Directorates
·	Director General's Office
·	Data Development
·	Strategic Evaluations
·	Partnership and Continuous Evaluations
Publications
·	Evaluation Reports
·	Evaluation Briefs
·	Labour Market Development Agreement Series
·	Lessons Learned Series
·	Evaluation Tool Kit Series
·	Evaluation Forum Newsletter
·	From Theory To Practice

In our empirical analysis below, we assess the consistency of the evidence with three simple accounts of what quit penalties do to separation behaviour: a "pure relabelling" scenario in which the penalties have no real effects but change the reported reasons for separation; an "inhibition" scenario in which these penalties reduce the total number of separations; and a naive, "no-response" model in which both the total volume and composition of separations is unchanged, with an attendant decrease in UI receipt by workers who quit or are dismissed from their jobs. In the remainder of this section, we discuss theoretical models of the separation process that might give rise to each of these three possible scenarios in turn.

A. Models Generating Pure Relabelling

A common characterisation of the efficient turnover view of labour markets (e.g. Becker et al. 1977; Burdett 1978; Jovanovic 1979; Mortensen 1988) is that there is no meaningful distinction between quits and layoffs. McLaughlin's (1991) model is perhaps the most noteworthy of this class, since it builds on earlier work. It interprets the quit or layoff label as summarising the labour market environment of a particular separation but posits that the label is economically inconsequential. Labour turnover is efficient and joint wealth maximising in the face of economic shocks, since wage revisions or side payments can always be used to prevent inefficient separations. A quit occurs when an upward revision of the wage requested by a worker on the basis of an outside offer is not met by the employer. Similarly, a layoff results when an employer initiated downward revision is refused by a worker.

Interestingly, in all of these models there is no financial cost or gain to either party attached to the relabelling of separations: in the absence of UI, a firm or worker that chooses to change the label publicly attached to a separation incurs no costs or benefits. If these models are taken literally, the introduction of a financial gain to relabelling, via the UI system, should cause the complete elimination of quits from the economy.³ Further, there should be no change in the actual separation behaviour of firms and workers in response to a UI quit penalty⁴Pure relabelling models also predict that UI quit penalties will have no effects on government UI expenditures, since those who would have quit will continue to claim as "layoffs" after the policy change.

B. Models Generating an Inhibition of Separations

In any model where workers' ability to have their separations relabelled is limited, UI quit penalties will raise the costs of employee-initiated separations, and should lead to a reduction in total separations from the firm. In this subsection we make note of three broad classes of factors that might limit the extent of such relabelling. One of these is the fact that, if asymmetric information is important in labour markets, separation labels might convey useful information to other firms or workers in the labour market and in this sense be valuable in themselves. An example of this class of models is Gibbons and Katz's (1991) "layoffs and lemons" hypothesis, where a worker would prefer to be identified as separating as a result of a mass layoff than an individual layoff, because of the negative signal of worker quality embodied in the individual layoff category. Analogous to Gibbons and Katz, one can easily conceive of a signalling model in which workers prefer other firms to know that they quit their former jobs and were not laid off (because it signals greater ability). Similarly, firms might not want to develop a reputation for laying workers off repeatedly. In both these cases, the parties may have an interest in preventing the relabelling of quits as layoffs, even though there is a UI-related financial gain associated with doing so.⁵

A second model in which relabelling might be limited is one with some ex post wage rigidities (as in Hashimoto and Yu 1980; Hall and Lazear 1984), combined with some limitations on firms' and workers' abilities to make contractual commitments. For example, suppose that firms and workers agree, ex ante, to a real wage of w, despite the existence of uncertainty in the worker's future internal and external productivity ( ⁱ and ^e, respectively). Focus on a worker whose best outside option includes at least the possibility of an unemployment spell, let ^e + UI be her outside option if she is laid off, and let q^e be her outside option if she quits (quitters are ineligible for UI). Ex post, suppose it happens that ^e + UI > ⁱ > w > ^e . In this situation, firm-plus-worker wealth is maximised by dissolving the firm-worker match and calling the separation a layoff. Firms faced with this situation, however, will be reluctant to lay the worker off unless they can be guaranteed a sufficiently large share of the value of the UI payment to cover the loss of ( ⁱ - w) and any transaction costs. In the absence of such a side payment, the firm can make positive profits by retaining the worker at the wage w, and the worker cannot profitably quit, given this wage and the UI quit penalty. Thus, unless firms can credibly precommit to relabel all separations as layoffs, the introduction of UI quit penalties will cause some workers who would otherwise leave to remain with their former employers.

Finally, the economic theories discussed to this point include only two types of players: firms and workers. The environment under consideration, of course, has other important players, notably the ministry that runs the UI system and the Canadian legislature, which originated the Bills being studied. A stated purpose of each piece of legislation is to reduce the cost of operating the UI system. These additional players, therefore, have an incentive to ensure that massive relabelling does not occur. What efficient turnover models euphemistically call side payments, are, in some of their incarnations, labelled as bribes by those who drafted the legislation. While verification of relabelling is difficult, given that penalties were introduced in Bill C-21 and HRDC employs a number of inspectors to guard against, among other things, false reporting, there is at least the threat of penalties being sufficient to stop relabelling.

C. Models Generating No Response

While it is hard to think of a plausible economic model in which a large change in the UI eligibility of quitters has no effect at all on the number of separations labelled as quits, a number of factors might act to limit firms' and workers' responses to the legislation to the point where it cannot be detected in national separation rates. Aside from the factors mentioned in the previous section, these might include the fact that quitters are more likely than other workers to move directly from one job to another without an intervening unemployment spell. To the extent that some quitters are certain that no such spell will occur, this should make them immune to the effects of UI quit penalties, rendering relabelling unnecessary. Another would be a combination of good enforcement of HRDC's "false reporting" provisions and a relatively small number of workers who are on the margin between quitting and staying with the firm. Thus, despite the fact that such behaviour may be hard to rationalise, it is important to take note of this possibility in our analysis of the data.

Footnotes

3	McLaughlin (1991, p 11) obviously does not intend his model to be taken quite so literally, but in addressing the implications of efficient turnover for UI he posits an incentive for quits to be relabelled so as to make them UI eligible. Side payments, or arrangements of some type, are proposed as mechanisms by which the worker could entice the firm to agree to relabelling.
4	More precisely, comparing a world in which all separations are treated equally by the UI system to one where only layoffs are UI eligible, efficient labour contracts should generate identical patterns of separations in the two worlds. The only difference between the two worlds will be that in the second, all separations will be labelled as quits, since it is jointly inefficient (from the firm and worker's joint perspective) to forsake UI benefits. It may also be worth noting that both worlds will have more separations than a world with no UI at all.
5	Of course, in some environments, the argument might run in the opposite direction: workers might be reluctant to have their separations labelled as quits because it signals less expected labour force attachment. The main point of the argument is that, with signalling, there are factors other than UI eligibility affecting the optimal labelling of separations and that the presence of these other considerations is likely to dampen the "extreme" amount of relabelling predicted by the pure efficient separations model.