The Cost of Code Violations: How Building Codes Shape Residential Sales Prices and Rents

ABSTRACT

Existing literature suggests a positive correlation between building codes and housing prices. Yet studies rarely differentiate between resolved and unresolved code violations, or between residential sales prices and rent prices. As such, there are gaps in our knowledge about the landscape of housing regulations, which have particular relevance for understanding barriers to housing affordability and equity. To begin to fill these gaps, I present statistical analyses of building code violations data and housing market data in Chicago. Whereas resolving building violations does increase rents, I find no significant effect on residential sales price. And, although unresolved code violations decrease residential sales price, there is no significant effect on rent prices. Considering these results, I suggest that code violations reinforce the divide between wealthy and poor homeowners and exacerbate the existing lack of affordable housing options for renters. Overall, the article draws attention to the variation in effects of housing regulations in practice. I contend that it is crucial to understand the varied relationships between regulations and the housing market to make a dent in housing inequality.

KEYWORDS:

• Affordability
• rental housing
• housing
• policy
• real estate
• regulation

Acknowledgments

For suggestions, comments, and guidance, I thank Lincoln Quillian, Mary Pattillo, Wendy Griswold, Xiaojin Chen, Japonica Brown-Saracino, Melissa Pearson, Jordan Conwell, Catherine Gillis, Vincent Yung, Gary Fine, Laurie Edelman, Laura Beth Nielson, Jeffrey Parker, Melissa Osborne, Anya Degenshein, Kevin Loughran, Stefan Vogler, and participants in Northwestern’s Culture Workshop, Ethnography Workshop, Legal Studies Workshop, the University of Chicago’s Urban Workshop, and the Paris Law and Society Graduate Student Workshop. For assistance with obtaining data, I am indebted to Anthony DeFusco, Kelsey Rydland, Jessica Ruminski, and Jay Cao. I also thank reviewers at Housing Policy Debate.

Disclosure Statement

No potential conflict of interest was reported by the author.

Notes

1. Illinois General Assembly. 1994. “Residential Real Property Disclosure Act.” http://www.ilga.gov/legislation/ilcs/ilcs5.asp?ActID=2152&ChapterID=62 Accessed August 27, 2018.

2. This applies to landlords of buildings with six or more units (see City of Chicago, 2019).

3. Conservation inspections do not cover new construction or zoning.

4. Chicago Department of Buildings. 2016. “Top Ten Building Code Violations” https://twitter.com/AMesseSupply/status/747451967782289413 Accessed August 27, 2018.

5. Chicago Municipal Code, Sections 15-8-370, 15-8-380 through 15-8-430; Chicago Municipal Code Section 13-19-540(c).

6. I matched building violation data to their corresponding block groups to calculate the purview of violations. Block groups are the smallest geographical units for which the Census Bureau publishes sample data. Block groups are groupings of census blocks based on population characteristics and are relatively homogeneous; there are typically between three and five block groups to a tract.

7. I selected this neighborhood as it receives approximately the mean number of complaints and it is diverse in terms of demographics, tenure, and building types. The ward is home to approximately 56,000 residents, 54% of whom are White, 17% Black, 14% Hispanic, and 14% Asian. Median household incomes range widely in the ward, from $71,101 in one census tract to $12,485 in another. Whereas 32% of the ward’s housing stock is owner occupied, tracts range from 100% rentals to 76% owner occupied. One-unit buildings (i.e., single-family homes) make up 11% of housing stock in the ward, 2–9-unit buildings comprise 36%, and 53% of housing stock is over 10 units. Median ages of buildings per tracts range from pre-1939 to 1973. I calculated these figures using census tract data from the American Community Survey’s 2012 5-year summary. I matched complaints to violations to outcomes by address for one neighborhood in the city. Doing this for the whole city is beyond the scope of this study.

8. Administrative hearings officers do not ask for proof of resolution. Fines are at the discretion of the hearings officer and are collected up to the date of the hearing or of a follow-up inspection.

9. A further 10% of inspections in the example neighborhood resulted in both an administrative hearings case and, later, a housing court case.

10. I obtained this database from Midwest Real Estate Data, a real estate data aggregator and distributor that provides the Chicagoland MLS to brokers and appraisers across the Midwest. The data are limited to properties listed by realtors. In email correspondence, an MLS staff member suggested that nonlocal landlords are more likely to list their units with the service. As buildings owned or managed by nonlocal landlords are more likely to receive code violations (Travis, 2019), my data may overestimate the number of violations in the city’s rental stock. The benefit of this data source, however, is that it lists unit numbers. Other data sources (e.g., webscraping Craigslist listings) rarely include unit numbers (or even addresses) and would not allow me to ensure I was capturing changes in rent for the same unit in a building. This 5-year period afforded me data that was the most removed from (a) the housing crash, and (b) the date of analysis (2017). I matched housing market data to violations data from as early as 2006, however, to allow me to capture the effects of violations before a first sale or rental for properties that may have sold or rented in 2010 through 2015. Doing so affords a 4-year gap, for both a property that first rented or sold in 2006 and 2015.

11. The CoreLogic database is a record of all property transactions across the country. I selected data on sales in Chicago.

12. A lack of data means that I am unable to capture (a) properties that initially comply with building codes before an inspection or a citation; and (b) violations that are addressed but not recorded as such by an inspector.

13. For a discussion of the advantages of fixed effects, see England et al. (1988).

14. Property owners often upgrade their buildings without adding rooms. This is the only variable in the rental listings data set, however, that captures renovations.

15. I used the coefficient for complied violations in Model 3 and the following calculation to obtain this figure: 100*(exp(0.00512*10) − 1) = 5.25.

16. To clarify, rents will increase for the next tenant rather than an existing tenant because my data come from rental listings.

17. There may be other plausible explanations for this positive correlation between rents and resolved code violations. The increase in rents may stem from landlords opting to address violations in buildings they view as the most profitable, for example. If that were the case, however, then higher renting apartments might have a stronger correlation with code resolution. Whereas coefficients lose statistical significance when I only include rental units under $1000 or $1500, they barely change when I only include more expensive rental units. Future research could include instrumental variables to eliminate endogeneity.

18. I obtained similar results when I included properties with more than 30 violations. However, the coefficients were slightly smaller in magnitude and less significant.

19. When Desmond (2016) examined the distribution of rent in Milwaukee, Wisconsin, for example, only $260 separated the 90th percentile from the 10th percentile. For example, he notes, a two-bedroom apartment in the poorest neighborhood in Milwaukee—where more than 40% of people live below the poverty rate—is only $50 less a month than the citywide median.

20. To test the representativeness of my transactions data (i.e., whether repeat sales in this period are representative of all sales), I calculated the median sale price for all property transactions between 2010 and 2015 and the median sale price for the first and second sales in my data set of repeat sales. The median values of both the first and second sales fall within one standard deviation of the median of the whole sample. Thus, my data set is reasonably representative of general sales.

21. In both the rental and residential sales models, I exclude buildings that were listed as vacant as these are qualitatively different cases with separate sections of the building code, their own set of inspectors, and different court processes.

22. I obtain similar results when I include properties with more than 30 violations; however, the coefficients are slightly smaller in magnitude and less significant.

23. I used the coefficient for unresolved violations in Model 2 and the following calculation to obtain this figure: 100*(exp(− 0.00344*10)-1) = − 3.38.

24. I also ran models for rental units that only included units with at least one violation recorded between the two rents. Whereas the effects are similar to those for units with violations at any point in time, the coefficients are smaller in magnitude and less statistically significant (p = .055). This suggests that, unlike the case for residential sales prices, the effects of addressing violations on rent are not shaped by when the initial violations occurred. Property transactions are more sensitive to the recentness of violations.

25. This is not the case for rental properties; coefficients lose statistical significance when I only include rental units under $1000 or $1500, for example.

26. The citywide vacancy rate is relatively low (8.95% in 2006 and 5.22% in 2015), but Chicago’s vacant properties tend to concentrate in distinct neighborhoods (Mallach, 2018).

27. Chicago’s Department of Buildings runs workshops for landlords and advises screening potential tenants. The effects of this kind of screening are stark. In Milwaukee, for example, Desmond et al. (2015) found that renters whose previous move was involuntary were 25% more likely than similar renters to experience long-term housing problems.

28. Block group-level variables for race are not significant when I include them in my rent models.

29. Landlords who maintain their buildings—without the impetus of a building code violation—may also cover costs by increasing rents. In this way, my findings may illuminate a broader issue in the private housing market rather than pertaining only to building code violations.

30. Other lenders may also be less likely to furnish loans, and appraisers for lending institutions may be unwilling to take on a building with code violations as it is not worth their time. Prospective buyers may also be reluctant to buy a property with a long list of violations. The property transactions data set does not capture homes that are on the market but do not sell, or owners who are put off even trying to sell.

Additional information

Notes on contributors

Robin Bartram

Robin Bartram is an assistant professor of Sociology at Tulane University.