class: center, middle, inverse, title-slide # In Search of Peace and Quiet: The Heterogeneous Impacts of Short-Term Rentals on Housing Prices ## MEA 2021 ### Brett Garcia, Keaton Miller & John M. Morehouse --- # Research Overview __Research Q:__ How do short-term rentals (STRs) impact housing prices? → _A lot of papers:_ more STRs `\(\implies\)` higher prices! (why?)<sup>__†__</sup> .footnote[ __†__ See Calder-Wang (2019), Fonsesca (2019), Barron et al. (2020), Garcia-López et al. (2020), Valentin (2021) for examples ] --- count: false # Research Overview __Research Q:__ How do short-term rentals (STRs) impact housing prices? → _A lot of papers:_ more STRs `\(\implies\)` higher prices! (why?)<sup>__†__</sup> .footnote[ __†__ See Calder-Wang (2019), Fonsesca (2019), Barron et al. (2020), Garcia-López et al. (2020), Valentin (2021) for examples ] → Option value of owning a home increases `\(\implies\)` higher demand `\(\implies\)` higher prices. QED --- count: false # Research Overview __Research Q:__ How do short-term rentals (STRs) impact housing prices? → _A lot of papers:_ more STRs `\(\implies\)` higher prices! (why?)<sup>__†__</sup> .footnote[ __†__ See Calder-Wang (2019), Fonsesca (2019), Barron et al. (2020), Garcia-López et al. (2020), Valentin (2021) for examples ] → Option value of owning a home increases `\(\implies\)` higher demand `\(\implies\)` higher prices. QED ## Is that it? → Could STRs lower housing values under any circumstances? We think so --- count: false # Research Overview __Research Q:__ How do short-term rentals (STRs) impact housing prices? → _A lot of papers:_ more STRs `\(\implies\)` higher prices! (why?)<sup>__†__</sup> .footnote[ __†__ See Calder-Wang (2019), Fonsesca (2019), Barron et al. (2020), Garcia-López et al. (2020), Valentin (2021) for examples ] → Option value of owning a home increases `\(\implies\)` higher demand `\(\implies\)` higher prices. QED ## Is that it? → Could STRs lower housing values under any circumstances? We think so → .orange_b[Key idea:] The effect of STRs on residential amenities is ambiguous (more later) --- # Plan __Goal:__ Demonstrate theoretically that the effect of STRs on housing prices is ambiguous. Empirically substantiate this claim. --- count: false # Plan __Goal:__ Demonstrate theoretically that the effect of STRs on housing prices is ambiguous. Empirically substantiate this claim. 1) .orange_b[Background] 2) .orange_b[Highly stylized model] .ul[ → Illustrate a potential mechanism for ambiguity ] 3) .orange_b[Panel Regression]: demonstrate heterogeneous effects of Airbnb listings on housing pirces .ul[ → use a (not novel) instrument for Airbnb listings ] 4) .orange_b[Event Study]: .ul[ → Hilarious descriptive evidence for proposed mechanism ] --- count: false # Plan __Goal:__ Demonstrate theoretically that the effect of STRs on housing prices is ambiguous. Empirically substantiate this claim. 1) .orange_b[Background] 2) .orange_b[Highly stylized model] .ul[ → Illustrate a potential mechanism for ambiguity ] 3) .orange_b[Panel Regression]: demonstrate heterogeneous effects of Airbnb listings on housing pirces .ul[ → use a (not novel) instrument for Airbnb listings ] 4) .orange_b[Event Study]: .ul[ → Hilarious descriptive evidence for proposed mechanism ] __What we won't do:__ provide _causal_ evidence for our mechanism --- # STRs in the News <img src="figures/combined.png" width="80%" height="80%" style="display: block; margin: auto;" /> <img src="figures/rosville_ban.png" width="50%" height="50%" style="display: block; margin: auto;" /> --- # Regulating STRs → Neither California nor the U.S. federal government explicitly regulates STRs --- count: false # Regulating STRs → Neither California nor the U.S. federal government