rbraman

Study Proposal submitted to Griffin Housing Authority (GHA) on April 28, 2014.

Principal Investigator

Jeff Jordan, Professor, Agricultural & Applied Economics, University of Georgia

Co-principal Investigator

Abdul Munasib, Research Scientist, Agricultural & Applied Economics, University of Georgia

I. Scope

Meriwether Redevelopment (MR), the ‘intervention’, provides a natural experiment where the former residents of Meriwether constitute the treatment group. The control group consists of the residents of Fairmont and public housing applicants on the GHA waiting list. This study will assess various impacts on the families of those former residents of Meriwether Homes by comparing their experiences with those at Fairmont Homes as well as people who are currently on the GHA waiting list utilizing this natural experiment setting.

MR provides an opportunity to examine two kinds of effects. First, as the voucher recipients choose a neighborhood (henceforth referred to as ‘move-out’), we can examine the impact of the chosen neighborhood on her and her children’s (if any) observed outcomes. Secondly, once the redevelopment is completed, conditional on the return of the resident (henceforth referred to as ‘move-back’), we can again assess the impact of redevelopment on her and her children’s observed outcomes.

II. Outcomes of Interest

1. Relocation decisions: Both ‘move-out’ and ‘move-back’ decisions are prompted by the intervention (MR). In case of ‘move-out’ we can examine the neighborhood choice resulting from the voucher program. Following up on the ‘move-back’ decision, we can examine the impact of the improved environment resulting from the redevelopment.
2. Self-sufficiency: The goal of the voucher program is often to prompt greater self-sufficiency on the part of the voucher recipient. The ‘move-out’ portion of MR gives us an opportunity to examine the validity of this expectation. Even for the ‘move-back’ portion of MR, we can examine if the improved neighborhood environment could have a similar impact.
3. Family structure: We are interested in studying what impact these relocation decisions and subsequent changes in environment have on family structure. In particular, we plan to explore the issue of male role model.
4. Children’s education: As in the case of self-sufficiency, both the ‘move-out’ and the ‘move back’ portions of MR merit examination of their respective impacts on the educational outcomes of the children of the treated families. In this regard, we plan to acquire supplementary information from the school district.
5. Health: Some of the most robust effects estimated in the HUD-sponsored ‘Moving to Opportunity’ (MTO) studies are those related to health. These studies found evidence of improved physical and mental health for those who left public housing. We intend to re-examine these effects for both adults and children, especially in the context of the presence of social interaction interference (see section IV below).
6. Safety: While both Meriwether and Fairmont Homes are relatively safe environments, the neighborhoods surrounding them are less so. The ‘move-out’ portion of MR provides us with an opportunity to examine if the voucher recipient relocates to a relatively safer neighborhood, and if so, whether their perception of safety (as well as realized safety) is affected as a result. As for the redevelopment – which is generally aimed to impact not only the housing complex but also the surrounding neighborhood – the ‘move-back’ part of MR provides us with an opportunity to examine the direct impact of a safer neighborhood on the perception of (and the actual) safety of the treated families.
7. Quality of Life: Some of the most important potential neighborhood effects may be in the form of intangibles such as quality of life, perception of fulfillment, and happiness. Of particular importance, especially for the children of the treated families, are aspirations and expectations. These are some of the quality of life aspects that we are interested in investigating.

While it may not be feasible to cover each and every one of these topics, we plan to accommodate as many of these issues as allowed by the scope and feasibility of the information gathering process.

III. Information Gathering

Some of the basic information would be collected from the Griffin Housing Authority (GHA), the primary partner of this study, and the Griffin-Spalding County School System. We also plan to conduct a number of surveys.

