Contract Costing
and the
Campaign for Reduced Working Time

Tom Walker

At it's April 8th, 1997 meeting, the Canadian Labour Congress's Ad-Hoc Committee on Working Time raised the question of how working time issues could better be brought to the membership and to the bargaining table. One strategy that has not yet been widely recognized would be to adopt the use of contract costing methods that are more accurate and more sensitive to the effects of changes in work hours, paid time off and overtime.

A review of contract costing methods used by some unions shows that these methods introduce substantial errors into the calculation of contract costs. Such errors are invariably biased against reduced work time. They tend to understate the value of paid time off and of reductions in working hours and they tend to overstate the value of overtime work. But, aside from their errors in calculating the relative benefits of working time proposals, these methods tend to significantly overstate the total cost of a settlement.

Miscalculations in costing contract proposals weaken the labour movement's pursuit of shorter working time through collective bargaining. And they may contribute to settlements that contain unnecessary concessions to management. More accurate and time sensitive contract costing could help avoid these pitfalls.

From a broader perspective, more accurate contract costing also could provide new insights into the efficacy of strategies that have long dominated the labour movement's pursuit of shorter work time. These strategies include demands for reductions in work time with no loss in pay and legislative advocacy of increases in overtime premiums. In the concluding section of this brief, these strategies will be re-examined in light of the data from more accurate contract costing.

Some of the suggestions for new strategic directions may be controversial within the labour movement, so it must be emphasized that there are two distinct and equally important reasons to pursue better contract costing:

For both reasons, it is urged that the Ad-hoc Committee on Working Time give careful consideration to the analysis presented below.

Annual paid hours versus annual worked hours

The most serious problems with the way that contract costing is currently done stem from the use of "annual paid hours" as the divisor for calculating the cost per hour of the items in a proposal. The rationale for using annual paid hours is understandable and seems to be based on common sense. First, the use of annual paid hours as the divisor generates a gross calculation of cost per hour that equals the wage rate. That is, with a wage of $20 an hour, annual earnings of $41,600 and 2080 annual paid hours, the cost per hour also calculates as $20. Second, the use of annual paid hours simplifies calculations in that the standard doesn't fluctuate with changes in paid holidays or vacation entitlements.

The alternative to using annual paid hours as the divisor is to use "annual hours worked". Before showing why this is the correct number to use, let's look at the reasons some unions would object to using it.

First, using hours worked seems to "inflate" the cost per hour. A $20 an hour wage generates a $22.61 cost per hour, after accounting for, say, three weeks of paid vacation, 10 days of holidays and five days of sick pay. Here is the calculation:

	$20.00 per hour x 2080 hours paid = $41,600 annual earnings
	2080 hrs. - (120 hrs. +80 hrs. + 40 hrs. ) = 1840 hours worked.
	$41,600/1840 hrs. = $22.61 cost per hour
Second, using hours worked as the divisor requires that all figures must be recalculated whenever there is a change in provisions for paid time off. Finally, and perhaps less urgently, using hours worked as the standard requires making judgements about what to include or exclude (e.g. should paid breaks or travel time be counted as hours worked).

The tendency to want the cost per hour (of annual earnings) to equal the hourly wage is understandable until it is realized that the "hour" in one and the "hourly" in the other are shorthand for two entirely different measures of time. The hourly wage refers to the worker's time -- the time for which the worker is paid whether he or she is at work, on holiday or home with the flu. But the cost per hour refers to the employer's time -- the time during which the worker is available to perform tasks on behalf of the employer. Paid time off may indirectly contribute to worker productivity by making the worker more effective during actual work hours, but it is only during those work hours that production occurs and hence that greater productivity can be realized.

If wages were the only item on the bargaining table, it would hardly matter which standard was used to calculate cost per hour, so long as the same standard was used to calculate both current and proposed wages. However, it is in the calculation of non-wage benefits, and particularly in the calculation of paid time off, that the use of paid hours as the divisor introduces substantial errors.

Consider what happens in the calculation of the cost of an additional week of vacation, using hours paid as the divisor (method 1):

Current contract			 

(three weeks vacation)	hours	annual cost	cost per hour 
Direct wage costs	1840	 $36,800  	$20  
Paid time off		 240	 $4,800  	N/A 
Total			2080	 $41,600  	$20  
Proposed contract			 

(four weeks vacation)	hours	annual cost	cost per hour 
Direct wage costs	1800	 $36,000  	$20  
Paid time off		 280	 $5,600  	N/A 
Total			2080	 $41,600  	$20  
The above calculation leads to the conclusion that an additional week of vacations costs the employer nothing. Because the idea of zero-cost-per-hour vacations is so clearly wrong, some followers of the hours paid method try to remedy this by adding in the vacation cost per hour to the base cost per hour. This "correction" only compounds error and confusion.

Below is how the same calculation would be performed based on the method described in a pamphlet produced by the Oil, Chemical and Atomic Workers' Union (method 2):

Current contract			 
(three weeks vacation)	annual cost	cost per hour*	 
Direct wage costs	$41,600  	$20.00 	 
Paid time off	 	$4,800  	$2.31 	 
Total		 	$46,400  	$22.31 	 
Proposed contract			 
(four weeks vacation)	annual cost	cost per hour*	 
Direct wage costs 	$41,600  	$20.00 	 
Paid time off		$5,600  	$2.69 	 
Total	 		$47,200  	$22.69 	 
*defined as annual cost divided by 2080 hours 		
While the above calculation does show an increased cost per hour resulting from the additional week of vacation, it does so by double counting the vacation cost and thereby overstating the total annual cost of the contract by over 13%.

This double counting is not unique to the OCAW pamphlet. The same procedure is specified in the manual, How to cost a union contract: a guide for union negotiators published by the Center for Labor Research and Education at the University of California. Discussions with sources familiar with union contract costing practice confirm that a similar approach is widely used. In the CLRE manual, the double counting of paid time off benefits also overstates the employer's total annual cost by 9%(see appendix).

It's conceivable that a negotiating team could use method 1 to calculate total annual cost and method 2 to calculate cost per hour. But, the disadvantage of such a mixed method is that the two sets of calculations don't "add up" -- that is, there is no way to reconcile the two separate sets of calculations. And, while the double counting of paid time off is readily apparent in the simple example presented above, it would be easy to overlook in a bargaining-unit wide calculation involving a complex benefit package and a variety of pay grades and vacation entitlements.

Here's how the same cost comparison looks when hours worked is used as the divisor for calculating cost per hour (method 3):

Current contract			 

(three weeks vacation)	hours	annual cost	cost per hour** 
Wages for hours worked	1840	 $36,800  	$20.00  
Paid time off		 240	 $4,800  	$2.61  
Total			2080	 $41,600  	$22.61  

**defined as annual cost divided by 1840 hours 			 

Proposed contract			 

(four weeks vacation)	hours	annual cost	cost per hour*** 
Wages for hours worked	1800	 $36,000  	$20.00  
Paid time off		 280	 $5,600  	$3.11  
Total			2080	 $41,600  	$23.11  

***defined as annual cost divided by 1800 hours 
Note that using method 3, the total annual cost of the current and proposed contracts has remained the same, $41,600, as has the wage. But the added cost of an extra week of vacation is shown in the increase in the cost per hour from $22.61 to $23.11. Sure enough, the cost per hour using hours worked is $.42 (or about 2%) higher than when hours paid is used as the divisor ($23.11 vs. $22.69). But the total annual cost is $5,600 (13%) lower.

Even it union negotiators are able to avoid the worst miscalculations, the hours paid method contains a significant bias in the treatment of paid time off. This bias is evident when we compare the vacation cost per hour calculations for method 2 and method 3:

Paid time off	Cost per hour		 
		Method 2	Method 3	 
Existing	$2.31 		$2.61 	 
Proposed	$2.69 		$3.11 	 
Increase	$0.38 		$0.50 	 
As the above table shows, method 2 (based on hours paid) consistently assigns a lower cost per hour to paid time off than does method 3 (based on hours worked). Furthermore, the gap between the two methods grows as the amount of paid time off increases. Using hours paid for the calculation, an extra week of vacation would be comparable to a $.38 an hour wage increase. The hours worked calculation, however, would value the extra week as comparable to a $.50 an hour wage hike -- a difference of over 30%!

What about the argument that using annual hours worked as the divisor is disadvantageous for unions because it inflates the employer's cost per hour? Wouldn't it be harder to bargain for the $.50 an hour increase indicated by method 3 than for method 2's $.38 an hour increase? This is a difficult claim to answer, because it relies on a mathematical tautology. In a trivial sense, using hours worked does appear to raise the cost per hour simply because a smaller divisor will always result in a larger quotient. But it's another matter whether or not the difference is a disadvantage to unions.

The usefulness of a calculation depends on the logic and persuasiveness of the terms, not on the absolute size of the answer. One could make the cost per hour appear even smaller by arbitrarily adding another 1000 hours to the divisor or even by taking the total number of hours in a year (8760) as the divisor. However, employers are not going to be persuaded by lower cost per hour figures arrived at through the use of a conveniently large divisor. In practice, large discrepancies between union and company estimates would probably pose a bigger obstacle to settlement than would a slightly higher, but mutually agreed upon, cents per hour figure.

On the management side, the accounting literature (Granof 1973, Hazelton 1979, Lau and Nelson 1981, Sullivan 1980) is clear on the importance of using "productive hours" or hours actually worked as the divisor. For example, Hazelton states flatly, "Gross hours per year cannot be used as a divisor to arrive at a meaningful cost per hour figure. The hours worked calculation offers a more reasonable solution."

But even this consensus is relatively recent. As late as 1979, Hazelton observed that reports of national, local and industry-wide wage rates and benefit costs were "relatively useless" because of inconsistencies in calculating the costs of various benefits.

For their part, union negotiators have shown a well-founded skepticism toward costing as it is practiced by management. Former CUPE research director Gilbert Levine summed up his skepticism when he commented that "From what little I have observed about costing of collective agreement settlements, I am convinced that it has been used by management as a sophisticated device to deprive workers of a fair and just settlement."

Unfortunately, this otherwise healthy skepticism often leads union negotiators to be satisfied with an informal approach to costing. Or, what is worse, to rely on questionable costing formulas that produce misleading results, such as the 13% overstatement of employer costs arrived at in the previous example using method 2.

Another contract costing dilemma -- the overtime premium as "benefit"

Besides the main problem of using the wrong divisor, there are at least four other aspects of collective agreements that can create confusion when costing them:

All of these problem areas can result in misleading calculations that have an impact on the issue of working time. But we will here concentrate on problems in costing payments for overtime work that derive from a combination of the above difficulties. The errors invariably overstate the value of overtime.

A common practice is to treat the overtime premium as a benefit in calculating cost per hour and to use a historical average from recent experience to estimate the number of overtime hours. This practice embeds a "usual amount" of overtime in a collective agreement. However, because the calculated values of cost items are interrelated -- and because some cost items are fixed, per worker costs -- the treatment of "usual hours of overtime" as a benefit reduces the apparent value of other benefits.

In this case, the problem is not a mistake in calculation, it's the misleading use of labels. It would be more accurate to describe a "usual amount of overtime" as a tacit agreement to lengthen the workweek at management's discretion. Any lengthening of the workweek decreases the per hour cost to the employer of fixed cost benefits because it spreads those costs over a greater number of hours. In effect, the overtime premium captures these employer cost savings and falsely reports them as a new benefit.

The distortion can be seen if we compare calculations based on different assumptions regarding usual hours of overtime.

0 hours of overtime	hours	annual cost	cost per hour worked 
Wages for hours worked	1840	 $36,800  	$20.00  
Paid time off		 240	 $4,800  	$2.61  
Overtime premium	   0	 $0.00 		$0.00  
Total			2080	 $41,600  	$22.61  
50 hours of overtime	hours	annual cost	cost per hour worked 
Wages for hours worked	1890	 $37,800  	$20.00  
Paid time off		 240	 $4,800  	$2.54  
Overtime premium	  50  	 $500 		$0.26  
Total			2130	 $43,100  	$22.80  
100 hours of overtime	hours	annual cost	cost per hour worked 
Wages for hours worked	1940	 $38,800  	$20.00  
Paid time off		 240	 $4,800  	$2.47  
Overtime premium	 100	 $1,000  	$0.53  
Total			2180	 $44,600  	$22.99  

As the "usual amount of overtime" increases, the calculated cost per hour (and consequently the apparent value) of paid time off decreases. Meanwhile the calculated cost per hour and presumed value to the workers of the overtime premium appears to increase directly in proportion to the number of overtime hours worked.

What is happening here? Essentially, a constant cost is being transferred from one account to another -- like taking a dollar out of one pocket and putting it into another pocket. A portion of the overtime premium (in the above example, 28%) is therefore not a new "benefit" at all, it's just a new -- and misleading -- name for an old benefit.

Again, an argument can be made that the lower cost per hour of the other benefits could make improvements easier to sell to an employer. In effect, though, such an argument is at the same time an argument to the employer against reducing the use of overtime.

During their April 8 meeting, members of the Ad-hoc Committee on Working Time discussed why overtime persists in the face of high overtime premiums. Some of the factors mentioned were bad planning on the part of management; the desire of supervisors for overtime pay and collusion between supervisors and workers to increase overtime and ration it. A more likely explanation is that the "high overtime premiums" are an illusion. These nominal premiums aren't really a new cost to the employer. They are a cost saving that has been recycled as a way to make longer hours of work appear more attractive to workers.

Costing for reducing work time and sharing work

Up to this point, we have dealt with how various contract costing methods can affect the evaluation of a given wage and benefit package for a given number of workers. The same methods of costing contract proposals can be used to evaluate the potential for reducing work time and sharing the work. However, such an analysis requires making many more assumptions about total employer costs (aside from direct labour costs), anticipated changes in productivity and the divisibility of jobs.

Furthermore, costing of shorter work time proposals is a one-sided exercise as long as it looks only at the employer side of the ledger. This is because what the employer pays is NOT what the worker gets. Payroll and income tax policies have created a bias against reductions in the full time workweek. This bias is accentuated if unions and workers focus exclusively on gross earnings as the measure of employment compensation.

Presentation of a work sharing example would require a separate paper to work through all the assumptions and variables. Aside from the complexity of the calculation, the main difference between evaluating a work sharing proposal and evaluating the usual proposal for wages or benefits is that the work sharing proposal should start with the result of a calculation dealing with the current work force and hours and work backward to examine the changes that would occur "automatically" and the adjustments that would have to be made to accommodate a different mix of workers and hours.

For example, consider what happens if we take a work force of 100 workers working a 40 hour week and putting in an average of about two hours of overtime a week. Our goal is to reduce work time and create 20 new jobs. To work the same total number of annual hours, the new workweek would be 35 hours. With no loss in pay or benefits, this seven hour reduction in the work week would cost the employer nearly 20% more (assuming no change in productivity, etc.). On the other hand, if the entire cost of the work time reduction were borne by the workers, the original 100 workers' gross compensation would fall by nearly 17%.

Such a take-it-or-leave-it scenario would be unlikely to win acceptance from either employers or workers. Nor would a crude "split-the-difference" solution (say a 9% increase in employer cost combined with an 9% reduction in individual gross pay and benefits) seem to offer much greater chances of success.

It is only when we begin factoring in different assumptions about productivity change and tax impacts that the gap starts to narrow to the point where agreement becomes conceivable. For example, because of tax impacts, the net (after tax) loss of compensation for existing workers would be only about 60% of the gross loss. Productivity gains in the range of 5% to 10% would, correspondingly reduce the cost impact on the employer.

Using these methods, it is not as difficult to arrive at a proposal in which workers can gain a large reduction in work time and a significant increase in employment in exchange for a small sacrifice in net pay. Employers would face cost increases no greater than those associated with typical wage and benefit proposals. Governments could be pressured to change payroll tax policies to reward unions and companies that create jobs in this way.

Strategic considerations

The above analysis hints at some of the strategic implications of better contract costing. It would perhaps be useful to restate these implications clearly and to briefly re-examine the strategies that have dominated union campaigns for shorter work time for several decades.

One mainstay of union strategy for reducing work time has been to call for increasing the overtime premium to serve as a greater disincentive to employers to schedule overtime. Obviously the overtime premium is a two-edged sword. While it may serve as a disincentive for employers to schedule overtime, it also acts as an incentive that makes workers desire the premium pay that comes with overtime work

But our example showed how traditional costing practices overstate the cost to the employer and the "benefit" to workers of the overtime premium. The problem with the overtime premium, then, is not simply that it is too small but also that there is a large gap between the actual net cost of the overtime premium to the employer and its perceived value to workers. Raising the overtime premium to, say, double-time would not eliminate that gap and could even increase it if -- as seems likely -- employers took measures to offset the impact of a new premium structure.

