Incidence, allocation, and efficiency costs of tenancy rent control

Hauck, Lukas; Stalder, Nicola; Büchler, Simon; von Ehrlich, Maximilian Working Paper Incidence, allocation, and efficiency costs of tenancy rent control Discussion Papers, No. 25-07 Provided in Cooperation with: Department of Economics, University of Bern Suggested Citation: Hauck, Lukas; Stalder, Nicola; Büchler, Simon; von Ehrlich, Maximilian (2025) : Incidence, allocation, and efficiency costs of tenancy rent control, Discussion Papers, No. 25-07, University of Bern, Department of Economics, Bern This Version is available at: https://hdl.handle.net/10419/333527 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Faculty of Business, Economics and Social Sciences Department of Economics Incidence, Allocation, and Efficiency Costs of Tenancy Rent Control Lukas Hauck, Nicola Stalder, Simon Büchler, Maximilian von Ehrlich 25-07 August, 2025 Schanzeneckstrasse 1 CH-3012 Bern, Switzerland http://www.vwi.unibe.ch DISCUSSION PAPERS Incidence, Allocation, and Efficiency Costs of Tenancy Rent Control Preliminary version – August 15, 2025 Lukas Hauck r ○b, Nicola Stalder r ○b,c, Simon B¨uchler r ○a,b, Maximilian v. Ehrlich r ○b aFarmer School of Business, Miami University, MIT Center for Real Estate bUniversity of Bern cIAZI AG – CIFI SA Abstract Tenancy rent control limits rent increases for sitting tenants while allowing market resets at vacancy. When demand grows or household composition differs across segments, spillovers raise rents in the unregulated market. We study its general equilibrium effects in Switzerland, where a nationwide regime meets large spatial variation. Linking administrative records on all households from 2010–2022 to detailed unit data and market rents, we estimate a structural sorting model with heterogeneous preferences, correcting for selection and price endogeneity. Counterfactual simulations show unregulated rents would be 8–21 percent lower, with the largest drops in supply-inelastic cities. Older, lower-income, and less educated households gain most, while newcomers face higher entry rents. The policy reduces mobility and induces space overconsumption, generating efficiency losses. Keywords: Rent Control, Residential Mobility, Inequality. JEL-codes: H7, H72, R23, R31, R38. ⋆We thank Mathias Amb¨uhl for his support in implementing the Hungarian algorithm. We benefited from numerous comments by discussants and participants at UEA European (2024, 2025) and North American (2024) meetings; AREUEA International Conference (2024), ARES 2024; ERES 2024; SSES Annual Congress (2024, 2025); the 2025 Lisbon Urban and Public Economics Workshop; the 2024 Swiss Workshop on Local Public Finance and Regional Economics; the 2024 CRED-MIT Workshop on Urban Economics and the Economics Brown Bag Seminar at the University of Bern. We thank the Swiss Federal Statistic Office and IAZI AG for data access and related support. Email addresses: [email protected] (Lukas Hauck r ○), [email protected] (Nicola Stalder r ○), [email protected] (Simon B¨uchler r ○), [email protected] (Maximilian v. Ehrlich r ○) 1. Introduction Housing affordability and availability are persistent concerns worldwide. In response to sharply rising rents, many jurisdictions have adopted or expanded rent control. Although politically popular and often motivated by distributional goals, rent control can distort price signals (Olsen, 1972; Sims, 2007; Monras and Garcia-Montalvo, 2023) and lead to inefficient allocation of housing (Suen, 1989; Glaeser and Luttmer, 2003; Bulow and Klemperer, 2012; Favilukis et al., 2023; Diamond et al., 2019a). A common form is tenancy rent control1which limits rent increases for sitting tenants, often tying them to inflation, while allowing market resets at vacancy. This widespread regulation applies in roughly 55% of OECD countries (OECD, 2024). Although often perceived as a milder intervention, a defining feature is the gap it creates between incumbent and market rents. Quantifying its distributional consequences and the scale of the resulting misallocation is challenging. It requires detailed microdata and a structural framework to recover the full counterfactual reallocation: how heterogeneous households would choose locations and units without regulation. These policies generate price spillovers between regulated and unregulated segments and alter the composition of households across neighborhoods and market tiers. Because such effects shape equity and efficiency outcomes, understanding them is central to informed housing policy debates. This paper develops a framework to evaluate the general-equilibrium effects of rent regulation and to quantify its incidence across households, locations, and unit types. We implement the framework in a representative setting, combining a structural demand model with comprehensive household–unit microdata covering the entire rental market of Switzerland. Switzerland is an ideal setting for our analysis. Over the past decade, strong housing demand has generated sizable and geographically heterogeneous gaps between incumbent and market rents. As As Figure 1 shows, these gaps are large in many growing cities worldwide: Zurich stands out with a gap of almost 40%, while the national average is about 20%, comparable to estimates for Vancouver. Comparable measures are not systematically available in most countries, and nationwide microdata that permit decomposing these gaps across households and locations are exceptional. Our data make this possible. We move beyond averages to examine the full distribution of gaps. Figure 2 shows that the gaps widen mechanically with tenancy duration (Panel A). The descriptive patterns reveal strong heterogeneity. Older households (Panel B) and lower1This form of rent control policy is also termed third generation rent control (Arnott, 2003). 