Principal investigator: Omar Barroso Khodr
Authors: Sodini et al. (2023) – P. Sodini, SV Nieuwerburgh, R. Vestman and UvL Toal
Original title: Identifying the Benefits from Homeownership: A Swedish Experiment
Location of the Intervention: Sweden (Northern European country)
Sample Size: 1.911 families — 567 in the treatment group and 1.344 in the control group — over nine years, from 1999 to 2007.
Primary Variable of Interest: Family 'i' associated with
property ownership
Type of Intervention: Acquisition of residential properties.
Methodology: Causal Inference; Instrumental Variables (IV).
Summary
The acquisition of residential real estate is widely encouraged by public policies, but its economic effects are still poorly understood. Sodini et al. (2023) explore the near-random variation in residential real estate acquisition generated by privatization decisions of municipally owned buildings and use granular data on demographics, income, housing, financial assets, and debt, allowing us to construct high-quality spending measures. Residential real estate acquisition leads to wealth accumulation through real estate appreciation, increases consumption, and improves consumption smoothing over time and in different economic scenarios. As a result, the authors observe an increase in the mobility of young families, who move up the real estate ladder, and an amplification in wealth accumulation for older families, who take on more risk in their financial portfolio.
- Policy Problem
Sodini et al. are motivated by a central political dilemma: governments, both in advanced and developing economies—such as the United States, which spends approximately US$200 billion annually—heavily subsidize the acquisition of residential real estate, based on the belief that property ownership generates broad benefits for families. Despite this large fiscal commitment and broad political consensus, the authors explain that there is surprisingly little rigorous causal evidence to confirm such benefits. This gap between large-scale public intervention and the absence of solid empirical evidence constitutes the problem that the study seeks to address.
Thus, the main obstacle to generating this evidence is the difficulty of isolating the causal effects of homeownership. Direct comparisons between homeowners and tenants suffer from strong endogeneity, as these groups differ systematically in observable and unobservable characteristics (such as age, income, and financial literacy), in addition to occupying properties of different quality and location. The ideal solution—a randomized experiment assigning identical families to own or rent identical properties—is unfeasible from a fiscal, technical, and ethical standpoint. This study overcomes these limitations by exploring a quasi-unique experiment in Stockholm, where a change in political control and subsequent law... Stopplag They generated a plausibly random variation in the likelihood of tenants in similar buildings purchasing their homes.
Based on this natural experiment, the article provides reliable causal estimates on five central outcomes for the debate on housing policies: whether property ownership truly generates wealth; how it affects consumption and the ability to smooth it over time and in the face of shocks; its impact on geographic and upward mobility; and its influence on the composition of household financial portfolios. Answering these questions is essential to assessing whether large public subsidies for homeownership are effective and justifiable.
The findings have direct implications for public policy debates. The study offers new evidence on the role of housing as a guarantee, showing that it can function as a powerful mechanism for smoothing consumption, especially for young people and families subject to income shocks. It also engages with the literature on rent regulation—increasingly relevant in cities with affordability crises—by analyzing the conversion of regulated units into owner-occupied units. Finally, by documenting how property ownership causally affects wealth formation and portfolio choice, the study contributes to broader discussions on the determinants of wealth inequality and the distributive effects of housing policies.
- Policy Implementation Context
This study is situated within the unique institutional context of the Swedish real estate market, particularly in Stockholm. A central element is the rent regulation system in place since 1974, in which values are determined through negotiations between landlord-tenant associations and become binding on all landlords, both public and private. Crucially, the rents charged by municipal landlords serve as a benchmark for the entire city. To fulfill this role, these landlords are expected to maintain a diverse stock of housing in different locations, sizes, and qualities—a condition that underlies the quasi-experiment analyzed.
In this context, according to the authors, the local political scenario directly triggered the events studied. Stockholm's housing stock is roughly divided between cooperatives, municipal owners, and private owners. After the 1998 elections, a center-right coalition took over the municipal government with the aim of reducing the presence of public owners through the sale of properties. Between 1999 and 2004, more than 12 municipal apartments were converted into cooperatives, allowing tenants to become owners. The process accelerated rapidly, reaching its peak in 2001.
