1# RECORD d3cc2b5ffbf39645fc47562257a85c240474e3d72# TITLE: Investment incentives of rent controls and gentrification: Evidence from German micro data3# YEAR: 2024.0 VENUE: Real Estate Economics DOI: 10.1111/1540-6229.124784# AUTHORS: Vera Baye; Valeriya Dinger5# PAGES: 42 CANDIDATE TABLE PAGES (1-based): [8, 12, 21, 22, 23, 25, 33, 34, 35, 36, 37, 38, 39, 40, 41]6# ABSTRACT: We empirically document that the effectiveness of the German rent control introduced in 2015 in achieving rental housing affordability is limited. Exploring the reasons for this limited effectiveness, we focus on the impact of the rent control on the yield on rental housing investments proxied by rent‐price ratios, which we derive by predicting sale prices to rental objects based on a hedonic model using micro‐level quotes on rental and sale listing. Exploiting the temporal, regional, and object‐specific variation generated by the design of the rent control, we identify a causal negative effect of the rent control on the yield of rental objects subject to the regulation. Furthermore, we zoom into the spillovers across regulated objects and objects in the affected markets that were exempt from the regulation and find rising yields for the exempted objects, suggesting that the regulation contributed to gentrification via a shift of rental housing supply away from the regulated segment.789===== PAGE 1 (score 3, numeric-paren 0, table-words 0, var-words 3) =====1011DOI: 10.1111/1540-6229.1247812O R I G I NA L A RT I C L E13Investment incentives of rent controls and14gentrification: Evidence from German micro15data16Vera Baye117Valeriya Dinger1,2181Institute of Empirical Economic19Research, School of Business20Administration and Economics,21University of Osnabrueck, Osnabrueck,22Germany232Accounting and Finance Department,24Leeds University Business School, Leeds,25England26Correspondence27Vera Baye, University of Osnabrueck,28Osnabrueck, Germany.29Email: vera.baye@uni-osnabrueck.de30Abstract31We empirically document that the effectiveness of the32German rent control introduced in 2015 in achieving33rental housing affordability is limited. Exploring the34reasons for this limited effectiveness, we focus on the35impact of the rent control on the yield on rental hous-36ing investments proxied by rent-price ratios, which we37derive by predicting sale prices to rental objects based38on a hedonic model using micro-level quotes on rental39and sale listing. Exploiting the temporal, regional, and40object-specific variation generated by the design of the41rent control, we identify a causal negative effect of42the rent control on the yield of rental objects subject to43the regulation. Furthermore, we zoom into the spillovers44across regulated objects and objects in the affected mar-45kets that were exempt from the regulation and find rising46yields for the exempted objects, suggesting that the reg-47ulation contributed to gentrification via a shift of rental48housing supply away from the regulated segment.49This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and50reproduction in any medium, provided the original work is properly cited.51© 2024 The Authors. Real Estate Economics published by Wiley Periodicals LLC on behalf of American Real Estate and Urban Economics52Association.53Real Estate Econ. 2024;52:843–884.54wileyonlinelibrary.com/journal/reec55843565758===== PAGE 8 (score 23, numeric-paren 0, table-words 5, var-words 11) =====596085061BAYE and DINGER62TA B L E 163Descriptive statistics of rental and sale objects.64Variable65Mean66Std. Dev.67Min68p2569p5070p7571Max72N73A. Rental objects74Year of construction7519717634.80771800781957791975801997812020825,662,35883Living space8476.728531.57886108756.1688728991901000915,662,35892Number of rooms932.68940.9995196297398399101005,662,358101Rent per square meter1028.421033.031043.661056.101067.801071010820.451095,662,358110B. Sale objects111Year of construction112197811333.55114180011519651161984117200311820201195,609,255120Living space12185.3712238.3812310124611257912610112710001285,609,255129Number of rooms1302.941311.111321133213431353.5136101375,609,255138Price per square meter1392269140128414139614213161431970144290914577151465,609,255147Note: This table shows descriptive statistics of the quotes on rental objects and on objects for sale from the micro-level housing148dataset based on German residential real estate advertisements. The data are a systematic collection of all objects that were offered149for rent on the internet platform ImmobilienScout24 with a monthly frequency covering the period from January 2010 to December1502019.151For the empirical analysis, we use data on apartments for sale and apartments for rent. The152raw data provide a high number of observations. To ensure the quality of the analyzed dataset,153incomplete advertisements that do not contain a net rent could not be included in the analysis.154Moreover, we only consider objects with a listed postcode area,7 that were built in 1800 or later,155with a minimum number of rooms of one, reported living space of at least ten square meters, and156that do not belong to the cheapest or most expensive 1% in terms of price per square meter. The157average apartment for rent in our dataset is located in a building that was built in 1971, has three158rooms, and a living space of 76.72 square meters. The monthly net rent of this object would be159645.98 Euros, as the rent per square meter is 8.42 Euros (Table 1, Panel A). The average apartment160for sale has three rooms as well, is slightly newer, larger, and has an average sale price per square161meter of 2269 Euros (Table 1, Panel B).162To ensure the representativeness of our dataset, we compare the distribution of the key property163characteristics living space and year of construction in the latest available German census data164(German Federal Statistical Office, 2023) from the year 2011 to our rental and sale data from the165same year. Figure A1 shows that our data compare to the census data, thus, indicating that our166dataset includes a representative sample of the German housing market.8 Recent studies based on167these data were, for example, published by Breidenbach, Eilers et al. (2022) analyzing the temporal168dynamics of rent prices due to the rent control, Breidenbach, Cohen et al. (2022), Pommeranz and1697 The German postcode system is a pure number system consisting of five digits. By only considering objects in five-digit170postcode areas, the quality of the observations considered should be ensured. Studies using the same data set use similar171procedures for quality assurance (Breidenbach, Eilers, et al., 2022; Deschermeier et al., 2016; Eilers, 2017).1728 One deviation appears in the share of newer dwellings built after 2006 whose share is larger in the ImmobilienScout24173dataset than in the census data. These deviations could be expected as the census depicts the characteristics of the stock of174existing housing, while the ImmobilienScout24’s data are representative of the dwellings currently offered for sale or rent.175As newer and more standard properties are more likely to be sold and standard properties more likely to be rented out,176the fact that we observe these differences relative to the census data is not surprising. In addition, the ImmobilienScout24177dataset does not perfectly mirror the stock of housing units in very small municipalities where only a few dwellings are178offered for rent or sale. This applies especially to larger apartments.179 