2 Figure 1: Gap between private-sector rent-controlled (sitting-tenant) and market rents (new leases) 0 10 20 30 40 Rent gap (%) Paris Bern Switzerland Vancouver New York Toronto Los Angeles Stockholm Zurich Notes: Sources: Stockholm: Donner and Kopsch (2023); Los Angeles: Diamond et al. (2019a); Toronto and Vancouver: Canada Mortgage and Housing Corporation (2022); New York: NYC Department of Housing Preservation and Development (2021); Ireland: Residential Tenancies Board (2023); Paris: Observatoire des Loyers de l’Agglom´eration Parisienne (OLAP) (2022); Switzerland: own calculations. See Appendix A for details on the methodologies of each source. income households (Panel C) pay rents far below current market levels. These sizable gaps make it essential to understand both the distributional and efficiency effects of tenancy rent control. Crucially, such an assessment must account for spillover effects from controlled to uncontrolled markets (George Fallis, 1984; Early, 2000; Autor et al., 2014; Hahn et al., 2024). Incorporating these spillovers, this paper uses rich microdata and a novel empirical approach to estimate the full incidence of tenancy rent regulation. We study tenancy rent control under Switzerland’s stable federal regime. It allows free negotiation of initial rents but caps within-tenancy increases to mortgage rate passthrough, partial inflation adjustments, and value-enhancing investments. These rules shield sitting tenants from market pressures while newcomers pay market rents, creating 3 Figure 2: Gap between rent-controlled and market rents across household characteristics Panel A: Gap over Tenancy Duration 0 20 40 60 Occupation Bonus (as % of current rent) 0 10 20 30 40 Tenancy Duration Panel B: Gap over Age 10 20 30 40 Occupation Bonus (as % of current rent) 20 40 60 80 Age Panel C: Gap over Income Percentile 5 10 15 20 25 30 Occupation Bonus (as % of current rent) 0 2 4 6 8 10 Income Percentile Notes: Data from the Structural Survey (see Section 4). Observed rents are available for all households. We estimate a hedonic rent model using the subsample of households that moved within the survey year (“movers”) and use the estimated coefficients to impute market rents for units occupied by households that did not move (“non-movers”). Panels A–C report the difference between imputed market rent and actual rent paid, with 95% bootstrap confidence intervals. See Appendix B for details on methodology. 4 a persistent rent gap. Because the regulation applies nationwide, yet market conditions vary sharply across space, the setting provides rich variation for identifying generalequilibrium effects. Our empirical analysis combines administrative microdata on the universe of Swiss households from 2010–2022 with detailed housing unit characteristics and precise geolocation. We link the Population and Households Statistics to annual income records from the social security registry, structural dwelling attributes from the Federal Register of Buildings and Dwellings, and self-reported rents from the Structural Survey. This linkage allows us to observe household demographics, incomes, and exact rental payments for the regulated sector, and to impute market rents for these units. We build a structural residential choice model following McFadden (1978), Bayer et al. (2003), and Bayer et al. (2007). Households choose among a finite set of available units within their labor market region. Indirect utility depends on housing attributes, price, and a flexible measure of distance from the current residence, which interacts with life-cycle indicators such as having children or being retired. We allow for rich heterogeneity in tastes and price sensitivity across households. To avoid bias from regulated “stayers,” we estimate preferences using only movers, correcting for selection with a Heckman-style first stage. Using a Bartik shift–share instrument based on initial municipal employment shares and subsequent sectoral shocks, we address price endogeneity from unobserved quality. The estimated model yields household-level demand elasticities and willingness-topay measures for each unit. We then simulate the counterfactual without tenancy control by matching households to units through a Hungarian auction algorithm, which adjusts prices until markets clear. This procedure accounts for demand spillovers from regulated to unregulated segments, a key channel in the incidence of rent control. Comparing observed and counterfactual allocations gives the implicit subsidies accruing to incumbents, the corresponding burdens on newcomers, and the aggregate deadweight loss from misallocation. We decompose these outcomes by income, age, education, and geography, and study how regulation affects mobility and housing consumption. Our findings reveal sizable and uneven effects of tenancy rent control. The model estimates show large variation in households’ price elasticities and willingness to pay. Low-income and less-educated households are more sensitive to rent changes. In the counterfactual without regulation, and accounting for spillovers, market rents in the unregulated segment would be 8 to 21% lower than observed. The effects differ sharply across space. Urban labor markets with inelastic supply see the largest rent reductions. Rural areas experience smaller changes. Composition effects, where households with 5 low price elasticity cluster in the unregulated sector, amplify price pressures in cities. Supply responses mitigate but do not offset these pressures. The distributional impacts are pronounced. Incumbent tenants in regulated units capture large implicit subsidies. These subsidies are highest for older, lower-income, and less-educated households. Newcomers and mobile households bear the costs through higher rents. Across regions and demographic groups, the policy reshapes housing allocation and welfare distribution. Efficiency losses are concentrated in high-demand urban areas. This paper contributes to a growing body of literature analyzing the causal impacts of rent control policies on various outcomes.2Diamond et al. (2019a) show that San Francisco’s 1994 rent control law benefited incumbent tenants but imposed costs on future renters and unregulated units, effectively transferring wealth to long-term residents. Similarly, Ahern and Giacoletti (2022) find that St. Paul’s 2021 rent control policy sharply reduced property values, with wealthier tenants gaining the most, contrary to the policy’s intended redistributive goals. Mense et al. (2023) document that Germany’s rent cap lowered regulated rents but raised unregulated ones, reduced mobility, and led to inefficient redevelopment. Using a quasi-natural experiment, Cerqueiro et al. (2024) find that rent control removal disproportionately harms low-income workers, pushing them to city outskirts with higher rents and lower-quality jobs. By capping rents below market levels, regulation may also discourage new construction and maintenance (Downs, 1988) ultimately exacerbating housing shortages in the long run (Asquith, 2019). Our study extends this literature by quantifying how tenancy rent control redistributes housing consumption across income, education, and age groups, and by measuring its effects on mobility, composition of demand, and rural–urban price differentials. Studies by Nagy (1995); Gyourko and Linneman (1989); Ault et al. (1994); Munch and Svarer (2002) found reduced mobility among tenants in rent-controlled units. Further, this paper relates to the literature examining rent regulation externalities, such as the effects on land prices (Mense et al., 2019), housing quality (Olsen et al., 2004; Moon and Stotsky, 1993), crime (Autor et al., 2019), labor markets (Jiang et al., 2025), and gentrification (Autor et al., 2017). Our paper makes several contributions to the existing literature. First, we develop a structural framework that combines a residential sorting model with an assignment algorithm to recover general-equilibrium prices and allocations absent rent control. Second, we estimate the model using linked household–unit microdata covering the entire 2Malpezzi (2003) provides a concise literature review on the costs and benefits of rent control up to the early 2000s, while Kholodilin and Kohl (2023) present a more recent survey. 