In collaboration with Yi Li and Brian Zage
Affordable housing is a topic that is debated constantly among American policy makers. The implications of affordable housing and providing for the homeless cannot be understated, as millions of people require these services in order to survive. While great strides have been made in terms of offering help to those who need it, our country still faces this very serious and very real problem.
Homelessness has always been an issue in the United States, and has become a problem that government entities have tried to remedy. Policy makers have been tasked with the job of finding a way to successfully approach the problem while finding effective ways to eliminate or reduce the amount of homeless people that exist. Considering each individual in this world is born into a different situation, how do we determine who needs the most help and which people to target for affordable housing?
The U.S. Department of Housing and Urban Development as well as other agencies and organizations have all studied these variables to attempt to reduce the amount of people in our country who are homeless. While some have been successful in doing this, there are still a large amount of people who are not being provided with the support and aid that they need and warrant. In his study, Early (2004) attempts to further understand the causes of homelessness by finding the largest determinants and measuring how well current housing assistance programs are at addressing these issues.
When choosing who is eligible for different types of programs that are offered to the homeless and the poor, many variables are taken into account such as gender, family size, income level, and age. Policy has been created to target those in need, but the question of how well these policies are doing remains. Therefore, the goal of this study is to address how successful these policies are by figuring out if those who need assistance the most are actually getting it. Early (2004) was able to do this by studying poor households and the probability of homelessness among them. Combining this with information on those poor families who are in subsidized housing, we can then see how many families would become homeless without the assistance of their subsidy.
The study combines data from previous studies and other separate surveys to generate samples of homelessness and low-income households under unsubsidized renting housing. A logit regression is used to determine the probability of being homeless as a function of characteristics of households and the areas the households are located. Then the author simulates the results of the regression to a sample of households who are currently under subsidized programs to estimate the fraction of households who would be homeless under the absence of the subsidy. Observations of homelessness and poor households under unsubsidized housing are from the 1996 National Survey of Homelessness Assistance Providers and Clients (NSHAPC), and data was collected through 22 largest metropolitan areas. Finally, Early (2004) incorporates 1418 observations on homeless and 2069 observations on poor households in unsubsidized rental housing. In order to test the effectiveness of the current subsidy program, 540 observations of households under subsidized housing are drawn from SIPP and NSHAPC data sets. Early (2004) furthers the study by taking endogeneity of income into consideration. He applies a two-stage estimation model in control of how endogeneity of income would affect the result.
The three variables of household markets included in the regression model are price of substandard rental housing, the lowest rate needed to occupy a rental unit, and the vacancy rate of low cost housing. As there are no sources of these variables existing, the author uses a hedonic model as a function of structural characteristics of house unit, measure of unit quality, and neighborhood to estimate these variables.
Other area-specific variables included are measures of average level of aid directed toward low-income households, max AFDC benefit, amount spent per personal served by community health care clinics, and shortage of mental health care. The author believes all these factors are necessary to consider in the regression model. Further, Early (2004) put household-specific variables into consideration. These variables are income, household size, gender, race and ethnicity, age and head of the household. While inclusion of the above factors is obvious, Early (2004) has special interest in whether Vietnam veteran status is a factor. He believes Vietnam veterans are underrepresented in homelessness studies.
Due to the potential endogeneity of income, there are three specifications of the model that are presented. Model 1 assumes income is exogenous, Model 2 treats income as endogenous and employs a two-stage process to estimate the model, and Model 3 is a reduced form regression. Early (2004) substituted the equation explaining income for the equation explaining homelessness to remove income from the regression and add the explanatory variables used in the income equation. Therefore, when running the logit regressions, we are able to see the variables being studied and their levels of significance among all three models.
The findings were consistent among the three different modifications. Households headed by women and headed by persons more than 50 years of age were less likely to be homeless. Also, it was found that African-Americans are more likely to be homeless, when everything else was held constant. States with higher per capita spending on community health care resulted in less households likely to be homeless. Average minimum temperature was inversely related to the probability of being homeless, which surprised the researchers. While one would think the housing market in a given location would play a large role in homelessness, only vacancy rates for low rent units in the reduced form regression was a statistically significant factor. All of the other variables used to describe conditions of the housing market were not significant in the study.
After this initial analysis, Early (2004) moves the research to the next level, providing the predicted probability of being homeless of a hypothetical low-income household. As certain variables such as monthly income, age, married, and Vietnam veteran are changed, the numbers adjust and the percent change in the probability of being homeless fluctuates. The research shows that with other variables being held constant, changes in demographic characteristics largely influence the predicted probability of the household being homeless. Marital status, income, gender, and race were the variables with a strong pull, while variables such as increasing the price of substandard housing and rental vacancy rates for low-cost housing did not have much pull.
