## 8. Rental Index and Non-Primary Home Demand – IV 2SLS Estimations

## Source details

**Canonical URL:** [8. Rental Index and Non-Primary Home Demand – IV 2SLS Estimations](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020174-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020174-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020174-print-pdf.pdf.json)

---

### I. Introduction and research questions
- Context:
  - “Non-bank financial intermediation” and “shadow banking” used interchangeably; “shadow banks” defined as financial intermediaries other than banks and credit unions.
  - Prior to the GFC, mortgage loans intermediated by shadow banks represented more than 60 percent of total residential lending.
  - Transaction-level data from Zillow Transaction and Assessment (ZTRAX) include mortgage-financed and cash transactions over the 1996-2018 period (coverage described in detail below).
- Main research questions:
  - How are different non-primary home buyers affecting house price cycles?
  - Has the increasing share of non-bank lending affected the housing market?
  - Which investors are more relevant during boom-bust periods?
  - What is the link between credit supply by shadow banks and non-traditional buyers over the cycle?

### II. Key contributions and empirical findings (summary)
- Contributions:
  - Document evolution of non-traditional buyers and capital invested in US residential housing.
  - Explore link between credit supply, non-traditional buyers, and house prices with heterogeneity across zip codes.
  - Exploit a quasi-experimental IV based on the GSE First Look program to identify causal effects of investor demand on prices.
- Principal empirical findings (preserve numeric fidelity):
  - Corporate cash transactions increased substantially after the GFC and peaked in 2011.
  - Market-timing activity (transactions resold within 12 months) tripled in 2008-09 relative to pre-crisis levels.
  - More than 30 percent of residential home sales were transacted by non-primary home buyers during the 2008-12 period; total share declined to about 15 percent of total transaction volume at the end of 2018.
  - Second-home buyers and short-term investors amplify prices throughout the housing cycle.
  - Out-of-state buyers prop up house prices, especially during the bust when liquidity dries up.
  - Non-primary home buyers consistently negatively affected rent growth; largest effects for short-term and second-home investors, and to a lesser extent for out-of-state buyers.
  - Areas with higher penetration of shadow banking mortgage activity are disproportionately affected by non-traditional buyers, particularly second-home buyers and short-term investors.
  - A higher share of loans originated by non-bank financial intermediaries is typically associated with a short-term amplification of the housing cycle.
  - IV analysis exploiting GSE First Look program variation shows second-home investment demand had an economically and statistically significant impact on house prices.
  - Investment purchases helped decelerate the increase in rental prices, indicating a supply channel for rental properties.

### III. Instrumental strategy and identification
- Instrument:
  - First Look program: starting in 2009-10, GSEs allowed owner-occupiers to purchase foreclosed GSE-owned properties during a window of up to 20 days before other investors could bid (window later extended to 30 days in some areas).
  - Instrument = (share of recently foreclosed properties among total sales) × (share of conforming loans proxy where loan amount does not exceed $500,000) at zip code-quarter level.
- Identification advantage:
  - Zip code-level variation in the instrument allows causal interpretation linking investor activity (non-primary home demand) to house prices while addressing reverse causality and endogeneity.
- IV estimation period and related timings:
  - IV estimations: 2008Q1–2018Q4 (44 quarters).
  - PostFL dummy = 1 after 2009Q3.
  - Conforming-loan proxy threshold used: loan amount does not exceed $500,000.
  - First Look program timing preserved: started August 2009 by Fannie Mae; similar program in September 2010; window of 15 to 20 days (later extended to 30 days in some areas).

### IV. Definitions used for non-primary buyers, lending, and outcomes
- Total house sales: N_i,q = CC_i,q + LL_i,q (both cash and loan-financed transactions).
- Non-primary buyer categories (share weighted by sale price at zip code-quarter level):
  - Second-home buyers: SH_i,q / N_i,q (non-owner occupied transactions).
  - Short-term buyers: STR_i,q / N_i,q (same property bought and sold within a 12-month window).
  - Corporate cash buyers: LCC_i,q / N_i,q (non-individual buyer, no mortgage).
  - Out-of-state buyers: OSB_i,q / N_i,q (buyer tax address in a different state than property).
- Non-bank lending:
  - Loans partitioned as L_i,q = B_i,q + NB_i,q + O_i,q (banks and credit unions; non-bank financial intermediaries; other lenders).
  - Non-bank lending defined as share of mortgages originated by intermediaries other than banks and credit unions (includes mortgage lenders, finance companies, lending companies, insurers, foreclosure companies, subprime lenders).
- Foreclosed sales constructed from Zillow’s monthly data on percentage share of sales where the bank took possession following a foreclosure within prior year.
- Conforming loans approximated by loan amount limit of $500,000.

### V. Data coverage, aggregation, and salient descriptive statistics
- ZTRAX transaction-level data:
  - Covers more than 20,000 zip codes; almost 90 million transactions after data cleaning (overall transaction sample about 88 million observations).
  - Analysis focuses on 40 states with reliable data; dataset span mentioned as 1998–2018 (ZHVI) and transaction-level reference to 1996-2018 earlier.
  - About 50,000 observations (~5 percent) of data dropped due to non-representative zip codes or insufficient observations.
  - Excluded: refinancing, home equity loans, reverse mortgages, loan modifications.
- Aggregated counts and values:
  - Nearly 50 million loan-financed transactions out of roughly 88 million total observations.
  - Mean property sale price: $265,000.
  - Median property sale price: $195,000.
  - Typical loan amount: $215,000.
  - LTV ratio averages 84 percent.
  - LTV ratio of loans originated by banks is, on average, lower by about 3.5 percentage points than LTV ratio of loans originated by non-bank financial intermediaries.
- House price and rent indices:
  - Zillow Home Value Index (ZHVI): 17,216 US zip codes across 40 states from January 1998 to December 2018.
  - Zillow Rent Index (ZRI): similar geographical coverage, starts in September 2010.

### VI. Time patterns, lender composition, and transaction composition
- Cash and investor dynamics:
  - Cash purchases constitute roughly 30 percent of total residential volume; elsewhere noted cash transactions rose to almost 50 percent over 1998–2013 and declined to 30 percent by end-2018.
  - Total investor share peaks in 2011–12, after more than doubling pre-crisis levels.
  - Short-term investors topped 15 percent of total sales volume in 2008-09.
  - Corporate cash buyer inter-quartile range across states from 5 to 35 percent, broadening in 2008–12.
  - Average share of out-of-state buyers increased steadily since 1998.
- Mortgage origination and lender shares:
  - Share of loans originated by non-bank financial intermediaries plunged in 2007–08 and gradually increased to more than 50 percent by end-2018.
  - Non-bank intermediaries originate a higher share of riskier loans such as mortgages with an LTV ratio above 80 or 90 percent.
  - Differences between banks and non-banks less pronounced for ARMs.
  - Share of originated ARMs positively associated with short-term interest rates.