Furthermore, the authors indicate that the natural experiment arises from an abrupt reversal of this policy. In response to the accelerated pace of privatizations, the national social-democratic government passed the Stopplag Act, in effect from April 2002, with the aim of halting or slowing down conversions. The law introduced a new requirement: after an agreement between the cooperative and the owner, the sale needed the approval of the county council, responsible for assessing whether the transaction would compromise the municipal owner's ability to serve as a benchmark for rents. As this criterion was new and poorly defined, the councils were given broad discretion.
In summary, the margin of decision resulted in rejections that, in practice, were arbitrary. Analysis of the minutes shows that denials were frequently justified by the supposed “uniqueness” of some units within the building—such as large studios or apartments with patios—considered necessary to maintain the comparability of rents in the neighborhood. In some cases, nearly identical buildings received opposite decisions. Thus, after Stopplag, a set of buildings that had already begun the privatization process was divided essentially randomly: 13 buildings had their privatization approved (treatment) and 33 had their sale denied (control). This arbitrary division provides the study with a rare source of plausibly exogenous variation in residential property ownership.
- Evaluation Details
The unique dataset for this study was constructed by linking detailed information about tenants involved in the privatization of housing cooperatives in Stockholm, combined with their demographic and financial characteristics, as well as the attributes of their residences. This linking is possible thanks to the integration of data from five primary sources. These include the minutes and decisions of district council meetings, the municipal files of landlords—which contain all correspondence and transaction details related to each privatization attempt—and landlord databases, which provide personal identification numbers, rental values, and apartment square footage. This information allows for tracking families individually throughout the entire process.
The core of the household-level data comes from tax records from the Swedish National Institute of Statistics (Statistics Sweden), which provide comprehensive information on the demographics, income, and assets of all individuals residing in the buildings analyzed between 1999 and 2007. The asset data is exceptionally detailed, allowing for the construction of portfolio returns for each family. Furthermore, by combining information on disposable income, changes in debt, and changes in real estate and financial assets, the researchers construct a high-quality measure—based on administrative records—of each family's annual consumption. A key innovation of the study is the use of exact apartment transaction prices, obtained from tax records, to accurately assess units in housing cooperatives, avoiding the imprecise estimates usually employed by Statistics Sweden.
To complete the data set, the researchers collected additional data directly from cooperative boards on tenants who chose to remain as tenants after a successful privatization. They also digitized the cooperatives' annual reports to determine members' monthly housing costs after privatization and to verify the asking price by the landlord. The study defines the timing of the event as the year of the property transfer (for approved cooperatives) or the year of the district board decision (for rejected ones), using the previous year (year -1) as the baseline for sample formation. The final, unbalanced panel follows 1.911 families—567 in the treatment group (successful privatizations) and 1.344 in the control group (failed attempts)—over nine years, from 1999 to 2007.
- Method
Sodini et al. begin by establishing a theoretical framework to conceptualize the financial implications of the quasi-experiment. This model compares the lifetime budget constraints of a family that remains a tenant with those of a family that successfully privatizes and becomes a homeowner. The main conclusion is the identification of a wealth shock, understood as the net gain in present value that the family obtains from privatization. Although the immediate benefit is the discount granted by the landlord on the sale price, the model demonstrates that the effective wealth shock is smaller, as it must be adjusted for the loss of future benefits associated with regulated rents. This shock, which depends on factors such as the size of the discount, the future appreciation of the property, and the family's time horizon, constitutes a central variable in the subsequent empirical analysis.
Therefore, to identify the causal impact of privatization on household outcomes, the authors employ an event study. The model compares households residing in buildings where privatization was approved (treatment group) with those in buildings where it was rejected (control group). By tracking outcomes in the years before and after the decision (relative years), the method allows for the visual inspection of possible pre-existing differences in trends between the groups, which could compromise causal validity. Furthermore, it captures the dynamic effects of the event, including the immediate impact in the year of purchase (relative year 0) and the effects in subsequent years, recognizing that the financial shock can generate distinct responses over time.
