Quantifying economic outcomes for whites of minority-targeted programs
November 23, 2016 12:00 PM   Subscribe

In personal experience, many programs notionally or popularly perceived to be aimed or benefiting primarily minorities end up making me (white dude) better off as well, e.g. Section 8 housing assistance. I'm trying to identify data and research sources that either support or diminish this claim.

The hunch I have is that even as a white guy (therefore theoretically not the beneficiary), social programs with a disproportionate positive effect on minorities (e.g. disability payments, which do have a disproportionately high number of minority recipients) end up benefiting me too, either primarily or secondarily, and I'm trying to think of ways to ask that question that deal with publicly-available data (e.g. ACS).

For an example, I'm wondering if in counties/Census tracts/states with proportionately higher numbers of spending in disproportionately minority-benefiting programs (e.g. SNAP or disability payments) end up seeing better economic outcomes for whites on the whole (both those who do and do not receive benefits) correlated with the amount of spending overall, or the amount of spending that goes to minority beneficiaries.

All of the research that I've found through some cursory Google searches seems to lack the relationships to make this argument, either pro or con, but I'm hoping that some social scientist or macro economist has already looked into this. If that's true, I'd like links to that research.

If no one has done this research yet, what's the best way to do it? What datasets might contain this information? I recognize that there are tons of likely confounders — what should I be aware of if I try to find the data myself? How would you approach creating a research question for this?

Or is this something that, like many broad policy questions, just isn't answerable in a meaningful way from the information we have available?

Again, ultimately in my anecdotal experience, programs aimed at benefiting minorities and women end up also making my life better as a straight white dude. I'd like to be able to undermine the us-against-them narratives with solid quantitative evidence, but if it doesn't exist or the evidence contradicts my assumptions, I'd like to know that too.
posted by klangklangston to Science & Nature (9 answers total) 11 users marked this as a favorite
 
One serious problem you're going to have is that not every household is going to receive money from the same programs - they might have SNAP but not Section 8, or whatever, and how are you going to account for that? Is an average or a median benefit going to tell you much? SNAP beneficiaries spend their money on food, so if you own a grocery store that's probably going to be good for you, but Section 8 is going to have a greater impact on landlords. So you'd have to account for how many grocery stores are in that county vs. how many landlords live there, and also know the race of the store owners/landlords (probably impossible).

Perhaps a bigger problem is that you're going to have a disproportionate impact based on where the money the recipients spend is actually going. There's no reason to think that it necessarily stays in the county. You're not benefiting at all from your TANF recipient neighbor who orders diapers online. Maybe a tiny bit if you're a delivery driver, but again that's nearly impossible to tease out - if your neighbor was NOT a TANF recipient, would they order diapers online or not?

County level data is going to be way too broad anyway. We have $3 million dollar houses and decrepit $30k houses in my county. Perhaps census tract would be more accurate, but then the previous paragraph makes analysis impossible.

Sorry to rain on your parade and please correct me if I'm misunderstanding your hypothesis entirely.
posted by AFABulous at 12:26 PM on November 23, 2016


You'd have to define the question more precisely and even then it would be difficult to answer. But a related issue is the effect of inequality on broad measures of economic growth and health.

For decades, many researchers have found that higher levels of inequality result in worse population-wide health.

It's clear that too much equality reduces growth. For example, overly high tax rates reduce the incentive to work (though the debate about where "overly high" begins is, shall we say, vigorous). More recently, however, some economists have been concerned that too much inequality can reduce economic growth.
posted by Mr.Know-it-some at 12:50 PM on November 23, 2016


With the exception of a few childbearing-related programs like WIC, I'm not sure which programs are "targeted" at women and POC, to the exclusion of white men (historically, social welfare has excluded POC; see, e.g., the New Deal). Women and POC might be more likely to use a program, due to systemic differences in income level, but I'm not sure that translates to "targeting" those demographic groups, since a white man who found himself in similar economic circumstances wouldn't be turned away due to his race or gender.

