Ehrenberg's research on fringe benefits followed up on less systematic but widespread observations that were heard from the business press and government analysts in the early 1960s. Joseph Garbarino's 1964 article, "Fringe Benefits and Overtime as Barriers to Expanding Employment," referred to the growing "consensus [among business, government and labour] about the role of 'fringe benefits' in weighting the choice in favor of overtime." As an example of this emerging consensus, Garbarino cited a 1963 Business Week article, "Unions Mount Attack on Overtime," that reported even AFL-CIO officials concede what employers have long since held: It's cheaper to pay overtime rather than hire additional employees with the heavy fringe costs that go with each worker.

Chart 1 shows the growth of fringe benefits in the U.S. manufacturing sector for 1957-79 and 1981-93. Comparable data for Canada are not as readily available. Although the two sets of data differ in how they define employee benefits, the trends are clear. Fringe benefits rose steeply during the 1957-79 period and continued to rise at a slower pace from 1981 to 1993.

Other factors, besides fixed nonwage costs, may also be at work undermining the effectiveness of the overtime premium. In his examination of "The Effects of Overtime Pay Regulation on Worker Compensation," Stephen Trejo found that employers make wage adjustments to mitigate (although not completely offset) the effects of overtime premiums.[2] This is a special case of the expectation, commonly stated by economists, that employers pass on the cost of payroll taxes to workers in the form of lower wages. Trejo gave the following example of what he calls the "fixed-job" model in which lower wages could entirely offset the overtime premium:

suppose that in the absence of overtime pay regulation a job pays $550 per week and requires 50 hours of work, so the average wage is $11 per hour. An employer may satisfy the FLSA by reducing the straight-time wage to $10 per hour and paying time and a half for the 10 overtime hours. The employee works 50 hours and receives $550 per week in either case; consequently the law has no real effect.

Combining Ehrenberg's and Trejo's analysis produces a picture in which much of the overtime premium is offset by fixed nonwage labour costs and a smaller part is offset by wage adjustments. Outright noncompliance with the law is more common near the minimum wage level, where benefit substitution and wage cuts are not feasible. After accounting for fixed costs, wage adjustments and noncompliance what remains of the overtime premium is that it perhaps discourages some overtime in some low-wage jobs.

Nevertheless, when the issue of curbing overtime and redistributing work time is raised, proponents often seize the first policy tool at hand -- the overtime premium -- and recommend augmenting it. For example, John Zalusky, head of Wages and Industrial Relations at the AFL-CIO, has long championed increasing overtime compensation from time and a half to double or triple time. In Zalusky's view, the fundamental purpose of the overtime premium was to discourage regularly scheduled overtime. The fact that it no longer serves that purpose, he believes, can best be addressed by adjusting the size of the premium.

In the early 1960s, the AFL-CIO campaigned for an increase in the overtime premium, backing legislation introduced into the U.S. Congress. That bill would have required employers to pay double-time for overtime. Again, in the late 1970s, legislation was introduced to increase the overtime premium to double time and set a threshold of 35 hours after which the premium would apply. Mainstream economists routinely scoff at such proposals as unlikely to create as many jobs as hoped. Employers rail against them as too costly and as potentially harmful to job creation. Perhaps most importantly, they fail to inspire widespread public support to the issue of limiting work time.

But if the proponents of double time seem heedless of experience, the economists who criticize such proposals could possibly be too subtle and sophisticated in their analysis. A rule of thumb to guide policy analysis should be: an ounce of field work is worth a pound of theory, or: never do statistical regressions when a simple cost calculation will tell the whole story.

After Ehrenberg completed his study of the relationship between fringe benefits and overtime, he turned his attention to the job creating potential of increasing the overtime premium. Ehrenberg concluded that such proposals were unlikely to generate the number of jobs claimed, mainly because the resulting higher labour costs could lead employers to substitute capital for labour and because the higher prices resulting from higher labour costs could lead to lower total demand.

The puzzling thing about Ehrenberg's argument is that it doesn't address the possibility that an increase in the overtime premium could stimulate a further increase in fringe benefits, which could, in turn, offset the increase in the premium. However, if we accept that the rise of fringe benefits may have been -- at least in part -- a strategic response by employers to the overtime premium, then the best argument against increasing the premium would be its futility.

What evidence might there be for suspecting that the rapid rise in fringe benefits after World War II was a strategic response from employers to the overtime premium? To answer this question, we need to leave aside theoretical models and look at the formula management uses to cost labour contracts. For the rise in fringe benefits to have been a strategic response, doesn't require that it had to have been a conscious response. Strategic choices can be based on summary information without the decision maker being aware of precisely why one option seems to offer an advantage over the other.

While preparing a research proposal on the issue of work time redistribution, I spoke with Csaba Hajdu, research director for Forest Industrial Relations -- an employers' organization that conducts collective bargaining for member companies. Hajdu had no hesitation in stating what he saw as the relevant data: cost per hour actually worked and annual hours worked per employee. When Hajdu mentioned $38.80 as the cost per hour actually worked at the benchmark millwright's rate, I was immediately struck by how close this figure was to one and one half times the basic wage rate for that classification (around $25 an hour). As the Bob Dylan song went, "you don't need a weatherman to know which way the wind blows."

Rather than press Hajdu further on his exact method of calculation (and armed with a working knowledge of basic accounting and cost control principles), I spent a few hours building a spreadsheet model to cost a sample collective agreement between Forest Industrial Relations and the Industrial Woodworkers of America, Canada. I suspect my model is quite similar to Hajdu's -- my results were almost identical, even though my data was undoubtedly less complete.[3]

Working with a cost accounting model produces some remarkably clear insights into the structure of compensation packages -- insights that the econometric models are unlikely to produce. The model demonstrates the logic of the following proposition:

Given the following conditions:

• an overtime premium;
• a wage rate higher than the minimum wage by a factor at least equal to the overtime premium;
• workweeks that on average are higher than the legal maximum; and
• a fixed nonwage labour cost that is lower, as a proportion of total straight-time labour cost, than is the overtime premium as a proportion of overtime pay.[4]

This is just a another way of saying that the nominal offer doesn't take into account the overtime premium but the cost calculations do. The key issue here is not the arithmetic itself, but the rhetoric of calculation that diverts attention away from doing other arithmetic.

Chart 2, below, shows the curve for the cost per hour worked, given an overtime premium of time and a half and no fixed labour costs.

Chart 3 shows how a 1/3 proportion of fixed labour costs entirely offsets the time and a half premium.

To be sure, fringe benefits have other features to make them attractive to both employers and unions. For unions, a negotiating package of diverse benefits can help build a coalition of support among union members for a negotiating strategy. For employers, fringe benefits are a way of ensuring a more stable labour force and of moderating the costs of automatic wage escalation in a multi-year contract. But such other attractions of fringe benefits complement, rather than contradict, their appeal to employers as a way to save on overtime costs.