6 Swiss rental market from 2010 to 2022, yielding household-specific demand elasticities and willingness-to-pay measures. Third, we quantify spillovers from regulated to unregulated segments, finding that unregulated rents would be 8–21 percent lower in the counterfactual, with effects varying by location, demand growth, and supply elasticity. Fourth, we document large and uneven distributional impacts, as older and lowerincome households capture the largest subsidies. Fifth, we show that regulation reduces mobility and induces space overconsumption, generating substantial misallocation and deadweight loss. The remainder of the paper is structured as follows. Section 2 introduces the conceptual framework motivating our analysis. Section 3 derives the residential choice model. Section 4 describes the Swiss rent-regulation system and the household-level data. Section 5 reports the estimation results, and Section 6 presents the counterfactual analysis. Section 7 concludes. 2. Conceptual framework Rent control policies vary in design, particularly in terms of which market segments they target. The literature commonly distinguishes between two generations of rent control (Basu and Emerson, 2000; Arnott, 2003; Malpezzi, 2003). First-generation rent control policies apply broadly across the entire rental market and typically involve a strict rent freeze, setting rents below market-clearing levels (Arnott, 1995). In contrast, second-generation policies regulate only specific segments of the market, leaving other parts unaffected (George Fallis, 1984; Arnott, 1995; Basu and Emerson, 2000). We focus on tenancy rent control, a prevalent form of second-generation regulation (Basu and Emerson, 2000; Arnott, 2003). This approach restricts rent increases within an existing tenancy but permits landlords to reset rents to market levels between tenancies. It is widely used in jurisdictions like Germany, France, Ireland, Spain, Sweden, and several US and Canadian cities, including New York, San Francisco, Toronto, and Vancouver. As Arnott (2003) notes, tenancy rent control represents a compromise between heavy-handed regulation and unfettered market mechanisms, creating a distinct market equilibrium that balances reduced efficiency with increased tenure security. Tenancy rent control effectively segments the housing market into two distinct sectors: a regulated sector, in which incumbent tenants pay below-market rents (pr), and an unregulated sector, in which new tenants face market-clearing rents (pur). This segmentation generates spillover effects and leads to housing misallocation, such that pur exceeds the unified market-clearing price pthat would prevail in the absence of rent 7 The average utility component for housing unit hcan be further decomposed as δh= U X u=1 α0uxh,u −β0ph+ϕh,(4) where xh,u are the observable characteristics of the unit, phis its rent, and ϕhis an unobserved unit-specific error term. The coefficients α0ucapture how each attribute contributes to utility for the average household, while β0measures the average price sensitivity. 3.1. Estimation Selection stage: In the selection stage, we estimate the probability that household imoves with a logit model: M∗ i= K X k=1 γkzi k+νi,(5) where M∗ i= 1 if household imoves and 0 otherwise. The vector ziincludes current household characteristics and changes in these characteristics since the last period, and νiis an i.i.d. error term. The selection model implies P(Mi= 1 |zi, wi) = expz⊤ iγ+π wi 1 + expz⊤ iγ+π wi,(6) where wiis an excluded instrument that affects selection but is not included in the outcome equation and z⊤ iγ+π wi=PK k=1 γkzi k+π wi. From the estimated model, we compute the generalized IMR for household i: IMRi=   ϕ(z⊤ iγ) Φ(z⊤ iγ)if Mit = 1, ϕ(z⊤ iγ) 1−Φ(z⊤ iγ)if Mi= 0, (7) where ϕ(·) and Φ(·) are the probability density function and cumulative distribution function of the logistic distribution, respectively. We then include the IMR as an additional household characteristic in the residential choice model, following Heckman et al. (1998). First stage: We estimate (2) using a conditional logit model. For each household i, we draw n= 19 non-chosen alternatives from units vacated in the same year and 14 labor market region, and add the true choice to form the choice set.8We then estimate the parameters in λi hfrom (3) and the unit-specific fixed effects δh, maximizing the probability that each household chooses its observed unit h∗.9 The likelihood function is l=X iX h Ii hln Pi hwhere Pi h=exp (Vi h) Pj∈Ciexp Vi j,(8) where Ii hequals 1 if ichooses its true housing choice h∗and 0 otherwise, and Pi his the probability of individual ichoosing unit h. Following Bayer et al. (2003, 2007), we apply contraction mapping to ensure that demand for each unit does not exceed supply (see Berry et al., 1995). The marketclearing condition is X iPi h= 1 for all n(9) is met at every iteration of refitting the model. For the case at hand, the contraction mapping relevant for the presented application is simply δt+1 h=δt h−ln X i ˆ Pi h!.(10) Second stage: We decompose mean indirect utilities δhbased on the set of housing characteristics xhand prices ph. We estimate 4 while addressing the endogeneity of prices. Unobserved unit and neighborhood attributes (ϕh) may correlate with rents ph. To address this, we instrument phwith a shift–share variable based on sectoral employment in municipality min 2000. We combine the municipality’s sectoral employment shares with growth rates at the cantonal sector level. This approach uses nearly 20-year-old historical sectoral shares that interacted with aggregate sectoral shifts to generate exogenous variation in rents. Work on shift–share designs (Ad˜ao et al., 2018; Goldsmith-Pinkham et al., 2020; Borusyak et al., 2022) shows that validity requires predetermined exposure shares or random aggregate shocks. We rely on the first condition. Initial sectoral shares are persistent, shaped by natural amenities and market access, and unlikely to correlate with recent changes in local housing–supply shifters. 