A few years ago, Human Resources Development Canada (HDRC) ran an experimental project to see how single parents on welfare responded to changes in their budget constraint. It has long been known that single mothers on social assistance are particularly vulnerable to the welfare trap: not only are their payments clawed back as they earn wage income, they risk losing their non-monetary benefits. In many cases, these parents face marginal income tax rates of well over 100%.
So the HDRC came up with the Self-Sufficiency Project (the SSP), which provided income supplements to single parents (almost without exception, these were women) who, after having received social assistance for at least a year, found full-time work. These premiums were quite generous: a participant could receive up to $12,000 a year, thus doubling what she might otherwise have earned at a job that paid minimum wage.
Here's the graph of the employment status (source) of the randomly chosen group of single parents who were offered the chance to participate in the SSP, and of the employment status of the control group:
During the first 12 months of the program, those who were eligible for the SSP were less likely to find a job: since the SSP premiums were only available to those who had been on social assistance for a year, some of those who might have found a job within the first year decided to wait for the premium. Once they were eligible, the change in behaviour is dramatic: during the second year, some 25% of the SSP group found a job, compared with 8% of the control group.
But what's curious is the employment patterns after that. In the SSP group, the employment rate levels off, but the control group's rate continues to rise. After the fifth year of the program, there's little to choose between the two employment rates. So the long-run effect of the SSP program appears to be pretty small: all it really ended up doing was to move forward the decision to look for a job among those who would most likely have gone back to work anyway.
From a policy standpoint, the SSP experiment was a bit of a disappointment ('oh well, back to the drawing board'). But it's a neat illustration of how people respond to incentives.
[I'm taking a break from blogging over the holidays, and recycling
some earlier posts. This one was first published on April 23, 2006.]
Worthwhile Canadian Initiative 
What I get out of the SSP experiment is that if you are a single parent with young children (less than school age) it’s not worth leaving social assistance for work unless
– you have lots of earning potential or
– childcare is close to free (subsidized by the government or provided by granny/grandpa/sisters)or
– employment is subsidized (e.g. the Self-sufficiency project).
After a time the effects of the subsidy go away because the kids in the control group get older and need less care, freeing the parents for employment. Because being on social assistance is really not that much fun – if people have an option of working, they often will do so, even for the most minimal rewards.
So the interesting question from a policy point of view is, which is better,
a) subsidizing employment
b) subsidizing non-parental childcare
c) subsidizing parental childcare by paying single parents to look after their children at home
I think economists found something really appealing about the subsidizing employment option. But I think the average person feels that option (b) subsidizing non-parental childcare is in some sense more ‘fair’, in that funds go to children (for care) and it fills a need while encouraging work and responsibility. Subsidies, by way of contrast, income subsidies in some sense provide cash rewards for behaviour that is in some sense non-meritorious (being on social assistance in the first place). Personally, I don’t see what’s so wrong about (c) – if I was a parent with young children on social assistance, it’s what I would want.
I was just about to post this, went to empty the dishwasher, then saw Frances had beaten me to it:
I find it very puzzling that there was so little long run effect. Maybe their kids got older and went off to school, so the control group found work anyhow?
I’m also surprised that there’s not more discussion of this problem (100% effective marginal tax rates on some of the poor), from both left and right. Left and right argue over what is a good MTR at the top end of the income distribution. They ought to be able to find common ground over LTR at the lower end of the distribution.
I didn’t really find this surprising at all – I’ve had the impression from things I’ve read that in the case of single parents, welfare tends to be a temporary situation related to childcare needs, and that once children are old enough to go to school they tend to get back into the labour force and do fairly well. The SSP program gives them enough extra to buy childcare from others and still have an acceptable level of income from working. In the absence of a subsidy like that welfare is the more workable option, in particular in the early years when childcare tends to cost more.
Frances – your option (c) I think in some sense could be simply the welfare system as it is. I assume you were thinking of something in addition?
I think there is confusion about exactly what behaviour is being modified here. It’s not holding a job, it’s looking for a job. Job hunting is not free or easy. With high incentives large efforts produce quick results. With low incentives people wait until a job lands on them.
The other implication of the graph is that only 30% of the particular population can get a job, even with large incentives.
If learning-by-doing impacts productivity and labour market value, then moving the work decision forward may have benefits.
On a related theme, how do children fare? Better with single parents work or better with single parents that do not work?
If you’d have a flat tax of 50% and replaced all income support programs (welfare, unemplyment insurance, old age pensions, tc..) with a universal, unconditional, untaxed guaranteed minimum income of around $12 000 a year for everyone (including children, but the money would go to their parents), the net effect in terms of transfers and taxes would probably be roughly equivalent to what we have now, but you’d get rid of a lot of bureaucracy and eliminate a lot of the wierd incentives and disincentives we now have.