By H. Randolph Thomas,1 Member, ASCE
1Prof., Penn. Transp. Inst., Penn. State Univ., Res. Bldg. B, University Park, PA 16802.
Note. Discussion open until August 1, 1992. To extend the closing date one month, a written request must be filed with the ASCE Manager of Journals. The manuscript for this paper was submitted for review and possible publication on January 11, 1991. this paper is part of the Journal of Construction Engineering and Management, Vol ' 118, No. 1, March, 1992. (DASCE, ISSN 0733-9364/92/0001-0060/$1.00 + $.15 per page. Paper No. 1241.
|ABSTRACT: This paper reviews the construction literature on the effects on labor productivity of scheduled overtime. The literature is organized into three groups: studies based on data front project records, studies in which the sources of data are unknown, and studies done in the manufacturing sector. Analyses are also performed on the influence of the number of days per week and the number of hours per day. The literature on scheduled overtime was found to be very sparse; dated to the late 1960s and earlier; based on small sample sizes; and largely developed from questionable or unknown sources. The analysis reveals very few original data. Many studies reference other studies, giving the false appearance of originality. The analysis of data shows general consistency with respect to overall losses of efficiency. However, with respect to the loss of efficiency as a function of the number of hours per day and the number of days per week, many studies show that the effect of these two variables is negligible. The paper concludes that these studies provide strange and largely unbelievable results.|
Construction overtime has frequently been used as in inducement to attract labor and to accelerate schedule performance. Although there may be positive short-term benefits to working an overtime schedule, the long-term consequences are typically viewed as detrimental Understanding the effects of overtime scheduling is quite difficult because the factors affecting productivity in the overtime situation are numerous. Furthermore, if an overtime schedule must be used, there is little, if any, information available to aid in deciding which type of schedule to adopt (Thomas 1990).
Reliable studies of the effects of extended overtime on labor productivity are very difficult to produce because there are many factors that can affect productivity, some of which have nothing to do with the overtime situation. Examples include the character of work being performed and start-up and testing activities. While obvious, it is often forgotten that manpower is not the only resource or component that is consumed at an accelerated pace in an overtime situation. Materials are installed at a faster rate; consumables, tools, and construction equipment are in greater demand; and engineering questions and information demands must be processed at a faster rate. Absenteeism, accidents, and fatigue may become a growing problem, and the quality of workmanship may decline. Thus, if a project is behind schedule, working overtime may simply exacerbate the problem. Therefore, whenever overtime is discussed, the surrounding circumstances must be clearly understood.
The objective of this paper is to critique the literature describing the effects of an overtime schedule on construction-labor productivity. The need for such a summary arises because, relative to the amount of published literature, little is known about the origin of the overtime data, number and type of projects, and surrounding circumstances. Yet widespread assumptions have been made that the data and conclusions are reliable. This paper clarifies the aforementioned aspects for each data set.
This paper focuses on scheduled or extended overtime, that is, an overtime schedule that lasts longer than several weeks. Spot overtime, which is intermittent, is not covered, because the negative aspects are minor relative to the job as a whole.
The literature was collected from published and unpublished sources. The graphs are presented to illustrate the flaws in the data and should not be interpreted as validating or invalidating the effects of scheduled overtime on labor productivity.
In this paper, the following definitions are used:
"Extended overtime" is involved in a work schedule that extends over more than 40 hr of work per week. The schedule is planned in advance and lasts for at least three consecutive weeks, and typically longer. The term "overtime" is used interchangeably with "extended overtime."
"Labor productivity" is the ratio of the input in terms of labor hours to the output in terms of units of work.
STUDIES OF HOURS PER WEEK AND HOURS PER DAY
Various studies have reported losses of productivity caused by scheduled overtime. These studies are grouped in the following according to the source of the data and the applicability to construction situations.