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License180181182===== PAGE 9 (score 12, numeric-paren 0, table-words 1, var-words 10) =====183184BAYE and DINGER185851186Steininger (2021), Klick and Schaffner (2019), Eilers (2017), and Deschermeier et al. (2016) who187focus on recent developments in the housing market for rentals and sales.188We are aware that the asking prices might deviate from actual transaction prices, but as Dinkel189and Kurzrock (2012), Kholodilin et al. (2016), and Lyons (2013) emphasize, asking price data show190reliable price trends. Especially for the advertised rent prices, significant deviations from the trans-191actions do not need to be assumed because, as Zhu (2005) emphasizes, bargaining over rent prices192is relatively rare, especially in regions with a demand overhang. This is particularly true for Ger-193many, where it is not common to negotiate apartment rental prices (Breidenbach, Eilers et al.,1942022), and landlords generally obtain their asking prices (Deschermeier et al., 2016). Of course,195although we use a large and representative data set on micro-level, we might still miss some rental196agreements that may be used to bypass the regulation, as for example shadow rental agreements197and subletting.198We match the micro-level real estate data to self-collected data from the federal state’s laws on199the application of rent controls, which were introduced on the municipality level at various points200in time by the federal states. Further data on regional characteristics, local economic activity, and201socioeconomic variables at the municipality and district level9 are collected from the “Genesis”202regional data platform maintained by the German Federal Statistical Institute (Statistisches Bun-203desamt) and the German Federal Institute for Research on Building, Urban Affairs and Spatial204Development which offers indicators of spatial and urban development (INKAR). Broad sum-205mary statistics for regulated and unregulated municipalities of those variables used as controls in206our empirical analyses are provided in Table 2.2074208THE EFFECTS OF RENT CONTROL ON RENTAL HOUSING209AFFORDABILITY210As a point of departure, we first evaluate the effectiveness of the German rent control by analyzing211the effect of the introduction of this policy instrument on the rent-income ratio, which we use to212proxy the affordability of rental housing. We explicitly do not focus on the development of the rent213prices but rather on the rent-income ratio, because we intend to rule out an increased demand for214higher living space and housing standards which might occur due to growing income and focus215on the intended target of improving housing affordability. The variable displays the proportion of216the household income that is spent on rental payments proxied by the mean of the yearly net rent217of the newly offered flats per district.10 Our data show that despite the implementation of the rent2189 Municipalities are cities or towns with an own local government, they define the lowest level of territorial division in219Germany. In Germany exists more than 11,000 municipalities in 401 districts. The German districts are at an intermedi-220ate level of administration between the German federal states and the municipality governments. Cities with more than221100,000 inhabitants do not usually belong to a district, but form their own district.22210 As described above, this indicator is also used to identify tight housing markets (Simons et al., 2020).When using the223rent-income ratio in our analysis, we are aware of the concerns of Favilukis et al. (2023) who underline that this ratio224must be interpreted carefully because rent-income ratios reflect equilibrium rents and the income of those people who225have sorted themselves into each area in the spatial equilibrium. In their dynamic stochastic spatial equilibrium model, an226increasing rent-income ratio can be a sign of an effective regulation as misallocation decreases and low-income households227can move to more expensive areas, like city centers, where they could not get an apartment before a specific regulation228was implemented. However, there are various reasons why the use of the rent-income ratio is still reasonable in our anal-229ysis. First, the rent control in Germany is not targeted and does not include an income qualification of tenants, which is230crucial in the framework of Favilukis et al. (2023). Second, the rent control is introduced in tense housing markets with231 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License232233234===== PAGE 12 (score 22, numeric-paren 4, table-words 3, var-words 12) =====235236854237BAYE and DINGER238TA B L E 3239The effect of rent control on the rent-income.240(1)241(2)242Rent-income ratio243Rent-income ratio244ATT2450.0122***2460.0515***247(0.0028)248(0.0075)249Treatment variable250Discrete251Continuous252Note: This table presents the results of the staggered difference-in-differences analyses to estimate the effect of the introduction253of rent controls on the rent-income ratio. The dependent variable is the rent-income ratio, which displays the proportion of the254household income that is spent on rental payments proxied by the yearly mean net rent of the newly offered flats per district.255The included covariates cover region-specific dummy variables on urbanization and location in Western Germany, as well as the256population density, the number of students, and the number of building completions. The overall treatment effect in Column 1257is estimated using the doubly robust difference-in-differences estimator based on inverse probability and weighted least squares258estimator from Sant’Anna and Zhao (2020). In Column 2, we extend the staggered design from Column 1 with the treatment259intensity, which is modeled by the share of regulated objects per district. The sample covers the observation period from 2010260to 2019. Standard errors are displayed in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level,261respectively.262F I G U R E 1263Dynamic effects of rent control on rent-income ratio. [Color figure can be viewed at264wileyonlinelibrary.com]265level of income spent on rental payments, the estimator for the ATT will be significantly negative266or insignificant.267In Table 3, Column 1, the overall treatment effect of the introduction of rent control on the268rent-income ratio is reported, which is calculated as the average of all ATTs weighted by group269size. It shows that after the introduction of rent control, the rent-income ratio in regulated areas270increases on average by 1.22 percentage points. Thus, rental payments seem to increase faster271than household incomes in these tight markets after the implementation of the rent control and272housing affordability is not improved.273To test for the conditional parallel trends, we estimate Equation (1) including the covariates274captured in 𝑋across all cohorts for each period relative to the period first treated. Figure 1 shows275 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License276277278===== PAGE 13 (score 17, numeric-paren 0, table-words 1, var-words 15) =====279280BAYE and DINGER281855282the estimated ATTs. The coefficients suggest that in some pretreatment periods, the rent-income283ratio already lies above the level in the never-treated municipalities, which can be explained by284the fact that rent control was introduced in tense housing markets. However, these deviations are285of small scale, compared to the growing difference in rent-income ratios that occurs 1 year after286the implementation of the regulation and then increases from period to period. Thus, although287the preexisting differences reflect the fact that the treatment is not introduced randomly, after the288implementation of the treatment we see an increase in these differences in the rent-income ratios289of treatment and control districts.290In addition to the standard staggered difference-in-differences estimation, we modify the esti-291mation from Equation (1) and estimate the effect of the treatment intensity proxied by the292rent-controlled share of apartments in treated districts. The results presented in Table 3, Column2932, reveal that the increase of the rent-income ratio in treated districts increases with the share of294rent-controlled apartments in the district. We next zoom into the potential reasons for this failure295of the regulation and exploit the micro dimension of our dataset to investigate how rent control296affects the supply side of the housing market by affecting the yields on rental housing.2975298RENT CONTROL AND THE RENT-PRICE RATIO299Our results so far indicate that the effectiveness of rent control in terms of achieving affordabil-300ity of rental housing is limited. In this section, we explore the granularity of our real estate data301and identify a channel through which rent control might affect the supply side of the housing302market: the reduction of the yield of investments in regulated rental housing. As the provision303of affordable housing naturally depends on the supply of living space, we put the yield on rental304housing, proxied by the rent-price ratio, in the center of the analysis. As it indicates the attractive-305ness of investing in rental housing, the rent-price ratio reflects the investment incentives which306are needed to address supply shortages in the long term. Using this ratio allows us a more holistic307view of the dynamics of the incentives to supply rental housing relative to a simple focus on rent308levels. This is particularly the case as we look at periods when not only rents but also house price309levels are changing substantially.3105.1311Variables and descriptive analysis3125.1.1313The rent-price ratio314The rental housing yield is measured by the rent-price ratio. This is calculated for each rental315object from the reported yearly net rent and the potential sale price which we derive from esti-316mating a hedonic model since each dwelling in our dataset is advertised either as a rental or as a317sale object:318𝑝𝑟𝑖𝑐𝑒𝑖= 𝛽𝑋𝑖+ 𝛿319(320𝑅𝐶𝑖× 𝐴𝑑× 𝐵𝑦321)322+ 𝜀𝑖.323(2)324Here, 𝑝𝑟𝑖𝑐𝑒𝑖is log of the purchase price per square meter of object 𝑖and 𝑋includes a rich set of325property characteristics for each object in district 𝑑in year 𝑦. Specifically, we include the log of the326dwellings’ year of construction, the log of the number of rooms, a vector of dummy variables that327indicate the object condition, and if a garden or balcony belongs to the apartment. 𝛿estimates the328 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License329330331===== PAGE 21 (score 131, numeric-paren 104, table-words 0, var-words 15) =====332333BAYE and DINGER334863335TA B L E 5336The effect of the rent control on the rent-price ratio.337(1)338(2)339(3)340(4)341Variables342Rent-price343ratio344Rent-price345ratio346Rent-price347ratio348Rent-price349ratio350Treatment municipality3510.0260***352(0.0093)353Regulated objects in regulated mun354–0.1330***355–0.2788***356–0.1944***357–0.1471***358(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖)359(0.0192)360(0.0424)361(0.0378)362(0.0228)363Unregulated objects in regulated mun3640.2522***365(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖) (0.0288)366Object-specific variables:367Base yield3680.9551***3690.9754***3700.9809***3710.9768***372(0.0217)373(0.0282)374(0.0269)375(0.0162)376Year of construction377–0.0003378–0.0008***379–0.0014***380–0.0012***381(0.0002)382(0.0001)383(0.0002)384(0.0002)385Living space3860.00013870.0015*3880.0017*389–0.0012390(0.0009)391(0.0008)392(0.0009)393(0.0009)394Number of rooms395–0.2310***396–0.2333***397–0.2424***398–0.2210***399(0.0053)400(0.0073)401(0.0063)402(0.0052)403Basement (dummy)404–0.0066*** –0.0082***405–0.0084***406–0.0059***407(0.0013)408(0.0015)409(0.0016)410(0.0015)411Balcony (dummy)412–0.1203***413–0.1327***414–0.1337***415–0.1107***416(0.0104)417(0.0169)418(0.0142)419(0.0097)420Object condition4210.1536***4220.1491***4230.1549***4240.1024***425(1 = new; 10 = demolition)426(0.0046)427(0.0080)428(0.0070)429(0.0052)430Region-specific variables:431Urban area (dummy)4320.00974330.01484340.0169435–0.0052436(Regional centers)437(0.0141)438(0.0370)439(0.0398)440(0.0139)441City/metropolitan area (dummy)442–0.0278443–0.0665***444–0.0660***445–0.0324446(0.0180)447(0.0230)448(0.0237)449(0.0199)450West/East Germany (dummy)4510.0657**452–0.0873453–0.1180**454–0.4779***455(1 = West, 0 = East)456(0.0305)457(0.0584)458(0.0547)459(0.1544)460Socioeconomic variables:461Population density4620.00144630.00014640.00584650.0088466(0.0089)467(0.0117)468(0.0121)469(0.0094)470Population growth4710.3231***4720.16794730.12564740.1858**475(0.0746)476(0.1734)477(0.1692)478(0.0788)479Primary income per capita480–0.0063**481–0.0055482–0.0088***483–0.0082***484(0.0025)485(0.0035)486(0.0033)487(0.0026)488Students4890.42404900.52234910.60204921.2369**493(0.3168)494(0.5629)495(0.5244)496(0.4975)497(Continues)498 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License499500501===== PAGE 22 (score 63, numeric-paren 27, table-words 6, var-words 21) =====502503864504BAYE and DINGER505TA B L E 5506(Continued)507(1)508(2)509(3)510(4)511Variables512Rent-price513ratio514Rent-price515ratio516Rent-price517ratio518Rent-price519ratio520Unemployment rate5213.0265**5223.7454*5234.8997**5242.2513525(1.4089)526(2.0504)527(1.9070)528(1.7935)529Construction completions5300.0682*5310.1633**5320.04895330.1118534(0.0365)535(0.0633)536(0.0593)537(0.0682)538Social assistance recipients539–0.0017***540–0.0012541–0.0022**542–0.0020**543(0.0006)544(0.0012)545(0.0011)546(0.0009)547Constant5480.46785491.3856***5502.4277***5513.1244***552(0.3272)553(0.3565)554(0.3428)555(0.4064)556Observations5574,619,7275582,406,1605592,247,5015603,739,728561R25620.61925630.58755640.57865650.6228566Year FE567Yes568Yes569Yes570Yes571District FE572Yes573Yes574Yes575Yes576Sample577Full578Regulated579municipalities580Regulated581municipalities582w/o newbuilds583Regulated584municipalities585w/o renovated586Note: This table presents the results of the OLS regression modeling the determinants of housing yield described in Equation (3)587using micro data. The dependent variable is the rent-price ratio, that is, the yearly net rent divided by the objects’ potential sale588price. The first explanatory variable is the dummy variable (𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚), which divides the sample into a treatment and a con-589trol group depending on the application of the rent control. The interaction terms of the dummy variable 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞,590which equals 1 if the rent control applies in the municipality at the point of time of the observation, and the dummy variables591𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖and 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖, which indicate if the object itself is regulated due to its year of construction and condition, dis-592play the effects of the introduction of the rent control on the rent-price ratio depending on the individual objects’ regulation593status.594In the estimation of Column 1, regulated and unregulated municipalities are considered. In Columns 2, 3, and 4, we conduct595within-municipality analyses; that is, we only include regulated municipalities in our estimation. In Column 3, objects that are596categorized as first occupancy are excluded from the sample. In Column 4, objects that are categorized as first occupancy after597reconstruction and like new are excluded.598The control variables can be categorized as object-specific, region-specific, and socioeconomic variables. The object-specific con-599trol variables include the base yield, that is, the quarterly mean of the rent-price ratio in the object’s municipality, the object’s year600of construction, the living space, the number of rooms, dummy variables considering if the object has a basement and a balcony601or terrace and the object condition which can vary between 1 = new and 10 = demolition.602The region-specific variables indicate if an object is located in an urban area, a city, or a metropolitan area and in West or East603Germany. The considered socioeconomic variables include population density, population growth, primary income per capita,604the number of students, the unemployment rate, construction completions, and the number of social assistance recipients. Fixed605effects for years and districts are included.606The sample covers the observation period from 2010 to 2019. Robust standard errors clustered for districts are displayed in607parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively.608We review the results in several robustness analyses. Specifically, we show that the direction609and significance of the results are robust to including the year of construction and the object610condition as categorical variables (Table A2). Furthermore, we show that the estimates are robust611to an extension of the observation period from 2010–2019 to 2008–2019 (Table A3, A4, Column 2).612We also ensure that the results are not driven by a specific subsample, for example, in the largest613cities or housing markets with high construction activity.614 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License615616617===== PAGE 23 (score 151, numeric-paren 120, table-words 0, var-words 17) =====618619BAYE and DINGER620865621TA B L E 6622The effect of the rent control on the rental and sale prices.623(1)624(2)625(3)626(4)627Variables628Rent price629per sqm (log)630Rent price per631sqm (log)632Sale price633per sqm634(log)635Sale price per636sqm (log)637Treatment municipality6380.0632***6390.0850***640(0.0140)641(0.0245)642Regulated objects in regulated mun6430.0236***644–0.00646450.0987*** 0.0735***646(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖)647(0.0055)648(0.0066)649(0.0099)650(0.0206)651Unregulated objects in regulated mun6520.0426***653–0.0093654(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖)655(0.0126)656(0.0082)657Year of construction (log)658–0.4313659–1.0019***6600.5455661–0.5379*662(0.2871)663(0.1922)664(0.7097)665(0.3098)666Living space (log)667–0.00136680.02096690.2474***6700.2512***671(0.0311)672(0.0362)673(0.0269)674(0.0289)675Number of rooms (log)676–0.0650***677–0.0703***678–0.1438*** –0.1403***679(0.0138)680(0.0171)681(0.0144)682(0.0151)683Basement (dummy)684–0.0018***685–0.0017***6860.00096870.0016688(0.0004)689(0.0005)690(0.0014)691(0.0014)692Balcony (dummy)6930.0200***6940.0169***6950.0329***6960.0262***697(0.0035)698(0.0052)699(0.0039)700(0.0037)701Urban area (dummy)7020.0442*7030.1112***7040.04547050.1922**706(Regional centers)707(0.0246)708(0.0229)709(0.0465)710(0.0908)711City/metropolitan area (dummy)712–0.0090713–0.1104***714–0.0113715–0.1282***716(0.0325)717(0.0089)718(0.0442)719(0.0344)720West/East Germany (dummy)7210.09917220.2437***7230.0982*7240.1968***725(1 = West, 0 = East)726(0.0645)727(0.0508)728(0.0557)729(0.0381)730Population density (log)7310.0441***7320.0213***7330.0495***7340.0040735(0.0068)736(0.0079)737(0.0114)738(0.0137)739Population growth (log)7400.00147410.0037742–0.0069*** –0.0036743(0.0022)744(0.0049)745(0.0025)746(0.0054)747Primary income per capita (log)7480.2254***7490.1638*7500.6727***7510.5359***752(0.0845)753(0.0882)754(0.1092)755(0.0998)756Students (log)757–0.0092758–0.0071759–0.01237600.0007761(0.0107)762(0.0152)763(0.0159)764(0.0129)765Unemployment rate (log)766–0.2109*767–0.3313**768–0.2579*** –0.3812***769(0.1154)770(0.1417)771(0.0826)772(0.0744)773Construction completions (log)7740.0157***7750.0113**7760.0272***7770.0164**778(0.0029)779(0.0055)780(0.0043)781(0.0079)782Social assistance recipients (log)783–0.0234784–0.05777850.1073*7860.1747*787(0.0527)788(0.0805)789(0.0625)790(0.0902)791Constant7920.92317936.6185***794–11.8200**795–2.7222796(3.0588)797(2.4338)798(5.1400)799(2.7286)800(Continues)801 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License802803804===== PAGE 24 (score 29, numeric-paren 0, table-words 5, var-words 19) =====805806866807BAYE and DINGER808TA B L E 6809(Continued)810(1)811(2)812(3)813(4)814Variables815Rent price816per sqm (log)817Rent price per818sqm (log)819Sale price820per sqm821(log)822Sale price per823sqm (log)824Observations8253,064,1528262,017,6868271,716,3178281,210,342829R28300.66188310.55148320.68958330.6125834Year FE835Yes836Yes837Yes838Yes839District FE840Yes841Yes842Yes843Yes844Sample845Full846Regulated847municipalities848Full849Regulated850municipalities851Note: This table presents the results of the OLS regression modeling the determinants of rental and sale prices described in Equa-852tion (4) using micro data. The dependent variable in Columns 1 and 2 is the log of the rent price per square meter and in Columns 3853and 4; it is the log of the sale price per square meter. The variable “Treatment municipality” is a dummy variable which divides the854sample into a treatment and a control group depending on the application of the rent control. The second and third reported esti-855mators belong to the interaction terms of the dummy variable 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞, which equals 1 if the rent control applies856in the municipality at the point of time of the observation, and the dummy variables 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖and 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖, which indi-857cate if the object itself is regulated due to its year of construction and condition. They display the effects of the introduction of rent858control on the rent-price ratio depending on the individual objects’ regulation status.859In the estimation of Columns 1 and 3, regulated and unregulated municipalities are considered. In Columns 2 and 4, we conduct860within-municipality analyses; that is, we only include regulated municipalities in our estimation. Thus, we cannot estimate coef-861ficients for “Treatment municipality” and for “Unregulated objects in regulated mun” as the unregulated objects are the control862group.863The control variables can be categorized as object-specific, region-specific, and socioeconomic variables. The object-specific con-864trol variables include the log of the object’s year of construction, the log of the living space, the log of the number of rooms,865dummy variables considering if the object has a basement and a balcony or terrace, and for the different object condition categories866(unreported).867The region-specific variables indicate if an object is located in an urban area, a