1. The Baseline: Surveys will be conducted with current Fairmont residents, former Meriwether residents (housing voucher recipients) and individual/families in the GHA waiting list to gather information pertaining to the pre-move baseline. All the individuals surveyed in the baseline will constitute the original sample.
2. First follow-up: This round of information gathering will take place within the first six months of the re-opening of Meriwether, for everyone in the baseline original sample (subject to availability).
3. Final follow-up: Approximately two years after the reopening of Meriwether, we plan to follow-up with everyone in the baseline sample (subject to availability).
4. Intervening periods: Between the first and the final follow-ups, we may consider a round of intervening survey. Similarly, in between the baseline and first follow-up, depending on how much time has passed, we may have to consider conducting another intervening survey.

IV. Expected Contributions

The benchmark study in this area of research is the Moving to Opportunity (MTO) project sponsored by the U.S. Department of Housing and Urban Development (HUD) that, in 1994, selected families with children in five sites (Baltimore, Boston, Chicago, Los Angeles, and New York) residing in public and assisted housing projects in census tracts with at least 40 percent poverty rate (See Goering et al. (1999) and Goering et al. (2002) for a description of the program). At each site, randomization was used to assign participants to,

1. a control group,
2. an experimental group receiving counseling and housing vouchers that could be used only tomove to a census tracts with poverty rates less than 10 percent, or
3. a Section 8 group receiving vouchers that could be used anywhere.

The objective was to design an experimental set up so that comparisons between the control group and the experimental (and/or the section 8) group would yield unbiased estimates of neighborhood effects. This can, however, be problematic when substantial fractions do not lease up (i.e., use their vouchers to move). The Chicago lease up rate, for instance, was 34 percent in the experimental group, and 66 percent in the Section 8 group (Goering et al. 1999).

In contrast, in case of Meriwether Redevelopment (MR) the treatment group, by construction, did not have a choice to not lease up. This is a particular advantage of this project that has the potential to offer a more accurate and unbiased estimate.

A specific methodological emphasis in this proposed MR study would be the issue of social interaction that contaminates estimate of neighborhood effects. Neighborhood effect estimates in experimental settings, such as the MTO project, are based on the implicit assumption of no interference between units – that is, a subject’s response to the intervention depends only on the treatment to which that subject is assigned, not on the treatment assignments of other subjects (Sobel 2006). For the MTO studies, this assumption is not reasonable because interference is common in these social settings (e.g., schools and social networks).

Consider, for example, the relocation decision of a former Meriwether resident – while looking for a rental unit she is likely to be heavily influenced by other Meriwether residents who are also looking for accommodations. Note that the relocation decision determines the neighborhood which, in turn, determines the degree and the nature of the neighborhood effects that we intend to estimate. Furthermore, whether she subsequently decides to return to Meriwether after the completion of the redevelopment may also be determined by other former residents’ decision to return. The social network developed at Meriwether, thus, could play a crucial role in the determination of the neighborhood effects.

These social network (or social interaction) effects are the so called interferences that breaks down the no-interference assumption. When data from experimental studies are analyzed under the no-interference assumption, very misleading inferences can result; a research project that fails to recognize this could easily infer that a treatment is beneficial when in fact it is harmful (Sobel 2006). Additionally, the consequences of interference (e.g., spillovers) are themselves of great substantive interest. This project aims to accommodate social interaction (or social network) effects in assessing the neighborhood effects.

V. Time Frame and Deliverables

Since the issuance of vouchers for the Meriwether residents started in November 2013, we plan to begin the baseline survey and basic information gathering immediately. With the final follow-up taking place two years after the re-opening of the redeveloped Meriwether, we plan to submit the final report within a year (12 months) of the completion of the final follow-up survey, which involves preparing the newly collected data, analyses of the data, testing of the hypotheses, and the writing of the report.

Additionally, within a year (12 months) of the completion of the first follow-up, we plan to prepare an intermediate report analyzing the ‘move-out’ part of MR which, again, involves preparing the newly collected data, analyses of the data, testing of the hypotheses, and the writing of the report. Throughout the study period we plan to prepare quarterly progress reports.

VI. Budget