Data from Project Records
An extensive review of the published literature and other sources yielded three studies of the effects of construction overtime in which the project records were the basis for the conclusions. Table 1 summarizes pertinent data about each study.
Proctor & Gamble
The most publicized study of construction overtime is that made by Proctor & Gamble at their Green Bay, Wisconsin, operation. The study was first published as a Business Roundtable (BRT) report ("Effect" 1974) and was reissued in 1980 as part of the Construction Industry Cost Effectiveness Project ("Scheduled" 1980). Figs. 1 and 2 show the reported effect on productivity of extended overtime for a period of 12 weeks. As stated in both reports, the figures represent "the reduction in productivity normally experienced on projects operated on a basis of 50 hours per week and 60 hours per week .These observations are on a weekly basis with all completed work recorded from physical count or measurement and the work hours expended obtained from actual payroll hours. The curves reflect the averages of many observations" [("Scheduled" 1980) page 10].
The data in Figs. 1 and 2 are based on a comparison of actual work hours expended to a fixed standard base called a "bogey." The "bogey" standard is for a straight-time schedule. Unfortunately, the data are not a comparison between actual productivity on straight time and overtime productivity. The 1974 and 1980 reports contain the following warning with respect to comparisons of various data sets: "Direct comparisons of various data are difficult since all measurement of productive effort is not referenced to a Fixed Standard. The Industrial firm's data on productivity is based on Fixed Standards and a performance of 1.0 may not be the same as a performance of 1.0 referenced to some other standard of comparison. As a result, a 30% reduction of productivity in one set of data could compare with a 15% reduction reflected in another set of data due to this difference" [("Scheduled" 1980) page 15].
What is little known about the Business Roundtable reports is that the data all originated from a single project (i.e., Proctor & Gamble's Green Bay operation). Also, the nature of the Green Bay construction activities is not known. The curves are a composite view of a series of comparably short jobs on scheduled overtime covering a 10-year period. The project operations were carried out in a tranquil labor climate and the field management was reported to be excellent.
A technical paper by L. V. O'Connor described the experiences of Foster Wheeler in constructing five large fossil boilers in the Ohio Valley between 1963 and 1968 (O'Connor 1969). The curve in Fig. 3 shows the results. The paper reported an average productivity decline of 7.9% per year during the period because of a variety of factors, including overtime, over-manning, and labor strikes. Although it is not explicitly stated, the reader is left with the impression that the conclusions are based on the boilermaker craft. No other information is given, and it is not known how the overall trend of productivity losses form the other causes was factored into Fig.3. The percentage loss of efficiency is consistent with the BRT report if the scheduled overtime period is about five to six weeks.
Construction Industry Institute
In 1984, the Construction Industry Institute (CII) sponsored a three-year study of construction overtime ("The Effects" 1988). This study represents the only source of original data since the 1960s. The study also differs from the previous two studies in several important respects. The data base includes multiple projects, and the data were collected by direct observation rather than from project records. The focus is on the work of a single crew as opposed to an entire craft, and the study reported results for a number of crafts. The analyses are independent of the project estimate.
The study included seven projects. As shown in Table 2, the two natural gas recovery projects were in the early stages of completion and the refinery expansion project was nearing completion. One was a shutdown project, and another (project 4) was experiencing considerable productivity problems aside from the overtime situation. Project 5 was on a rolling 4/10 (four days of 10 hr each) schedule. Only on projects 3 and 4 were data collected on straight-time and overtime schedules.
The limitations of the study include the inability to compare overtime productivity to straight-time productivity, incomplete data-collection procedures, the use of moving-average calculations, and the inability to correlate changes in productivity to other factors and job-site conditions. The problems caused by these limitations are illustrated in Figs. 4 and 5, which show the overtime and straight-time productivities of electricians on project 4. No discernible patterns can be identified. On project 3, the electrician productivity was better on a six-day, 10 hr schedule than on straight time.