city, or a metropolitan area and in West or East868Germany. The considered socioeconomic variables include population density, population growth, primary income per capita,869the number of students, the unemployment rate, construction completions, and the number of social assistance recipients. Fixed870effects for years and districts are included.871The sample covers the observation period from 2010 to 2019. Robust standard errors clustered for districts are displayed in872parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively.8735.2.3874Respective effects on rental and sale prices875In sum, our micro-level results so far are consistent with the district-level results reported in Sec-876tion 4 and suggest a supply-driven within-market shift toward high-priced newbuilds or renovated877objects in tense, regulated housing markets resulting in no improvement (at least in the short run)878of rental housing affordability. To examine if the effect on the rent-price ratio is driven by the rental879or sale prices, we separately estimate the effects on rents and sale prices with the simple model:880𝑝𝑖= 𝛼+ 𝛾𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚+ 𝛿1 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖881+ 𝛿2 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖+ 𝛽𝑋𝑖𝑚𝑑𝑦+ 𝐴𝑑+ 𝐵𝑦+ 𝜀𝑖,882(4)883where 𝑝𝑖is either the log of the rental or sale price of object 𝑖, 𝛾accounts for general differences884between the treated and the untreated municipalities 𝑚as 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚is the treatment group885indicator. The coefficient 𝛿1 (𝛿2) displays the effect of the introduction of the rent control on the886price variable of (un)regulated objects (in regulated municipalities). 𝑋contains several object-887 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License888889890===== PAGE 25 (score 54, numeric-paren 35, table-words 0, var-words 16) =====891892BAYE and DINGER893867894TA B L E 7895Spillover effects of rent controls in regulated areas to regulated and excluded objects.896(1)897Variables898Rent-price ratio899Direct treatment effect900–0.6154***901𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖902(0.0118)903Spillover effects to treated units9040.4010***905𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔1 × 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑚𝑦906(0.0174)907Spillover effects to control units9080.1370***909(1 −𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔1) × 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑚𝑦910(0.0052)911Object-specific variables:912Year of construction913–0.0038***914(0.0000)915Living space9160.0058***917(0.0000)918Number of rooms919–0.3487***920(0.0012)921Basement (dummy)922–0.0056***923(0.0004)924Balcony (dummy)925–0.1473***926(0.0015)927Object condition9280.1485***929(1 = new; 10 = demolition)930(0.0003)931Region-specific variables:932Urban area (dummy)9330.4525***934(Regional centers)935(0.0083)936City/metropolitan area (dummy)937–0.5006***938(0.0095)939West/East Germany (dummy)940–0.0739***941(1 = West, 0 = East)942(0.0154)943Socioeconomic variables:944Population density9450.1041***946(0.0026)947Population growth948–0.0007949(0.0749)950Primary income per capita951–0.0262***952(0.0007)953Students954–3.4645***955(0.1163)956Unemployment rate9571.9816***958(0.3575)959Construction completions9600.6494***961(0.0107)962(Continues)963 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License964965966===== PAGE 26 (score 49, numeric-paren 2, table-words 13, var-words 21) =====967968868969BAYE and DINGER970TA B L E 7971(Continued)972(1)973Variables974Rent-price ratio975Constant97613.3005***977(0.0443)978Observations9792,406,210980R29810.4266982Year FE983Yes984District FE985Yes986Note: This table presents the results of the spillover analysis based on Equation (5) using micro data. The dependent variable987is the rent-price ratio, that is, the yearly net rent divided by the objects’ predicted sale price. The dummy variable 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖988displays the direct treatment effect. The variable 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑚𝑞is the group-level average treatment intensity, that is, the proportion989of apartments in a municipality that is regulated, thus, not new or renovated. The interaction term with the treatment effect990reveals the spillover effects of rent control on regulated objects. To extract the spillover effects to the control units, we interact the991unregulated objects (1 −𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖) in the regulated area with the group-level average treatment intensity 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑚𝑞.992The control variables can be categorized as object-specific, region-specific, and socioeconomic variables. The object-specific con-993trol variables include the base yield, that is, the quarterly mean of the rent-price ratio in the objects’ municipality, the object’s year994of construction, the living space, the number of rooms, dummy variables considering if the object has a basement and a balcony995or terrace and the object condition which can vary between 1 = new and 10 = demolition.996The region-specific variables indicate if an object is located in an urban area, a city, or a metropolitan area and in West or East997Germany. The considered socioeconomic variables include population density, population growth, primary income per capita, the998number of students, the unemployment rate, construction completions, and the number of social assistance recipients. Year and999district fixed effects are included.1000The sample covers the observation period from 2010 to 2019. Only observations from regulated municipalities are included. Robust1001standard errors are displayed in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively.1002specific and region-specific control variables20 and we include district and year fixed effects in1003𝐴𝑑and 𝐵𝑦. We also estimate the effect that occurs within the regulated municipalities (Table 6,1004Columns 2 and 4).1005The results reveal that the rent price per square meter of regulated objects is 2.36% higher in1006regulated municipalities than the rent price per square meter of comparable objects in unregulated1007municipalities (Table 6, Column 1). This shows that the difference between the rents in regulated1008and unregulated municipalities increased after the introduction of the regulation. However, as1009the sale prices for regulated objects are 9.87% higher after the introduction of rent control in these1010municipalities (Table 6, Column 3), the rent-price ratio of regulated objects is smaller after the1011application of the regulation. The analysis from the previous section shows that the rent-price1012ratio of unregulated, new objects in regulated municipalities increased after the introduction of1013rent control. Since rents are 4.26% higher (Table 6, Column 1) and sale prices are not affected1014significantly (Table 6, Column 3), this coefficient can be explained by the higher rental prices for1015unregulated newbuilds and renovated objects in tense housing markets.1016If we conduct the analysis only in treatment municipalities where rent control is introduced1017during the observation period, we can confirm that compared to unregulated objects, the rent-1018price ratio of regulated objects is smaller (Table 5) because of increases in sale prices for these1019objects (Table 6, Column 4). If the regulation was successful, we would expect a significant102020 Included control variables are the logs of the year of construction, living space, and number of rooms, dummy variables1021for a basement, balcony, or terrace, the location in an urban area, city, and Western Germany, the logs of the socioeconomic1022factors population growth, primary income per capita, number of students, unemployment rate, and social assistance1023recipients in each district 𝑑and year 𝑦, and construction completions in each municipality 𝑚and year 𝑦.1024 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License102510261027===== PAGE 33 (score 57, numeric-paren 39, table-words 3, var-words 8) =====10281029BAYE and DINGER10308751031How to cite this article: Baye, V., & Dinger, V. (2024). Investment incentives of rent1032controls and gentrification: Evidence from German micro data. Real Estate Economics, 52,1033843–884. https://doi.org/10.1111/1540-6229.124781034APPENDIX1035TA B L E A 11036Number of building completions after introduction of rent control.1037(1)1038(2)1039(3)1040Variables1041Lag of1042implementation = 11043Lag of1044implementation = 21045Lag of1046implementation = 31047y = number of building completions1048(L = 1)1049(L = 2)1050(L = 3)1051𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚105222.3212**105327.7601***105431.8351***1055(Treatment municipality)1056(9.1955)1057(7.8060)1058(6.5848)1059𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚× 𝑝𝑒𝑟𝑖𝑜𝑑_𝑟𝑒𝑔𝑡−𝐿106057.6797**106155.9226*106263.7477*1063(implementation of rent control)1064(27.4879)1065(30.2519)1066(35.9447)1067Inhabitants10680.0031***10690.0031***10700.0031***1071(0.0003)1072(0.0003)1073(0.0003)1074Urban area (dummy)107515.8539107615.4476107715.12741078(regional centers)1079(12.3264)1080(12.2953)1081(12.2514)1082City/metropolitan area1083−199.5527***1084−198.9124***1085−198.5782***1086(dummy)1087(60.9356)1088(61.1287)1089(61.2912)1090Western/Eastern GER10917.1835**10927.2029**10937.1657**1094(1 = West, 0 = East)1095(2.8580)1096(2.8597)1097(2.8476)1098Constant1099−16.1400***1100−16.6407***1101−16.9681***1102(2.9808)1103(3.1131)1104(3.2374)1105Observations110645,905110745,905110845,9051109Number of municipalities1110720111117201111272011113Year FE1114Yes1115Yes1116Yes1117Municipality-type FE1118Yes1119Yes1120Yes1121Note: This table presents the results of the OLS regression modeling the determinants of the number of building completions1122described by the equation 𝑏𝑢𝑖𝑙𝑑𝑖𝑛𝑔𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑖𝑜𝑛𝑠𝑚𝑦= 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚+ 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚× 𝑝𝑒𝑟𝑖𝑜𝑑_𝑟𝑒𝑔𝑦−𝐿+ 𝑋𝑚𝑦+ 𝐵𝑦+ 𝜀.1123Building completions show the number of building completions per municipality 𝑚in year 𝑦. The dummy variable 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚1124divides the sample into a treatment (regulated) and a control group (unregulated) depending on the application of the rent control.1125The dummy variable 𝑝𝑒𝑟𝑖𝑜𝑑_𝑟𝑒𝑔𝑦−𝐿indicates the treatment period when the rent control was implemented in a specific munici-1126pality. The columns of the table vary by the considered lags (one, two, and three years) of the rent control implementation. Thus,1127the interaction term displays the effects of the introduction of the rent control on the number of building completions 1, 2, and 31128years after the application of the regulation.1129We control for inhabitants per municipality and region-specific variables which indicate if an object is located in an urban area, a1130city, or a metropolitan area and in West or East Germany. Year fixed effects and municipality-type fixed effects (city, large town,1131small town, mid-sized, rural municipality) are included in all specifications.1132The sample covers the observation period from 2010 to 2019. Robust standard errors clustered for districts are displayed in1133parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively.1134 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License113511361137===== PAGE 34 (score 73, numeric-paren 52, table-words 0, var-words 15) =====113811398761140BAYE and DINGER1141TA B L E A 21142Robustness test with categorical variables.1143(1)1144(2)1145Variables1146Categorical variables for1147object condition1148Categorical variables for1149object condition and year1150of construction1151y = rent-price ratio1152Treatment municipality11530.0297***11540.0382***1155(0.0097)1156(0.0101)1157Regulated objects in regulated mun1158−0.1910***1159−0.1870***1160(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖)1161(0.0186)1162(0.0190)1163Unregulated objects in regulated mun11640.3545***11650.3280***1166(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖)1167(0.0262)1168(0.0274)1169Object-specific variables:1170Base yield11710.9525***11720.9355***1173(0.0220)1174(0.0216)1175Year of construction1176−0.00011177(0.0003)1178Living space11790.00041180−0.00081181(0.0009)1182(0.0010)1183Number of rooms1184−0.2395***1185−0.2203***1186(0.0052)1187(0.0057)1188Basement (dummy)1189−0.0060***1190−0.0063***1191(0.0013)1192(0.0013)1193Balcony (dummy)1194−0.1212***1195−0.1159***1196(0.0102)1197(0.0105)1198Region-specific variables:1199Urban area (dummy)12000.004912010.00441202(Regional centers)1203(0.0149)1204(0.0143)1205City/metropolitan area (dummy)1206−0.02581207−0.0336*1208(0.0170)1209(0.0201)1210West/East Germany (dummy)12110.03391212−1.0584***1213(1 = West, 0 = East)1214(0.0298)1215(0.0498)1216Socioeconomic variables:1217Population density1218−0.000012190.0163*1220(0.0091)1221(0.0088)1222Population growth12230.3439***12240.2823***1225(0.0784)1226(0.0809)1227Primary income per capita1228−0.0076***1229−0.0119***1230(0.0026)1231(0.0035)1232Students12330.490912340.43761235(0.3353)1236(0.3646)1237(Continues)1238 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License123912401241===== PAGE 35 (score 43, numeric-paren 14, table-words 4, var-words 19) =====12421243BAYE and DINGER12448771245TA B L E A 21246(Continued)1247(1)1248(2)1249Variables1250Categorical variables for1251object condition1252Categorical variables for1253object condition and year1254of construction1255Unemployment rate12562.7590**12572.2974*1258(1.3092)1259(1.2633)1260Construction completions12610.0699*12620.06651263(0.0421)1264(0.0434)1265Social assistance recipients1266−0.0016**1267−0.0015**1268(0.0007)1269(0.0006)1270Constant12710.266112721.4412***1273(0.5183)1274(0.1291)1275Observations12764,619,72712774,619,7241278R212790.635212800.64921281Object condition1282Categorical1283Categorical1284Year of construction1285Categorical1286Year FE1287Yes1288Yes1289District FE1290Yes1291Yes1292Note: This table presents the results of the OLS regression modeling the determinants of housing yield described in Equation (3)1293using micro data. The dependent variable is the rent-price ratio, that is, the yearly net rent divided by the objects’ potential sale1294price. The first explanatory variable is the dummy variable (𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔𝑚), which divides the sample into a treatment and a con-1295trol group depending on the application of the rent control. The interaction terms of the dummy variable 𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞,1296which equals 1 if the rent control applies in the municipality at the point of time of the observation, and the dummy variables1297𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖and 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖, which indicate if the object itself is regulated due to