8We restrict non-chosen alternatives to the same labor market region, taking the decision of which labor market to move to as given. 9Bayer et al. (2007) and Bayer et al. (2003) provide a detailed description of the approach. 15 δh= U X u=1 α0uxh,u −β0bph+ϕh(11) 3.2. Willingness to pay With the estimated parameters in (3) and (4), we compute each household i’s willingness to pay (WTP) for each housing option h. To obtain the marginal WTP for a specific attribute x1, we hold utility constant and solve for the rent change ∆WTPithat offsets a change ∆x1. For the average household, WTP for unit hwith all characteristics identical to the average unit except x1is: WTPh= exp(¯p)×exp −αi 1 β0 (xh,1−¯x1),(12) where (xh,1−¯x1) is the deviation of unit hin characteristic x1from the mean, and ¯pis the average rent. If household idiffers from the average in characteristic z1by ∆z1, its WTP for unit h, again differing in x1from the mean, is: WTPi h= exp(¯p)×exp −α01 +α11∆z1 β0+β1∆z1 (xh,1−¯x1)(13) 3.3. Market clearing prices We compute equilibrium prices that clear the market for moving households, so that Dmove =Smove.10 We use a Hungarian auction algorithm, following Gilbert and French (2024) and Demange et al. (1986). Households value units according to the WTP parameters from the residential choice model. Households value units according to the WTP parameters from the residential choice model. The auction begins with all prices set to zero. Each household chooses the unit hthat maximizes their utility, i.e., the unit for which WTPih -phis largest. An equilibrium occurs when every unit is allocated to exactly one household, maximizing that household’s utility. If no such matching exists, the algorithm raises the prices of units with excess demand. It repeats this process, increasing the prices of scarce units, until it reaches an equilibrium allocation. The Hungarian auction is well-suited for this task because it ensures the final allocation is efficient and market-clearing. It finds a price vector and a matching where no householdunit pair would prefer to deviate, and no unit remains unmatched. 10We assume supply is limited to units vacated by movers and do not allow for new construction. 16 We add outside options with zero utility to capture the possibility that households leave or do not enter the regional labor market. We choose the number of outside options so that the algorithm’s mean market-clearing price matches the observed mean. In our preferred specification, about 3.5% of options are outside the mover market, consistent with average net immigration. We keep this share constant across simulations. The algorithm uses an I×Hmatrix of WTP values for all household-unit combinations within a labor market region. To speed computation, we randomly sample 800 i–h combinations per market, yielding an 800 ×800 matrix. We run the routine 20 times for each labor market region. 3.4. Overidentification check We interpret the prices obtained for movers as the unregulated market prices, denoted pUR h. To validate the model, we compare the predicted pUR hwith the observed rents for newly contracted leases on each unit. Tests for potential sampling bias ensure stable results. The panel structure of the dataset enables a stronger backtesting framework. The model is estimated for one year and used to predict pUR hfor the preceding or following year. Such out-of-sample predictions provide a robust check on the validity and stability of the estimates. 4. Institutional background and Data 4.1. Institutional background Switzerland, with over 60% of households renting, provides a pertinent setting for studying the effects of tenancy rent control. Table 1 compares tenure types in Switzerland, Germany, and the USA, highlighting notable differences. Switzerland’s high rental share stands in sharp contrast to the USA, which has a similar GDP per capita, and to Germany, its closest geographical and cultural neighbor.11 Renting in Switzerland spans all income levels: nearly 50% of households in the top income quintile rent, a pattern uncommon in the USA and Germany. Private households own almost half (47%) of rented apartments. Institutional investors hold 34%, cooperatives own 8%, real estate firms account for 7%, and public housing initiatives manage only 4% of the rental market. Swiss rental law, established in the 1980s and 1990s, builds on emergency rental regulations introduced during the two World Wars (Hausmann, 2016; Rohrbach, 2014). 11Werczberger (1997) and Bourassa and Hoesli (2010) discuss the drivers of Switzerland’s low homeownership rate. 17 Table 1: Comparison Tenure Types 2020 Income Quintile All households Lowest 2nd 3rd 4th Highest Switzerland Renter 60.8 68.9 65.8 63.1 57.4 49.0 Owner with mortgage 33.9 22.8 29.0 32.4 39.3 45.9 Owner outright 4.4 6.6 4.0 4.1 2.9 4.5 Other, unknown 0.9 1.6 1.2 0.4 0.4 0.6 Germany Renter 51.0 68.7 58.8 50.7 42.2 34.7 Owner with mortgage 25.6 11.8 17.4 26.0 33.4 39.1 Owner outright 19.8 14.4 19.3 19.5 21.6 24.2 Other, unknown 3.6 5.1 4.5 3.8 2.8 1.9 United States Renter 32.1 52.6 37.9 29.8 23.2 17.0 Owner with mortgage 40.4 17.4 31.1 42.9 52.4 58.4 Owner outright 25.7 26.2 28.8 25.9 23.5 24.1 Other, unknown 1.7 3.7 2.1 1.4 0.9 0.6 Source: OECD Affordable Housing Database. While landlords and tenants can freely negotiate initial rents, the law limits rent increases during existing tenancies.12 Rents within existing contracts can rise only under three conditions: (i) higher refinancing costs from rising mortgage rates, (ii) inflation adjustments, and (iii) passing on costs from value-enhancing investments such as major refurbishments or new installations. The Federal Agency for Housing sets a quarterly reference interest rate, which allows landlords to adjust rents when mortgage rates increase. Regulations also permit inflation adjustments, capped at 40% of the inflation rate. Tenants benefit from strong protections. Rental agreements are typically open-ended, and tenants can terminate with three months’ notice. Landlords face stricter rules and may terminate only for personal use or when major renovations require it.13 12Regulations nominally restrict rent increases between tenancies, but enforcement is weak. No central agency monitors compliance; households must appeal if they believe initial rents are excessive. With little access to information on previous rents, only 0.3% of new leases face legal challenges (BWO 2022). According to the common practice of arbitration courts, rent increases of up to 10% are not considered excessive. Even if a few appeals occur, regulation could still have a disciplinary effect on landlords. In this case, we would expect to observe bunching around the 10% cutoff. If so, we would expect rent changes to bunch around the 10% cutoff. Figure B.11 shows no such bunching, suggesting the regulation does not bind rent increases between tenancies. 13Cantonal and communal authorities can impose further restrictions, such as in Geneva, Basel, and Vaud. 18 4.2. Data sources We combine individual-level data from multiple administrative registries for 2010– 2022. Our core dataset is the Population and Households Statistics (STATPOP), which tracks all individuals and households in Switzerland over the 13 years. We link these records to yearly labor incomes from the Old Age and Survivors’ Insurance (AHV). Each individual is matched to their apartment, enabling us to merge structural characteristics from the Federal Register of Buildings and Dwellings maintained by the Swiss Federal Statistical Office (FSO). Further, we supplement this with the Structural Survey (SE), which covers a yearly sample of about 200,000 households. The SE provides socio-economic information such as education, residential status, and the net rent paid. To impute market rents for the unregulated segment (new contracts), we use the IAZI database on offered rents from 2004–2022.14 This dataset includes detailed housing characteristics, precise geocoordinates, and asking rents. Finally, we use a rich set of tenancy contracts curated by IAZI to back-test our results and assess model performance. Figure 4 shows the spatial distribution of average rents across Swiss municipalities. Large cities such as Zurich and Geneva, as well as their surrounding areas, have the highest rents. High rents also appear in municipalities near lakes and tourist destinations like Interlaken and Grindelwald. 14Appendix B details the imputation of market rents for the unregulated segment. 19 Figure 4: Spatial Distribution of Rents on Unregulated Market Segment 20 5. Results for the estimation of housing demand parameters We estimate heterogeneous housing preferences in three steps using (2) to (4). First, we model selection between movers and stayers and compute the Inverse Mills Ratio (IMR) for the next step. Second, we solve a conditional choice model by contraction mapping to recover heterogeneity in housing demand across household types. Third, we use the stage-two valuations to estimate inverse demand for the average household. These steps deliver willingness-to-pay (WTP) by household type and unit. We implement four specifications that vary housing characteristics xhand household characteristics zk. The comprehensive model uses seven household and 12 unit dimensions. The benchmark uses six and nine. The minimal uses four and seven. A data-based model retains only statistically significant variables from the first stage, excluding the IMR as a household characteristic. 5.1. Selection stage estimates Table 2 reports selection-model coefficients. Figure 5 plots IMR scores for movers and non-movers. The specification includes all household characteristics zkfrom (5) and indicators for changes in marital status (newly married or newly separated). The estimates show that transitions from single to married and married to divorced significantly raise moving propensity. Other coefficients align with expectations. The coefficient on retired may look surprising, but age fully identifies it. The grouped histograms reveal apparent IMR differences between movers and nonmovers. These differences are consistent with unobserved shocks tied to expectations, networks, or constraints. To address selection on unobservables when estimating the residential choice model on movers, we include the estimated IMR as a household characteristic in the indirect utility function, following Heckman (1979). The substantial overlap in IMR distributions provides common support for this control. 21 Figure 5: Distributions of inverse-Mills ratio Table 2: Selection model estimates Moved = True ∆ Separation 1.399∗∗∗ (0.012) ∆ Marriage 0.313∗∗∗ (0.006) Children U18 −0.042∗∗∗ (0.004) HH Income −0.044∗∗∗ (0.001) HH Age −1.332∗∗∗ (0.004) HH Size −0.084∗∗∗ (0.004) Retired 0.091∗∗∗ (0.005) Married −0.079∗∗∗ (0.003) Observations 115,933 Note: ∗p < 0.1; ∗∗ p < 0.05; ∗∗∗ p < 0.01 Figure 6: Predicted average valuations across models 5.2. First-stage estimates Table B.7 in Appendix B reports first-stage coefficients from (5). These coefficients are not directly interpretable in levels, but statistical significance reveals the direction 22 and relevance of unit and household characteristics. Higher-income households, households with children, and larger households demand more living space, ceteris paribus. Price elasticity is larger for households with children and smaller for higher-income and older households. Multicollinearity cautions against strict causal interpretation. For example, the positive link between having children and high-income neighborhoods weakens once we account for the negative link between household size and neighborhood income. Figure 6 compares recovered unit-level mean utilities δhfrom the baseline to three alternatives. The left panel uses the “data-based” model with only statistically significant interactions. The middle panel uses the minimal heterogeneity parameterization. The right panel uses the comprehensive heterogeneity specification. The δhalign tightly across panels. Correlations are high, and slopes are close to one. Unit rankings remain stable across specifications. This supports δhas a credible approximation to the latent mean indirect utility for the average household. Figure C.12 in Appendix B shows additional stability checks across specifications. 5.3. Second-stage estimates Table 3 decomposes the recovered δhinto observable unit characteristics. Column (1) reports OLS estimates, which suffer from rent endogeneity. Column (2) shows the first stage of the 2SLS in (11), instrumenting rent with a shift-share measure. Column (3) presents the second-stage 2SLS estimates. The instrument is strong (first-stage F= 612). The 2SLS price semi-elasticity is −7.9, so a 1% rent increase lowers demand by about 8%. Combining the price coefficient with the living-space coefficient yields an average substitution rate of 4.2/7.9≈0.53. Households pay about 0.53% higher rent for a 1% increase in living space. Other coefficients show clear trade-offs. Households pay about 0.14% more rent for a 1% increase in neighborhood income. They pay about 0.11% more rent for a 1% reduction in building age, consistent with a premium for newer stock. The small and statistically insignificant effect for distance to public transport reflects measurement limits that omit service frequency, travel times, and high baseline accessibility. These elasticities align with prior work. The estimated tax elasticity of rents is 0.28, within the Swiss range reported by Basten et al. (2017). 5.4. Willingness to pay estimates Table 4 reports willingness to pay (WTP) from the residential choice model. Column (1) gives the average household’s WTP for unit and neighborhood attributes in monthly 23 15% for Berner Oberland. Results from the second exercise reveal even bigger spillovers ranging up to 28% in Berner Oberland. Overall, the results for both exercises indicate sizable spillovers from regulation: by segmenting the market and protecting incumbent tenants, regulation pushes up rents in the unregulated segment by roughly 10% to 21%. Comparing this to the observed regulated–unregulated gap (pur/pr≈1.2) implies that, under a unified market, regulatedsegment households would face 8-16% higher rents, while unregulated-segment households would pay 10% - 21% less. Furthermore, the tenancy duration required to be an average net beneficiary of the regulation ranges between 8 and 12 years. Table 5: Median Simulated Prices and Relative Differences: Municipality Attachment Median pur 1−(p′/pur) 1 −(p′/pur) Only Units from All Units Unregulated Market Aareland 1 885 5% 10% Zentralschweiz 2 305 9% 12% Zurich 2 477 7% 10% Sopraceneri 1 680 2% 7% Sottoceneri 1 638 0% 5% Bodenseeregion 1 961 5% 6% Ostalpen 2 054 6% 4% Neuenburg 1 460 9% 9% Freiburg 1 916 5% 10% Biel-Jura 1 697 7% 10% Bern 1 757 12% 11% Westalpen 1 840 2% 1% Basel∗1 982 7% 9% Berner Oberland 1 726 15% 16% Swiss Median 2 088 8% 10% Note: The labor market regions of Geneva and Vaud are omitted because they have stricter rent regulations at the cantonal level. ∗The city of Basel introduced a stricter form of rent regulation in 2021; our analysis focuses on the year 2018. In the following, we explore the incidence of tenancy control in more detail across different household types and compute the welfare costs caused by misallocation induced by tenancy control. 6.1. Incidence To be completed: Incidence in different dimensions: at what age, what tenure duration, etc. We show for different labor market regions the incidence of tenancy control, i.e., at which duration of tenancy, what age, level of income, etc., households tend to be 30 Table 6: Median Simulated Prices and Relative Differences: Neighbourhood Attachment Median pur 1−(p′/pur) 1 −(p′/pur) Only Units from All Units Unregulated Market Aareland 2371 17% 21% Zentralschweiz 2856 22% 22% Zurich 3071 24% 24% Sopraceneri 2406 15% 18% Sottoceneri 2296 8% 11% Bodenseeregion 2544 23% 23% Ostalpen 2794 20% 18% Neuenburg 2005 24% 19% Freiburg 2506 19% 21% Biel-Jura 2173 22% 22% Bern 2397 22% 20% Westalpen 2496 23% 17% Basel∗2843 13% 14% Berner Oberland 2178 27% 28% Swiss Median 2731 21% 21% Note: The labor market regions of Geneva and Vaud are omitted because they have stricter rent regulations at the cantonal level. ∗The city of Basel introduced a stricter form of rent regulation in 2021; our analysis focuses on the year 2018. winners or losers of tenancy control. This is displayed in graphs similar to Figure 2 17 6.2. Misalloction To be completed: We follow two ways to compute misallocation: First, we measure the efficiency loss from tenancy control by quantifying the degree of overconsumption (e.g., in square meters) of regulated households compared to the counterfactual. More specifically, we compute x−x′and multiply with 1/2×(pUR −pR). Second, we compute a full utility based measure accounting for all dimensions of unit and households heterogeneity by computing the consumer surplus – estimates of WTP minus prices i.e. WTPi−pur,WTPi−pr,WTPi−p′. – in the observed and the counterfactual scenario. Therby, we quantify the full misallocation by labor market region, household type, and unit type. 17Essentially, the y-axis in Figure 2 panel (a) - (c) would be shifted down by 10-20 percentage points. Indicating that, e.g., households with a tenancy duration below 10 years are net losers of the regulation. 31 7. Conclusion Housing affordability and availability remain pressing concerns worldwide. Rent control policies continue to generate debate over their effectiveness and unintended consequences. This paper develops and implements a structural framework to quantify the general-equilibrium incidence of tenancy rent control. We combine administrative microdata on all Swiss households with a residential sorting model and a market-clearing assignment algorithm to recover counterfactual prices and allocations absent regulation. Accounting for spillovers, we find that unregulated rents would be 8–21 percent lower without regulation. Reductions are largest in urban areas with inelastic supply. Incumbent tenants capture large implicit subsidies, concentrated among older, lower-income, and less-educated households. Moving households face higher rents than they would in the absence of regulation. The policy reduces mobility and induces overconsumption of space, generating misallocation and deadweight loss, especially in tight markets. These results show that tenancy rent control shapes both the distribution and efficiency of housing consumption. It protects long-term tenants and delivers progressive transfers along some dimensions, yet it imposes substantial costs on mobile households and distorts allocation in high-demand areas. The proposed framework provides a transparent tool to evaluate these trade-offs and can be adapted to assess alternative policy designs or other institutional settings. 