Intersecting Disincentives

From WSJ:

Congress created the broadband subsidy with other Covid handouts in the December 2020 omnibus bill...

The 2021 infrastructure bill provided $14.2 billion to extend the $30 a month broadband discounts, and these funds are expected to run out this month….

Some 23 million households receive broadband subsidies, which is more than the number on food stamps. Households qualify if they earn 200% or less than the poverty line or participate in other welfare programs such as Medicaid. This includes relatively affluent households with public-school students in localities like New York City that provide universal free school meals.

Context:

Intersecting disincentives are a central problem of our means-tested (pretty much all) social programs. Most of them have, in isolation, medium disincentives. “Affordable housing” programs, for example, often require one to pay 30% of income, meaning an additional 30% marginal tax rate. But most people sign up for more than one means-tested program. With just three such programs, the face an effective 90% tax rate. (The tax rates add directly, because social program phaseouts are each based on total income, not adjusting for the loss of other benefits. They can easily exceed 100%.) That is, of course, on top of actual state and federal taxes, income-based phaseouts of the earned income tax credit and child tax credit, sales taxes, and so on. And many programs have cliffs, where you lose everything if you earn another dollar.

Roughly speaking, and including in-kind transfers (health insurance, housing), the average American earning from $0 to $60,000 faces a 100% marginal tax rate. Equivalently, we have a $60,000 universal basic income, with a 100% tax of actual income on top of that.

That claim is of course contentious. I take it from Gramm, Ekelund and Early, and Casey Mulligan. But there are hundreds of studies which come to different answers. There are hundreds of different programs, but most people do not sign up for more than two or three, and most studies only assess the impact of two or three. The tax rate varies a lot depending who you are and which you sign up for. Gramm et al. also include the value of in-kind transfers, where most studies only include cash transfers.

How many such programs are there? A lot. A small list from CBO: Medicaid, Low-income subsidy (LIS) for Part D of Medicare, earned income tax credit, child tax credit, Supplemental Security Income (SSI), Temporary Assistance for Needy Families (TANF), Supplemental Nutrition Assistance Program (SNAP, Food Stamps), Child nutrition programs, Housing assistance programs, Pell Grant Program.

To which I can add off the top, income-linked phaseouts of Obamacare subsidies, income-linked student loan repayment, income-limited education vouchers, unemployment insurance (duh), social security disability (work and lose it), and “affordable” housing: The government forces developers to build them, and then allocates apartments at “below market” rates to people below an income limit.

From HHS comes the following (still partial) list, based on who uses their definition of “poverty” for income limits. In addition to the above, and including both the enormous and the ridiculous: Children's Health Insurance Program, Consolidated Health Centers (CHCs), including Federally Qualified Health Centers (FQHCs), Maternal and child health services, Title X Family Planning Program, Older Americans Act Nutrition Program, Head Start, Health professions student loans and scholarships, Low-income Home Energy Assistance, National School Lunch Program, Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), Child and Adult Care Food Program, School Breakfast Program, Summer Food Service Program, Commodity Supplemental Food Program, Food Distribution Program on Indian Reservations (FDPIR), Senior Farmers' Market Nutrition Program (SFMNP), Special Milk Program for Children, TRIO Programs, Educational stipends for the Native Hawaiian Career and Technical Education Program (NHCTEP), Educational stipends for the Native American Career and Technical Education Program (NHCTEP), D.C. School Choice Incentives, Federal Student Aid-Income-Driven Plans, Weatherization Assistance Program, U.S. Citizenship and Immigration Services, Immigration Form Fee Waiver (Form I-912), Job Corps Workforce Innovation and Opportunity Act (WIOA), Senior Community Service Employment Program (SCSEP), Health Insurance Premium Tax Credits, Low-Income Taxpayer Clinics (LITC), Fee waiver for Offer in Compromise, Reduced user fee for Installment Agreements, Corporation for National and Community Service, Foster Grandparent Program, Senior Companion Program, LifeLine, Legal Services. Plus a few that do not use the HHS poverty definition, State/local-funded General Assistance (in most cases), Section 8 low-income housing assistance, Low-rent public housing.