### VII. Methodology — timing-assumption and IV approaches (preserved specifications)
- Timing-assumption main estimation:
  - Fixed-effect panel regressions covering 1998Q1–2018Q4 using zip code-quarter observations and one-quarter lagged regressors.
  - Dependent variable: Δln(hp)_i,t — quarterly log change in the median house value of each zip code.
  - Key lagged regressors: share of second home buyers; share of within-year resales; share of corporate cash transactions; share of out-of-state buyers; share of ARMs; share of mortgages issued by non-bank lenders; share of mortgages with an LTV higher than 90 percent.
  - Nonlinearity via crisis indicator equal to 1 during GFC years (2007 - 2009) and bust dummy 2007Q1 to 2009Q4 in some specifications.
  - Controls: zip code and quarter fixed effects; robustness: state-year fixed effects; standard errors clustered at zip code level.
  - Alternative dependent-variables: year-on-year price changes and rental index changes.
- IV approach:
  - Instrument: zip code-quarter measure combining foreclosure sales share and conforming-loan proxy (≤ $500,000).
  - Foreclosure data source: Zillow Research Data (monthly percentage share of sold properties foreclosed within prior year and bank owned).
  - IV estimations cover 2008Q1–2018Q4; first-stage F-statistics reported in Tables 7–8 confirm IV strength.

### VIII. Results — House prices and non-primary home buyers (timing-assumption estimates)
- General: non-primary buyers (second-home buyers, short-term investors, corporate cash buyers, out-of-state buyers) are significantly associated with the housing cycle, with asymmetric magnitudes across booms and busts.
- Selected quantitative findings (preserved exactly):
  - A one percentage point increase in the share of second-home buyers strengthens booming house prices by about 0.3 percent (column 2).
  - A one percentage point rise in the share of second-home buyers is associated with a 2.36 percent reduction (the sum of coefficients) in house prices during the GFC (2007Q1–2009Q4).
  - An increase in the share of short-term investors by one percentage point reduces house prices by about 1.5 percent during market deterioration.
  - A one percentage point increase in the share of second-home buyers or short-term investors is associated with about 0.4 percent decline in rental indices over the following quarter (Table 2.B; rental analysis covers period after 2010).
  - Increase in out-of-state buyers: a one percentage point increase in out-of-state buyers leads to 0.5 percent rise in house prices during the GFC.
  - Corporate cash transactions: generally negative and significant association with the housing cycle but weaker and less robust across stringent specifications.
- Representative regression coefficients (Table 2.A highlights):
  - House prices (lag, Δ%): coefficients ~0.6815*** to 0.5480*** across specifications.
  - Second Home (lag): examples include -0.0025*** (0.0004), 0.0030*** (0.0004); Crisis # Second Home (lag) = -0.0266*** (0.0010).
  - Short-term Buyers (lag): 0.0031*** (0.0002); Crisis # Short-term Buyers (lag) = -0.0182*** (0.0006).
  - Corporate Cash (lag): -0.0008*** (0.0001); Crisis # Corporate Cash (lag) = -0.0042*** (0.0003).
  - Out-of-state (lag): 0.0007*** (0.0001); Crisis # Out-of-state (lag) examples include 0.0016*** (0.0003).

### IX. Results — Role of non-bank lending (shadow banking heterogeneity)
- Zip codes split into High and Low groups by average share of non-bank lending (top and bottom quartiles):
  - High group average share of shadow bank lending: about 65 percent.
  - Low group average share of shadow bank lending: about 20 percent.
- Key contrasts (Table 3 preserved coefficients):
  - High group: a one percentage point increase in second-home buyer share leads to about a 0.5 percent increase in house prices during booms (0.0047*** (0.0014)); Crisis # Second Home (lag) = -0.0568*** (0.0035) during bust.
  - Low group: second-home rise associated with -0.0008 (0.0009) during booms and Crisis # Second Home (lag) = -0.0133*** (0.0020) during bust.
  - Short-term investors: High group boom impact 0.0093*** (0.0005); Low group boom impact 0.0023*** (0.0006). During busts, a 1 percentage point rise in short-term investor share leads to about 0.8 percent drop in both High and Low areas.
  - Corporate cash and out-of-state buyers significant primarily in High group in some specifications, but point estimates weaker than for second-home and short-term investors.
- Shadow banking amplification:
  - A one percentage point rise in fraction of shadow bank lending associated with less than 0.1 percent house price growth during boom times and about 0.25 percent decline during the bust (Table 4 summary: shadow banking (lag, share): 0.0003*** to 0.0007***; Crisis # shadow banking (lag): -0.0032*** (0.0002) to -0.0033*** (0.0002)).

### X. Results — Non-bank lending and riskier mortgages
- Shadow banking and risky loan features (Table 5 preserved results):
  - Robustly significant and positive relationship between shadow bank lending and the share of mortgages with LTV above 90 percent: LTV > 90% coefficient 0.1015*** (0.0087).
  - Mean LTV association: 0.0003*** (0.0000).
  - Significant and negative association between shadow banking and ARMs: Share ARM coefficient -0.0215** (0.0090) to -0.0223*** (0.0056).
  - ARM cycle effect: a one percentage point increase in share of ARMs is accompanied by a 0.2 percent drop in house prices over the following quarter during the bust (Table 4: ARM (lag, share): -0.0007*** (0.0001); Crisis # Share ARM (lag) = -0.0012*** (0.0002)).

### XI. Results — Instrumental Variable Analysis (First Look instrument)
- Time patterns and geography:
  - Share of foreclosure sales peaks in 2009–10 and decreases gradually after 2013.
  - Share of First Look properties follows a similar pattern; geographical distribution shows higher percentage shares in coastal areas and particularly in the Southwestern states.
- IV objective and setup:
  - Instrument the share of second-home investors and share of institutional cash buyers using First Look exposure; estimate effect on quarterly log change in median house value with zip code and quarter fixed effects and one-quarter lags; use one-year moving average shares.
- IV quantitative results (Tables 7–8 preserved exactly):
  - Table 7 (House Prices, IV 2SLS; Δ%, qoq):
    - IV estimates for Second Home (lag): 2.3650*** (0.2506), 1.2129*** (0.1227), 0.6569*** (0.0419) across specifications.
    - IV estimates for Corporate Cash (lag): 0.0146*** (0.0011), 0.0676*** (0.0017), 0.0369*** (0.0013), 0.0937*** (0.0036) in various specs.
    - Short-term Buyers (lag) example: 0.0048*** (0.0008) in one IV spec; -0.0704*** (0.0029) in another.
    - Two-stage evidence: A one percentage point expansion in the share of second home buyers increased house prices on average by 66 basis points per quarter (column 3 in linear IV summary).
  - Table 8 (Rental Index, IV 2SLS; Rents Δ%, qoq):
    - IV Second Home (lag): -1.5983*** (0.1632), -1.2786*** (0.1430), -0.5405*** (0.0588) across specifications.
    - IV Corporate Cash (lag): -0.0143*** (0.0013), -0.0495*** (0.0028), -0.0396*** (0.0027), -0.0844*** (0.0077).
    - Out-of-state (lag): -0.0096*** (0.0011) to -0.0084*** (0.0009).
    - Interpretation: Instrumented second-home buyers significantly and negatively affect rental growth (large magnitudes), consistent with a buy-to-rent supply channel.
- First-stage and program effects:
  - Table 6 (First Look Program and non-primary home demand; 1-year moving averages):
    - Share First Look coefficients:
      - Column (1) (Second Home): 0.0352*** (0.0060)
      - Column (2) (Second Home with Post # Share First Look): 0.0438*** (0.0059); Post # Share First Look = -0.0322*** (0.0057)
      - Column (3) (Corporate Cash): 0.7190*** (0.0642)
      - Column (4) (Corporate Cash with Post # Share First Look): 0.3223*** (0.0282); Post # Share First Look = -0.1969*** (0.0275)
    - Regression evidence: strong negative and significant correlations between share of properties affected by First Look and the share of buys that are non-primary home demand of household owner-occupiers, present only after the program started.
    - Graphical parallel trends: trends relatively parallel prior to August 2009; highest-exposure areas saw declines in second-home buys post-program versus lowest-exposure areas.

