled: 0.500*, 0.482*, 0.693*** (columns 4-6)

### Risks, distributional implications, and behavioral notes
- Potential unintended consequences:
  - Excessively high mortgage borrowing and leverage may emerge, sowing seeds for the next crisis similar to the subprime mortgage crisis during 2007-09.
  - This risk is particularly pronounced if households with low or volatile incomes have leveraged up more.
  - Inequality may worsen because beneficiaries of ultra-easy monetary and financial conditions are more likely to have high/stable incomes.
- Age and sectoral exposure highlighted:
  - The median age of US homebuyers is 47, while the median age of people working in restaurants is 29, implying that a relatively small number of potential near-term homebuyers lost their jobs.

### Policy implications and recommendations
- Macroprudential measures to complement monetary easing
  - Enforce loan documentation requirements.
  - Impose stringent limits on loan-to-value (LTV) ratios, debt-to-income (DTI) ratios, and debt-service-to-income (DSTI) ratios.
  - Close scrutiny of bank and nonbank mortgage lending to avoid loosening underwriting standards that could fuel unstable credit expansion.
- Trade-offs and distributional considerations
  - Trade-off: tightening underwriting standards reduces systemic risk but may block distressed or income-disrupted households from benefiting from low rates.
  - Recommendation: consider "streamlined" refinance programs (e.g., allowing existing homeowners to refinance without employment/income documentation) to enable low-income homeowners to benefit from lower monthly payments while avoiding relaxation of underwriting for new loans.
  - Targeted renter assistance: if low-income homeowners (not renters) capture much of the price gains, targeted financial assistance to renters (e.g., renter direct payment/voucher programs or payments directly to property owners) could reduce within-group inequality between low-income homeowners and low-income renters.
- Rationale and caution
  - The rapid price acceleration and strong demand response to mortgage rates raise financial-stability concerns given historical precedents (lead-up to the 2007–09 GFC involved low rates, relaxed lending standards, and implicit guarantees).
  - The US banking system showed resilience in stress tests (noted stress test results on June 25, 2020), but vigilance is warranted given the exceptional policy environment and distributional patterns observed.
- Research priorities for policy design
  - Empirically distinguish Scenario (a) (speculative FOMO) vs Scenario (b) (fundamental demand shift) to assess sustainability of price rises and implications for household leverage, financial stability, sectoral allocation, and medium-to-long-run growth.
  - Merge homebuyer income data at the zip code level to identify whether increased demand in low-income areas reflects local homeowners or purchases by wealthier households.
  - Control for omitted local economic variables (COVID case growth, unemployment, local disruptions) to refine estimates of mortgage-rate effects.

### Appendix: observations and totals (after cleaning)
- Monthly observations (Frequency, Percent, Cumulative) from 201707 to 202008 shown in Appendix Table 1 (selected entries):
  - 201707 — 12,927; 2.6; 2.6
  - 201808 — 13,205; 2.7; 36.4
  - 201912 — 13,623; 2.8; 79.3
  - 202002 — 13,262; 2.7; 84.8
  - 202004 — 12,918; 2.6; 90.0
  - 202006 — 12,477; 2.5; 95.1
  - 202008 — 11,936; 2.4; 100.0
- Key statistics and totals:
  - Total — 493,956; 100.0

*Italic: Source — wpiea2020212-print-pdf (Section V and supporting analysis and appendix content).*

### References.....................................................................................................42

### wpiea2020212-print-pdf - References.....................................................................................................42

### Key findings
- On March 3, 2020, the Fed cut the benchmark rate by 50 basis points to 1-1.25 percent; on March 15, 2020, the Fed cut the rate to 0-0.25 percent.
- On June 10, 2020, the “dot plot” indicated that the rate would be kept in the range of 0-0.25 percent through 2022 and Chairman Powell announced that the purchases of MBS would continue, triggering the 30-year fixed mortgage rate to hit a record low of 2.97%.
- On July 16, the rate dropped again and marked the seventh record low this year.
- According to the Case-Shiller National Home Price Index, real house prices rose 45 percent from February 2012 through May 2020.
- The year-on-year growth rate of the median price per square foot in the recent four months (April-August 2020) has accelerated faster than any four-month period in the lead-up to the 2007-09 global financial crisis.
- The acceleration in median price per square foot in April-August 2020 is observed across urban, suburban, and rural areas.