And it continues, right down to the SF Bay Area low income bus pass, 50% off up to 200% of the poverty line, and half off express lanes and toll debt forgiveness also up to 200% of the poverty line (HHS missed those).

And, finally to get to the point, $30 a month off broadband up to 200% of the poverty line (also missing from the HHS list).

Of course not many people are going to turn down a job or a promotion over losing $30 a month. But it adds up. The increasing number of programs that eliminate all benefits at 200%, with no phase-out, create a larger cliff out of small programs.

*****

Each program might be small, but the disincentive comes from the combined effect of all programs taken together.

For now, most people only sign up for a few programs, and many programs only attract a few percent of eligible people. Heaven help them, and the budget, if they start signing up for more.

Signing up is a pain in the butt. I had to research the list of programs. The average beneficiary is going to have trouble even finding out about many programs. I think that’s a rough and ready rationing mechanism to control the disincentives of signing up for too many programs, but we can do a lot better than rough and ready, both in helping people who need it and in controlling disincentives.

Implied taxes add directly. If you pay a 50% income tax, and then a 10% sales tax, your total tax rate is only 55%, not 60%, because the 10% only applies on what you have left over after they take 50%. If you have one program that phases out 50% of your income, and another that phases out 10% of your income, your marginal tax rate is 60%.

Income phaseouts apply to market income, not program income. If you earn an extra legal dollar, you lose x% of your benefits. If you join another program (low income bus pass) the extra income does not count against your other benefits, so there is zero marginal tax rate. Getting a subsidized bigger apartment is just free. Getting a better paying job and paying for that apartment pays taxes and reduces other benefits.

I am led to the view that the government needs to integrate all means-tested programs in one place. At least, then, we can know the actual marginal tax rate that each person faces and redesign programs to control the combined disincentive. We might also then control the disincentives. How about, “Here is the menu of programs available to you. Choose which ones you want, but the calculator will keep track of the all in marginal tax rate and make sure it’s below 50%.” Or perhaps it can offer a new personalized subsidy/tax credit to smooth the disincentives. Given all the programs you’re in, you get something like the EITC with a phase-out that makes sure your all-in marginal tax rate is below 50%. I once suggested a similar “alternative maximum tax” on the high end: since those disincentives also depend on all taxes together, nobody has to pay more than (say) 70% all-in tax rate. The same principle applies. Or work requirements can be tailored to the all-in marginal tax rate.

****

Which brings me finally to the quotation. We are starting to integrate programs, but in a way that seems to multiply rather than control the all-in disincentive.

Households qualify if they earn 200% or less than the poverty line or participate in other welfare programs such as Medicaid…or free school meals.

Starting at least with food stamps in the Obama Administration, there has been a concerted effort to sign people up. The political and bureaucratic forces that like these programs naturally want more people in them, and view two percent sign up rates as a human tragedy.

My “one stop shopping” vision, run by them (naturally) would be an incentives and budget disaster: Walk in (or visit the website) and the AI helper will instantly sign you up for every single program you’re eligible for, right down to the low-income bus pass. The marginal tax rate would be astronomical, and the same AI helper will easily tell you “if I report $100 more income what do I get?” Don’t do it.

You see in the quote a major step in that direction. Rather than have each program’s bureaucracy independently verify income, people automatically cross-qualify once they are accepted into one large program, such as Medicaid or free school meals. Obviously, the incentive to fudge one application rises dramatically. The people checking your income for free school lunches (who really cares if a few extra people get free school lunches?) now control huge sums, and huge disincentives.

Income is not an innate immutable characteristic. Incentives matter. The main problem with our social programs is not cost, but intersecting disincentives.

****

An example: