How to House the Homeless by Ellen & O'Flaherty (Book Review)
The relationship between housing policy and the macroeconomy has become more important for economists to study as the costs of spatial misallocation have become increasingly clear. For a middle class family, a poorly functioning housing market results in underhousing. My own story is a good example of this phenomenon. Despite earning a salary in the top quintile of the income distribution, I lived in bachelor or small one-bedroom apartments throughout my twenties. Were I living in a cheaper housing market (like Montreal or Saskatoon), I would have rented larger and nicer apartments. When higher income individuals purchase or rent the smallest sized apartments, the remainder of the population has to respond by reducing their consumption of housing services or increasing their share of income dedicated to its acquisition. Reductions of housing consumption can be accomplished through additional roommates or relocating to undesirable neighbourhoods. There is an effective limit, however, in how far housing consumption can be reduced.
Increasing housing demand on limited supply causes a cascade of price effects that are one of the many topics discussed in How to House the Homeless – a volume of policy papers edited by Ingrid Gould Ellen and Brendan O’Flaherty. The relationship between housing policy and homelessness is difficult to study because, like extreme value theory, the tails of the distribution have few observations and are difficult to model. The first section of the book discusses the most effective service models that homeless individuals can receive. The consensus from researchers is that a Housing First (HF) approach is the most effective policy. Traditional models of homelessness alleviation combine tiered housing with sobriety obligations. HF focuses on getting a homeless person into a housing unit with no strings attached. Drug and alcohol treatment program are optional, and the participants can access them whenever they want to. There is growing evidence that unconditional cash transfers can be remarkably effective. I am not surprised that “unconditional” housing transfers have a similarly salubrious effect.
There are two extremes in the policy debates around homelessness. Housing market impotence is the belief that people are homeless because of complex personal pathologies. Even in a world of abundant and cheap housing, most of today’s homeless would remain so given their myriad of complex problems. On the opposite side, housing market omnipotence views homelessness as simply a problem of a housing shortfall. A famine is the result of too few calories being available, or the distribution of those calories being poorly shared. Homelessness, according to housing market omnipotence, is a famine of shelter. The editors of this volume have selected a combination of articles that straddle these two extremes.
Because homelessness is difficult to predict using individual level features, housing market impotence seems unable to fully explain the data we observe. If homelessness is a completely random event, then individual behaviours are irrelevant. Alternatively, if there does exist a set of theoretically observable (but otherwise unmeasurable) features in the data, then no policy exists which could accurately target the individuals most at need of intervention. At the same time, the housing market omnipotence theory seems insufficient in being able to explain the temporal trends of homelessness in the United States.
Aggregate volumes of homelessness seem to be poorly correlated with aggregate volumes of the conditions that make individuals in a given population more likely to be homeless – for instance, being poor, being male, being a substance abuser, being mentally ill. This appears true both in a number of detailed empirical studies (for a review, see O’Flaherty 2004) and in trends over time. In 1972, for instance, the United States had staggering numbers of alcohol abusers by today’s standards, a raging heroin epidemic, de-institutionalization that was well on its way to completion for the non-elderly mentally ill, no Supplemental Security Income (SSI), poverty rates comparable to today’s rates–and almost no homelessness.
A lack of food leads to starvation because one cannot divide a calorie in half. Housing markets are much more dynamic. Individuals can couch surf, relocate to the cheaper parts of the city, or move into an apartment with numerous roommates to lower their costs. Interventions that affect the supply-side of the market can also lead to incentives, which on the margin, may or may not alleviate homelessness.
Real housing markets are not stagnant glasses of water where all you have to measure is the distance to the brim. Instead, they are oceans with tides, ever-shifting currents both horizontal and vertical, evaporation, rain-falls, and waves. You can’t control the height of the ocean with a hand pump and a bucket. Housing market players adjust to changing environments by moving in with relatives or moving out; by building, renovating, or abandoning; by exercising more or less caution in screening potential tenants; by looking more or less assiduously for a smaller or larger place to live and accepting more or fewer inconveniences; by raising and lowering advertising budgets; by putting on a new roof, adding a new bathroom, or letting the boiler deteriorate. Players have to adjust to the adjustments that other players make, too, so that in the end the final outcome of a policy’s implementation may look nothing like its initial thrust or intended goals.
The case of rent control is interesting. For tenants lucky enough to be in the rent-controlled sector, their chance of homelessness decreases because rental price shocks are taken off the table. However rent control almost certainly decreases the churn of new apartments available and therefore increases the price volatility for non-rent controlled apartments. Rent control also decreases the cost that landlords face when engaging in discriminatory practises. If single-mothers, for example, are assumed to be riskier tenants, then picking the next-best tenant (from the landlord’s perspective) will lead to a smaller price decrease than what would otherwise be needed in an uncontrolled market. There is conflicting evidence whether rent control, on net, decreases homelessness.
The Wharton Residential Land Use Regulatory Index was released in 2006 and has allowed housing economists to better understand the relationship between housing market outcomes and regulatory factors. In the past decade three robust findings have been established.
