«reNtal HousiNg Policy iN tHe uNited states Volume 13, Number 2 • 2011 U.S. Department of Housing and Urban Development | Office of Policy ...»
The strongest result on this score is that exposure to lead in childhood is very bad (Reyes, 2007) (for reasons of education as well as crime). Children can be exposed to lead in paint and lead in the atmosphere. HUD guidelines that prohibit leaded paint—a structural issue—are thus a major crime-fighting tool. Since leaded gasoline was phased out in the 1980s, the author is unaware of how atmospheric lead concentrations vary. It would be good to know this. Treating atmospheric lead as a nuisance in the ways described in the section on health could thus cause a long-run reduction in crime, if there still are meaningful differences in atmospheric lead concentration.
Another strong result is that education reduces future crime (Lochner, 2010; Lochner and Moretti, 2004). The steps outlined in the section on education therefore reduce crime.
Aside from these two results, there are no other strong results about childhood experiences that cause future criminality. In particular, nothing about architecture, poverty concentration, or public housing seems to make kids grow up to be criminals.
Thus, to reduce long-run criminality, HUD should continue to be vigilant about structural lead paint, penalize atmospheric lead, and reward good schools in the ways that have been discussed.
Short-Run Criminogenic Influences Other neighborhood features may increase or decrease the total volume of crime more immediately.
Landlords should be rewarded for locating in neighborhoods that have good features, and penalized for locating in neighborhoods that have bad features.
Unfortunately, not a lot is known about what these good and bad neighborhood features are.
Numerous papers report an association between liquor stores and bars on the one hand and crime on the other, and many assume that churches help to reduce crime. Yet, no hard evidence exists demonstrating that these property uses actually shape levels of crime. (Gyimah-Brempong (2006) tried to connect liquor stores and crime but did not have a convincing identification strategy.) Wilson and Kelling (1982) argued that visible disorder in a neighborhood (for example, broken windows) encourages crime, but this hypothesis has not fared well empirically (Fagan and Davies, 2000; Harcourt and Ludwig, 2006). DiTella and Schargrodsky (2004) showed that police patrol reduces crime in a natural experiment. But police patrol levels are usually correlated with unobserved features that increase crime—police patrol more in dangerous neighborhoods—and so it makes little sense to reward landlords in neighborhoods with greater police presence.
Dahl and Della Vigna (2009), however, found that violent movies tend to incapacitate violent people while they watch them, and that these people do not compensate fully for the period of
incapacitation after the movies. Perhaps landlords should be rewarded for locating near theaters that show violent movies.
In general, incentives should be based on evidence, not speculation. Hence, only violent movies should even be considered at this point as a short-run criminogenic influence.
Short-Run Neighborhood Effects “Not in my backyard” (NIMBY) fits in this section. People who live in subsidized housing may tend to commit more crimes than wealthier Americans, and so their unsubsidized neighbors may be upset about their presence. Perhaps victimizing neighbors may be an external cost of subsidized housing that should be internalized.
This reasoning, however, is incomplete. Suppose that some subsidized tenants have a high propensity to commit index crimes against their neighbors. Moving them from neighborhood A to neighborhood B hurts the people in neighborhood B, but helps the people in neighborhood A.
To the extent that the location of subsidized housing affects merely the location of crime, not its volume or severity, it should be of minimal social concern (although it could affect proper allocation of police resources). (An analogy is domestic violence: to a first approximation at least, where a family is living when a domestic violence incident occurs is of no concern.) It is possible to offer various hypotheses about the type of neighborhoods where crime should be highest, but there is little empirical evidence about the causal influence of neighborhood conditions. For instance, bringing low-income people into a rich neighborhood might increase burglary because there is more to steal, but it might decrease motor vehicle theft because cars are more likely to be in garages at night. White neighborhoods might encourage robbery because evidence suggests that White people are less likely to resist, but research finds that Black people tend to carry more cash, making them potentially more attractive targets (O’Flaherty and Sethi, 2008).
