Is Ignatieff playing with the right Employment Insurance instrument?

Apparently we're going to have yet another political psychodrama, this time over Employment Insurance (EI). Michael Ignatieff is threatening to force an election if the government doesn't go along with the Liberal proposal to make everyone eligible for EI after 360 hours of work. (I should explain to our non-Canadian readers that we've set the bar pretty low when it comes to psychodramas: the last one was about subsidies for political parties.)

There have been any number of reforms to EI (for example, it's now called Employment Insurance, and not Unemployment Insurance) since the mid-1990's, but the new regime hasn't yet had to deal with a real recession. It's a good idea to keep a close eye on how the new system deals with the current labour market conditions, but that doesn't mean we should ignore what we've learned from the past. I thank my colleague Guy Lacroix for pointing me to the research that is discussed below the fold.

Ignatieff's choice of policy instrument - the number of hours worked before becoming eligible for EI - is somewhat puzzling. The econometric evidence suggests - not surprisingly - that people respond to incentives in this area: employers and employees will game the eligibility requirements if they can. And the eligibility requirement is an easy criterion to manipulate.

The most striking evidence for this sort of behaviour is documented by David Green and Craig Ridell (Jstor):

This paper examines the impact of the Ul entrance requirement (ER) on the distribution of employment durations by studying an accidental experiment that took place in Canada's Ul system in I990. In that year, as a result of a dispute between Canada's Senate and House of Commons, the Variable ER was suspended. We focus on the hazard rate out of employment for the 'maximum entitlement regions' - those in which the ER increased from 10 to 14 weeks and no other features of the UI system changed - and identify the ER effects by comparing the 'experimental' period (most of 1990) to the equivalent part of 1989.

We find evidence of the effects of the ER change using both simple empirical hazard (Kaplan-Meier) estimates as well as duration model estimates that control for the influences of covariates. Between I989 and I990, there is a substantial decline in the hazard in the 10-13 week range, with a particularly noticeable decline at 10 weeks. These changes are offset by an increase in the spike in the hazard at 14 weeks and by increases in the hazard between 14 and 20 weeks. Plots of the empirical hazard for 'control group' regions - those in which the ER changes very little or not at all between the two years - show no significant difference between I989 and I990. This evidence strongly suggests that the patterns observed in the maximum entitlement regions can indeed be attributed to the increase in the ER.

In a year where workers had to work 14 weeks to obtain benefits instead of 10, the frequency of 10-week employment spells fell, and those with a duration of 14 weeks increased. [Insert cynical remark here.]

Okay, so we're in a recession; maybe we should be more flexible in responding to new circumstances. But just what problem is Ignatieff's measure supposed to solve? The wave of workers who were hired in the last couple of months - during the worst labour market in a generation - and are now facing unemployment?

It seems to me that the focus of attention should be on the benefits period. In a study conducted for Human Resources Canada, Thomas Lemieux and Bentley McLeod conclude:

The majority of individuals who used the UI system between 1972 and 1992 received benefits no more than one to three times. If the objective of the system is to target a group that needs occasional support while discouraging cycling by a minority of individuals, then the best policy would be to lengthen the qualification period. Since lowering the replacement rate uniformly affects both high and low users, significant changes in this area would likely result in high levels of opposition. Given that the benefit period is more important to the occasional user than to the frequent user, decreasing this level of benefits would have a disproportionate impact on the low-use group. We conclude, therefore, that the qualification period is the instrument of choice if one wishes to target users who need UI during occasional periods of job loss.

The only explanation I can think of for choosing the eligibility criterion as a battleground is that it would have an immediate effect. And since Michael Ignatieff is a politician whose attention is focused on the very short term, that seems to be reason enough. He's only responding to the incentives he's facing.

I should make it clear that I support efforts to shield workers from the worst of the recession. But incentives matter. Thought must be given to make sure that a policy has good outcomes, and not just good intentions.

Comments

[David]

I think Ignatieff's on the right track. There is a moral hazard problem if we're not careful about people gaming the system. But it might resurrect consumer confidence. Right now, people are afraid to spend because they might lose their job and be screwed for income. Lower sales translate into employers actually laying off workers. It's a vicious cycle. But if we strengthen EI eligbility, maybe people will be less scared about the consequences of losing their job and return to normal spending habits.

[Neil]

I strongly suspect that this is a long-term play to try to break down CPC support in Alberta. There's been a lot of criticism here of late that - because of the tendency of workers here to swap jobs very frequently - a disproportionately high number of Albertans don't qualify for EI. If the Liberals can be seen as pushing for "pro-Alberta" changes, it might finally quiet down some of the "remember the NEP" crowd. Regardless of its economic merit.

[Patrick]

Given that many households are liquidity constrained even with two regular incomes and it quickly becomes desperate if someone losses their job, I think it would make sense to shorten the waiting period before benefits kick-in. Maybe in periods of high unemployment it could automatically be dropped altogether. Doing so would prevent people from running up credit cards and using quasi-loan sharking services like The Cash Store.

Neil - I believe that the criteria is not contiguous hours for a single employer, it's X hours worked over some time period. Job swapping, if employment is uninterrupted, shouldn't have an effect on eligibility.

Here in AB, it's generally believed that CPC is bankrupt. A little PR to dispel that myth would be money well spent, IMO.

[Northern Observer]

I disagree with Ignatieff. I think he should be focusing on net benefit paid, not time to eligiblillity. 9 weeks 14 weeks what's the diff? By lowering it to 9 weeks Iggy is just encouraging employers to devolop spotty jobs. What I would like to see is that if you do get laid off from your job of ten years, you don't fall to 55% of paycheck. Any additional amount would help. 60%, 65% 75%. My guess is that this is not being considered as it would be very expensive. But I have lived with the 55% benefit back in 94 and it was a bit stressful. Distracting from the job search really.

[Patrick]

"But I have lived with the 55% benefit back in 94 and it was a bit stressful. Distracting from the job search really."

Trying being 'self-employed' and going to zero. That's loads of fun.

And self-employed happens more than you might think. I know lots of IT contractors who can only get contract work and they don't get EI if the economy tanks and that next gig is nowhere to be found. Neither do they get get pensions, health care benefit, limits on working hours (in AB IT workers are excluded from limits on working hours) etc ... White collar slavery. No wonder they flock to Ayn Rand and Ron Paul; Atlas Shrugged is a story of sweet revenge.

Anyway, I shouldn't get started...

[Phillip Huggan]

A news blurb suggested the policy is tied to in the future harmonizing eligible hours across all regions. If true need to keep that in mind when discussing this.