But, let's focus on WIC, since at least part of the program is targeted to pregnant and breastfeeding women (in a perfect world, it would also extend to trans men who get pregnant, but I'm not sure whether that's the case in any state currently). The CBPP has a bunch of stats on the effects of WIC. Effects for non-program-participants include:
-Within months of implementation, WIC-approved convenience and grocery stores in Connecticut, especially those in low-income areas, offered more and a wider variety of healthy foods, especially whole-grain products.

-While supermarkets and larger grocery stores were likely to carry a wide range of healthy foods prior to the policy change, a natural experiment in New Orleans found large and significant increases in the percentage of small stores that carried nutritious foods, such as whole-wheat bread and brown rice.

-The availability of fruits, vegetables, and whole grains in small stores across Colorado, New Hampshire, Pennsylvania, and Wisconsin increased significantly. The availability of low-fat (1 percent) milk increased in New Hampshire and Wisconsin, which did not allow participants to purchase reduced-fat (2 percent) milk.

-Connecting women and children to the health care system may increase short-term costs associated with the prevention, diagnosis, and treatment of disease. But underutilization of health care in early childhood can lead to more health problems — and costs — when children go undiagnosed and untreated. And if participation in WIC contributes to better birth outcomes and healthier babies, as the evidence reviewed here suggests it does, then WIC has the potential to reduce costs associated with hospitalization and post-natal care.

-In addition, to the extent that WIC increases total food expenditures, WIC benefits the country’s farmers. USDA estimates that farmers received almost $1.3 billion from the sale of commodities used in producing the $4.6 billion in WIC retail food sales (after rebates) in fiscal year 2008. This amounts to a net addition of $331 million to farm revenues after accounting for the food purchases participants would have made without WIC.

-In the mid-1990s, the General Accounting Office conducted a comprehensive meta-analysis of 17 studies that examined the impact of WIC on Medicaid costs. It concluded that prenatal WIC participation reduced post-partum medical costs in the first year by more than enough to offset the entire cost of the prenatal WIC program. Research conducted for USDA similarly concluded that every dollar spent on the prenatal component of WIC was associated with Medicaid savings during the first 60 days after birth, ranging from $1.77 to $3.13.

posted by melissasaurus at 1:33 PM on November 23, 2016 [2 favorites]


Just bouncing in for a sec to clarify and respond to a couple things:

"County level data is going to be way too broad anyway. We have $3 million dollar houses and decrepit $30k houses in my county. Perhaps census tract would be more accurate, but then the previous paragraph makes analysis impossible.

Sorry to rain on your parade and please correct me if I'm misunderstanding your hypothesis entirely.
"

Somewhat, in that you're both looking for a granularity that I'm not tremendously concerned about off the bat (i.e. it doesn't matter to me what the specific programs are so much — figuring out whether SNAP or Section 8 had a bigger impact would be a question after seeing whether there's an impact at all), and seem to be treating benefits as microeconomic questions — for the example of the neighbor buying diapers online, if they would have otherwise paid for those same online diapers out of their own pocket, giving a subsidy to the purchase means that they will have more money to spend on other things, that the parent will be less stressed, etc. Whether that would have a demonstrable economic effect for e.g. white people that could be correlated to the amount of spending on non-white people.

"You'd have to define the question more precisely and even then it would be difficult to answer. But a related issue is the effect of inequality on broad measures of economic growth and health. "

Yeah, that's what I'm trying to get a handle on — defining the question so that it is something that could be measured, and that there is likely to be data on — and why I'm hoping some other people already have, to give me a jumping off point.

"With the exception of a few childbearing-related programs like WIC, I'm not sure which programs are "targeted" at women and POC, to the exclusion of white men (historically, social welfare has excluded POC; see, e.g., the New Deal)."

First, there are tons of programs that are specifically targeted at women and minorities, from federally-funded scholarships to business loans and contractor preferences. But I was also asking about things that would be covered by symbolic racism complaints about government spending, e.g. TANF, Title 1 spending, SSI, etc., because I know that white people benefit from them too, even though some of them are or are popularly perceived to be disproportionately benefiting minorities.
posted by klangklangston at 3:20 PM on November 23, 2016


A bigger issue than finding data, for pretty much any kind of statistical approach, is that this is almost an archetypal example of the identification problem.