In conclusion, the study shows that female-headed households, households with older heads, and nonminority households are less likely to be homeless. This result mirrors the study done in mid-1980s. The simulation of results of the regression model compared to the current households under subsidized programs indicates that housing authorities do not target those who are at the most risk of being homeless. Early (2004) states that while men make up a large fraction of homeless population, housing subsidy programs tend to support households with children. Moreover, those who are at most risk to be homeless tend to have alcohol and drug addiction problems. This would limit chances of being subsidized. The author asserts that housing authorities should do more to target those who are at most risk of being homeless without relaxing the support on families with children. Early (2004) argues that the current income eligibility for being subsidized is too strict and that it does not effectively target those who are at the most risk of being homeless. According to Early (2004), based on setting preferences used for housing authorities to select who should receive assistance, nothing should prevent them from targeting those who are most likely to be homeless in the absence of subsidy. However, the fact is that 27 percent of housing staff authorities tend to assist people who are in school, working, and under job training programs, and less than 12 percent of them follow the hardship measurements of the selecting process. Early (2004) finally assists that the federal guidelines should require housing authorities to target the poorest of the poor households, and that income eligibility requirements should be lowered.
In their book, House of Debt, Mian and Sufi (2014) investigate reasons causing the great recession. They insisted that the subprime mortgage market crash in 2007 caused deep reduction in household spending, which induced the great recession. They also demonstrate the concentration losses on debtors and the amplification effect due to foreclosure. A lot of people became unemployed and had become homeless without housing subsidies during the great recession. Before the Great Recession, Early (2004) argued that the targeting of housing assistance is not sufficient because the minimum income requirement is too strict. Therefore, after the Great Recession, this only becomes more true as millions of jobs were lost, and the average median income level among the lower class was only further reduced.
Early (2004) mentions several data limitations which would bias or underestimate the results. According to Early (2004), the NSHAPC gathered data on homelessness at one point of time instead of a period of time. This fails to indicate households who would remain homeless in the near future. Also, the one-point prevalence of homelessness would underestimate the effectiveness of the subsidy program. Furthermore, studies tend to sample those who spend a long time being homeless, and these households are inclined to associate with certain factors (severe chemical addiction and problem with mental illness). This disproportion of data would be biased to show a strong relationship between homelessness and these certain factors. Furthermore, the one-point prevalence targeting of homelessness would underestimate the effectiveness of current subsidized programs. Moreover, the author suggests that the length of duration households have received housing subsidy may alter characteristics of households. Since no such measure is included in this study, further possibilities could not be explored.
Research should be conducted to show the timetables of the families in question, and how often they slip in and out of homelessness. The data that Early (2004) analyzed does not take into account how volatile the life of a homeless family can be. More studies can be done to measure how successful affordable housing programs and subsidized housing are at keeping families off the streets in the long term. Appropriately so, many housing programs are targeted at supporting families who have dependent children. Unfortunately, this prevents many young-adult men from getting the assistance they need. Often removed from their homes for a variety of reasons, young men have the tendency of getting lost in the process. Therefore, future research should also take into consideration these nonelderly men who are forced to provide for themselves, and can determine a way to create better programs for them with the hopes of possibly saving their lives.
-
1. Through the study, Early (2004), which households were found to be less likely of being homeless?
a. Households headed by women
b. Households headed by persons over the age of 50
c. Households headed by young men
d. Both (a) and (c) are correct
d. All of the above
Explanation: The correct answer to this question is (d). Through the regression, the study found that households who were headed by women and headed by persons over the age of 50 were less likely to be homeless. Surveys have found that the most dependable households are those who are headed by older adults, while women have also been found to be more dependent than men. Many housing programs are targeted at providing households of families with children with assistance and aid, which has made them less likely of being homeless than those households made up of single young men.
-
2. According to Early (2004), which of the following explains why housing authorities fail to target those who are at the most risk of being homeless in the absence of subsidized housing?
a. The income eligibility requirements set by housing subsidy programs is higher than median annual income of homeless families
b. They tend to target those with severe alcohol and drug use
c. They have preference to subsidy households who are working or are in school and under job training
d. All of the above
e. Only (a) and (c) are correct
Explanation: The right answer is (e). Only (a) and (c) are correct. The author mentions in the article that federal guideline for eligibility for public housing and Housing Choice Voucher Program is defined to be less than 80 percent of area median income. Recent legislation has taken consideration of extreme low-income households, and the requirement is set to be less than 30 percent of area median income. However, the median income of homeless is much lower than defined limit of even extremely low income in 22 metro areas, which makes harder for those who are at most risk to be homeless to receive assistance. (b) is not correct, because severe alcohol and drug use actually confound housing authorities in targeting these households who are most risk of being homeless. (c) is correct, as less than 13 percent of housing authorities follow hardship measurements of who should be assisted and who should be not. They tend to subsidy those who are in school or at work and under job training, believing they are worth subsidizing.
-
3. According to Early (2004), all of the following variables played a large role in determining homelessness EXCEPT:
a. Variables describing conditions of the housing market
b. States with high per capita spending on community health care
c. Monthly household income
d. Household size
e. Both (a) and (d) are correct
Explanation: The answer is (a), variables describing conditions of the housing market. While we would think that the housing market plays a very large role in determining homelessness, this is not entirely the case. The problem does not simply lie with the amount of affordable housing that is offered, but is more related to whom the subsidized housing is offered to. The regression shows that (b), (c), and (d) are all variables that are statistically significant in determining homelessness, while variables associated with the housing market are not.
References
Mian, A. and A. Sufi (2014), “House of Debt,” University of Chicago Press: Chicago, USA.
Early, D. (2004), “The Determinants of Homelessness and the Targeting of Housing Assistance,” Journal of Urban Economics 55, 195–214.