its year of construction and condition, dis-1298play the effects of the introduction of the rent control on the rent-price ratio depending on the individual objects’ regulation1299status.1300The control variables can be categorized as object-specific, region-specific, and socioeconomic variables. The object-specific con-1301trol variables include the base yield, that is, the quarterly mean of the rent-price ratio in the objects’ municipality, the object’s1302year of construction, the living space, the number of rooms, and dummy variables considering if the object has a basement and1303a balcony or terrace. In this specification, we include the object condition, which can vary in ten categories between new and1304demolition, as a categorical variable. Moreover, in Column 1, the year of construction is included as a continuous variable and in1305Column 2 as a categorical variable (grouped by ten years, i.e., 1800–1810).1306The region-specific variables indicate if an object is located in an urban area, a city, or a metropolitan area and in West or East1307Germany. The considered socioeconomic variables include population density, population growth, primary income per capita, the1308number of students, the unemployment rate, construction completions, and the number of social assistance recipients.1309Year fixed effects are included in all specifications.1310The sample covers the observation period from 2010 to 2019. Robust standard errors are displayed in parentheses. *, **, and ***1311indicate statistical significance at the 10%, 5%, and 1% level, respectively.1312 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License131313141315===== PAGE 36 (score 112, numeric-paren 92, table-words 0, var-words 10) =====131613178781318BAYE and DINGER1319TA B L E A 31320Robustness analysis with different subsamples.1321(1)1322(2)1323(3)1324(4)1325(5)1326Variables1327Baseline1328Baseline1329Top 7 cities1330Little1331construction1332completions1333Many1334construction1335completions1336y = rent-price ratio1337Treatment municipality13380.0260***13390.0260***13400.0346**13410.0364***1342(0.0093)1343(0.0094)1344(0.0141)1345(0.0106)1346Regulated objects in regulated mun1347−0.1330***1348−0.1365***1349−0.0893***1350−0.1096***1351−0.1871***1352(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖)1353(0.0192)1354(0.0188)1355(0.0227)1356(0.0185)1357(0.0243)1358Unregulated objects in regulated mun13590.2522***13600.2620***13610.2275***13620.2224***13630.2709***1364(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖)1365(0.0288)1366(0.0299)1367(0.0469)1368(0.0299)1369(0.0427)1370Object-specific variables:1371Base yield13720.9551***13730.9547***13740.9919***13750.9664***13760.8980***1377(0.0217)1378(0.0226)1379(0.0238)1380(0.0246)1381(0.0097)1382Year of construction1383−0.00031384−0.00031385−0.0009***1386−0.00021387−0.0005**1388(0.0002)1389(0.0002)1390(0.0002)1391(0.0002)1392(0.0002)1393Living space13940.000113950.000213960.0031***13970.00061398−0.0016*1399(0.0009)1400(0.0009)1401(0.0005)1402(0.0011)1403(0.0009)1404Number of rooms1405−0.2310***1406−0.2320***1407−0.2172***1408−0.2359***1409−0.2004***1410(0.0053)1411(0.0057)1412(0.0130)1413(0.0050)1414(0.0089)1415Basement (dummy)1416−0.0066***1417−0.0043***1418−0.0105**1419−0.0065***1420−0.0050**1421(0.0013)1422(0.0012)1423(0.0030)1424(0.0015)1425(0.0021)1426Balcony (dummy)1427−0.1203***1428−0.1159***1429−0.1100***1430−0.1039***1431−0.1507***1432(0.0104)1433(0.0104)1434(0.0179)1435(0.0102)1436(0.0070)1437Object condition14380.1536***14390.1572***14400.1393***14410.1474***14420.1544***1443(1 = new; 10 = demolition)1444(0.0046)1445(0.0049)1446(0.0130)1447(0.0055)1448(0.0034)1449(Continues)1450 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License145114521453===== PAGE 37 (score 96, numeric-paren 76, table-words 0, var-words 11) =====14541455BAYE and DINGER14568791457TA B L E A 31458(Continued)1459(1)1460(2)1461(3)1462(4)1463(5)1464Variables1465Baseline1466Baseline1467Top 7 cities1468Little1469construction1470completions1471Many1472construction1473completions1474Region-specific variables:1475Urban area (dummy)14760.009714770.009114780.016414790.00311480(Regional centers)1481(0.0141)1482(0.0139)1483(0.0150)1484(0.0167)1485City/metropolitan area (dummy)1486−0.02781487−0.0289*1488−0.02281489−0.03311490(0.0180)1491(0.0172)1492(0.0166)1493(0.0322)1494West/East Germany (dummy)14950.0657**1496−0.6544***1497−0.008114980.0958***14991.9781***1500(1 = West, 0 = East)1501(0.0305)1502(0.0399)1503(0.3333)1504(0.0103)1505(0.0716)1506Socioeconomic variables:1507Population density15080.00141509−0.00131510−0.4700**1511−0.009615120.0353***1513(0.0089)1514(0.0090)1515(0.1392)1516(0.0099)1517(0.0120)1518Population growth15190.3231***15200.3134***1521−0.4874*15220.1940***15230.4364***1524(0.0746)1525(0.0816)1526(0.2443)1527(0.0611)1528(0.1336)1529Primary income per capita1530−0.0063**1531−0.0078***1532−0.0412***1533−0.0076**1534−0.00361535(0.0025)1536(0.0019)1537(0.0102)1538(0.0031)1539(0.0027)1540Students15410.424015420.399915430.708715440.7587**1545−0.25031546(0.3168)1547(0.2858)1548(0.7883)1549(0.3555)1550(0.5304)1551Unemployment rate15523.0265**15532.3026**1554−10.8256**15552.008615566.8074***1557(1.4089)1558(1.0735)1559(3.2604)1560(1.5577)1561(1.0968)1562Construction completions15630.0682*15640.2007***1565−0.0958*15660.1005**15670.10101568(0.0365)1569(0.0355)1570(0.0398)1571(0.0418)1572(0.0823)1573Social assistance recipients1574−0.0017***15750.0017*1576−0.0010**1577−0.0038***1578(0.0006)1579(0.0008)1580(0.0005)1581(0.0008)1582(Continues)1583 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License158415851586===== PAGE 38 (score 25, numeric-paren 8, table-words 4, var-words 8) =====158715888801589BAYE and DINGER1590TA B L E A 31591(Continued)1592(1)1593(2)1594(3)1595(4)1596(5)1597Variables1598Baseline1599Baseline1600Top 7 cities1601Little1602construction1603completions1604Many1605construction1606completions1607Constant16080.467816091.1814***16103.7102***16110.27071612−0.8682**1613(0.3272)1614(0.3233)1615(0.8117)1616(0.3574)1617(0.3663)1618Observations16194,619,72716205,111,96316211,140,28416223,244,78616231,767,4761624R216250.619216260.615516270.543116280.620916290.64641630Year FE1631Yes1632Yes1633Yes1634Yes1635Yes1636District FE1637Yes1638Yes1639Yes1640Yes1641Yes1642Note: This table presents the results of the OLS regression modeling the determinants of housing yields described in Equation (3) using micro data. The dependent variable is the rent-price ratio.1643Column 1 shows the baseline specification as described in Section 5.2. Column 2 displays the results of the estimation with an extended observation period, already beginning in 2008. Column 31644shows the subsample analysis of the seven biggest cities (Hamburg, Berlin, Duesseldorf, Frankfurt, Stuttgart Cologne, and Munich). Because of the reduction of the sample, all municipalities are1645subject to rent control; thus, the dummy to differentiate between treatment and control groups cannot be included. The variables that determine if an object is located in an urban area, a city, or1646a metropolitan area are not included as well. For the results in Columns 4 and 5, the sample is split by the median of the yearly construction completions measured by square meters of newbuild1647living space per capita on municipality level.1648Year and district fixed effects are included in all specifications. The sample covers the observation period from 2010 (2008) to 2019. Robust standard errors clustered for districts are displayed