32 References Ad˜ao, R., Koles´ar, M., and Morales, E. (2018). Shift-share designs: Theory and inference. NBER Working Paper No. 24944. Ahern, K. R. and Giacoletti, M. (2022). Robbing Peter to Pay Paul? The Redistribution of Wealth Caused by Rent Control. SSRN Electronic Journal. Arnott, R. (1995). Time for Revisionism on Rent Control? Journal of Economic Perspectives, 9(1):99–120. Arnott, R. (2003). Tenancy rent control. Swedish Economic Policy Review. Asquith, B. (2019). Do Rent Increases Reduce the Housing Supply Under Rent Control? Evidence from Evictions in San Francisco. Ault, R. W., Jackson, J. D., and Saba, R. P. (1994). The Effect of Long-Term Rent Control on Tenant Mobility. Journal of Urban Economics, 35(2):140–158. Autor, D. H., Palmer, C. J., and Pathak, P. A. (2014). Housing Market Spillovers: Evidence from the End of Rent Control in Cambridge, Massachusetts. Journal of Political Economy, 122(3):661–717. EMPIRICS. Autor, D. H., Palmer, C. J., and Pathak, P. A. (2017). Gentrification and the Amenity Value of Crime Reductions: Evidence from Rent Deregulation. Autor, D. H., Palmer, C. J., and Pathak, P. A. (2019). Ending Rent Control Reduced Crime in Cambridge. AEA Papers and Proceedings, 109:381–384. Basten, C., von Ehrlich, M., and Lassmann, A. (2017). Income Taxes, Sorting and the Costs of Housing: Evidence from Municipal Boundaries in Switzerland. The Economic Journal, 127(601):653–687. Basu, K. and Emerson, P. M. (2000). The Economics of Tenancy Rent Control. The Economic Journal, 110(466):939–962. Bayer, P., Ferreira, F., and McMillan, R. (2007). A Unified Framework for Measuring Preferences for Schools and Neighborhoods. Journal of Political Economy, 115(4):588– 638. 33 Bayer, P., McMillan, R., and Rueben, K. (2003). An Equilibrium Model of Sorting in an Urban Housing Market: A Study of the Causes and Consequences of Residential Segregation. Berry, S., Levinsohn, J., and Pakes, A. (1995). Automobile Prices in Market Equilibrium. Econometrica, 63(4):841–890. Borusyak, K., Hull, P., and Jaravel, X. (2022). Quasi-experimental shift-share research designs. The Review of economic studies, 89(1):181–213. Bourassa, S. C. and Hoesli, M. (2010). Why Do the Swiss Rent? The Journal of Real Estate Finance and Economics, 40(3):286–309. Bourassa, S. C., Hoesli, M., and Peng, V. S. (2003). Do housing submarkets really matter? Journal of Housing Economics, 12(1):12–28. Bulow, J. and Klemperer, P. (2012). Regulated prices, rent seeking, and consumer surplus. Journal of Political Economy, 120(1):160–186. B¨uchler, S., von Ehrlich, M., and Sch¨oni, O. (2021). The amplifying effect of capitalization rates on housing supply. Journal of Urban Economics, 126:103370. Canada Mortgage and Housing Corporation (2022). Rental market survey, 2022. Average rents for turnover and non-turnover units in Toronto and Vancouver. Cerqueiro, G., Hacamo, I., and Raposo, P. S. (2024). Priced-Out: Rent Control, Wages, and Inequality. Demange, G., Gale, D., and Sotomayor, M. (1986). Multi-Item Auctions. Journal of Political Economy, 94(4):863–872. Diamond, R., McQuade, T., and Qian, F. (2019a). The effects of rent control expansion on tenants, landlords, and inequality: Evidence from San Francisco. American Economic Review, 109(9):3365–3394. Diamond, R., McQuade, T., and Qian, F. (2019b). The Effects of Rent Control Expansion on Tenants, Landlords, and Inequality: Evidence from San Francisco. American Economic Review, 109(9):3365–3394. EMPIRICS. Donner, H. and Kopsch, F. (2023). An income-distributional analysis of the rent control subsidy. Journal of Housing and the Built Environment, 38(4):2729–2749. 34 Downs, A. (1988). Residential rent controls. Washington, DC: Urban Land Institute. Early, D. W. (2000). Rent Control, Rental Housing Supply, and the Distribution of Tenant Benefits. Journal of Urban Economics, 48(2):185–204. EMPIRICS. Favilukis, J., Mabille, P., and Van Nieuwerburgh, S. (2023). Affordable Housing and City Welfare. The Review of Economic Studies, 90(1):293–330. George Fallis, L. B. S. (1984). Uncontrolled Prices in a Controlled Market: The Case of Rent Controls. American Economic Review. Gilbert, V. and French, R. (2024). Suburban Housing and Urban Affordability: Evidence from Residential Vacancy Chains |Joint Center for Housing Studies. Glaeser, E. L. (2003). Does Rent Control Reduce Segregation? SSRN Electronic Journal. Glaeser, E. L. and Luttmer, E. F. P. (2003). The misallocation of housing under rent control. American Economic Review, 93(4):1027–1046. Goldsmith-Pinkham, P., Sorkin, I., and Swift, H. (2020). Bartik instruments: What, when, why, and how. American Economic Review, 110(8):2586–2624. Gyourko, J. and Linneman, P. (1989). Equity and efficiency aspects of rent control: An empirical study of New York City. Journal of Urban Economics, 26(1):54–74. Hahn, A. M., Kholodilin, K. A., Waltl, S. R., and Fongoni, M. (2024). Forward to the past: Short-term effects of the rent freeze in berlin. Management Science, 70(3):1901– 1923. Hausmann, U. (2016). Vertragsfreiheit im Schweizer Mietrecht von 1804 bis 2014 unter besonderer ...(Hausmann, Urs) - Schulthess Buchhandlungen - Kommentare, Repetitorien, Fachinformationen. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica: Journal of the econometric society, pages 153–161. Heckman, J. J., Ichimura, H., and Todd, P. (1998). Matching As An Econometric Evaluation Estimator. The Review of Economic Studies, 65(2):261–294. Hilber, C. A. L. and Vermeulen, W. (2016). The impact of supply constraints on house prices in England. The Economic Journal, 126(591):358–405. Publisher: Oxford Academic. 35 Jiang, H., Quintero, L., and Yang, X. (2025). Does rent control increase tenant unemployment? Journal of Urban Economics, 149:103790. Kholodilin, K. A. and Kohl, S. (2023). Do rent controls and other tenancy regulations affect new construction? Some answers from long-run historical evidence. International Journal of Housing Policy, pages 1–21. EMPIRICS. Malpezzi, T. . (2003). A review of empirical evidence on the costs and benefits of rent control. Swedish Economic Policy Review. McFadden, D. (1978). MODELING THE CHOICE OF RESIDENTIAL LOCATION. Transportation Research Record, (673). Mense, A., Michelsen, C., and Kholodilin, K. A. (2019). The Effects of SecondGeneration Rent Control on Land Values. AEA Papers and Proceedings, 109:385–388. EMPIRICS. Mense, A., Michelsen, C., and Kholodilin, K. A. (2023). Rent Control, Market Segmentation, and Misallocation: Causal Evidence from a Large-Scale Policy Intervention. Journal of Urban Economics, 134:103513. EMPIRICS. Monras, J. and Garcia-Montalvo, J. (2023). The effect of second generation rent controls: New evidence from Catalonia. CEPR Discussion Paper Series. THEORY + EMPIRICS. Moon, C.-G. and Stotsky, J. G. (1993). The Effect of Rent Control on Housing Quality Change: A Longitudinal Analysis. Journal of Political Economy, 101(6):1114–1148. Munch, J. R. and Svarer, M. (2002). Rent control and tenancy duration. Journal of Urban Economics, 52(3):542–560. Nagy, J. (1995). Increased Duration and Sample Attrition in New York Citys Rent Controlled Sector. Journal of Urban Economics, 38(2):127–137. NYC Department of Housing Preservation and Development (2021). Nyc housing and vacancy survey 2021. Median rent comparison between rent-stabilized and unregulated private units. Observatoire des Loyers de l’Agglom´eration Parisienne (OLAP) (2022). Rapport annuel 2022. Data on average rents per square meter for new and ongoing private tenancies in Paris. 