explicitly regulates STRs → We focus on Santa Monica's .orange_b[Ordinance 2484CCS], which was adopted by its City Council on May 12, 2015. Went into effect in June - According to staff reports and the text of the measure STRs removed "needed permanent housing from the market" and transient visitors could ".orange_b[disrupt the quietude]... of the neighborhoods and adversely impact the community" - Nominally banned owner-absent STRs, while allowing owner-present STRs to continue with additional costs (taxes, reporting, etc) --- count: false # Regulating STRs → Neither California nor the U.S. federal government explicitly regulates STRs → We focus on Santa Monica's .orange_b[Ordinance 2484CCS], which was adopted by its City Council on May 12, 2015. Went into effect in June - According to staff reports and the text of the measure STRs removed "needed permanent housing from the market" and transient visitors could ".orange_b[disrupt the quietude]... of the neighborhoods and adversely impact the community" - Nominally banned owner-absent STRs, while allowing owner-present STRs to continue with additional costs (taxes, reporting, etc) → Airbnb (and other platforms) quickly sued the city, which made enforcement difficult. City prevailed. --- class: inverse, center, middle # Model --- # Overview .orange_b[Goal]: Demonstrate that the effects of STRs on housing prices is ambiguous in _as parsimonious_ of a framework as possible .ul[ → Intentionally abstract from anything except our __main mechanism__: the interplay between STRs and residential amenities ] --- count: false # Overview .orange_b[Goal]: Demonstrate that the effects of STRs on housing prices is ambiguous in _as parsimonious_ of a framework as possible .ul[ → Intentionally abstract from anything except our __main mechanism__: the interplay between STRs and residential amenities ] __Main Ingredients:__ .ul[ → Static, discrete choice over nbhd , `\(j\)` , and owner-status `\(k \in \{o, a\}\)`. → Fixed quantity of housing in each nbhd, `\(H_j\)` and an exogenous number of rep. agents in market `\(N\)` ] --- # Model: Utility Utility for owning in nbhd `\(j\)`: `\begin{align*} u_{i,j,o} = \xi_j(k_j, f(str_j), g(str_j)) - P_j + \epsilon_{i,j,o}\\ u_{i,j,a} = \frac{R_j}{1-\delta} - P_j+ \epsilon_{i,j,a} \end{align*}` Where: `\(P_j\)` is the housing price, `\(R_j\)` is the rental price, `\(\delta\)` is the discount rate, and `\(\epsilon\)` is an iid preference shock --- count: false # Model: Utility Utility for owning in nbhd `\(j\)`: `\begin{align*} u_{i,j,o} = \xi_j(k_j, f(str_j), g(str_j)) - P_j + \epsilon_{i,j,o}\\ u_{i,j,a} = \frac{R_j}{1-\delta} - P_j+ \epsilon_{i,j,a} \end{align*}` Where: `\(P_j\)` is the housing price, `\(R_j\)` is the rental price, `\(\delta\)` is the discount rate, and `\(\epsilon\)` is an iid preference shock `\(\xi_j:\mathbb{R}^3 \to \mathbb{R}\)`: maps three local features to a scalar amenity value. .ul[ → `\(k_j\)`: fixed, time-invariant amenity level unrelated to STRs → `\(f(str_j)\)`: the "good" amenities that come with STRs (_e.g_ extra restaurants). __Assume:__ `\(f'>0\)` → `\(g(str_j)\)` the "bad" amenities that come with STRs (more partying?). __Assume:__ `\(g'>0\)` ] --- # Amenities Under the assumption that STRs impact local amenities positively and negatively, it follows that: `\begin{align*} \frac{\partial \xi_j }{\partial str_j} = \underbrace{\frac{\partial \xi_j}{\partial f}}_{+} \times\underbrace{f'(str_j)}_{+} + \underbrace{\frac{\partial \xi_j }{\partial g }}_{-}\times\underbrace{g'(str_j)}_{+} \end{align*}` --- count: false # Amenities Under the assumption that STRs impact local amenities positively and negatively, it follows that: `\begin{align*} \frac{\partial \xi_j }{\partial str_j} = \underbrace{\frac{\partial \xi_j}{\partial f}}_{+} \times\underbrace{f'(str_j)}_{+} + \underbrace{\frac{\partial \xi_j }{\partial g }}_{-}\times\underbrace{g'(str_j)}_{+} \end{align*}` .orange_b[Key Idea]: __The net impact of STRs on residential amenities is ambigious__ → STRs may have positive impacts on residential amenities (added restaurants) → STRs also may have negative impacts on residential amenities (more noise) --- # Partial Derivatives Consider a regulation that changes the return on holding a STR. What happens to equilibrium housing prices? $$ \frac{\partial P_j^\star}{\partial R_j} = \frac{ 1}{ \exp( \frac{R_j}{1-\delta} ) + \exp( \xi_j(\cdot) ) ) } \left( \frac{\exp( \frac{R_j}{1-\delta})}{1-\delta} + \exp( \xi_j(\cdot)) \times \frac{\partial \xi_j}{\partial str^\star} \times \frac{\partial str^\star_j}{\partial R_j} \right) \\ $$ __3 (non-trivial cases):__ .orange_b[Case 1:] `\(\frac{\partial \xi_j}{\partial str^\star} > 0 \implies \frac{\partial P_j^\star}{\partial R_j}>0\)`: - _STRs create net-positive amenities_ --- count: false # Partial Derivatives Consider a regulation that changes the return on holding a STR. What happens to equilibrium housing prices? $$ \frac{\partial P_j^\star}{\partial R_j} = \frac{ 1}{ \exp( \frac{R_j}{1-\delta} ) + \exp( \xi_j(\cdot) ) ) } \left( \frac{\exp( \frac{R_j}{1-\delta})}{1-\delta} + \exp( \xi_j(\cdot)) \times \frac{\partial \xi_j}{\partial str^\star} \times \frac{\partial str^\star_j}{\partial R_j} \right) \\ $$ __3 (non-trivial cases):__ .orange_b[Case 1:] `\(\frac{\partial \xi_j}{\partial str^\star} > 0 \implies \frac{\partial P_j^\star}{\partial R_j}>0\)`: .orange_b[Case 2:] `\(\frac{\partial \xi_j}{\partial str^\star} < 0\)` and `\(\frac{\exp( \frac{R_j}{1-\delta})}{1-\delta} > \left | \exp( \xi_j(\cdot)) \times \frac{\partial \xi_j}{\partial str^\star} \times \frac{\partial str^\star_j}{\partial R_j} \right | \implies \frac{\partial P_j^\star}{\partial R_j}>0\)`: - _STRs create net-negative amenities and the magnitude of the change in the marginal benefit to absentee landlords exceeds the magnitude of the change in marginal benefit to owner-occupiers_ --- count: false # Partial Derivatives Consider a regulation that changes the return on holding a STR. What happens to equilibrium housing prices? $$ \frac{\partial P_j^\star}{\partial R_j} = \frac{ 1}{ \exp( \frac{R_j}{1-\delta} ) + \exp( \xi_j(\cdot) ) ) } \left( \frac{\exp( \frac{R_j}{1-\delta})}{1-\delta} + \exp( \xi_j(\cdot)) \times \frac{\partial \xi_j}{\partial str^\star} \times \frac{\partial str^\star_j}{\partial R_j} \right) \\ $$ __3 (non-trivial cases):__ .orange_b[Case 1:] `\(\frac{\partial \xi_j}{\partial str^\star} > 0 \implies \frac{\partial P_j^\star}{\partial R_j}>0\)`: .orange_b[Case 2:] `\(\frac{\partial \xi_j}{\partial str^\star} < 0\)` and `\(\frac{\exp( \frac{R_j}{1-\delta})}{1-\delta} > \left | \exp( \xi_j(\cdot)) \times \frac{\partial \xi_j}{\partial str^\star} \times \frac{\partial str^\star_j}{\partial R_j} \right | \implies \frac{\partial P_j^\star}{\partial R_j}>0\)`: .orange_b[Case 3:] `\(\frac{\partial \xi_j}{\partial str^\star} < 0\)` and `\(\frac{\exp( \frac{R_j}{1-\delta})}{1-\delta} < \left | \exp( \xi_j(\cdot)) \times \frac{\partial \xi_j}{\partial str^\star} \times \frac{\partial str^\star_j}{\partial R_j} \right | \implies \frac{\partial P_j^\star}{\partial R_j} < 0\)`: - __STRs create net-negative amenities and the decrease in the marginal benefit to owner-occupiers exceeds the decrease in the marginal benefit to absentee landlords.