in1649parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively.1650 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License165116521653===== PAGE 39 (score 133, numeric-paren 111, table-words 0, var-words 10) =====16541655BAYE and DINGER16568811657TA B L E A 41658Robustness analysis with subsamples divided by building permits.1659(1)1660(2)1661(3)1662(4)1663(5)1664(6)1665Variables1666Little building1667permits1668Many building1669permits1670Little building1671permits1672Many building1673permits1674Little building1675permits1676Many building1677permits1678y = rent-price ratio1679Lag = 1 year1680Lag = 1 year1681Lag = 2 years1682Lag = 2 years1683Treatment municipality16840.0274**16850.0341***16860.0277**16870.0340***16880.0282**16890.0336***1690(0.0109)1691(0.0122)1692(0.0110)1693(0.0121)1694(0.0110)1695(0.0121)1696Regulated objects in regulated mun1697−0.0897***1698−0.1705***1699−0.0886***1700−0.1717***1701−0.0878***1702−0.1731***1703(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑟𝑒𝑔𝑖)1704(0.0235)1705(0.0246)1706(0.0235)1707(0.0244)1708(0.0237)1709(0.0244)1710Unregulated objects in regulated mun17110.1948***17120.2764***17130.1945***17140.2774***17150.1939***17160.2773***1717(𝑚𝑢𝑛𝑖𝑐𝑖𝑝_𝑟𝑒𝑔_𝑎𝑝𝑝𝑙𝑖𝑒𝑑𝑚𝑞× 𝑜𝑏𝑗𝑒𝑐𝑡_𝑢𝑛𝑟𝑒𝑔𝑖)1718(0.0395)1719(0.0344)1720(0.0396)1721(0.0347)1722(0.0397)1723(0.0350)1724Object-specific variables:1725Base yield17260.9574***17270.9222***17280.9576***17290.9225***17300.9574***17310.9225***1732(0.0214)1733(0.0226)1734(0.0214)1735(0.0226)1736(0.0214)1737(0.0225)1738Year of construction1739−0.00021740−0.0006***1741−0.00021742−0.0006***1743−0.00021744−0.0006***1745(0.0003)1746(0.0002)1747(0.0003)1748(0.0002)1749(0.0003)1750(0.0002)1751Living space17520.00021753−0.000717540.00021755−0.000717560.00021757−0.00071758(0.0011)1759(0.0008)1760(0.0011)1761(0.0008)1762(0.0011)1763(0.0008)1764Number of rooms1765−0.2409***1766−0.1989***1767−0.2410***1768−0.1990***1769−0.2410***1770−0.1989***1771(0.0096)1772(0.0070)1773(0.0096)1774(0.0070)1775(0.0096)1776(0.0069)1777Basement (dummy)1778−0.0063***1779−0.0059***1780−0.0063***1781−0.0058***1782−0.0062***1783−0.0057***1784(0.0016)1785(0.0018)1786(0.0016)1787(0.0018)1788(0.0016)1789(0.0018)1790Balcony (dummy)1791−0.1078***1792−0.1358***1793−0.1080***1794−0.1357***1795−0.1080***1796−0.1358***1797(0.0109)1798(0.0111)1799(0.0109)1800(0.0111)1801(0.0109)1802(0.0111)1803Object condition18040.1514***18050.1483***18060.1515***18070.1487***18080.1516***18090.1489***1810(1 = new; 10 = demolition)1811(0.0031)1812(0.0066)1813(0.0031)1814(0.0066)1815(0.0031)1816(0.0066)1817(Continues)1818 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License181918201821===== PAGE 40 (score 111, numeric-paren 88, table-words 0, var-words 11) =====182218238821824BAYE and DINGER1825TA B L E A 41826(Continued)1827(1)1828(2)1829(3)1830(4)1831(5)1832(6)1833Variables1834Little building1835permits1836Many building1837permits1838Little building1839permits1840Many building1841permits1842Little building1843permits1844Many building1845permits1846Region-specific variables:1847Urban area (dummy)18480.02381849−0.006018500.02521851−0.007318520.02471853−0.00681854(Regional centers)1855(0.0160)1856(0.0133)1857(0.0159)1858(0.0133)1859(0.0160)1860(0.0132)1861City/metropolitan area (dummy)1862−0.02561863−0.03151864−0.02661865−0.03111866−0.02681867−0.03201868(0.0179)1869(0.0273)1870(0.0180)1871(0.0264)1872(0.0181)1873(0.0264)1874West/East Germany (dummy)1875−0.022918760.1362***1877−5.4607***18782.1477***18790.4122***1880−0.04881881(1 = West, 0 = East)1882(0.0150)1883(0.0329)1884(0.3978)1885(0.0676)1886(0.0699)1887(0.0421)1888Socioeconomic variables:1889Population density1890−0.001918910.0310**1892−0.002118930.0316**1894−0.001918950.0316**1896(0.0085)1897(0.0126)1898(0.0084)1899(0.0126)1900(0.0084)1901(0.0127)1902Population growth19030.2555***19040.3645***19050.2438***19060.3666***19070.2345***19080.3640***1909(0.0704)1910(0.1122)1911(0.0684)1912(0.1138)1913(0.0669)1914(0.1101)1915Primary income per capita1916−0.0065**1917−0.0058**1918−0.0065**1919−0.0059**1920−0.0065**1921−0.0060**1922(0.0031)1923(0.0027)1924(0.0031)1925(0.0027)1926(0.0031)1927(0.0026)1928Students19290.7227**1930−0.293719310.7196**1932−0.280219330.7255**1934−0.28481935(0.3258)1936(0.5549)1937(0.3253)1938(0.5532)1939(0.3239)1940(0.5482)1941Unemployment rate19422.651619435.4674***19442.681419455.4439***19462.648119475.6040***1948(1.6867)1949(1.1005)1950(1.6872)1951(1.0817)1952(1.6814)1953(1.0707)1954Construction completions1955−0.046919560.05501957−0.042119580.05861959−0.037119600.06521961(0.0795)1962(0.0651)1963(0.0801)1964(0.0645)1965(0.0788)1966(0.0645)1967Social assistance recipients1968−0.0027***1969−0.00061970−0.0027***1971−0.00061972−0.0028***1973−0.00071974(0.0007)1975(0.0006)1976(0.0007)1977(0.0006)1978(0.0007)1979(0.0006)1980(Continues)1981 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License198219831984===== PAGE 41 (score 26, numeric-paren 10, table-words 4, var-words 6) =====19851986BAYE and DINGER19878831988TA B L E A 41989(Continued)1990(1)1991(2)1992(3)1993(4)1994(5)1995(6)1996Variables1997Little building1998permits1999Many building2000permits2001Little building2002permits2003Many building2004permits2005Little building2006permits2007Many building2008permits2009Constant20100.328120110.9773***20125.7616***2013−1.0371***2014−0.106320151.1421***2016(0.4163)2017(0.3150)2018(0.4908)2019(0.3030)2020(0.3929)2021(0.3229)2022Observations20233,005,47620242,006,78620253,005,98820262,007,39820273,007,33920282,008,8812029R220300.599020310.651320320.598520330.651620340.598320350.65222036Year FE2037Yes2038Yes2039Yes2040Yes2041Yes2042Yes2043District FE2044Yes2045Yes2046Yes2047Yes2048Yes2049Yes2050Note: This table presents the results of the OLS regression modeling the determinants of housing yields described in Equation (3) using micro data. The dependent variable is the rent-price ratio.2051The sample is divided into different subsamples depending on the median of building permits, measured by authorized living space (thousand square meters) per capita on municipality level,2052with no lag (Columns 1 and 2), 1-year lag (Columns 3 and 4), and a 2-year lag (Columns 5 and 6).2053Year and district fixed effects are included in all specifications. The sample covers the observation period from 2010 (2008) to 2019. Robust standard errors clustered for districts are displayed in2054parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively.2055 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License205620572058===== PAGE 42 (score 5, numeric-paren 0, table-words 0, var-words 5) =====205920608842061BAYE and DINGER20620206352064102065152066Proportion (in %)2067< 30 sqm206830 - 39 sqm206940 - 49 sqm207050 - 59 sqm207160 - 69 sqm207270 - 79 sqm207380 - 89 sqm207490 - 99 sqm2075100 - 109 sqm2076110 - 119 sqm2077120 - 129 sqm2078130 - 139 sqm2079140 - 149 sqm2080150 - 159 sqm2081160 - 169 sqm2082170 - 179 sqm2083> 180 sqm2084020855208610 15 202087Proportion (in %)2088until 191920891919 - 194920901950 - 195920911960 - 196920921970 - 197920931980 - 198920941990 - 199920952000 - 200520962006 and later2097Census data2098Rentals ImmobilienScout242099Sales ImmobilienScout242100F I G U R E A 12101Distribution of apartments over size and year of construction. [Color figure can be viewed at2102wileyonlinelibrary.com]2103 15406229, 2024, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1540-6229.12478 by University Of Leeds The Brotherton Library, Wiley Online Library on [22/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License2104