### Mechanisms and interpretation
- Downward pressure on housing prices:
  - COVID-19 is an unprecedented negative shock to households’ incomes (both current and expected incomes due to massive layoffs), leading to significant downward pressure on housing prices via depressed demand.
- Upward pressure on housing prices:
  - (a) Higher housing demand induced by the central bank’s unprecedented monetary easing, part speculative housing demand (“fear of missing out”) and part demand from previously liquidity-constrained low-income households (who were borrowing up to the debt-service limits before the monetary easing).
  - (b) Lower housing supply due to stay-at-home orders and sellers’ reluctance to supply houses out of COVID infection fears.
- Net outcome remains unclear a priori because these forces operate in opposite directions.

### Risks, distributional implications, and behavioral notes
- Potential unintended consequences:
  - Excessively high mortgage borrowing and leverage may emerge, sowing seeds for the next crisis similar to the subprime mortgage crisis during 2007-09.
  - This risk is particularly pronounced if households with low or volatile incomes have leveraged up more.
  - Inequality may worsen because beneficiaries of ultra-easy monetary and financial conditions are more likely to have high/stable incomes.
- Age and sectoral exposure highlighted:
  - The median age of US homebuyers is 47, while the median age of people working in restaurants is 29, implying that a relatively small number of potential near-term homebuyers lost their jobs.

### Data, methods, and paper structure
- Data and methods:
  - The paper uses zip code-level data and nonparametric estimation.
- Paper structure:
  - Section II reviews related literature.
  - Section III describes the data and methodologies.
  - Section IV presents the results, including the aggregate evidence, distributional evidence, and results after accounting for the urban status of the zip code, the composition change in housing supply, and the behavioral change of households.

### Figures and tables (as listed in the source)
- Figures include distributions of outliers, median housing price series (July 2017 – August 2020), comparisons of housing price growth between GFC and COVID, housing supply/listings time series, hotness of housing demand time series, nonparametric estimations of price per sqft vs median family income and views per property vs median family income, Google Mobility in the US: February 15 – September 11, 2020, and geographic medians for prices, demands, and supplies.
- Tables include summary statistics before and after cleaning, panel regressions for full and subsamples, falsification tests, and regressions incorporating mobility indicators and PMI.

*wpiea2020212-print-pdf - References.....................................................................................................42*

### Section V concludes and discusses some policy implications. The appendix collects some

### wpiea2020212-print-pdf - Section V concludes and discusses some policy implications. The appendix collects some

### Key empirical findings (stylized facts)
- Fact #1 (aggregate price rebound)
  - After a temporary slow-down in March and April 2020, the median growth rates (year-on-year and month-on-month) of the median price per sqft bounced back and exceeded pre-COVID levels.
  - Year-on-year growth: 5.0 in June 2020, 5.7 in July, and 6.6 in August, all higher than the 4.8 in February 2020.
  - Some zip codes experienced year-on-year growth of more than 30 percent in each month since March 2020.
- Fact #2 (acceleration vs pre-GFC)
  - The year-on-year growth rates of median price per sqft accelerated by 2.6 percentage points from 4.0 percent in April 2020 to 6.6 percent in August 2020 (April–August 2020).
  - By comparison, the fastest four-month pre-GFC acceleration was 1.8 percentage points, from 8.1 percent year-on-year to 9.9 percent (February 2004 to June 2004).
  - After the interest rate cut in March 2020, the price-growth acceleration entered the steady phase in one month; after the November 2002 cut it took about eight months.
- Fact #3 (supply decline amplified)
  - Median year-on-year growth rate of active listings was negative since July 2019 and dropped sharply since March 2020.
  - Month-on-month growth rates after April 2020 became more negative, contributing to rapid price growth.
- Fact #4 (demand hotness surge)
  - Median views per property rose by 99 percent in July 2020 and 121 percent in August 2020 (year-on-year), record highs since data began in July 2017.
- Fact #5 (structural break in demand response to mortgage rate)
  - Zip code-level panel regressions show a structural break in March 2020 in the elasticity of housing demand to mortgage rates (Chow test rejects no-break).
  - From July 2017 to February 2020 the 30-year fixed mortgage rate was negatively associated with year-on-year growth of online views but statistically insignificant.
  - From March 2020 onwards: as the 30-year fixed mortgage rate drops by 1 percentage point, the associated year-on-year growth of online views per property increases by about 267 percentage points (Table 3 coefficient: -266.940*** in column (3); pooled column (6): -264.443***).
  - Falsification tests using March 2019 as breakpoint do not show this pattern.
- Fact #6 (price growth concentrated in lower-to-middle income zip codes)
  - Nonparametric estimates (March–August 2020 average per zip code) show year-on-year price growth decreases with median family income: about 15 percent in zip codes with median family income of $25,000 down to close to 0 percent in zip codes with $225,000, then increasing again for very rich zip codes.
- Fact #7 (U-shaped demand response across income)
  - Year-on-year growth of views per property displays a U-shaped relationship with median family income (March–August 2020): high at low incomes, declines through the middle, then rises at the top end.
  - Example: from around 58 percent in zip codes with median family income of $45,000 to about 37 percent in zip codes with $200,000, then rising to about 50 percent at the very top (full-sample description).
- Fact #8 (broad geographic reach)
  - Changes in housing price, demand, and supply since April 2020 are broad-based across metropolitan, micropolitan, small-town, and rural areas with similar magnitudes.
  - Examples:
    - Metropolitan and micropolitan: median housing price growth (y-o-y) accelerated from 4–4.5 percent in April 2020 to about 7 percent in August 2020 (acceleration of 3–3.5 percentage points).
    - Small towns: 3.5 percent to 5.3 percent (1.8 percentage point acceleration).
    - Rural areas: 3.5 percent to 6.0 percent (3.5 percentage point acceleration).
    - Demand (views): metropolitan/micropolitan/small towns accelerated from about 0 percent in April 2020 to about 125 percent in August 2020; rural accelerated to 178 percent in August.
    - Supply (active listings): metropolitan from -18 percent in April 2020 to -28 percent in August 2020; micropolitan and rural from about -10 percent to about -30 percent; small towns from -11 percent to -24 percent.

### Robustness and additional controls
- Composition control
  - Results hold after restricting to zip codes with year-on-year median growth rates of median sqft between -5 and 5 percent (similar house sizes sample).
  - Example: With similar house sizes, the four-month acceleration after Fed easing was 2.4 percentage points (from 3.9 percent in April 2020 to 6.3 percent in August 2020).
- Mobility controls (Google mobility data)
  - Merged county-level Google mobility (workplace, residential, overall) with zip code housing data (Feb–Aug 2020).
  - Workplace mobility positively correlated with online views; residential mobility negatively correlated.
  - In fixed-effect regressions (Table 8), FRM_30Yr coefficients remain large and negative (e.g., -264.645*** in column (1)).
  - After adding normalized PMI year-on-year growth (Table 9), mortgage-rate coefficients become even larger in magnitude (e.g., FRM_30Yr -313.395*** to -340.559*** across specifications), while workplace mobility remains positively associated and residential mobility negatively associated with views.
  - Interpretation: mobility patterns during stay-at-home orders tended to reduce online views, reinforcing that the observed demand surge is linked to low mortgage rates rather than only to increased online activity during lockdown.
- Omitted variable considerations
  - Local COVID case growth, unemployment, and economic disruption variables (if controlled) would likely require an even stronger interest-rate response to explain the observed increase in views (direction of omitted variable bias uncertain for other omitted factors like subjective valuation of owning a house).

### Behavioral scenarios consistent with findings
- Scenario (a): Fear of Missing Out (FOMO)
  - Potential homebuyers responded strongly to historically low mortgage rates, rushing to buy to avoid missing low financing costs and rising prices.
- Scenario (b): COVID-induced fundamental preference/behavior changes
  - Stay-at-home orders and work-from-home expectations raised the value of owning larger homes, increased forced saving and down-payment ability, and shifted search behavior to online channels.
- These scenarios have different implications for sustainability and financial stability; further empirical work is needed to disentangle them.

### Key regression coefficients and summary statistics (exact values preserved)
- Table 3 (panel regressions, full sample): FRM_30Yr coefficients
  - FE All: -22.927* (p-value 0.056)
  - FE Until Feb: -1.007 (p-value 0.807)
  - FE Since March: -266.940*** (p-value 0.006)
  - Pool All: -22.235* (p-value 0.059)
  - Pool Until Feb: -0.703 (p-value 0.863)
  - Pool Since March: -264.443*** (p-value 0.004)
- Table 6 (panel regressions, similar house sizes): FRM_30Yr coefficients
  - FE All: -23.799** (p-value 0.042)
  - FE Until Feb: -2.232 (p-value 0.561)
  - FE Since March: -268.867*** (p-value 0.004)
  - Pool All: -22.271* (p-value 0.058)
  - Pool Until Feb: -1.292 (p-value 0.754)
  - Pool Since March: -269.286*** (p-value 0.004)
- Table 8 (mobility controls, similar house sizes, Feb–Aug 2020): FRM_30Yr and mobility
  - FRM_30Yr: -264.645*** (FE, column 1), -223.346*** (FE, col 2), -201.779*** (FE, col 3)
  - Workplace: 2.531*** (FE, col 1)
  - Residential: -5.236*** (FE, col 2)
  - Overall: 1.990*** (FE, col 3)
- Table 9 (mobility + PMI): FRM_30Yr and PMI_yy
  - FRM_30Yr: -313.395*** (FE, col 1), -261.913*** (FE, col 2), -312.828*** (FE, col 3); pooled: -334.613***, -311.265***, -340.559*** respectively
  - PMI_yy: 0.315 (FE, col 1), 0.208 (FE, col 2), 0.554** (FE, col 3); pooled: 0.500*, 0.482*, 0.693*** (columns 4-6)