__ --- # Recap Built a parsimonious model that suggests the effects of STRs on housing prices is ambiguous .ul[ → Intentionally made it as simple as possible. .orange_b[Minimal assumption:] STR impact on amenities is ambiguous → Model makes it clear that that `\(\frac{\partial P_j^\star}{\partial R_j}<0\)` is an edge case but still possible → __No sharp predictions about__ `\(\xi\)` ] --- count: false # Recap Built a parsimonious model that suggests the effects of STRs on housing prices is ambiguous .ul[ → Intentionally made it as simple as possible. .orange_b[Minimal assumption:] STR impact on amenities is ambiguous → Model makes it clear that that `\(\frac{\partial P_j^\star}{\partial R_j}<0\)` is an edge case but still possible → __No sharp predictions about__ `\(\xi\)` ] ## Question → Is this just a theoretical curiosity? We turn to test our theory empirically. --- class: inverse, center, middle # Empirics: Panel Regressions --- # Data overview We combine data from multiple sources: - .orange_b[Zillow]: Monthly housing price indices (ZHVI) at _zip code level_ - .orange_b[Inside Airbnb + Tomslee]: Publicly available, scraped Airbnb listings .ul[ → Scraped at irregular intervals -- combine them to get largest possible sample → Characteristics of listings, location accurate to within 500m ] __Focus Area:__ Los Angeles County. Estimation window: July 2015-June 2017 --- # Specification Using our model to guide the empirics, we estimate: `\begin{align} \log(ZHVI_{zjt}) = \beta_{0j} + \beta_{1j}\log(listings_{zjt}) + FX + \epsilon_{zjt}%\label{eq:est_eqn-1} \\ %\log(ZHVI_{jct}) = \beta_0 + \beta_1\log(R_{jct}) + FX + \eta_{jct}\label{eq:est_eqn-2} \end{align}` where: → `\(ZHVI_{jct}\)` is the Zillow Home Value Index for zip code `\(z\)` in jurisdiction `\(j\)` at year-month time `\(t\)` → `\(listings_{zjt}\)` is the number of Airbnb listings → `\(FX\)` is a set of fixed effects → `\(\epsilon_{zjt}\)` is an unobservable We use an instrument from Barron et. al (2020) for listings: - Interact google search hits for Airbnb, `\(g_t^{air}\)`, with num. of restaurants and accomodations estab (NAICS 72) in 2010, `\(b_{zj}^{2010}\)` --- # Results <img src="figures/iv_results.png" width="75%" height="75%" style="display: block; margin: auto;" /> --- # Results <img src="figures/coefficient_histogram.png" width="80%" height="80%" style="display: block; margin: auto;" /> --- class: inverse, center, middle # Empirics: Event Study Evidence --- # Idea and Data We provide _descriptive_ evidence of our proposed mechanism, using calls to police → .orange_b[Hypothesis:] Nuisance calls to police decline after STRs are regulated in SM (may in part negative estimated coefficient) → .orange_b[Data:] Santa Monica Open Data Project for 2013--2019 .ul[ → Geolocated calls with reason for the call → Define a call `\(k\)` as being party related if it was for `loud music`, `public intoxication` or `noise complaint` → Event study with pre-post as policy date STR regulation went into place ] --- # Listings over time <img src="figures/num_listings.png" width="100%" height="100%" style="display: block; margin: auto;" /> --- # Event Study <img src="figures/crime_event.png" width="80%" height="80%" style="display: block; margin: auto;" /> --- # Public Intoxication <img src="figures/public_intox.png" width="80%" height="80%" style="display: block; margin: auto;" /> --- # Conclusion __Main takeaways__ → Literature has exclusively focused on STRs and _rising_ housing prices → We point out that this is probably right .orange_b[on average] .ul[ - Averages mask heterogeneity! - Less important if this heterogeneity means `\(\frac{\partial P_j^\star}{\partial str_j^\star}\)` still always positive ] __Policy implication__: regulating STRs in the name of housing affordability _may_ backfire. → Much more work to be done here, though --- class: middle, inverse, center # Thank you!! Questions? Comments? Concerns? jmorehou@uoregon.edu https://www.johnmmorehouse.com/ --- exclude: true