OMB Justification

Follow-up to questions raised by OMB during March 20, 2017 meeting

Disaster Planning Module

There were no questions in the Disaster Planning module related to climate change and the phrase “climate change” is not mention in the module.

Motivation for Eviction Module

Several researchers have noted that housing insecurity is a major problem in the U.S. and negatively impacts health outcomes¹. To help shed light on this issue, the 2013 AHS included a series of questions on housing insecurity, including questions on doubled-up households; questions on difficulty paying rent, mortgages, and utilities; and questions on likelihood of eviction or foreclosure.

Since 2013, new research from Matthew Desmond, a sociologist at Harvard University, has revealed a major data gap in our understanding of housing insecurity: the prevalence, causes and consequences of evictions. Although prior AHS instruments included a question about eviction, a thorough review of the existing AHS question, coupled with review of Desmond’s research, led HUD to conclude the AHS was likely not producing accurate national counts of evictions. Moreover, the AHS was not capturing any reasons why evictions occur.

To fill the data gap, HUD developed the Eviction module. The Eviction module includes questions on the causes, process, and effects of formal (i.e., court ordered) and informal eviction. There are questions on the types of eviction warnings or notices received, what led to the eviction (i.e., raised rent, property condemned, etc.), and whether children had to switch schools because of the eviction. The content of these questions was taken from Desmond’s Milwaukee Area Renters Study (MARS). The AHS eviction questions will be limited to respondents who moved from a place that was either rented or occupied without payment of rent.

Motivation for Commuting Cost Module

Quantifying housing costs is one of HUD’s strategic goals for the AHS. Housing costs typically include rent or mortgage, upkeep, and utility costs. However, many researchers have noted housing costs and commuting costs are closely related because a household’s choice of housing location impacts the household’s commuting costs. According to the Consumer Expenditure Survey, households spend more than 43 percent of their income on housing and commuting costs.

A HUD goal for the 2017 AHS was to estimate commuting costs for AHS households. HUD explored a few options for producing commuting cost estimates. The first option was to apply estimates from the Consumer Expenditure Survey to AHS households. However, the Consumer Expenditure Survey does not account for workplace location. As such, a 2-person household living in Fairfax county, VA, would have the same estimated commuting cost as a 2-person household living in downtown Washington, D.C. Moreover, the Consumer Expenditure Survey does not account for mode of transportation or portion of transportation costs devoted to commuting.

A second option was to apply ACS/SIPP-based estimates developed by Edwards, McKenzie, and Short. They combined the American Community Survey (ACS) and data from the Survey of Income and Program Participation (SIPP) and other sources². This method has a lot of appeal in that it addresses within-metropolitan area variation in commuting costs and variation in commuting costs across types of commuting modes. However, the Census Bureau posited that the ACS/SIPP-based methodology is limited in that it is a derived estimate of commuting cost using proxy variables rather than a direct measure. For example, for ACS, home and workplace locations are used to estimate distance to work, then distance is used to calculate cost using a mileage reimbursement rate. This methodology makes several assumptions about travel route, the price of gas, congestion, auto efficiency, and other factors that influence commuting cost. Further, the place of work information used to approximate travel distance is not available for all workers, so about one-quarter of respondents are omitted from the commuting cost calculation.

In order to apply ACS/SIPP-based estimates, HUD would have to make a few assumptions about AHS households, including where they work, how they get to work and how often they report to a workplace location. This is certainly not difficult to do. One option would be to use a propensity scoring model to match an AHS household to a similar ACS household (the ACS includes distance to work and commuting mode), then use apply the ACS/SIPP-based estimates to the AHS household.

After consideration of alternatives and consultation with the Census Bureau, HUD ultimately elected to develop our own commuting cost module for two principle reasons. First, subject matter representatives from the Census Bureau have expressed their support for the 2017 AHS commuting cost module, and they have indicated that these questions would fill a longstanding void in information directly related to commuting cost. Specifically, the AHS provides an opportunity to directly ask respondents about commuting cost, providing a more accurate and relevant measure of this concept. HUD developed the module in conjunction with the Journey to Work team at the Census Bureau (Brian McKenzie). The 2017 AHS commuting cost module includes crucial questions related to household commuting expenses that are not available from the ACS or other large household surveys. The AHS commuting cost module asks about individual components of commuting cost associated with several travel modes. For example, gas, tolls, and parking costs are captured for those who drive to work, and transit fares are captures for transit users. Respondents also directly report travel distance, an important component of travel cost and overall burden that is not captured by the ACS.

Second, HUD believes the commuting cost module presents an important opportunity to conduct commuting cost research to support future surveys. To that end, HUD plans to compare commuting cost estimates for the 2017 AHS with alternative model-based estimates, such as those suggested above. If it turns out the model-based estimates are good predictors of actual commuting costs, HUD can permanently add commuting cost estimates to the AHS without needing to collect additional data. Moreover, the Journey to Work team at Census would have additional evidence to evaluate their ACS/SIPP-based estimation methodology. On the other hand, if it turns out the ACS/SIPP-based estimates applied to the AHS are not good predictors of commuting costs for AHS respondents, then HUD will be able to make a better-informed decision in the future regarding if they should include the commuting cost module in future surveys.

1 For example: https://www.cdc.gov/pcd/issues/2015/14_0511.htm; https://www.huduser.gov/portal/periodicals/insight/insight_2.pdf

2 https://www.census.gov/hhes/povmeas/publications/SGEworkexpense.pdf