In addition to estimating the average effect, the methodology was designed to uncover the economic mechanisms that drive household responses, especially the role of real estate ownership itself versus the magnitude of the asset shock. This investigation begins with an analysis of heterogeneous treatment effects, dividing households into groups—for example, younger versus older heads of household—to test hypotheses such as that younger families, potentially more subject to credit constraints, respond more intensely to the new possibility of using real estate as collateral.
Finally, to more precisely separate the effect of property ownership from the effect of wealth shock, the study uses an instrumental variables (IV) approach. The challenge stems from the fact that wealth shock is not random and may be correlated with unobserved characteristics of the families. The researchers address this issue by constructing a hypothetical wealth shock for each family, based on apartment size and prevailing market prices in the neighborhood at the time of the decision. This instrument is relevant because it predicts the actual wealth shock for the treated families, but assumes a value of zero for the control group. The IV model uses this instrument to isolate the causal impact of both property ownership and wealth shock on outcomes such as consumption, offering a more rigorous test of the theoretical prediction that only wealth shock—and not homeownership itself—should drive changes in spending.
- Main results
The study results demonstrate that privatization generated an immediate and substantial change in household asset balances. The rate of homeownership among the treated households jumped from almost zero to 88%, significantly increasing homeownership and net worth thanks to the purchase discount and subsequent appreciation of the properties. The control group allows these effects to be isolated from potential selection biases.
Thus, the main evidence is that real estate ownership causes a large and persistent increase in consumption. In the year of privatization, consumption grew by 7,8%, and in subsequent years by 18,5%. This effect far exceeds what was predicted by a simple wealth effect model. Initially, the increase is financed by financial assets; later, by loans secured by the property, which then functions as a liquidity reserve.
Thus, the analysis separates the effect of the wealth shock from the effect of property ownership itself. Contrary to the standard model, families that received smaller shocks increased their consumption more. Using instrumental variables, the authors show that there is no positive causal relationship between the marginal wealth shock and consumption, indicating that the driver of increased spending is the easing of credit restrictions provided by property ownership.
In this context, the evidence is especially strong for young families. Those under 40 increased their consumption by almost 31%, despite experiencing smaller shocks, suggesting that they use the property as collateral to anticipate consumption. Housing also acts as a buffer: in the face of negative income shocks, homeowners maintain consumption through debt, while tenants reduce spending.
Finally, homeownership increases mobility among young families, who become more likely to move, especially to better neighborhoods—an upward mobility effect. It also alters portfolio composition: older families and those who remain in the property increase their allocation to riskier assets, amplifying the financial benefits of homeownership.
- Lessons in Public Policy
Based on the study's findings, several important lessons emerge for formulating public policies aimed at promoting the acquisition of residential real estate. First, the research provides robust causal evidence that homeownership generates substantial wealth gains, raising the families studied from the 54th to the 71st percentile of the wealth distribution. This validates a central premise behind government subsidies for home purchases.
Secondly, the study shows that one of the main mechanisms through which property ownership improves well-being is not only the increase in wealth, but also the easing of credit restrictions. By offering access to real estate collateral, property ownership allows families—especially younger ones—to smooth consumption throughout the life cycle and protect themselves against negative income shocks, maintaining consumption levels that renters cannot sustain.
Third, contrary to the common belief that property ownership reduces mobility, the results indicate that it increases upward mobility among young families, making them more likely to move to higher-quality neighborhoods.
Finally, the study highlights that the benefits of homeownership depend heavily on market conditions: capital gains are amplified by property appreciation and leveraged returns. This implies that policies incentivizing homeownership can have distinct effects depending on the real estate cycle. Taken together, the evidence suggests that such policies can generate significant economic benefits, especially when combined with measures that facilitate access to real estate assets for young families with liquidity constraints.
References
P. Sodini, SV Nieuwerburgh, R. Vestman and U. v. L.Toal. The American Economic Review, Vol. 113, No. 12 (2023), pp. 3173-3212.