Basically, the process that determines level of welfare spending in a region is (obviously!) not independent from the level of economic well-being (prior to said spending), however measured: Lower income areas will tend to have higher levels of welfare spending, by definition. This leads you to a chicken-and-egg problem: You want to find out if welfare spending raises future economic wellbeing, but welfare spending is influenced by past economic wellbeing, which is obviously a major determinant of future wellbeing. So if there's a correlation, is that correlation going from welfare spending to the future, or from the past to both of them?

If you frame your question very very carefully or you luck into finding a dataset where there *is* random variance among welfare spending that is guaranteed to be roughly independent of economic wellbeing (say state by state acceptance of the Medicaid expansion), then you can attack this. Or get a PhD in econometrics and learn about difference-in-differences and such things.
posted by PMdixon at 5:22 PM on November 23, 2016 [1 favorite]


"Basically, the process that determines level of welfare spending in a region is (obviously!) not independent from the level of economic well-being (prior to said spending), however measured: Lower income areas will tend to have higher levels of welfare spending, by definition. This leads you to a chicken-and-egg problem: You want to find out if welfare spending raises future economic wellbeing, but welfare spending is influenced by past economic wellbeing, which is obviously a major determinant of future wellbeing. So if there's a correlation, is that correlation going from welfare spending to the future, or from the past to both of them?"

The two things that I was hoping would help me differentiate are relative proportion of minority recipients (i.e. do white people do better overall at a similar level of spending if there are more targeted programs for minorities?) and abrupt changes in program funding (here in California, the drastic cuts and restoration of funding related to the recession seemed like it might give some opportunities to isolate things.

But I do appreciate that answer for giving me something else to think about when trying to construct a specific question.
posted by klangklangston at 9:58 PM on November 23, 2016


If you're not a social scientist or economist yourself with a lot of free time and energy to properly delve into this very complex issue than I think you're on the right track if you look for others who have already done this work.

Take a look at Scott Page's work. He's an economist (among other things) who has written about the benefits (and challenges) of diversity. He's focused primarily on problem solving and the kinds of problems that are more easily or better solved by diverse groups but he approaches this from an econometrics angle so I bet that you'd find references to the kind of work for which you're looking. And it's really interesting stuff in its own right anyway.
posted by ElKevbo at 8:04 PM on November 24, 2016


I'll delve further into Page's work — I'm familiar with him after coming across his map-agent game theory arguments for diversity in cognition providing synergistic benefits. I wonder if I should just try writing him to see if he has any recommendations.
posted by klangklangston at 1:20 PM on November 25, 2016


It's not exactly what you're looking for but Jones (2014) looks at something similar(ish). The paper looks at who benefits from EITC by matching CPS data with tax records during the Great Recession. It might be a useful jumping off point and gives an idea of what data is available. Robert Moffitt at Johns Hopkins also does related work which might help you refine your question or empirical strategy. He has a paper which describes the state of the social safety net in the US during and after the Great Recession.

Another approach, although more of a back of the envelope type calculation, would be to find relevant estimates for the fiscal multiplier related to social safety net expenditures and you could at least try to ballpark the increase in GDP that comes from the specific program. This certainly isn't a rigorous approach -- and there's lot of disagreement on the sign and magnitude of the multiplier -- but it's a start. A few relevant papers might include: the local fiscal multiplier estimates given by Suarez and Wingender (2014); Monacelli et al (2010)'s estimates for fiscal spending on the labor market; or Wilson's literature review of the multiplier literature at the San Francisco Fed.
posted by An Economist at 2:04 PM on November 25, 2016 [1 favorite]


« Older Non-FPS Xbox (or PS?) game suggestions for 16...   |   How contagious is shingles? Newer »
This thread is closed to new comments.