The CII study could not develop defensible conclusions relative to the effects of an overtime schedule, because the productivity trends for the same crews were not consistent. The report did conclude that productivity does not necessarily decrease on an overtime schedule. At best, fatigue did not seem to be a factor.
Other Studies of Effects on Construction Productivity
Five other sources of information summarizing the effects of construction overtime were identified in the literature. In each of these, the source of the data is either unknown, is considered to be more opinion than factual, or is a republication of other data. Table 3 summarizes these sources. As can be seen, most of this information is also significantly dated.
National Electrical Contractors Association
In 1969, the National Electrical Contractors Association (NECA) published the results of a 1964 study of overtime done by the NECA Southeastern Michigan Chapter (Overtime 1969, 1989). The findings for electricians for 50 hr and 60 hr worksheets showing productivity losses as a function of the schedule used and the duration are presented in Figs. 1 and 2. The origin of the data is unknown.
NECA also published the results of a survey of 289 NECA members regarding their experiences with reduced productivity associated with overtime (Overtime 1962). These data are generally consistent with the Foster Wheeler data except for the 6/10 schedule. They are somewhat consistent with the 1969 NECA Michigan Chapter study except for several of the more demanding schedules.
C.F. Braun Inc.
Unpublished material from the C. F. Braun Inc. construction guide (unpublished 1979) contains data that are remarkably similar to the Foster Wheeler data and therefore raise doubts about their originality. No information is available about the origin of the data.
J. J. Adrian
Adrian [(1987) page 971 reported productivity losses on concrete activities as part of an analysis of a contractor claim. The project was in Chicago, Ill., and the work was performed in 1982 under ideal weather conditions (60-80° F). The losses of productivity are shown in Fig. 6.
Qualified Contractor (Howerton 1969) published statistics on an overtime study conducted in 1964 [(Oglesby et al. 1989) page 260]. The data are plotted in Fig. 7. As the figure illustrates, the data are similar to those of Adrian (1987).
Mechanical Contractors Association of America
The Mechanical Contractors Association (MCA) provides information to its membership in the form of management methods bulletins. Bulletin No. 18A (dated January 1968) ("How" 1968) and Bulletin No. 20 (dated November 1968) ("Tables" 1968) were issued to assist contractors in the preparation of claims and change orders relative to overtime inefficiencies. Fig. 8 illustrates the losses that can be expected. As noted, the data are based on U.S. Department of Labor Bulletin 917 (Kossoris 1947,i,b). As is indicated later, Bulletin 917 summarizes data from the manufacturing sector, not from construction. Therefore, the MCA bulletins ("How" 1968; "Tables"' 1968) do not contain original data.
American Subcontractors Association
The Associated General Contractors, the American Subcontractors Association, and the Associated Specialty Contractors, Inc., jointly published a primer on overtime (Owner's 1979). The data are identical to those published by the Business Roundtable ("Effect" 1974; "Scheduled" 1980). Thus, the ASA primer contains no original data.
American Association of Cost Engineers
The American Association of Cost Engineers published a bulletin on overtime ("Effects" 1973). The data are identical to those published by the Business Roundtable ("Effects" 1974; "Scheduled" 1980). Therefore, the AACE bulletin contains no original data.
STUDIES IN MANUFACTURING SECTOR
A cursory review of the literature of overtime effects in manufacturing and other industries yielded very little quantitative information. The most notable reference is Bureau of Labor Statistics (BLS) Bulletin 917, published in 1947 (Kossoris 1947a,b). This report has been Mdely cited as a reliable source relative to construction ("Effect" 1974; "Scheduled" 1980; Owiler's 1979; "How" 1968; "Tables" 1968).
The 1947 BLS study involved 2,445 men and 1,060 women and covered 78 case studies at 34 facilities in a wide variety of manufacturing industries and settings such as foundries, machine shops, product packaging, and assembly and production lines. Production products included engines, airplanes, piston rings, metal bearings, airfield landing mats, hats and clothing, rubber hoses, office supplies, and cigars. Packaging activities included biochemicals, pharmaceuticals, and cough drops. Most of the work was indoors, some was machine paced, and most was highly repetitive. There were no construction activities included in the study.
The results of the study showed that efficiency was impaired as the work schedule exceeded 40 hr/week. The average efficiency for 50 hr, 60 hr, and 70 hr weeks was 0.92, 0.84, and 0.78, respectively. The degree of inefficiency was affected by the work schedule, the physical exertion required, and the pace of the machine. Comparisons to prewar situations in which the same work schedule was used were possible in 15 cases. The study found that efficiencies were better in all cases during the wartime period. The gain in efficiency ranged from 0.7% to 29.0%. The average gain was 13.6%. Thus, it appears that patriotism was a significant variable, and that the study results may not be applicable to other situations.
ANALYSIS OF HOURS PER DAY
Aside from determining the hours per week to be worked, one must also decide on the number of work hours per day. Common sense dictates that the rate of loss efficiency should accelerate nonlinearly as the length of the workday increases and as the number of hours per week increases.
Figs. 9-12 show the reported loss of efficiency as a function of the length of the workday and the number of hours per week. These figures provide somewhat conflicting and irregular results. Fig. 9 shows patterns that would be expected, although some might expect the differential between the 10 hr and 12 hr days to be greater. In Fig. 9, the differences appear to be almost linear. However, Figs. 10-12 show that the loss of efficiency is not related to the hours worked per day. In fact, Adrian's (1987) data show that in some ranges the longer workday is more efficient. Thus, the curves in Figs. 11 and 12 do not seem consistent with expectations.
The literature review reveals very little original data relative to construction overtime activities. Except for the CI I ("The Effects" 1988) and Adrian (1987) studies, all of the data originate from the 1960s or earlier. Very little is known about how the data were collected or the conditions under which the work was performed. Fig. 13 shows the reported efficiency for various 50 hr, 60 hr, and 70 hr workweeks. The basis for comparison is a 10 hr workday.
The results are generally consistent in that there is about a 10% increase in efficiency losses for each additional 10 hr per week added to the schedule beyond 40 hr. However, the conclusions that can be drawn from these data are somewhat uncertain, especially for the longer schedules.
Figs. 1 and 2 show the change in efficiency as a function of time for 50 hr/week and 60 hr/week schedules. The findings are generally consistent, although there is very little data on which to base conclusions.
The literature is inconsistent relative to the loss of efficiency as a function of the length of the workday. In one instance, the decrease in efficiency loss appears linearly related to the length of the workday; whereas in three other instances, loss of efficiency appears to be unrelated.
STUDIES OF DAYS PER WEEK
An increase in the number of hours worked can be made by increasing the hours worked per day, the number of days worked per week, or both.
Considerable uncertainty exists as to which is the most effective work schedule, although there is research to support the notion that workers need at least one day per week to relax (Kossoris 1947a,b).
Five-, Six-, and Seven-Day Workweeks
Common sense suggests that the longer the workweek and the more days per week, the greater the inefficiencies. The daily inefficiencies of starting up and winding down have been noted in other literature. Therefore, in choosing between two overtime schedules, both with the same number of hours per week, the better choice should logically be the one with the fewest days per week.
A number of the studies addressed the inefficiencies of working five days or more per week. These can be grouped into three general categories. The first category are those showing the expected trend of greater inefficiencies as the schedule extends to six and seven workdays. Fig. 14 shows the data from the NECA study (Overtime 1969). Using a 60 hr week as the basis for comparison, it shows that the seven-day schedule is about 7% less efficient than the six-day schedule.
The second category included data from the Foster Wheeler (O'Connor 1969), C. F. Braun Inc. (unpublished 1979), Adrian (1987), and Bureau of Labor Statistics (Kossoris 1947a,b) studies. Fig. 3, from the Foster Wheeler study, is typical; it shows that the number of days per week has little influence on the efficiency of construction operations. Figs. 15 and 16 actually show that the five-day schedule is less efficient than the six-day schedule when the number of hours worked per week is less than 55.
