Government Assistance and Backwards Incentives

The other day Stefanie, the author of a cool personal finance blog set in New York City titled "The Broke and Beautiful Life", wrote a post on unemployment and part time work. I loved the post because it was a detailed and transparent look at one woman's experience with government benefits. The part I found most interesting is how the system reduced her benefits ($251 a week, before taxes) by $63 in one week, as a result of earning just $40 after selling an article. In effect, by working, she actually reduced her income. Had she sold another article on another day that week, regardless of the amount paid for the work, her benefits would be reduced by another 25%. At that rate, the more she worked, the less she earned. This is a case of improper incentives: for every dollar Stefanie earned in this scenario, she lost more than a dollar in unemployment benefits. Rather than providing an economic incentive to work, the system is, in this case, ironically punishing the very behavior it wishes to encourage. So, unless employment with an income that significantly exceeds the weekly benefit becomes available (or if there is some other long term payoff, like a significant raise or health benefits), it makes little financial sense to accept the offer of work. It simply results in a lower income.


Before We Start, a Caveat:
The goal of this post is not to discuss politics, but rather, to discuss incentives. I'm not all that interested in the philosophical or moral arguments that typically surround discussions of social programs like unemployment, welfare and the like. It's all been said before. But I am interested in the idea of systemic incentives, and whether they are effective at driving beneficial behavior. Namely, how do you provide benefits to unemployed or underemployed citizens in a manner that effectively increases the desired behavior (i.e. - full, or fuller, employment)?

Do Incentives Even Work?
This is a thorny issue. Logically, incentives should have some impact on our behavior. Raise prices or provide discounts and spending behavior changes. Why can't the same logic apply to employment? Still, some people believe additional incentives are superfluous, since making more money is its own incentive: you don't need to convince anyone of that. Nearly all citizens would love to make more money, and the issue is not one of proper motivation. (The root cause is something else, like education, or macroeconomic conditions, social pressures, etc. etc.)

The problem with this issue (and with just about every other social issue that I find interesting) is that the discussion usually stays pretty academic. We exchange competing anecdotes and worldviews, but real data is hard to find because government entities are, for some reason, not keen on using their poorest citizens in social experiments.

An Experiment
But wouldn't it be great if some country, say, I don't know, Canada, did actually perform a randomized experiment using actual recipients of government assistance? And then published the results on the internet for free so that a frugal blogger could steal the information? Wouldn't that be neat?

Ask and ye shall receive. Let's go back to a simpler time. Let's go back to the days when I could wear baggy flannel shirts and torn jeans and part my wavy shoulder length hair right down the middle and somehow could think that I looked flipping awesome. Let's go back to 1996, and look at a study titled, "Do Financial Incentives Encourage Welfare Recipients to Work: Evidence from a Randomized Evaluation of the Self Sufficiency Project." (Another caveat: this experiment looks at welfare recipients, which is very different from those who receive unemployment benefits. Still, the programs are similar one respect: one of the stated goals of both programs is for recipients to find employment and, over time, no longer require benefits. So my interest is simply in seeing whether financial incentives are successful in achieving that goal.)

"ABSTRACT: This paper reports on a randomized evaluation of an earnings subsidy offered to long-term welfare recipients in Canada. The program -- known as the Self-Sufficiency Project (SSP) -- provides a supplement equal to one-half of the difference between a target earnings level and a participant’s actual earnings. The SSP supplement is similar to a negative income tax with two important differences: (1) eligibility is limited to long-term welfare recipients who find a full-time job; and (2) the payment depends on individual earnings rather than family income. Our evaluation is based on a classical randomized design: one half of a group of single parents who had been on welfare for over a year were eligible to receive the SSP supplement, while the other half were assigned to a control group. Results for an early cohort of SSP participants and controls suggest that the financial incentives of the Self-Sufficiency Program increase labor market attachment and reduce welfare participation."

Hooray for social experiments! Here is a quick rundown of the paper. The Self-Sufficiency Program was a new approach to welfare, in which an economic incentive (one-half of the difference between a single parent's actual earnings and a target earning level) was provided to long term welfare recipients who find employment for at least 30 hours per week. This was in contrast to the existing Income Assistance (IA) program, which reduced benefits dollar-for-dollar when income was earned, and provided "a strong disincentive to work for many IA recipients". (In contrast, the SSP only reduced the benefits by $.50 for each $1.00 in earnings.) Researchers followed the recipients in the program for 18-24 months.

So, what happened? Here are some quotes that illustrate their findings:

"The patterns for the control group in Figures 5 to 8 reveal a more-or-less steady trend toward greater labor market activity in the months following baseline. This trend is clearly accelerated among the program group. The earnings impacts (Figure 5) are statistically significant from months 5 through 17 and increase gradually from about $40 in month 5 to a peak of about $140 in month 14. This is a sizeable impact (roughly 60 percent) relative to mean earnings of the control group in the same month...."

"Hours of work follow a pattern similar to earnings. The hours impacts (Figure 6) are statistically significant in months 5 through 17, and rise gradually from around 6 hours per month in month 5 to 21 hours in month 14. Again, this is a sizeable impact (about a 70 percent increase) compared to the mean hours of the control group...."

"The overall employment rate (Figure 7) follows a pattern similar to earnings and hours. The employment impacts are statistically significant from months 6 through 17 and rise gradually to a peak impact of 14 percentage points in month 14 before falling slightly. As shown in Figures 8 and 9, almost all of this impact is accounted for by a rise in full-time employment and a drop in non-employment." (Quotes are from pages 18-19 of the study.)

The incentives from the Self-Sufficiency Program resulted in greater labor market activity, more hours worked, and higher overall employment rate when compared to the baseline group in Income Assistance. Roughly twice the number of participants left welfare and found full time jobs when compared to the group in IA. Simply put, the incentives drove the desired behavior. However, the conclusions are not entirely rosy. It appears that those who left welfare took low wage jobs, roughly $1.00 to $3.00 above minimum wage. Unless wages increase over time for these recipients, the authors conclude that many of these recipients may go back to welfare if and when the incentives end.

Conclusions:
The thing that I take away from this study, and from Stefanie's story about unemployment benefits, is that people react well to proper financial incentives. While financial incentives are no silver bullet and do not apparently solve the issue of low wages, they can be effective in driving desired behavior. Additionally, the traditional model of taking away $1.00 in benefits (or, shockingly, more in Stefanie's case) for $1.00 of earned income is a misguided and sub-optimal approach. The desired outcome of increased employment is better achieved if some benefits remain, even when income is achieved, providing a clear financial incentive to work more hours and/or earn more dollars.

Thanks, as always, for reading.

Photo of carrots is from www.metamophoricalplatypus.com at Flickr Creative Commons.