Kevin Martin · Wages and Working Families
The Benefit Cliff: Earn One More Dollar, Lose Three Hundred
Kevin Martin on why federal benefit programs can punish a raise, and how to fix the math so working more always leaves a family better off.
There is a problem in how federal support programs are built that I do not think most people know about, and once you see it you cannot unsee it.
It is called a benefit cliff.
Here is how it works. A family qualifies for help — childcare assistance, food assistance, a health subsidy — based on income. There is a threshold. Earn a dollar under it, you qualify. Earn a dollar over it, you do not.
So, somebody takes a raise, or an extra shift, and crosses the line. The raise is fifty dollars a month. The benefit they just lost was worth three hundred.
They did the thing we tell people to do and they came out behind. The motivation to work harder, smarter, to improve oneself is lost.
Why this bothers me specifically
Because it is a design flaw, not a philosophical dispute, and it is the kind of thing that would get caught immediately in any organization that modeled its own incentives.
If I built a commission structure where a salesperson made less money for selling more, I would not blame the salesperson for stopping at the threshold. I would fix the structure. I would also expect to answer for having built it.
We have built exactly that, across multiple programs, and then we hold conversations about whether people want to work.
Basic math is not lost on those who have to count every dime and penny needed to survive from paycheck to paycheck. They can do arithmetic. That is the whole explanation.
The Federal Reserve Bank of Atlanta's Policy Rules Database called PRD is the authority on the source for modeling Georgia families' benefit cliff across several specific programs as related to SNAP, Medicaid, CCDF, Section 8, TANF, WIC, SSI, SSDI, EITC, and CTC. And by defining the family type, it can simulate benefit changes as income rises and identify the cliff and magnitude. Families can see this drop, but it is usually after the fact and causes a negative response on trying to escape the government "free money". (source - Federal Reserve Bank of Atlanta and www.thepovertysolution.com)
What Congress actually controls
Congress writes the eligibility rules for federal benefit programs and the phase-out structure inside them. Whether a benefit ends at a threshold or tapers gradually is a drafting choice.
Congress also decides whether programs coordinate. A large part of the cliff problem is that several programs use different thresholds set at different times for different reasons, and a family can cross two or three at once without any of them knowing about the others.
Nobody designed that. It accumulated. Which is worse in a way, because there is no argument to win — just work nobody has done and it appears that nobody wants to take ownership and correct.
What needs to change
Taper instead of cutting off. Benefits should phase out gradually enough that another dollar earned always leaves a household better off. This is a technical fix with a real cost and it is worth paying.
Coordinate the thresholds. If several programs phase out, they should not all do it in the same narrow income band. These programs and thresholds should converge on one point: an income level at which the benefits being received total a specific dollar amount (which should be above the poverty level). That is the point where benefits should stop, without punishing a family for making $1 above the "cliff point."
Aim support at people who are working. Support tied to work and phased out slowly rewards the behavior we say we want. That is a conservative principle, and it is also just good design.
Model it before passing it. Every eligibility change should come with a chart showing effective marginal rates across the income range. If the chart has a cliff in it, fix the bill. Ask the question, WHY. Why is there a cliff and make corrective actions. Follow up and audit the programs and benefits and verify with actual spending costs data of families under these programs.
Why this is the same argument I keep making
Because it is about knowing the number.
Nobody intended to build a system that punishes a raise. It happened because programs were written separately, thresholds were set at different times, and nobody put the whole picture on one page and looked at it.
That is what I do. Put the whole picture on one page and look at it, and then say the uncomfortable thing about what it shows. Stop using "what if scenarios" and use actual data and results. Benefits should be based on normal accepted 40-hour work week and should not include overtime or earned bonuses. Simply put, base wage.
A person who takes an extra shift should end up with more money. If our system does not do that, it is our system that is wrong, not the person who noticed.