### Policy implications and recommendations
- Macroprudential measures to complement monetary easing
  - Enforce loan documentation requirements.
  - Impose stringent limits on loan-to-value (LTV) ratios, debt-to-income (DTI) ratios, and debt-service-to-income (DSTI) ratios.
  - Close scrutiny of bank and nonbank mortgage lending to avoid loosening underwriting standards that could fuel unstable credit expansion.
- Trade-offs and distributional considerations
  - Trade-off: tightening underwriting standards reduces systemic risk but may block distressed or income-disrupted households from benefiting from low rates.
  - Recommendation: consider "streamlined" refinance programs (e.g., allowing existing homeowners to refinance without employment/income documentation) to enable low-income homeowners to benefit from lower monthly payments while avoiding relaxation of underwriting for new loans.
  - Targeted renter assistance: if low-income homeowners (not renters) capture much of the price gains, targeted financial assistance to renters (e.g., renter direct payment/voucher programs or payments directly to property owners) could reduce within-group inequality between low-income homeowners and low-income renters.
- Rationale and caution
  - The rapid price acceleration and strong demand response to mortgage rates raise financial-stability concerns given historical precedents (lead-up to the 2007–09 GFC involved low rates, relaxed lending standards, and implicit guarantees).
  - The US banking system showed resilience in stress tests (noted stress test results on June 25, 2020), but vigilance is warranted given the exceptional policy environment and distributional patterns observed.
- Research priorities for policy design
  - Empirically distinguish Scenario (a) (speculative FOMO) vs Scenario (b) (fundamental demand shift) to assess sustainability of price rises and implications for household leverage, financial stability, sectoral allocation, and medium-to-long-run growth.
  - Merge homebuyer income data at the zip code level to identify whether increased demand in low-income areas reflects local homeowners or purchases by wealthier households.
  - Control for omitted local economic variables (COVID case growth, unemployment, local disruptions) to refine estimates of mortgage-rate effects.

*Italic: Source — wpiea2020212-print-pdf (Section V and supporting analysis and appendix content).*

### Appendix Table 1. Number of Observations across All Months (after Cleaning)

### Appendix Table 1. Number of Observations across All Months (after Cleaning)

### Monthly observations (Frequency, Percent, Cumulative)
- 201707 — 12,927; 2.6; 2.6
- 201708 — 12,961; 2.6; 5.2
- 201709 — 12,979; 2.6; 7.9
- 201710 — 12,988; 2.6; 10.5
- 201711 — 12,483; 2.5; 13.0
- 201712 — 12,655; 2.6; 15.6
- 201801 — 12,479; 2.5; 18.1
- 201802 — 12,533; 2.5; 20.7
- 201803 — 12,664; 2.6; 23.2
- 201804 — 12,755; 2.6; 25.8
- 201805 — 12,886; 2.6; 28.4
- 201806 — 13,105; 2.7; 31.1
- 201807 — 13,175; 2.7; 33.7
- 201808 — 13,205; 2.7; 36.4
- 201809 — 13,265; 2.7; 39.1
- 201810 — 13,233; 2.7; 41.8
- 201811 — 13,183; 2.7; 44.4
- 201812 — 13,113; 2.7; 47.1
- 201901 — 12,954; 2.6; 49.7
- 201902 — 12,890; 2.6; 52.3
- 201903 — 13,089; 2.7; 55.0
- 201904 — 12,959; 2.6; 57.6
- 201905 — 13,063; 2.6; 60.2
- 201906 — 13,219; 2.7; 62.9
- 201907 — 13,265; 2.7; 65.6
- 201908 — 13,374; 2.7; 68.3
- 201909 — 13,418; 2.7; 71.0
- 201910 — 13,565; 2.8; 73.8
- 201911 — 13,610; 2.8; 76.5
- 201912 — 13,623; 2.8; 79.3
- 202001 — 13,738; 2.8; 82.1
- 202002 — 13,262; 2.7; 84.8
- 202003 — 13,022; 2.6; 87.4
- 202004 — 12,918; 2.6; 90.0
- 202005 — 12,790; 2.6; 92.6
- 202006 — 12,477; 2.5; 95.1
- 202007 — 12,195; 2.5; 97.6
- 202008 — 11,936; 2.4; 100.0

### Key statistics and totals
- Total — 493,956; 100.0

*Source: Author’s calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020212-print-pdf.pdf_