Dense neighborhoods present more criminals with more targets but also confront them with more potential witnesses.
The Moving to Opportunity (MTO) experiment sheds some light on this issue, but not much.
Young men who moved to richer neighborhoods committed a few more crimes than those who stayed in lower income neighborhoods, and this evidence suggests that richer neighborhoods are relatively criminogenic in the short run. But the net change in crime in either set of neighborhoods is not known (the extent to which crimes committed by MTO teens would have been committed by someone else if the MTO teens were not around). Nor does MTO tell us much about older potential criminals.
Until more research is done, then, it seems best to consider criminal effects on neighbors as essentially a wash in social terms.
Index Crime Between Tenants The same conclusion applies to index crimes between tenants. If some prospective tenants are likely to commit crimes against their neighbors, it does not matter who their prospective victims are: HUD has no stake in deciding who the victims are. (Indeed, fairness suggests that 140 Rental Housing Policy in the United States Rental Housing Assistance for the 21st Century if HUD should protect someone, it is those low-income people who are not lucky enough to receive subsidies; hence, HUD should not be eager to encourage potential criminals to move into neighborhoods with the unlucky low-income people who do not get subsidies, as seems to be the consequence of the “one-strike” policy, for instance.) Street Vice Street vice means illegal commercial transactions involving a willing seller and a willing buyer, where the seller deals with many buyers, but has ongoing relationships with few of them, and where buyer and seller must come together in close physical proximity (O’Flaherty and Sethi, 2010). Open-air, anonymous drug selling is the variety of street vice that receives the most attention, and presents special issues for HUD.
Street vice is a business (and almost certainly a business smaller than clandestine, relationshipbased drug-selling) that locates where it is most profitable to locate.4 O’Flaherty and Sethi (2010) set out several reasons why street vice tends to be concentrated in African-American neighborhoods, even though drug demand is not concentrated in these neighborhoods.
Within any neighborhood, the best locations for street vice depend on physical features that have not been studied—perhaps easy access to highways or clear sightlines in many directions, for instance. Hence, in many neighborhoods, HUD-assisted housing, particularly public housing, may be among the best locations for street vice. It would be good to know this for sure.
Some clear solutions to this problem would be legalizing most currently illicit drugs or subsidizing the development of good substitutes. Such a program, however, is not within HUD’s purview.
This situation presents two kinds of problems for HUD.
One problem is how to reduce street vice in developments that have not been built yet. Obviously, research needs to be done on the structural and locational correlates of street vice. Future developments should be designed with these in mind.
To some extent, of course, better architecture will just shift street vice to less lucrative locations;
if that were the case, the investment in architecture would be misdirected. As long as the supply of street vice sites is not perfectly elastic, however, there will be real effects. While the elasticity of demand for illicit drugs is low, it is not zero, and the elasticity of demand for anonymously purchased illicit drugs is almost certainly higher than the overall elasticity of demand (Becker, Murphy, and Grossman, 2006). Hence, making HUD’s buildings less attractive places for street vice may not just dump the externalities on someone else. (Clandestine drug sales have considerably lower external costs than anonymous sales.) Moreover, to the extent that HUD-assisted developments are more densely populated than other neighborhoods where street vice might locate in the same city, moving street vice away from these developments reduces the external costs that street vice produces, even if the total volume does not change.
Buildings that have already been built present a different issue. Street vice is a neighborhood blight just like air pollution, and so the basic response should be to reduce landlord subsidies when For estimates on the relative size of the clandestine drug market, see O’Flaherty and Sethi (2010).
street vice is occurring nearby. The difficulty with this approach is measurement: people cannot be rewarded or punished for something that is not credibly and verifiably measured. Arrests, for instance, are evidence of action being taken against drug-selling, not of street vice or even drugselling. HUD, however, could employ testers to try on a random basis to buy drugs anonymously in or near assisted housing, or subsidize local police departments to do so. Testing programs might create big risks for landlords if they sampled too little, and would be very expensive if they sampled too much; the definition of “near” would also produce other tradeoffs. But sampling programs are a straightforward attempt to provide the right incentives, and so some decent tradeoff might be found.
Maybe a better measurement strategy would be to look at reported violent outdoor index crime (excluding rape and domestic violence) and shootings in the vicinity of HUD-assisted housing.
This is actually measured, and may be closer to the thing that should be measured. The external costs of street vice are the problem, not street vice itself, and so landlords should have incentives to minimize these costs. (As technology becomes cheaper, HUD might want to install shot-monitoring devices on all assisted housing; this could serve deterrence as well as incentive purposes.) One-Strike Rules Direct incentives like these are likely to be more effective than one-strike rules because they address the real problem—index crime and street vice near HUD-assisted housing—rather than some variant— index crime and street vice by HUD-assisted tenants. No known serious empirical evaluation of onestrike rules has ever been attempted, and theory suggests that their effectiveness is probably tiny.
To understand the theory, consider this scenario. Suppose that Congress were dominated by vegetarians who had not studied basic economics. To discourage meat-eating, they order periodic surprise raids on McDonald’s restaurants. In these raids, they detain all the employees. Any HUDassisted tenants among the employees are evicted immediately; other employees are blacklisted so they may never receive HUD assistance in the future.
What does this policy do? It raises the price of hamburgers a little bit and raises the wage of McDonald’s employees, but many substitutes for HUD tenants and aspiring tenants are available, and so the effect is not large. Most importantly, it does not substantially change the locations that McDonald’s chooses for its restaurants. If McDonald’s found it profitable to put a restaurant near or inside a HUD-assisted project before the vegetarians took over, it would almost certainly continue to find it profitable.
Since it appears that labor is supplied to street vice pretty elastically (Reuter, MacCoun, and Murphy, 1990), the drug-selling one-strike rule should have the same effect on street vice locations—approximately nothing. That is why a serious empirical evaluation would be helpful.
(Essentially, the question is whether the elasticity of land supply to anonymous drug-selling is less than the elasticity of labor supply to anonymous drug-selling.) The current one-strike rule, moreover, imposes real costs on tenants and prospective tenants— breaking up families, for instance. Treating young single adult minority males as pariahs contributes to many social problems that have large external costs—homelessness and homicide, for instance.
In thinking about the role of subsidized housing in the larger society, HUD may want to move toward a more goal-oriented and less soundbite-oriented policy. Landlords in a goal-oriented regime may very well bar felons in many cases, but they would be doing so for real reasons.
Conclusion Not all these goals need to be accomplished immediately. Just remember that a safety net today is different from a safety net in the 1930s, and externalities today are different too. All the rest follows.
Acknowledgments The author has benefited from helpful discussions with Rosanne Haggerty, Ingrid Ellen, Marah Curtis, Joseph Tracy, and staff at the New York City Department of Homeless Services and HUD.
Any remaining errors are the author’s.
Author Brendan O’Flaherty is a professor at Columbia University.
References Ambrose, Brent W. 2005. “A Hazard Rate Analysis of Movers and Stayers in Assisted Housing Programs,” Cityscape 8 (2): 69–93.
Becker, Gary S., Kevin M. Murphy, and Michael Grossman. 2006. “The Market for Illegal Goods:
The Case of Drugs,” Journal of Political Economy 114: 38–60.
Benjamin, John D., Kenneth M. Lusht, and James D. Shilling. 1998. “What Do Rental Contracts Reveal About Adverse Selection and Moral Hazard in Rental Housing Markets?” Real Estate Economics 26 (2): 309–329.
Bentolila, Samuel, and Andrea Ichino. 2008. “Unemployment and Consumption Near and Far Away From the Mediterranean,” Journal of Population Economics 21 (2): 255–280.