### XII. Selected summary statistics (zipcode-level, quarterly frequency; preserved exactly)
- N for Δprice (q-on-q, %): 1,046,854; mean 0.7437; sd 2.1114; min -14.25; p50 0.8173; max 12.4
- N for Δprice (y-on-y, %): 1,023,660; mean 2.941; sd 7.4883; min -51.256; p50 13.2389; max 38.5
- N for Δrent (q-on-q, %): 332,854; mean 0.4469; sd 2.6455; min -35.7366; p50 0.4623; max 26.2
- Cash (share, %): N 1,154,922; mean 48.9122; sd 37.1178; min 0; p50 38.48; max 100
- Mortgage (share, %): N 1,154,922; mean 51.0878; sd 37.1178; min 0; p50 61.51; max 100
- Second Home (share, %): N 1,141,653; mean 1.5001; sd 4.7898; min 0; p50 0; max 40.2
- Within-year Resale (share, %): N 1,145,061; mean 6.0864; sd 9.0110; min 0; p50 3.01; max 65.1
- Corporate Cash (share, %): N 1,154,922; mean 13.4638; sd 23.4881; min 0; p50 3.70; max 100
- Out-of-State Buyers (share, %): N 1,154,922; mean 5.1341; sd 14.4430; min 0; p50 0; max 100
- Bank Lender (share, %): N 729,496; mean 55.0128; sd 28.9705; min 2.39; p50 91.00
- NonBank Lender (share, %): N 729,496; mean 44.9858; sd 28.9677; min 70; p50 47.60; max 82.10
- LTV (mean): N 824,963; mean 0.88290; sd 0.1138; min 0; p50 0.91; max 1.5
- LTV > 90%: N 831,046; mean 49.363; sd 31.2893; min 0; p50 49.93; max 100
- Float Rate (share, %): N 723,871; mean 27.0979; sd 35.6142; min 0; p50 8.57; max 100
- Conforming Loans (share, %): N 831,046; mean 31.8134; sd 34.1021; min 2.18; p50 100
- Foreclosed Loans (share, %): N 623,509; mean 4.6465; sd 8.8612; min 0; p50 0; max 100

### XIII. Policy-relevant implications and recommendations
- Macroprudential stance and monitoring:
  - Rapid growth in securitization and shadow banking activity coincides with stronger co-movement of household indebtedness and housing cycle; areas with high shadow banking penetration see disproportionate investor activity and short-term price amplification.
  - Strong case for development of a complete macroprudential policy framework for non-bank financial intermediaries to address emerging risks and limit policy leakages.
- Data and regulatory focus:
  - Importance of including cash transactions and deed-level transaction data for comprehensive assessment of investor activity and its effects on prices and rents.
- Targeted interventions and place-based policies:
  - Geographically granular instruments (zip code-quarter variation) can identify causal investor effects and could inform place-based policies addressing investor-driven price amplification.
- Interaction considerations:
  - Policies should consider interactions between non-primary buyer demand (second-home buyers, corporate cash buyers, short-term investors) and shadow banking activity, given amplified price and rent effects in areas with higher shadow bank lending.

### XIV. Conclusion — main contributions and mechanisms
- Documentation: four types of non-traditional housing buyers and their post-GFC dynamics; rebound in shadow banking origination after the mortgage crisis; shadow bank lending historically associated with lower lending standards.
- Core mechanisms identified:
  - Second-home buyers and short-term investors amplify the housing cycle in booms and busts; out-of-state buyers help prop up prices during the bust.
  - Areas with higher share of shadow bank lending are disproportionately affected by non-traditional buyers; during the GFC sensitivity of house prices to second-home buyers in high shadow-bank areas is substantially larger than in bank-dominated areas.
  - First Look program provides a plausibly exogenous instrument to identify causal effects of buyer-pool composition on prices; IV results show second-home buyers significantly and positively affect house prices after First Look implementation, while increases in second-home investments and corporate cash purchases reduce the rental index in the most affected areas.

*Source: wpiea2020174-print-pdf — "8. Rental Index and Non-Primary Home Demand – IV 2SLS Estimations" (excerpts and tables as provided).*

### 1. Non-Primary Home Buyers and Transaction Volume ...........................................................5

### 1. Non-Primary Home Buyers and Transaction Volume ...........................................................5

### Major sections
- 1. Non-Primary Home Buyers and Transaction Volume ...........................................................5
- 2. Foreclosure Sales and First-Look Properties .......................................................................19
- 3. First Look Program and Non-Primary Home Demand ........................................................21

### Tables listed in the content unit
- 1. Summary Statistics...............................................................................................................23
- 2.A. House Prices and Non-Primary Home Buyers .................................................................24
- 2.B. Housing Rent and Non-Primary Home Buyers ................................................................25
- 3. Non-Bank Lending, Non-Primary Home Buyers, and House Prices ...................................26
- 4. Non-Bank Lending, Riskier Mortgages, and House Prices .................................................27
- 5. Non-Bank Lending and Riskier Mortgages .........................................................................28
- 6. First Look Program and Non-Primary Home Demand ........................................................29
- 7. House Prices and Non-Primary Home Demand – IV 2SLS Estimations ............................30

*Source: wpiea2020174-print-pdf - 1. Non-Primary Home Buyers and Transaction Volume ...........................................................5*

### 8. Rental Index and Non-Primary Home Demand – IV 2SLS Estimations .............................31

### 8. Rental Index and Non-Primary Home Demand – IV 2SLS Estimations

### I. Introduction and research questions
- Context:
  - “Non-bank financial intermediation” and “shadow banking” used interchangeably; “shadow banks” defined as financial intermediaries other than banks and credit unions.
  - Prior to the GFC, mortgage loans intermediated by shadow banks represented more than 60 percent of total residential lending.
  - Proprietary transaction-level data from Zillow Transaction and Assessment (ZTRAX) are used to include both mortgage-financed and cash transactions over the 1996-2018 period (transaction-level coverage further described below).
- Main research questions:
  1. How are different non-primary home buyers affecting house price cycles?
  2. Has the increasing share of non-bank lending affected the housing market?
  3. Which investors are more relevant during boom-bust periods?
  4. What is the link between credit supply by shadow banks and non-traditional buyers over the cycle?

### II. Key contributions and empirical findings (summary)
- Threefold contributions:
  - Document the evolution of non-traditional buyers and capital invested in US residential housing.
  - Explore link between credit supply, non-traditional buyers, and house prices with heterogeneity across zip codes.
  - Exploit a quasi-experimental instrumental variable based on the GSE First Look program to identify causal effects of investor demand on prices.
- Main empirical findings:
  - Corporate cash transactions increased substantially after the GFC and peaked in 2011.
  - Market-timing activity (transactions resold within 12 months) tripled in 2008-09 relative to pre-crisis levels.
  - More than 30 percent of residential home sales were transacted by non-primary home buyers during the 2008-12 period; total share declined to about 15 percent of total transaction volume at the end of 2018.
  - Second-home buyers and short-term investors amplify prices throughout the housing cycle.
  - Out-of-state buyers prop up house prices, especially during the bust when liquidity dries up.
  - Non-primary home buyers consistently negatively affected rent growth, with the largest effects for short-term and second-home investors, and to a lesser extent for out-of-state buyers.
  - Areas with higher penetration of shadow banking mortgage activity are disproportionately affected by non-traditional buyers, particularly second-home buyers and short-term investors.
  - A higher share of loans originated by non-bank financial intermediaries is typically associated with a short-term amplification of the housing cycle.
  - Instrumental variable (IV) analysis exploiting GSE First Look program variation shows second-home investment demand had an economically and statistically significant impact on house prices.
  - Investment purchases helped decelerate the increase in rental prices, indicating a supply channel for rental properties.

### III. Instrumental strategy and identification
- Instrument:
  - First Look program: starting in 2009-10, government sponsored enterprises (GSEs) allowed owner-occupiers to purchase foreclosed GSE-owned properties during a window of up to 20 days before other investors could bid.
  - Geographic variation in share of foreclosure sales owned by GSEs used as an instrument at zip code-quarter level.
- Advantage:
  - Zip code-level variation in the instrument allows causal interpretation linking investor activity (non-primary home demand) to house prices while addressing reverse causality and endogeneity.

### IV. Definitions used for non-primary buyers, lending, and outcomes
- Total house sales:
  - 푁푁푖,푞 = 퐶퐶푖,푞 + 퐿퐿푖,푞 (both cash and loan-financed transactions)
- Non-primary home buyer categories (each share weighted by sale price at zip code-quarter level):
  - Second-home buyers: 푆푆푆푆푖,푞 / 푁푁푖,푞 (non-owner occupied transactions)
  - Short-term buyers: 푆푆푆푆푖,푞 / 푁푁푖,푞 (same property bought and sold within a 12-month window)
  - Corporate cash buyers: 퐿퐿퐿퐿퐶퐶푖,푞 / 푁푁푖,푞 (non-individual buyer, no mortgage)
  - Out-of-state buyers: 푂푂푆푆퐵퐵푖,푞 / 푁푁푖,푞 (buyer tax address in a different state than property)
- Non-bank lending:
  - Loans partitioned as 퐿퐿푖,푞 = 퐵퐵푖,푞 + 푁푁퐵퐵푖,푞 + 푂푂푖,푞 (banks and credit unions; non-bank financial intermediaries; other lenders)
  - Non-bank lending defined as share of mortgages originated by intermediaries other than banks and credit unions (includes mortgage lenders, finance companies, lending companies, insurers, foreclosure companies, subprime lenders).
- Conforming loans:
  - Mortgages meeting FHFA criteria approximated by a loan amount limit of $500,000.
- Foreclosed sales:
  - Constructed from Zillow’s monthly data on percentage share of sales where the bank took possession following a foreclosure within prior year.

### V. Data coverage, aggregation, and salient descriptive statistics
- ZTRAX transaction-level data:
  - Covers more than 20,000 zip codes; almost 90 million transactions after data cleaning (overall transaction sample about 88 million observations).
  - Analysis focuses on 40 states with reliable data; dataset span mentioned as 1998–2018 (ZHVI) and transaction-level reference to 1996-2018 earlier in the text.
  - About 50,000 observations (~5 percent) of data dropped due to non-representative zip codes or insufficient observations.
  - Excluded: refinancing, home equity loans, reverse mortgages, loan modifications.
- Aggregated sample counts and values:
  - Nearly 50 million loan-financed transactions out of roughly 88 million total observations.
  - Mean property sale price: $265,000.
  - Median property sale price: $195,000.
  - Typical loan amount: $215,000.
  - LTV ratio averages 84 percent.
  - LTV ratio of loans originated by banks is, on average, lower by about 3.5 percentage points than LTV ratio of loans originated by non-bank financial intermediaries.
- House price and rent indices:
  - Zillow Home Value Index (ZHVI): 17,216 US zip codes across 40 states from January 1998 to December 2018.
  - Zillow Rent Index (ZRI): similar geographical coverage, starts in September 2010.

### VI. Time patterns, lender composition, and transaction composition
- Non-primary buyer and transaction shares over time:
  - Cash purchases constitute roughly 30 percent of total residential volume (statement emphasizing importance of including cash transactions); elsewhere cash transactions rose to almost 50 percent over 1998–2013 and declined to 30 percent by end-2018.
  - Total investor share peaks in 2011–12, after more than doubling pre-crisis levels.
  - Short-term investors topped 15 percent of total sales volume in 2008-09 (panel summary).
  - Corporate cash buyer inter-quartile range across states from 5 to 35 percent, broadening in 2008–12.
  - Average share of out-of-state buyers increased steadily since 1998 (panel summary).
- Mortgage origination and lender shares:
  - Share of loans originated by non-bank financial intermediaries plunged in 2007–08 and gradually increased to more than 50 percent by end-2018.
  - Non-bank intermediaries originate a higher share of riskier loans such as mortgages with an LTV ratio above 80 or 90 percent, indicating looser lending standards relative to banks.
  - Differences between banks and non-banks less pronounced when considering adjustable rate mortgages (ARMs).
  - Share of originated ARMs positively associated with short-term interest rates (panel summary).

### VII. Policy-relevant implications (inferred from findings in text)
- Market monitoring and macroprudential stance:
  - Rapid growth in securitization and shadow banking activity coincides with stronger co-movement of household indebtedness and housing cycle; areas with high shadow banking penetration see disproportionate investor activity and short-term price amplification.
- Data and regulatory focus:
  - Importance of including cash transactions and deed-level transaction data for a comprehensive assessment of investor activity and its effects on prices and rents.
- Targeted interventions:
  - Geographically granular instruments (zip code-quarter variation) can identify causal investor effects and could inform place-based policies addressing investor-driven price amplification.

*Source: wpiea2020174-print-pdf - 8. Rental Index and Non-Primary Home Demand – IV 2SLS Estimations (excerpts).*

### Annex  Figure  3  highlights  the  geographical  dispersion  of  shadow  banking  origination  across  US  counties  in

### wpiea2020174-print-pdf - Annex (Figure 3 context and methodological annex)

### Methodology — Timing-Assumption Approach
- Main estimation: fixed-effect panel regressions covering 1998Q1–2018Q4 using zip code-quarter observations and one-quarter lagged regressors.
- Dependent variable: 훥훥푙푙푙푙(ℎ푝푝)푖푖,푡푡 — the quarterly log change in the median house value of each zip code.
- Key regressors (lagged one quarter):
  - 푆푆푆푆푆푆푖푖,푡푡 — zip code-level share of second home buyers.
  - 푆푆푆푆푆푆푖푖,푡푡 — zip code-level share of within-year resales (short-term buyers).
  - 퐿퐿퐿퐿퐶퐶푖푖,푡푡 — share of transactions by corporate cash within the zip code.
  - 푂푂푆푆퐵퐵푖푖,푡푡 — share of transactions by out-of-state buyers.
  - 퐴퐴퐴퐴푀푀푖푖,푡푡 — share of adjustable rate mortgages within the zip code.
  - 푁푁푁푁 푙푙퐵퐵푁푁푙푙푘푘푖푖,푡푡 — share of mortgages issued by non-bank lenders.
  - 퐿퐿푆푆퐿퐿푖푖,푡푡 — share of mortgages with an LTV higher than 90 percent.
- Nonlinearity captured via crisis indicator 푆푆푡푡퐶퐶퐶퐶푖푖퐶퐶푖푖퐶퐶 equal to 1 during GFC years (2007 - 2009) and dummy for bust period 2007Q1 to 2009Q4 in some specifications.
- Controls: zip code and quarter fixed effects; robustness checks include state-year fixed effects.
- Standard errors: clustered at zip code level in most regressions; alternate clustering at FIPS and state level for robustness.
- Alternative dependent variable specifications: year-on-year changes in prices (instead of quarterly) and changes on rental indices (dependent-variable alternative).