36 OECD (2024). Ph6.1 rental regulation. Technical report, OECD Directorate of Employment, Labour and Social Affairs. Accessed: 2025-08-15. Olsen, E. O. (1972). An econometric analysis of rent control. Journal of Political Economy, 80(6):1081–1100. Olsen, E. O., Tyler, C. A., King, J. W., and Carrillo, P. E. (2004). The Effects of Different Types of Housing Assistance on Earnings and Employment. Poterba, J. (1984). Tax subsidies to owner-occupied housing: An asset-market approach. The Quarterly Journal of Economics, 99(4):729–752. Poterba, J. and Sinai, T. (2008). Tax expenditures for owner-occupied housing: Deductions for property taxes and mortgage interest and the exclusion of imputed rental income. American Economic Review, 98(2):84–89. Quigg, L. (1993). Empirical testing of real option-pricing models. The Journal of Finance, 48(2):621–640. Residential Tenancies Board (2023). Rtb rent index q2 2023. Median rents for existing and new tenancies in Ireland’s Rent Pressure Zones. Rohrbach, H. (2014). Die Entwicklung des schweizerischen Mietrechts von 1911 bis zur Gegenwart. Rosen, H. S. (1985). Housing subsidies: Effects on housing decisions, efficiency, and equity. In Handbook of Public Economics, volume 1, pages 375–420. Elsevier. Saiz, A. (2010). The geographic determinants of housing supply. The Quarterly Journal of Economics, 125(3):1253–1296. Sims, D. P. (2007). Out of control: What can we learn from the end of Massachusetts rent control? Journal of Urban Economics, 61(1):129–151. Suen, W. (1989). Rationing and rent dissipation in the presence of heterogeneous individuals. Journal of Political Economy, 97(6):1384–1394. Werczberger, E. (1997). Home ownership and rent control in Switzerland. Housing Studies, 12(3):337–353. 37 Appendix A. Data Sources and Methodologies for Rent Gap Estimates Figure 1 shows estimates of the rent gap between market (new-lease) rents and controlled (sitting-tenant) rents from reputable sources, each drawing on different data and methods. Donner and Kopsch (2023) measure the Swedish gap by comparing regulated rents with estimated market-clearing rents using a hedonic regression that controls for unit characteristics and location. Diamond et al. (2019a) estimate the Los Angeles gap as the average rent discount for rent-controlled tenants under the city’s Rent Stabilization Ordinance, applying a hedonic regression to adjust for differences in units and neighborhoods. Canada Mortgage and Housing Corporation (2022) report the gaps for Toronto and Vancouver using a fixed-sample approach that compares average rents for the same set of units across years, distinguishing between turnover and non-turnover units. NYC Department of Housing Preservation and Development (2021) provide the New York estimate by reporting median rents for rent-stabilized and unregulated units. Residential Tenancies Board (2023) give the Ireland figure as median rents in existing versus new tenancies in Rent Pressure Zones. Observatoire des Loyers de l’Agglom´eration Parisienne (OLAP) (2022) supply the Paris figure, reporting average rent per square meter for new versus ongoing private tenancies, grouped by unit size category without hedonic controls. We calculate the Swiss gap ourselves from administrative rental data, comparing the average rent paid by sitting tenants to the estimated market rent for the same units. These estimates differ in methodology, but all reflect robust, locally accepted approaches. Readers should keep this heterogeneity in mind when comparing gaps across locations. 38 Appendix B. Data Appendix B.1. Estimating rents in unregulated segment for units in regulated segment First, we estimate how much a household benefits from the Swiss rent control policy. To do so, we calculate the gap between the (regulated) rent a household pays under its current lease agreement and the rent it would have to pay in the unregulated market segment. The latter represents the household’s replacement costs (e.g., renting its unit under current conditions). As the corresponding rent on the unregulated market segment is unknown, for each observed rent within an existing contract (pR ht), we estimate the rent on the unregulated market segment (pht) based on machine learning: We train a machine-learning model (XGBoost) to predict market rents. Below, we present a stylized version of the model to convey to the reader an intuitive understanding: PUR ht =β0+δt+θ′Xht +λ′Nht +κ′Mht +ϵht (B.1) The model is trained to predict the market rent (PUR ht ) for unit hat time t. It takes into consideration dwelling characteristics (X), neighborhood characteristics (N), and municipality characteristics (M) as well as year-fixed effects (δ). The model is trained on over 900,000 units advertised for rent online. The IAZI training data is used to determine parameters (ˆ β0,ˆ δt,ˆ θ, ˆ λand ˆκ). We use the trained model parameters to predict rents on the unregulated market segment for households within an existing tenancy: GAPht =ˆ PUR ht −PR ht (B.2) GAPht represents the benefit that the household living in unit hin neighbourhood n in municipality mat time tdraws from the rent regulation. It is the difference between the rent the household would pay under current conditions at time ton the unregulated segment and the rent the household pays. To document the distributional consequences of the regulation, we are interested in estimating the following stylized regression. We replace subscript h(for unit) by i, indicating the household living in unit h, where Zit is the households dimension of interest (e.g. income, age). GAP∗ ht =α+xhtβ+εht (B.3) However, we only have a noisy measurement of GAP∗ i: GAPht =GAP∗ ht +uht (B.4) 39 Appendix E. Further Results Table E.8: Determinants of Moving Propensity (1) (2) (3) GAP -0.662*** -0.394*** -0.263*** (20.02) (11.36) (7.08) ∆ Income 0.0581*** 0.0574*** 0.0362** (5.53) (5.41) (3.21) ∆ Children 0.498*** 0.400*** 0.375*** (12.39) (9.92) (9.25) Age -0.0194*** -0.0127*** -0.0140*** (27.97) (17.11) (18.32) Rooms -0.281*** -0.254*** -0.254*** (31.46) (28.32) (27.06) Tenancy Duration -0.0478*** -0.0435*** (29.04) (26.07) Labor Market, Education and Residence Permit FE Yes Observations 120,761 120,761 120,761 Notes: Robust t-statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01. Based on specification (3): At the 25th percentile of the GAP variable, the predicted probability of moving is 12.7%; at the 75th percentile, it decreases to 11.7%. 46 Figure E.14: Model evaluation – backtesting Panel (a): Predicted vs. observed rents by unit Panel (b): Predicted vs. observed housing expenditure share by household 47 Appendix F. More Descriptives 48