Jonathan Barazzutti

There is a line of implicit thought I often see within policy discussions, particularly amongst those who are more sympathetic to redistributive policies. It is represented by a desire to institute as many different welfare programs as possible, always examining how we can add something, without considering if anything can be removed or optimized in some way. Such advocates wish simultaneously for a minimum wage, universal basic income, food stamps, and the list goes on.

But certainly these programs conflict with one another to some degree. Firstly, to whatever degree any of these particular policies work in terms of improving social welfare, the amplification of one reduces the necessity for others. If one’s concern is, for instance, reducing the share of individuals who have annual incomes 40% of the median income within a country, the implementation or amplification of a particular program reduces the necessity for other programs.

Furthermore, programs necessarily come fiscally at the expense of one another. Every dollar one spends on universal basic income, or on enforcing minimum wage legislation, is one dollar not spent on food stamps. In fact, because every program requires to some degree or another the creation of new infrastructure to administer social benefits or enforce regulations, the same amount of money spent on several programs will oftentimes be significantly less efficient than money spent on a single.

Part of this is likely a product of how politics works with regards to spending. Due in part to anchoring bias, citizens are naturally predisposed to increases in spending rather than cuts to spending, meaning that it is naturally easier to implement a new program compared to removing it. This means that spending will inevitably increase over time in the absence of any hard pushes, which will naturally create inefficiencies within a system as bloat expands with the spending.

There are numerous studies which suggest that increased social spending leads to reductions in poverty and wealth inequality. This fact is clear. The question is what policy arrangement most effectively achieves society’s goals in that domain, and how much spending should be adopted in pursuit of such policies.

Some research currently exists which attempts to compare various policies in terms of their effectiveness in reducing poverty. For example, there is substantial literature comparing the efficacy of an Earned Income Tax Credit (EITC) compared to a minimum wage. Minimum wages, due to their poor targeting, tend to perform poorer compared to EITCs. Many minimum wage workers are not at or below the poverty line, and so having a boost to those people’s incomes, while progressive, is not the most efficient targeting. Meanwhile, the EITC specifically targets individuals on the basis of their income, directly targeting the basis upon which poverty and/or inequality exists, which is income differentials. Furthermore, the EITC has positive effects on labor force participation amongst its recipients, while minimum wage increases generally have the opposite effect.

Similarly, there exists research examining taxation, and how much taxation is too much. The Laffer curve is a hypothesized inverse-u shaped relation between the tax rate on income and the revenue generated from such a tax. Comparing a 0% tax rate to a 10% tax rate, the 10% tax rate will certainly generate more revenue than the 0% rate, because there is at least some revenue being collected. But as the tax rate increases, the number of market distortions which produce deadweight losses increases. Eventually, an increase in the tax rate will start to decrease tax revenue because its effect on total income to be taxes will be greater than its effect on generating further revenue from income that exists. At a 100% tax rate, tax revenue will similarly be zero, at least in the long run, as the economy would be destroyed. Empirically, the question arises as to when these conflicting effects equal out and thus create a peak tax rate at which tax revenue will be maximized. There have been attempts to empirically measure this across a number of countries.

The Laffer Curve. Source

But a particularly fascinating paper I discovered on social spending in particular examines the marginal value of public funds (MVPF) of various policies. The paper, written by economists Nathaniel Hendren and Ben Sprung-Keyser, is entitled “A Unified Welfare Analysis of Government Policies”. A corresponding website was made (policyinsights.org) to display the data in a more accessible fashion.

The MVPF of a particular policy is calculated by dividing the beneficiaries’ willingness to pay for the policy by the net cost to the government. A negative MVPF implies that the program has a net cost to the government and that it leaves beneficiaries on net worse than otherwise; A MVPF of zero implies that the program has no benefit to the beneficiaries while having a net cost to the government; A MVPF between zero and one implies that the beneficiaries’ willingness to pay is less than the cost to the government; A MVPF above one implies that the beneficiaries’ willingness to pay is greater than the cost to the government (meaning the program helps the beneficiaries more than it costs to the government); and finally a MVPF of infinity implies that there is a positive effect on beneficiaries with zero net cost to the government.

Certainly, the most desirable policy choice is one in which there is an infinite MVPF, as it means that there is in effect a free benefit being attained. And Hendren and Keyser’s study actually did find that there are several general policy areas where the MVPFs tend to be infinite. These are policies which target low-income children throughout childhood, including with childhood education, child health insurance expansions, and college policies. Such policies appear to generate revenue for the government in the long run.

I was initially a bit surprised about college policies in particular having an infinite MVPF. I generally subscribe to a signaling view of post-secondary education in which studying for a degree doesn’t give the average person significant workplace-relevant knowledge but rather signals to employers certain traits such as cognitive ability and conformity, which employers value. I may write an article in the future going over the implications of this idea and my justifications for believing in it. However, even if post-secondary educational bloat may create certain market inefficiencies, within the existence of the market inefficiency allowing more people to get into postsecondary could at least in the short run improve economic output. This is due to it giving those people more opportunities to provide value within workplaces. Hence, under this view, college policies nonetheless can have infinite MVPFs because of their downstream effects on income which generates tax revenue.

These policies overall are areas in which it is absolutely in society’s best interest to fund. Not only are they distributively equitable, appealing to the desires of many to reduce poverty and wealth inequality, but they also actively pay for themselves and then some. A policy which costs $10 and makes back $15, has sufficient money to pay back the initial revenue it took from people while also keeping extra. Furthermore, the downstream effects of these policies limit the necessity for other programs which have lower MVPFs, which generally target those at an older age. Hence, they can help reduce concerns of administrative bloat that often come with the implementation of social programs.

There are many other policy categories which generally had positive MVPFs, but they didn’t pay for themselves. In these cases, tradeoffs between different parties in society necessarily have to be made when deciding whether they should be implemented. To discuss the costs and benefits of these types of programs goes beyond the scope of this article.

Overall, Hendren and Keyser’s article contributes massively to addressing the question of how to optimize social services. Further research that could provide some more information should probably examine the marginal effects of per-person spending on the MVPF of a policy. Certainly, there would come a point at which spending would no longer continue to pay for itself, at which point further spending may not be regarded as valuable. This may not necessarily be completely feasible in the same way it has been done with the Laffer curve, but attempting to uncover pieces of information which at least may help to answer the question would massively help to advance our knowledge surrounding welfare economics: How we optimize our social services, which programs work and which don’t, and how much social spending is too much.