### Methodology — Instrumental Variable (IV) Approach
- Instrument: zip code-quarter measure = (share of recently foreclosed properties among total sales) × (share of conforming loans proxy where loan amount does not exceed $500,000).
- Data source for foreclosure share: Zillow Research Data (monthly percentage share of sold properties that were foreclosed within the prior year and bank owned).
- Motivation: Fannie Mae and Freddie Mac First Look program (started August 2009 by Fannie Mae; similar program in September 2010) gave households a 15 to 20 days window to bid on foreclosed properties owned by the GSEs; later extended to 30 days in some areas.
- Identification: where instrument is high (higher share of First Look properties), expect a lower share of non-owner-occupied homes and lower speculative activity after program implementation.
- IV estimations cover 2008Q1–2018Q4 (44 quarters) for foreclosure-data-limited specifications.
- IV specification: two-stage least squares where shares of second-home investors and institutional cash purchases are instrumented by the First Look measure; PostFL dummy = 1 after 2009Q3.

### Results — House Prices and Non-Primary Home Buyers (timing-assumption estimates)
- General: non-primary buyers (second-home buyers, short-term investors, corporate cash buyers, out-of-state buyers) are significantly associated with the housing cycle, with asymmetric magnitudes across booms and busts.
- Selected quantitative findings:
  - A one percentage point increase in the share of second-home buyers strengthens booming house prices by about 0.3 percent (column 2).
  - A one percentage point rise in the share of second-home buyers is associated with a 2.36 percent reduction (the sum of coefficients) in house prices during the GFC (2007Q1–2009Q4).
  - An increase in the share of short-term investors by one percentage point reduces house prices by about 1.5 percent during market deterioration.
  - A one percentage point increase in the share of second-home buyers or short-term investors is associated with about 0.4 percent decline in rental indices over the following quarter (Table 2.B; rental analysis covers period after 2010).
  - Increase in out-of-state buyers: positive association with house prices during busts but much weaker magnitudes; a one percentage point increase in out-of-state buyers leads to 0.5 percent rise in house prices during the GFC.
  - Corporate cash transactions: generally negative and significant association with the housing cycle but weaker and less robust across stringent specifications.

### Results — Role of Non-Bank Lending (shadow banking heterogeneity)
- Zip codes split into High and Low groups by average share of non-bank lending (top and bottom quartiles).
  - High group average share of shadow bank lending: about 65 percent.
  - Low group average share of shadow bank lending: about 20 percent.
- Key contrasts:
  - In High group areas, a one percentage point increase in second-home buyer share leads to about a 0.5 percent increase in house prices during booms (column 1).
  - In High group areas, a one percentage point rise in second-home buyers is associated with about a 5 percent slump in house prices during the bust (column 1).
  - In Low group areas, a one percentage point rise in second-home buyers is associated with around 1.3 percent fall in house prices during the bust (column 2).
  - For short-term investors: High group — a one percentage point increase in short-term investor share associated with about 1 percent price growth during booms (column 3); Low group — about 0.25 percent during booms (column 4). During busts, a 1 percentage point rise in short-term investment share leads to about 0.8 percent drop in both High and Low areas.
  - Corporate cash and out-of-state buyers: significant primarily in High group in some specifications, but point estimates weaker than for second-home and short-term investors.

### Results — Non-Bank Lending and Riskier Mortgages
- Shadow bank origination and amplification:
  - A one percentage point rise in the fraction of shadow bank lending is associated with less than 0.1 percent house price growth during boom times and about 0.25 percent decline in house prices during the bust.
- ARMs and cycle:
  - A one percentage point increase in the share of ARMs is accompanied by a 0.2 percent drop in house prices over the following quarter during the bust.
- Shadow banking and risky loan features (Table 5):
  - Robustly significant and positive relationship between shadow bank lending and the share of mortgages with LTV above 90 percent (columns 2, 4, and 6).
  - Significant and negative association between shadow banking and ARMs (columns 7 and 8).

### Results — Instrumental Variable Analysis (First Look instrument)
- Time patterns:
  - Share of foreclosure sales peaks in 2009–10 and decreases gradually after 2013.
  - Share of First Look properties follows a similar pattern.
  - Geographical distribution shows higher percentage shares in coastal areas and particularly in the Southwestern states (Figure 2, Panels 3–4).
- IV objective: disentangle causal effects of non-primary home demand on house prices and rents by leveraging quasi-exogenous variation in buyer composition induced by the First Look program.
- IV estimations: instrument the share of second-home investors and share of institutional cash buyers; estimate effect on quarterly log change in median house value (Δln(hp)it) with zip code and quarter fixed effects and one-quarter lags; use one-year moving average shares.

### Key quantitative/temporal precisions preserved from the analysis
- Sample and period notes:
  - Fixed-effect panel estimations: 1998Q1–2018Q4.
  - Instrumental-variable estimations (due to foreclosure data availability): 2008Q1–2018Q4 (44 quarters).
  - Crisis indicator (GFC years): 2007 - 2009.
  - Bust-period dummy used in some regressions: 2007Q1 to 2009Q4.
  - PostFL dummy (First Look program period): equal to 1 after 2009Q3.
  - First Look program timing: started August 2009 by Fannie Mae; similar strategy in September 2010; window of 15 to 20 days (later extended to 30 days in some areas).
  - Conforming-loan proxy threshold used in instrument construction: loan amount does not exceed $500,000.
  - Rental-index analysis limited to period after 2010 (due to zip code-level rental index availability).

### Analytical implications (as presented in the source)
- Non-primary buyers (second-home buyers and short-term investors) amplify the boom-bust cycle in local house prices, with substantially stronger adverse effects during busts than supporting effects during booms.
- Areas with higher shadow bank lending (non-bank origination) are more prone to amplification of the housing cycle linked to non-primary buyers.
- Shadow bank activity is associated with a higher incidence of riskier mortgages (higher LTV shares) and with cyclical amplification, suggesting a channel from non-bank credit supply to both loan composition and price dynamics.
- The First Look program provides a plausibly exogenous instrument to identify causal effects of buyer-pool composition on prices by altering the share of owner-occupier opportunities in foreclosure-markets.

*Source: IMF working paper annex (methodology and results sections summarized from the provided content).*

### 1. The share of foreclosure sales  2. The share of First Look properties

### 1. The share of foreclosure sales  2. The share of First Look properties

### Data and figure description
- Sources: Zillow’s ZTRAX data and authors' calculations.
- Note: Panels 1 and 2 present the share of foreclosed and First Look properties in total sales, respectively, for the median, 75th, and 90th percentile zip code ranked quarterly.
- Note: Panels 3 and 4 depict a heatmap of the average shares of foreclosed and First Look properties in total sales by FIPS, during 2008 – 2010 period.

### Direct impact of the First Look program on speculative activity
- Research focus:
  - Investigate whether the program affected the share of second-home buyers and corporate buyers in cash.
  - Focus on these measures because the program was designed to affect non-owner-occupier investment demand.
- Method:
  - Zip codes sorted into quintiles by the share of First Look properties (as measured by the authors' instrument), for every quarter.
  - Time series of the shares of second-home buyers and corporate buyers in cash plotted for highest and lowest quintiles (Figure 3).
- Graphical findings:
  - Panel 1: Trends were relatively parallel prior to the enactment of the program in August 2009; areas in the highest quintile of exposure saw a drop in the share of second-home buys compared with the lowest quintile areas.
  - Panel 2: Parallel trend assumption supported prior to the program; post-program effect on corporate cash demand is less evident than for second-home demand, but a steeper decline is observed in the zip codes of highest exposure.

### Regression evidence on non-primary home demand (Table 6)
- Estimation:
  - Linear regressions around the time of program implementation.
- Findings:
  - Strongly negative and significant correlations between the share of properties affected by the First Look program in a zip code area and the share of buys that represent non-primary home demand of household owner-occupiers.
  - This negative relationship is present only after the program started, evidenced by the coefficient on the interaction term of post-First Look with the measure of investment activity.
  - Parallel trends assumption verified graphically in Figure 3, though a positive correlation between the share of First Look properties and non-primary home buyers prior to the program indicates higher ex-ante investment activity in targeted areas.
  - Key insight: The decline in non-primary home buyer demand corresponds closely with the program and is most prevalent in the most affected areas.
  - Interpretation: Areas with high exposure to the First Look initiative (measured by a higher share of foreclosed properties with conforming mortgages) saw a disproportionate decrease in second-home buys and purchases from corporate buyers in cash after enactment.

### Causal analysis using First Look exposure as an instrument (two-stage least squares; Tables 7–8)
- Instrument: Share of properties within a zip code affected by the First Look program.
- Note: F-statistics of the IV’s first stage, reported in Tables 7 and 8, confirm the validity of the instrument.18
- Second-stage results (Table 7):
  - Instrumented variable: share of second-home buyers (columns 1–3).
  - Instrumented variable: corporate transactions made in cash (columns 4–6).
  - Quantitative results:
    - A one percentage point expansion in the share of second home buyers increased house prices on average by 66 basis points per quarter (column 3).
    - Corporate purchases in cash raised house prices by about 9 basis points per quarter on average for each standard deviation increase.
  - Interpretation:
    - Instrumented measure of second home buyers significantly and positively affects house prices, providing causal support that second home demand played a significant role in driving up prices during the post-crisis recovery.
    - Evidence supports a role played by institutional buyers in cash, consistent with Lambie-Hanson and others (2019) and with anonymity-seeking capital (Hundtofte and Rantala 2018).

### Impact on rental prices (Table 8)
- Dependent variable: quarterly percentage changes in the Zillow Rental Index for the zip code area.
- Findings:
  - An increase in investment demand, proxied by the share of second home and corporate purchases in cash, helped decelerate the growth in the rental index in the most affected areas.
  - Interpretation: Consistent with a buy-to-rent logic where investors purchase properties and subsequently increase supply in the market for rentals.

*Source: Zillow’s ZTRAX data and authors' calculations, as presented in the PDF content unit.*

### 1. Share of second-h ome buyers 2. Share of corporate purchases in cash

### 1. Share of second-home buyers 2. Share of corporate purchases in cash

### V. Conclusion — main contributions and mechanisms
- Examines role of real estate investors using transaction-level data to study relationship between different types of non-primary home buyers and house price cycles, and causal impact of non-primary home buyers on rents.
- Documents four types of non-traditional housing buyers and their post-GFC dynamics:
  - Corporate cash buyers and short-term investors surged in the aftermath of the GFC, supporting housing market liquidity.
  - Out-of-state buyers and transactions involving non-primary home buyers dropped and slowly recovered after the crisis.
- Documents rebound in shadow banking origination activity after the mortgage crisis; shadow bank lending historically associated with lower lending standards (for example, high LTV mortgages).
- Finds that second-home buyers and short-term investors amplify the housing cycle in booms and busts; out-of-state buyers help prop up prices during the bust.
- Areas with higher share of shadow bank lending are disproportionately affected by non-traditional buyers; during the GFC the sensitivity of house prices to second-home buyers in areas with higher shadow bank lending activity is four times larger than in bank-dominated areas.
- Uses an instrumental variable approach exploiting zip code heterogeneity of exposures to the First Look program implemented by the two main GSEs after the financial crisis; finds compelling evidence that second-home buyers significantly and positively affect house prices after First Look implementation. Corporate purchases in cash also amplify house prices after First Look, but with weaker magnitude than second-home buyers.
- Novel finding: increases in second-home investments and corporate cash purchases reduce the rental index in the most affected areas.

### Key empirical findings (selected)
- Second-home buyers and short-term investors: amplify house price cycles during booms and busts.
- Corporate cash purchases: amplify house prices post-First Look, but less than second-home buyers.
- Shadow banking share: associated with riskier mortgage attributes (higher mean LTV, higher share LTV > 90%, and effects on ARM shares) and strengthens the price impact of non-primary buyers.
- Second-home investors and corporate cash buyers reduce rental returns in affected zip codes.

### Selected quantitative statistics and regression estimates (preserved exactly)
- Summary statistics (zipcode-level, quarterly frequency):
  - N for Δprice (q-on-q, %): 1,046,854; mean 0.7437; sd 2.1114; min -14.25; p50 0.8173; max 12.4
  - N for Δprice (y-on-y, %): 1,023,660; mean 2.941; sd 7.4883; min -51.256; p50 13.2389; max 38.5
  - N for Δrent (q-on-q, %): 332,854; mean 0.4469; sd 2.6455; min -35.7366; p50 0.4623; max 26.2
  - Cash (share, %): N 1,154,922; mean 48.9122; sd 37.1178; min 0; p50 38.48; max 100
  - Mortgage (share, %): N 1,154,922; mean 51.0878; sd 37.1178; min 0; p50 61.51; max 100
  - Second Home (share, %): N 1,141,653; mean 1.5001; sd 4.7898; min 0; p50 0; max 40.2
  - Within-year Resale (share, %): N 1,145,061; mean 6.0864; sd 9.0110; min 0; p50 3.01; max 65.1
  - Corporate Cash (share, %): N 1,154,922; mean 13.4638; sd 23.4881; min 0; p50 3.70; max 100
  - Out-of-State Buyers (share, %): N 1,154,922; mean 5.1341; sd 14.4430; min 0; p50 0; max 100
  - Bank Lender (share, %): N 729,496; mean 55.0128; sd 28.9705; min 2.39; p50 91.00
  - NonBank Lender (share, %): N 729,496; mean 44.9858; sd 28.9677; min 70; p50 47.60; max 82.10
  - LTV (mean): N 824,963; mean 0.88290; sd 0.1138; min 0; p50 0.91; max 1.5
  - LTV > 90%: N 831,046; mean 49.363; sd 31.2893; min 0; p50 49.93; max 100
  - Float Rate (share, %): N 723,871; mean 27.0979; sd 35.6142; min 0; p50 8.57; max 100
  - Conforming Loans (share, %): N 831,046; mean 31.8134; sd 34.1021; min 2.18; p50 100
  - Foreclosed Loans (share, %): N 623,509; mean 4.6465; sd 8.8612; min 0; p50 0; max 100

- Table 2.A (House prices and non-primary buyers; key coefficients, Δ%, qoq regressions):
  - House prices (lag, Δ%): coefficients ~0.6815*** to 0.5480*** across specifications (standard errors reported).
  - Second Home (lag, share of total): examples include -0.0025*** (0.0004), 0.0030*** (0.0004), and specification showing Crisis # Second Home (lag) = -0.0266*** (0.0010).
  - Short-term Buyers (lag, share of total): coefficients include 0.0031*** (0.0002); Crisis # Short-term Buyers (lag) = -0.0182*** (0.0006).
  - Corporate Cash (lag, share of total): -0.0008*** (0.0001); Crisis # Corporate Cash (lag) = -0.0042*** (0.0003).
  - Out-of-state (lag, share of total): 0.0007*** (0.0001); Crisis # Out-of-state (lag) examples include 0.0016*** (0.0003).

- Table 2.B (Housing Rent and non-primary buyers; Rents Δ%, qoq):
  - Rents (lag, Δ%): ~0.2365*** to 0.2371***.
  - Second Home (lag, share of total): -0.0036** (0.0018) to -0.0039** (0.0018).
  - Short-term Buyers (lag): -0.0038*** (0.0007).
  - Corporate Cash (lag): coefficients not significant in shown specs: -0.0005 (0.0003), -0.0002 (0.0004).
  - Out-of-state (lag): -0.0017** (0.0007) to -0.0019*** (0.0007).

- Table 3 (Non-Bank lending split high/low; House prices Δ%, qoq):
  - In high-shadow-banking zip codes:
    - Second Home (lag, share): 0.0047*** (0.0014)
    - Crisis # Second Home (lag): -0.0568*** (0.0035)
    - Short-term Buyers (lag): 0.0093*** (0.0005)
    - Corporate Cash (lag): 0.0031*** (0.0003)
  - In low-shadow-banking zip codes:
    - Second Home (lag, share): -0.0008 (0.0009)
    - Crisis # Second Home (lag): -0.0133*** (0.0020)
    - Short-term Buyers (lag): 0.0023*** (0.0006)
    - Corporate Cash (lag): -0.0004 (0.0003)

- Table 4 (Shadow banking, riskier mortgages and house prices; House prices Δ%, qoq):
  - Shadow banking (lag, share): 0.0003*** to 0.0007*** across specs.
  - Crisis # shadow banking (lag): -0.0032*** (0.0002) to -0.0033*** (0.0002).
  - High LTV (lag, share of total) sometimes positive and significant: 0.0005*** (0.0001); Crisis # share high LTV (lag) = 0.0026*** (0.0002).
  - ARM (lag, share of total): -0.0007*** (0.0001); Crisis # Share ARM (lag) = -0.0012*** (0.0002).

- Table 5 (Non-Bank Lending and mortgages; zip code-level outcomes):
  - Shadow Banking associations:
    - Mean LTV: 0.0003*** (0.0000)
    - LTV > 90%: 0.1015*** (0.0087)
    - Share ARM: -0.0215** (0.0090) to -0.0223*** (0.0056)

- Table 6 (First Look Program and non-primary home demand; 1-year moving averages):
  - Share First Look coefficients:
    - Column (1) (Second Home): 0.0352*** (0.0060)
    - Column (2) (Second Home with Post # Share First Look): 0.0438*** (0.0059); Post # Share First Look = -0.0322*** (0.0057)
    - Column (3) (Corporate Cash): 0.7190*** (0.0642)
    - Column (4) (Corporate Cash with Post # Share First Look): 0.3223*** (0.0282); Post # Share First Look = -0.1969*** (0.0275)

- Table 7 (House Prices and Non-Primary Home Demand – IV 2SLS; Δ%, qoq):
  - IV estimates for Second Home (lag): 2.3650*** (0.2506), 1.2129*** (0.1227), 0.6569*** (0.0419) across specifications.
  - IV estimates for Corporate Cash (lag): 0.0146*** (0.0011), 0.0676*** (0.0017), 0.0369*** (0.0013), 0.0937*** (0.0036) in various specs.
  - Short-term Buyers (lag) example: 0.0048*** (0.0008) in one IV spec; -0.0704*** (0.0029) in another.

- Table 8 (Rental Index and Non-Primary Home Demand – IV 2SLS; Rents Δ%, qoq):
  - IV Second Home (lag): -1.5983*** (0.1632), -1.2786*** (0.1430), -0.5405*** (0.0588) across specifications.
  - IV Corporate Cash (lag): -0.0143*** (0.0013), -0.0495*** (0.0028), -0.0396*** (0.0027), -0.0844*** (0.0077) across specifications.
  - Out-of-state (lag): -0.0096*** (0.0011) to -0.0084*** (0.0009) in IV rent specs.

### Policy implications and recommendations
- Macroprudential, monetary, and fiscal policies are central levers to mitigate excessive house price and rent appreciation through channels including credit availability, housing demand, and balance sheet effects.
- Strong case made for development of a complete macroprudential policy framework for non-bank financial intermediaries:
  - Expand scope of macroprudential toolkit to address emerging risks and limit policy leakages (cites: Adrian and Ashcraft, 2012; Claessens, 2015 as context in source).
- Policies should consider interactions between non-primary buyer demand (second-home buyers, corporate cash buyers, short-term investors) and shadow banking activity, given amplified price and rent effects in areas with higher shadow bank lending.

*Sources: Zillow’s ZTRAX data and authors' calculations; content from the specified IMF working paper chapter and tables.*

### REFERENCES

### wpiea2020174-print-pdf - REFERENCES

### References
- Adelino, M., A. Schoar, and F. Severino, 2016. “Loan originations and defaults in the mortgage crisis: The role of the middle class,” Review of Financial Studies, 29(7), 1635-70.
- Adrian, T. and Ashcraft, A.B., 2012. “Shadow banking regulation,” Annual Review of Financial Economics, 4(1), pp.99-140.
- Albanesi, S., G. DeGiorgi, and J. Nosal, 2017. “Credit growth and the financial crisis: A new narrative,” National Bureau of Economic Research Working Paper 23740.
- Alter, A., and E. Mahoney, 2020. “Household debt and house prices-at-risk: A tale of two countries,” IMF Working Paper 20/42.
- Alter, A., J. Dokko, and M. Seneviratne, 2018. “House price synchronicity, banking integration, and global financial conditions,” IMF Working Paper 18/250.
- Arellano, M., and S. Bond. 1991. “Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations,” Review of Economic Studies 58: 277–297.
- Ari, A., D. Puy, and Y. Shi, 2020. “Are foreigners driving local house prices? Evidence from the U.S.,” IMF Working Paper 20/43.
- Armona, L., Fuster, A. and Zafar, B., 2019. “Home price expectations and behaviour: Evidence from a randomized information experiment,” Review of Economic Studies, 86(4), pp.1371-1410.
- Badarinza, C., and T. Ramadorai, 2018. “Home away from home? Foreign demand and London house prices,” Journal of Financial Economics, 130(3), 532-55.
- Bhutta, N., 2015. “The ins and outs of mortgage debt during the housing boom and bust,” Journal of Monetary Economics, 76, 284-98.
- Buchak, G., Matvos, G., Piskorski, T. and Seru, A., 2018. “Fintech, regulatory arbitrage, and the rise of shadow banks,” Journal of Financial Economics, 130(3), pp.453-483.
- Burnside, C., Eichenbaum, M. and Rebelo, S., 2016. “Understanding booms and busts in housing markets,” Journal of Political Economy, 124(4), pp.1088-1147.
- Chinco, A., and C. Mayer, 2016. “Misinformed speculators and mispricing in the housing market,” Review of Financial Studies, 29(2), 486-522.
- Claessens, S., 2015. “An overview of macroprudential policy tools,” Annual Review of Financial Economics, 7, pp.397-422.
- DeFusco, A.A., C.G. Nathanson, and E. Zwick, 2017. “Speculative dynamics of prices and volume,” National Bureau of Economic Research, No. w23449.
- Demyanyk, Y., and O. Van Hemert, 2011. “Understanding the subprime mortgage crisis,” Review of Financial Studies, 24(6), 1848-80.
- Favilukis, J.Y., S.C. Ludvigson, and S. Van Nieuwerburgh, 2017. “The macroeconomic effects of housing wealth, housing finance, and limited risk sharing in general equilibrium,” Journal of Political Economy, 125 (1), 140–223.
- Favilukis, J.Y., and S. Van Nieuwerburgh, 2017. “Out-of-town home buyers and city welfare,” CEPR Discussion Paper No. DP12283.
- Foucault, T., Sraer, D. and Thesmar, D.J., 2011. “Individual investors and volatility” Journal of Finance, 66(4), pp.1369-1406.
- Fuster, A., and others, 2019. “The role of technology in mortgage lending,” Review of Financial Studies, 32(5), 1854-99.
- Gao, Z., M. Sockin, and W. Xiong, 2020. “Economic consequences of housing speculation,” Review of Financial Studies, 1-40.
- Greenwald, D., and A. Guren, 2019. “Do credit conditions move house prices?” Unpublished manuscript, MIT.
- Hundtofte, S., and V. Rantala, 2018. “Anonymous capital flows and U.S . housing markets,” University of Miami Business School Research Paper, (18-3).
- Justiniano, A., G.E. Primiceri, and A. Tambalotti, 2019. “Credit supply and the housing boom,” Journal of Political Economy, 127(3), 1317-50.
- Keys, B.J., and others, 2010. “Did securitization lead to lax screening? Evidence from subprime loans,” Quarterly Journal of Economics, 125(1), 307-62.
- Lambie-Hanson, L., W. Li, and M. Slonkosky, 2019. “Leaving households behind: Institutional investors and the US housing recovery,” Federal Reserve Bank of Philadelphia Working Paper No. 19-01.
- Mian, A., and A. Sufi, 2009. “The consequences of mortgage credit expansion: Evidence from the US mortgage default crisis,” Quarterly Journal of Economics, 124(4), 1449-96.
- Mian, A., and A. Sufi, 2018a. “ Credit supply and housing speculation,” National Bureau of Economic Research Working Paper No. w24823.
- Mian, A., and A. Sufi, 2018b. “Finance and business cycles: the credit-driven household demand channel,” Journal of Economic Perspectives, 32(3), 31-58.
- Mills, J., R. Molloy, and R. Zarutskie, 2019. “Large-scale buy-to-rent investors in the single-family housing market: The emergence of a new asset class,” Real Estate Economics, 47(2), 399-430.
- Nathanson, C.G., and E. Zwick, 2018. “Arrested development: Theory and evidence of supply-side speculation in the housing market,” Journal of Finance, 73(6), 2587-2633.

### Annex — Annex Table 1. Summary Statistics (Transaction level)
Sources: Zillow’s ZTRAX data and authors' calculations.
Note: “N” denotes number of observations and “sd” is the standard deviation; “min,” “p50,” and “max” denote the minimum, the median, and the maximum, respectively.

- Sales Price Amount88,063,434   2342172959622600 1650001000000049,763,801   264128272411260019500010000000
- Loan Amount86,589,692   1233241936000784621000000049,763,801   214585213621116690010000000
- Bank Lender88,063,434   0.19280.394500149,763,801   0.33750.4729001
- NonBank Lender88,063,434   0.23540.424300149,763,801   0.41440.4926001
- Cash Transactions88,063,434   0.41250.492300149,763,801   00000
- Corporate (cash)88,063,434   0.11130.314500149,763,801   00000
- Corporate88,063,434   0.13190.338300149,763,801   0.03220.1766001
- Out-of-State Buyers88,063,434   0.05680.231400149,763,801   0.04790.2135001
- Bank Float Rate88,063,434   0.05220.222400149,763,801   0.09220.2893001
- Bank Fixed Rate88,063,434   0.13850.345500149,763,801   0.24150.4280001
- Bank Interest Mortgage88,063,434   0.18970.392100149,763,801   0.33200.4709001
- NonBank Float Rate88,063,434   0.05700.231900149,763,801   0.10080.3011001
- NonBank Fixed Rate88,063,434   0.17840.382800149,763,801   0.31350.4639001
- NonBank Interest Mortgage88,063,434   0.23540.424300149,763,801   0.41440.4926001
- Float Rate88,063,434   0.10820.310700149,763,801   0.19130.3934001
- Fixed Rate88,063,434   0.31690.465300149,763,801   0.55500.4970011
- LTV48,470,597   0.83890.170800.821.548,470,597   0.83890.170800.822
- Bank LTV16,170,040   0.82030.172600.801.516,170,040   0.82030.172600.802
- NonBank LTV20,309,603   0.85580.156300.901.520,309,603   0.85580.156300.902
- Within-year Resale88,063,434   0.09570.294200149,763,801   0.03880.193100.001
- Conventional Loans49,769,959   0.30560.460700149,763,801   0.30560.460600.001
- Conforming Loans49,763,801   0.28610.451900149,763,801   0.28610.451900.001

(All Transactions | Loans Only)

### Annex Figures — Descriptions and Notes
- Annex Figure 1. Various Types of Non-Primary Home Buyers
  - Panels:
    1. Non-primary home buyers (percent of total; total volume)
    2. Share of short-term investors (within-year resales)
    3. Share of cash transactions
    4. Cross-state distribution of corporate cash
    5. Cross-state distribution of out-of-state buyers
    6. Cross-state distribution of second-home buyers
  - Sources: Zillow’s ZTRAX data and authors' calculations.
  - Note: Panel 1 depicts each type of non-traditional investments as a share of total transactions, weighted by the sales price. Corporate cash transactions are a subset of all cash transactions. In panels 4-6, 25th and 75th represent the 25th percentile and 75th percentile of the cross-state distribution.

- Annex Figure 2. Residential Mortgage Lending
  - Panels:
    1. All mortgage originations: shares by lender type
    2. All mortgage originations: banks vs. non-banks
    3. All mortgage originations: loan-to-value (LTV)
    4. Originations of risky mortgages by lender type
    5. Cross-state distribution of adjustable rate mortgages (ARMs)
    6. Adjustable rate mortgages (ARMs): banks vs. non-banks
  - Sources: Zillow’s ZTRAX data and authors' calculations.
  - Note: Non-banks include insurance, other financial intermediaries, subprime lenders, mortgage and finance companies. In panel 5, 25th and 75th represent the 25th percentile and 75th percentile of the cross-state distribution.

- Annex Figure 3. Shadow Banking Activity across US Counties
  - Panels:
    1. Share of shadow banking activity: 2008
    2. Share of shadow banking activity: 2018
  - Sources: Zillow’s ZTRAX data and authors' calculations.
  - Note: “Shadow banking” refers to non-deposit taking financial intermediaries such as mortgage and finance companies.
  - Legend bands shown: >90%, 76%-90%, 61%-75%, 46%-60%, 31%-45%, 16%-30%, 0-15%, No data

*Sources: References list and Annex (Zillow’s ZTRAX data and authors' calculations) as provided in the content unit.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020174-print-pdf.pdf_