Legislative efforts to increase the overtime premium have been notable for their almost total failure. Employers and their economists argue vehemently that increasing overtime premiums would have dire economic consequences. Whether or not such predictions would come true is probably a moot point -- governments are not going to take that chance in the face of such unwavering opposition from business. There's little chance of unions winning the battle for higher overtime premiums. But it's unlikely that a victory on this front would make any difference anyway.

A second pillar of union strategy has been to call for reducing work time with no loss in pay.

The Achilles heel of this strategy has long been that small reductions in work time with no loss in pay are entirely feasible (UAW speaker Nat Weinberg pointed this out to the 1956 AFL-CIO Conference on Shorter Hours of Work). Such reductions could typically be accompanied by an increase in the intensity of work and hence in productivity and would therefore create few new jobs.

On the other hand, reductions in work time large enough to create significant numbers of new jobs could only be achieved either through some sacrifice in the pay of existing workers or through mass political mobilization of the workers. In effect, then, the call for reduced work time with no loss in pay is a call for a revolution with no revolutionary strategy to back it up.

Defenders of these two traditional strategies will point to existing union members' concerns about loss of income or "cash flow" and consequent decline in purchasing power and standard of living. "In order to maintain the support of existing members for shorter work time," the argument goes, "we must assure them that they personally have nothing to lose." This presents an important dilemma: is it the union's role to represent the individual interest of each member or to represent the collective interests of the membership as a whole? What happens when the individual interests of a member conflict with the collective interest?

The traditional strategies evade the question of conflict between individual and collective interests by proclaiming them to be inherently compatible. It is as if an "invisible hand" of solidarity is expected to reconcile the (short term) selfish interest of each member with the general good of the membership, just as the invisible hand of market mythology supposedly guarantees the socially beneficial outcome of individuals maximizing their personal utility.

Contrary to this presumption of benign harmony, a careful evaluation of contract proposals reveals that some outcomes involve trade-offs between individual and collective gains. To allow gross annual individual earnings to dominate such trade-offs is to abandon the union's role as advocate of the collective interest. Not only does the obsession with gross annual earnings neglect the collective interest, it discards the genuine, long-term welfare of individual members in favour of an illusory immediate fix.

By contrast, accurate and flexible costing models show how large reductions in work time and increases in employment could be gained in exchange for very modest sacrifices in the net annual earnings of existing workers. Furthermore, such changes can also accommodate increases in individual members' hourly earnings and substantial increases in the net earnings of the membership taken as a whole. There is no guarantee that employers would readily agree to proposals constructed according to these principles of collective well-being. But there is a much better chance of winning support from both the union membership and the public for such practical, "moderate" proposals than for all-or-nothing, pie-in-the-sky demands.


Given the distortions introduced by commonly used contract costing methods is it any wonder that the union membership often resists shorter working time and welcomes the 'opportunity' for more overtime hours? Or, that once they've experienced shorter hours (and learned first hand that the impact on their pocketbook was much less than they'd expected), they don't want to go back?

Unions have historically supported the reduction of the workweek and the working day. Yet little progress has been made on this front over the past half century. Indeed, over the past two decades, full-time work weeks have been increasing in length at the same time that millions of people who want full-time jobs can't find any work at all or are forced to accept part-time employment.

We have argued here that part of the reason for this reversal is that many unions use flawed contract costing methods, and as a result fail to assign full value to the working time provisions of their collective agreements. Another part of the failure to advance shorter work time may be the reliance on strategies that sound militant yet are inflexible, incapable of inspiring broad support and that rely on obsolete assumptions about the relationships between working time, employer costs and workers' earnings. A more accurate and work time sensitive approach to contract costing can help achieve better results from current strategies and it can contribute to the re-evaluation of those strategies and the adoption of better strategies.


Given the enormity of the problem of the distribution of working time and given the labour movement's desire to address it, unionists generally and negotiators in particular have every reason to adopt contract costing methods that support the push for shorter work time instead of helping to undermine it. To facilitate this, the Canadian Labour Congress's Ad-Hoc Committee on Working Time should: