Which policies improve social well-being the most?

We examine 133 historical policy changes in the U.S. over the past half century. We analyze policies in 4 categories:

1. social insurance (e.g. health and disability insurance)
2. education and job training (e.g. preschool spending and college subsidies)
3. taxes and cash transfers (e.g. top tax rate changes or expansions of the earned income tax credit)
4. in-kind transfers (e.g. housing vouchers and food stamps)

We measure the benefit that each policy provides to its recipients compared to how much it costs the government. We call this number the policy’s Marginal Value of Public Funds, or its “MVPF.”

Here we plot the MVPFs of each of these policies by the year they were enacted. For example, the Massachusetts Adams Scholarship, studied in Cohodes and Goodman (2014) has an MVPF of 0.72. This means that $1 of net spending on this scholarship leads to $0.72 in benefits to the recipients.

Some policies yield very large benefits per dollar of government cost. We estimate that in certain cases these policies pay for themselves in the long run. Such policies have infinite MVPFs, which we plot along the top horizontal line in the graph.

How are these numbers calculated? Click below to learn more. Or, continue to scroll to see our main results.

Learn how an MVPF is calculated

We can learn more about these high-return policies by plotting each policy’s MVPF by the age of its beneficiaries. The cluster in the upper left helps to illustrate a key finding: direct investments in low-income children have historically had the highest MVPFs.

It is important to note that there is considerable statistical uncertainty contained in these estimates. Here we report 95% confidence intervals, which characterize the margin of error in each of these estimates.

To reduce this uncertainty, and to better understand the broad patterns in our estimates, we group our MVPFs by policy category. Here we report average MVPFs for each category and their 95% confidence intervals.

We find high MVPFs for policies targeting low-income children throughout childhood, including child health insurance expansions and many college policies. This suggests that the returns from investment in children do not appear to decline with age.

To see this more clearly, we change the vertical axis to report the net cost to the government for every dollar of upfront spending. In the case of four major Medicaid expansions to children studied in previous literature, we estimate that each $1 of initial spending was fully repaid and that the policy returned an additional $0.78 to the government.

The story is different for policies targeting adults: On average, these policies have not paid for themselves.

For policies targeting adults, we generally find MVPFs around 0.5-2. This does not mean these expenditures are undesirable. Rather, these policies involve budgetary tradeoffs. The MVPF helps measure these tradeoffs.

There are important caveats to the general patterns we find. Not all policies targeting youth have high MVPFs. For example, many job training programs such as Job Corps have much lower MVPFs. The same is true for the provision of Supplemental Security Income (SSI) to disabled children and their families. Importantly, this finding does not suggest that these policies are undesirable, rather, that they involve budgetary tradeoffs.

Lastly, we find that expenditures on adults can have high MVPFs if they produce positive spillovers on children. For example, the Moving to Opportunity Experiment, which improved family neighborhood environments, has an infinite MVPF.

Explore the data yourself to learn more about the welfare impacts of historical policy changes in the last half-century.