In the third category, Fig. 17, from the 1969 NECA survey, shows the longer workday schedules to be more efficient. The five-day schedule is less efficient than the six-day schedule. The curve for the seven-day schedule shows a 70 hr/week schedule to be more efficient than a 56 hr schedule.
In reviewing these data, one can readily discard the NECA survey data [(Overtime 1969) (Fig. 17)] as erroneous. While the results from the second category cannot be discarded, they can also be viewed with some degree of suspicion. Only Fig. 14 shows results that are consistent with generally accepted expectations and knowledge in the construction industry.
The Four-Day, 10 hr Schedule
The four-day, 10 hr/day schedule has gained popularity in the construction industry. An extension to this schedule is the rolling 4/10 schedule, in which two work forces alternate; that is, each work force is scheduled for four days on and three or four days off. In this way, the project can be manned continuously.
Based on the experiences of Daniel International (McConnell 1982), the 4/10 schedule has the following advantages.
|More productive work hours (less work starts and stops).|
|Reduced travel cost.|
|Increased employee morale.|
|Attraction of workers from a larger labor market.|
|Lower absenteeism and turnover.|
|The option to make up rain-out days on Friday, thus avoiding the expense of weekend work schedules.|
The disadvantages of the 4/10 schedule are the following.
|Engineering firms and the owner's administrative personnel operate on a different schedule.|
|Owner personnel may have to adjust the normal work schedule.|
Two studies have cited certain advantages and disadvantages of the 4/10 schedule in quantitative and qualitative terms. These are as follows.
First, an in-house study by Daniel International (The Four- Ten 1979 based on 10 years of experience with the 4/10 schedule, concluded that the 4/10 schedule was superior to other schedules. This conclusion resulted from both quantitative and qualitative analyses.
The quantitative analysis was a 20-week work-sampling study at a test site in the southeastern United States during the period from November 1975 to mid-March 1977. During this time, two schedules were used: a 4/10 and a 5/8. The work-sampling results are summarized in Table 4.
The study also examined the level of activity during the beginning and end of the shift. The percentage of direct activity during the first and eighth hours on the 5/8 schedule was no different than the ninth and tenth hours on the 4/10 schedule. While direct work activity is not well correlated to labor productivity (Thomas 1991), it would appear that differences in the level of activity on the site as a function of the schedule are insignificant.
Second, as part of the overtime study by CII ("The Effects" 1988), one project in the study used a rolling 4/10 schedule. The project was the construction of a chemical processing unit. Data were collected on crews installing pipe, conduit, tubing, terminations, cables, wire, and feeders.
The study made no comparison between the 4/10 schedule and any other. The results showed that during the seven-month duration of the study, the general productivity trend for all of the crews did not change, although there were brief periods of improvement and deterioration. The report made no assessment of the efficiency of the 4/10 schedule compared to other schedule alternatives.
Overall, the studies of different work schedules are inconclusive as to which alternative schedule is most efficient. Several studies show that a shorter workweek is best; and one study shows that longer workweeks are preferable. However, most studies show very little difference.
With respect to the 4/10 schedule, only one source was identified. Daniel International (The Four- Ten 1979) concluded that the 4/10 schedule was more efficient than a normal schedule of five 8 hr days.
The literature on scheduled overtime was found to be very sparse; dated to the late 1960s and earlier; based on small sample sizes; and largely developed from questionable or unknown sources. Although there appears to be a number of data sources, this is an illusion because many of the articles and publications quote other sources while providing no new data or insight. Where the data source is known, other pertinent information, such as the environmental and site conditions, quality of management and supervision, and labor situation, is unknown. The various graphs and data that have been published are inherently unreliable, except perhaps to suggest an upper bound on the losses of efficiency that might be expected. The literature offers no guidance as to what circumstances may lead to losses of efficiency.
With respect to the loss of efficiency as a function of the number of hours per day and the number of days per week, the literature provides strange and largely unbelievable results. The results from the southeastern Michigan NECA study (Overtime 1969) and as reported in Qualified Contractor (Howerton 1969) both show losses of efficiency as the length of the workday increases and as more days per week are worked. These data suggest that the six-day week is about 7% (absolute) more efficient than the seven-day week. The 10 hr workday compared to a 9 hr workday results in a loss of efficiency of about 4% (absolute). The 12 hr day results in a loss of efficiency of another 7-8% (absolute).
The studies by Foster Wheeler, C. F. Braun Inc., Adrian, the Bureau of Labor Statistics, and the Southeastern Michigan NECA survey show that efficiency is not related to the number of work hours per day or the number of workdays per week. Since it is not possible to increase the total hours per week in any other way, it is concluded that these studies are flawed.
This paper is based on research sponsored by the Construction Industry Institute (CII). The support of the Overtime Task Force is gratefully appreciated.
Adrian, J. J. (1987). Construction productivity improvement. Elsevier Science Publishing Co., New York, N.Y.
"Effects of scheduled overtime on construction projects." (1973). Report, American Association of Cost Engineers, Morgantown, W.Va.
"Effect of scheduled overtime on construction projects." (1974). Coming to grills with some major problems in the construction industry, Business Roundtable, New York, N.Y., 1-14.
Howerton, J. (1969). "Do you know the hidden costs of overtime'?" Qualified Contractor, (Mar.).
"How much does overtime really cost." (1968). Bulletin No. 18A, Mechanical Contractors Association of America, Washington, D.C.
Kossoris, M. (1947a). "Studies of the effects of the long working hours." Bulletin 917, Bureau of Labor Statistics, Washington, D.C.
Kossoris, M. (1947b). "Studies of the effects of the long working hours." Bulletin 917A, Bureau of Labor Statistics, Washington, D.C.
McConnell, G. (1982). "Evaluation of six alternative work schedules." Fluor-Daniel Construction Co., Inc., Greenville, S.C.
O'Connor, L. V. (1969). "Overcoming the problems of construction scheduling on large central station boilers." Proc., American Power Conf., (31), 518-528.
Oglesby, C., Parker, H., and Howell, G. (1989). Productivity improvement in construction. McGraw-Hill Book Co., Inc., New York, N.Y.
Overtime and productivity in electrical construction. (1969). National Electrical Contractors Association, Washington, D.C.
Overtime and productivity in electrical construction, 2nd Ed. (1989). National Electrical Contractors Association, Bethesda, Md.
Overtime work efficiency survey. (1962). National Electrical Contractor Association, Washington, D.C.
Owner's guide on overtime, construction costs and productivity. (1979). American Subcontractors Association, Association General Contractors of America, and Associated Specialty Contractors, Inc., Washington, D.C.
"Scheduled overtime effect on construction projects." (1980). Construction Industry Cost Effectiveness Task Force Report C-2, Business Roundtable, New York, N.Y. "Tables for calculation of premium time and inefficiency on overtime work." (1968). Bulletin No. 20, Mechanical Contractors Association of America, Washington, D. C.
"The effects of scheduled overtime and shift schedule on construction craft productivity." (1988). Source Document 43, Construction Industry Institute, Austin, Tex. 'The four-ten hourlday work week. (1979). Daniel International Construction Company ' Inc., Greenville, S.C.
Thomas, H. R. (1990). "Effects of scheduled overtime on labor productivity: A literature review and analysis." PTI Report No. 9107, The Pennsylvania State University, University Park, Pa.
Thomas, H. R. (1991). "Labor productivity and work sampling: The bottom line." J. Constr. Engrg. and Mgmt., ASCE, 117(3), 423-444.