- Higher levels of regulation lead to higher costs for housing. This result is found even after accounting for the cost of land and building materials. Regulation acts as a tax that increase the general equilibrium price of housing.
- Higher levels of regulation lead to lower supply elasticities. e.g. Home building is less responsive to price increases in highly regulated markets, on average.
- The price level increases associated with the regulation tax are regressive. In other words, housing units at the bottom of the distribution see the largest price increases, relative to what would occur in a less regulated market.
The last stylized fact is often overlooked. Media attention tends to focus on dwellings that could be formerly afforded by the middle class. Crowding out at the bottom of the housing distribution may occur for several reasons. First, land costs are probably a smaller share of the overall cost for small apartments, meaning the regulatory tax becomes a larger share of the cost equation by algebra alone. Second, underhousing operates in one direction. As mentioned before, when higher income workers move down the housing ladder, there is a floor in housing services which means that demand will begin to pool for this minimum.
What does a regulatory tax embody? Some of the price increase is caused by the provision of amenities whose existence or scale would be less than what a developer would provide for in an unregulated market including private kitchens, complete plumbing, multiple exits, or parking space. There are many direct regulatory costs a developer must face just to have the right to build including time to prepare permits, legal fees, appealing zoning board rulings, and uncertainty around the length of the approval process.
Zoning regulation often restricts the amount of land within a municipality available for residential development and then dictates the density and quality of the housing that can be built. Growth controls, growth moratoria, exaction fees levelled on new development, and lengthy and complex project approval processes tend to discourage new housing construction and the nature of new housing that is ultimately supplied to the local market.
In Chapter 6 Steven Raphael shows that higher regulation leads to higher prices and higher prices leads to more homelessness. Specifically, the component of price changes explained by regulation is sufficient (e.g. statistically significant) to explain a portion of the change in homelessness through the price channel. Though the estimated effect size seems quite large, and I am often sceptical of putting too much weight on statistically significant results, the hypothesized mechanism seems likely to be a real one. It should be noted that the direction of the causal mechanism can be disputed since it is uncertain whether regulations may be a response to higher prices, rather than their cause.
Jane Jacobs famously argued that older buildings are beneficial, among other reasons, for their ability to keep prices affordable. While she was right that older buildings tend to have cheaper rents, were cities to stop building new apartments, the result would not be a lower overall price level. Higher income individuals purchase more housing services, which includes the quality of the building. Newer buildings are more likely to be occupied by a wealthy person because they are nicer. One way of viewing the housing market is through a filtering mechanism whereby the poor are the last to get their pick of a housing unit.
Thus the supply of lower-cost affordable housing is linked dynamically to the supply of higher-quality housing through filtering and depreciation.
The last chapter in the book, Homelessness as Bad Luck, does a good job at explaining why homelessness is likely difficult to model econometrically given what we know about consumption theory. Homelessness is very bad. Individuals therefore have a strong incentive to prevent it from happening. If negative shocks cannot be predicted, then the level of housing consumption today is a sufficient statistic for its level tomorrow. Alternatively, if negative shocks can be predicted, but individuals still become homeless, then there must be a set of frictions that prevent people from saving. Both forces are likely to be at play. Saving and borrowing for the very poor is difficult because savings accounts are only economic for banks to provide when a customer is able to maintain a minimum balance. The interest rates charged by pay-day lenders are notoriously large. More generally “hedging” against real estate price changes is difficult. For example the only futures market in the world that exists for home price indices is the CME Case Shiller, and the market has proven to be quite thin.
Ellen and O’Flaherty suggest there are three main takeaways from the book. The first is that housing matters. The focus on “housing” ranges from programs that prioritize giving the homeless access to physical apartments to government regulations that drive up the cost and availability of housing for the lower end of the market. Second, who is eligible for a program is critical. Programs which target those most at risk of being homeless are, unsurprisingly, the most effective at reducing homelessness. However, the third takeaway is that the degree of targeting is inversely proportional to the amount of moral hazard created. If benefits are removed the moment an individual moves outside the “target” zone, there will be an incentive to remain there. As housing prices continue to outpace incomes in most metropolitan areas, it is an unfortunate reality that homelessness research will increase in social importance.
As always, even if the “sign” of the effects are known, their magnitude and heterogeneity may be epistemologically limiting. ↩
On a technical level, the only methodology that would be accepted as a credible tracker of home prices is repeat sales. However this approach only works for markets where a sufficient number of observations can reduce the level of noise to an acceptable level. ↩
This is similar to the principle of claw-backs that poor individuals on welfare experience when they begin to earn a labour market income. For example the Ontario Disability Support Program (ODSP) levies a 75% effective marginal tax rate for individuals earning more than \$300 a month. Imagine an individual works one day a week and earns \$300 a month. If their employer offers them two shifts a week, they can increase the labour input by 100% for the privilege of a 25% increase in take-home pay. It would therefore not be surprising if most people on ODSP decided to work enough hours to earn around \$300 a month! ↩