desert_back.gif (3849 bytes)

Action-Response Model and Loss of Productivity in Construction

by David W. Halligan,1 Member, ASCE

Laura A. Demsetz,2 Associate Member, Member, ASCE,

James D. Brown,3 Member, ASCE, and Clark B. Pace4

 

ABSTRACT: Unanticipated conditions on a construction project sometimes result in a significant loss of activity. The conditions that may affect productivity adverse weather, scheduled overtime, and material shortages, to name a few examples-have received considerable attention in the literature. However, both experience and a detail examination of this literature reveal that these conditions do not necessarily lead to a loss of productivity. Furthermore, when such losses are observed, their extent varies from project to project, from activity to activity, and from crew to crew. Other factors and processes must therefore be involved. T'he action-response model of productivity loss in construction identifies these factors and processes. The model graphically depicts how a variety of factors may interact to cause a loss of productivity; how a crew is influenced by these factors; and how the management of crews can mitigate, eliminate, initiate, Or exacerbate any particular loss of productivity. By applying the model to three case studies, a new approach to evaluating loss of productivity is illustrated.

1Proj. Controls Mgr., O'Brien-Kreitzberg & Assoc., Inc., Bayside Plaza, 188 The Embarcadero, San Francisco, CA 94105.

2Asst. Prof., Dept. of Ov. Engrg., Univ. of California, Berkeley, CA 94720.

3Vice Pres., O'Bfien-Kreitzberg & Assoc., Inc., Bayside Plaza, 188 The Embarcadero, San Francisco, CA 94105.

4Grad. Student, Dept. of Civ. Engrg., Univ. of California, Berkeley, CA 94720.

Note. Discussion open until August 1, 1991. 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 November 12, 1992. This paper is part of the Journal of Construction Engineering and Management, Vol. 120, No- 1, March, 1994. CASCE, ISSN 0733-9364194/0001-0047/$2.00 + $.25 per page. Paper No. 5094.

 

INTRODUCTION

Unanticipated conditions on a construction project sometimes result in a significant loss of productivity. Adverse weather, scheduled overtime, and material shortages, for example, have been shown to result in levels of craft productivity lower than would otherwise be expected. However, both experience and the research literature show that a loss of productivity is not always or necessarily the result of such unanticipated conditions.

Given a particular unanticipated condition on a construction project, and given an observed loss of productivity, it is easy to conclude that the condition or event in question led to the loss. Such conclusions may be in error. The cause-and-effect relationships leading to productivity loss in construction are complex, and the methods of analysis commonly used, while comprising an essential step in the analytical process, often fail to account for this complexity.

This paper provides a model for analyzing the complex cause-and-effect relationships that may lead to a loss of productivity in construction. This model is called the action-response model because productivity, or the lack of it, is a crew's response to external events and the actions of others. The action-response model provides a new approach to evaluating specific occurrences of a loss of productivity.

This paper begins by defining productivity and productivity loss. Two common approaches to identifying the causes of productivity loss-plots of productivity over time and reference to published studies-are examined in detail to illustrate that current approaches often oversimplify the complex nature of productivity loss in construction. The action-response model of productivity loss in construction is then presented. Its structure and underlying principles are explained, and its ability to graphically represent a loss of productivity is shown. Finally, three case studies are presented to illustrate how the action-response model has been used in practice.

PRODUCTIVITY

Productivity can be defined in many ways. In construction, productivity is usually taken to mean labor productivity, that is, units of work placed or produced per man-hour. This measure of productivity has several advantages: the meaning of the term labor productivity is relatively well understood; labor productivity is often the greatest source of variation in overall construction productivity; and the productivity of other inputs can often be measured with respect to labor productivity. The inverse of labor productivity, man-hours per unit (unit rate), is also commonly used.

Labor productivity is sometimes measured by proxy. One such proxy is installation rate, that is, units of work in place per unit time. Another common proxy is to measure man-hours expended per week or month. However, it is important to note that such proxies do not directly measure productivity and can therefore give misleading results if not carefully used. For example, man-hours per week or month indicates only the intensity of effort, not productivity.

Here productivity is defined as labor productivity, units of work accomplished per man-hour. In the literature, the term productivity is often used interchangeably with the term efficiency. In keeping with this convention, this paper makes no distinction between productivity and efficiency.

Labor productivity for a particular activity is often treated as a single value. However, productivity is better understood as a quantity that varies throughout the duration of an activity. Fig. 1 illustrates this distinction. A single-value estimate of productivity is typically used in preparing a bid;. In contrast, the measured value of productivity will vary throughout the job, at any given time the measured productivity may be close to the estimate or very different. Measured productivity is sometimes shown as a smooth curve; however, such curves are actually the result of a series of discrete measurements that are typically made at fixed intervals, for example, daily or weekly.

LOSS OF PRODUCTIVITY

Loss of productivity is defined here as the reduction in productivity caused by unanticipated conditions. Such conditions may include adverse weather, scheduled overtime, and material delivery problems. In other words, a loss of productivity is the difference between the productivity actually observed and the productivity that might reasonably have been expected if not for the unanticipated condition. It should be note6 that an overly optimistic or unreasonable estimate can lead to the appearance of a loss of productivity when, in fact, no such loss has occurred. In these situations, observed productivity is the productivity that should have been anticipated by a realistic estimate; despite appearances, no loss of productivity has occurred.

Two steps are required to evaluate a loss of productivity. First, it must be demonstrated that a loss occurred. Several techniques exist for demonstrating such losses. These techniques and their relative merits are described elsewhere (Demsetz et al. 1992). The second step in evaluating a loss of productivity is to determine what event or events caused the loss. In other words, it is necessary to show that a particular event led to the observed loss of productivity-to establish cause and effect.

Common techniques for establishing cause and effect are to generate a productivity-time plot and to reference published productivity studies. These techniques are now examined in detail.

PRODUCTIVITY-TIME PLOTS

A productivity time-plot has already been presented in Fig. 1. This figure shows variation of productivity over time. In this instance, the measured productivity is close to the estimated productivity. Fig. 2 shows an example where the measured productivity, curve A, is significantly lower during the second half of the performance period than during the first half. Note that the transition from expected to lower than expected productivity corresponds to the occurrence of some unspecified event. If curve B represents a realistic estimated productivity, and if the event shown led to the loss of productivity, then curve C may represent a reasonable approximation of what the measured productivity might have been if not for the occurrence of the event. It should be noted that curve C cannot be measured but only estimated or approximated. This curve therefore represents what is called here an as-might-have-been productivity.

plfig1.tif (63482 bytes) plfig2.tif (74654 bytes)

A productivity-time plot is sometimes developed to evaluate a loss of productivity without consideration of the following two points. First, it must be established that the estimate, curve B, was reasonable (Demsetz et al., in press, 1994). Second, when an event corresponds in time with the loss of productivity, it must be shown that the event noted caused the loss.

The simultaneous existence of an observed loss of productivity and some event does not establish that the event caused the loss. To establish a connection between an event and an observed loss of productivity, it is common for reference to be made to published productivity studies of similar events. However, the fact that on other projects a particular event has been shown to cause loss of productivity does not imply that the same loss has occurred on the project under consideration.

POSSIBLE CAUSES OF PRODUCTIVITY LOSS

Factors and events frequently cited as causing a loss of productivity include adverse weather, scheduled overtime, disruption, out-of-sequence work, congestion, dilution of supervision, and unavailability of manpower. Other factors frequently cited are those associated with worker motivation. Summaries of these and other factors are available (Borcherding and Alarcon 1991; Thomas 1991; Demsetz et al., in press, 1994).

Here, selected studies on adverse weather, scheduled overtime, and disruption are examined to show that the results of these studies vary widely and are in some instances contradictory. It is therefore unclear how to apply the conclusions of any one study or group of studies to any particular loss of productivity. Additional considerations in using such studies are described elsewhere (Demsetz et al. 1992). While useful for drawing some general conclusions or for estimating or forecasting the potential effect that certain factors might have on productivity, productivity studies are typically of limited value for establishing cause and effect for specific losses. Depending on individual project circumstances, any particular factor may or may not result in a loss of productivity. When such losses are observed, their magnitude varies from project to project, from activity to activity, and from crew to crew.

This section also reviews studies of worker motivation and productivity. Worker motivation in conjunction with or in the absence of other factors can play a significant role in determining the productivity experienced on any particular project. Therefore, it will often be insufficient to consider any one factor without also investigating the impact of worker motivation.

Adverse Weather and Environmental Conditions

Precipitation, wind, and extremes of temperature and humidity may lead to reduced performance due to both physiological and psychological factors. Common sense and experience show that it is harder to work in conditions that are very hot, very cold, or very humid, or when it is raining, snowing, or extremely windy. The extent to which productivity is affected by adverse weather depends on several factors, including the severity of conditions, the nature of the task, the individuals involved, their acclimatization and training.

Several studies have been undertaken to quantify the possible effects of adverse weather. These studies report widely varying effects and different "ideal" ranges of temperature and humidity. A 1966 study of residential construction found that work was frequently halted due to bad weather. But that no reduction in productivity occurred when conditions were such that work continued in spite of bad weather [Clapp 1966, cited in Yiakoumis (1986)]. In a 1972 study, the output of masons constructing test walls was measured over a nine-month period. Productivity was found to decline as much as 40% as conditions deviated from 75'F (24'C) and 60% relative humidity (Grimm and Wagner 1974). In another study, electricians installed duplex receptacles in an environmental chamber where temperature and humidity could be controlled. Productivity was found to decrease at temperatures above 80 F (27 C) and below 40 F (4 C), and at relative humidities above 80% ("The effect" 1974). More recently, in a several-month study of productivity in the installation of structural steel, masonry, and formwork, it was found that the "ideal" temperature was 55 F (13 C), with relative humidity having marginal effects below 80%, but reducing productivity above this level (Yiakoumis 1986). Additional reports on the possible effects of adverse weather or environmental conditions have been published by others (Dallavia 1954; Doyle 1979; Neil 1982; Neil and Knack 1984; Koehn 1985).

The influence of temperature and humidity varies a great deal by individual, and is affected by attitude and motivation ("The effect" 1974) and by the type of work being carried out (Oglesby et al. 1989). Worker acclimatization can also be an important factor. For example, acclimatization techniques have been used in mining to increase productive time by as much as 100% at 95 F (35 C) (Stewart 1982).

Scheduled Overtime

A number of publications report a loss of productivity when work is scheduled beyond 40 hr per week and/or beyond 8 hr per day. However, the extent of the reported productivity loss varies from study to study. Losses are generally attributed to fatigue and to a decrease in worker motivation. Therefore, to the extent that effective management can keep workers motivated, it should be possible to minimize the impact of overtime. Studies of overtime address only scheduled overtime occurring over an extended period of time. The issue of sporadic or occasional overtime has not been addressed. It has not been shown that sporadic or occasional overtime has an effect on productivity.

The premiere overtime studies, upon which much other published work relies, were carried out by Kossoris for the Bureau of Labor Statistics (BLS) in the mid- to late-19408-. Theme studies surveyed the effects of scheduled overtime on workers in six metal-working plants during World War 11, and Kossoris found that efficiency dropped with overtime in four of the six plants, but did not change in the other two ("Studies" 1944a). Studies of six additional plants showed a decrease in productivity with longer hours that was more noticeable among workers whose wages were based on output than among those working for an hourly wage ("Studies" 1944b). After studying 35 plants in various manufacturing industries, Kossoris concluded that

The results of the same increase in hours may vary widely according to the physical exertion required and the degree of control which the worker has over the job. [I]f the worker can retard or speed up production only during relatively minor intervals of time ... the controlling factor is the machine. If, on the other hand, the machine works as fast or faster than the operator, then efficiency definitely depends on the exertion of the worker. It is under these types of conditions, or the similar ones in which no machines are involved at all, that the fatiguing factor of long hours comes into play (Hours 1947).

However, Kossoris also noted that

There is no such thing as an "optimum hour schedule" for all of industry. What appears to be a satisfactory schedule of hours for a plant with light machining operations may be economically wasteful in a foundry. Further, there is a marked difference in the performance of men working under wage incentives and those working at straight hourly rates. Much depends on the type of work and the requirements it exacts from worker, the degree to which workers can control the speed of operations, and the incentives which motivate them whether volume of pay, participation in the war effort, labor relations, or working conditions generally (Studies 1944b).

Kossoris studied manufacturing operations nearly 50 years ago during wartime, a very different situation than construction work today. In addition, all of Kossoris's studies were based on extended, scheduled overtime, and exclude a one- to three-month "transition period" between different work schedules. Omission of this transition period may be significant in extending Kossoris's results to construction, since the period of scheduled overtime in construction is often less than three months. Many publications on overtime in construction provide tables and figures based on Kossoris's studies of manufacturing (How Much 1987).

In 1969 and again in 1989, the National Electrical Contractor's Association (NECA) reported the results of a 1964 survey conducted by their Southeastern Michigan Chapter. The survey addressed changes in productivity for up to a month of scheduled overtime. Productivity was reported to decrease as the number of work hours per week increased and as the number of weeks on overtime increased ("Overtime" 1989). The report notes that these results are based on "analysis of records"; however, no further information could be obtained from NECA as to the method of analysis used, the types of jobs considered, or authorship of the report.

The frequently cited Business Roundtable report on overtime ("Scheduled" 1980) relies upon data from an early study of industrial fatigue [Florence 1918, in Scheduled (1989)], from the BLS studies by Kossoris ("H-Ours" 1947), and from a survey conducted by the National Constructor's Association in the late 1960s. In some of the examples given, overtime is said to have caused such a drop in productivity that the total output becomes lower than it would have been under a standard work week. In this case, continuing with scheduled overtime would appear to be a poor management decision.

More recently, the Construction Industry Institute (CII) sponsored a study on the construction of seven industrial projects using scheduled overtime. The study concluded that productivity does not necessarily decrease under scheduled overtime ("The effects" 1988). Even on the same project, productivity trends of individual crews were not consistent, reinforcing Kossoris's comments on the importance of motivation and specific work conditions. The CII study postulated that productivity losses under overtime schedules may in large proportion be due to the effects of poor management. Additional reports on the possible effects of scheduled overtime on worker productivity have been published by others (O'Connor 1969; "Factors" 1976; "Modification" 1979; Doyle 1979; Neil 1982; Neil and Knack 1984). Additional analysis of the published research on the effects of scheduled overtime emphasizes the tenuous relationship between scheduled overtime and loss of productivity (Thomas 1992).

Disruptions

Interruptions to work in progress can reduce productivity. It is useful to consider two types of disruptions: short duration and long duration. In a study of short duration disruptions of piping insulation installation, productivity was reduced by 70% when work was disturbed by two or more interruptions per section of pipe (Hester 1987). The relationship between productivity and length of a short disruption will vary according to many factors, including the type of work involved (Thomas 1990). Interruptions of longer duration are also cited as a cause of productivity loss. A 1980 survey of mechanical and electrical contractors found that 87% of total cost overruns were due to delays or disruptions (Guide 1980). However, it should be noted that of 7,500 surveys sent out as part of this study, only 77 were returned.

The term disruption, in and of itself, is overly broad. In establishing a loss of productivity, the specific impacts of a disruption need to be identified. For example, a long disruption or delay may interrupt productivity increases due to learning, or it may cause the most skilled workers to leave the job and become unavailable for rehire. It is these latter, consequential effects of a disruption that need to be shown and quantified in establishing a loss of productivity.

The effect that disruptions can have on productivity may in many cases be the result of a failure to properly manage the unanticipated conditions. For example, if disrupted material deliveries are causing crew idle time, and a resultant loss of productivity, it may be possible to reduce crew size to eliminate the idle time. The activity duration may be extended, but no loss of productivity would result. The study of insulation installation found that many of the disruptions were due to crew reassignments as a result of unanticipated situations requiring immediate attention. By creating a separate crew whose sole purpose would be to respond to these urgent situations, one of the major causes of disruption to the production work could have been eliminated (Hester 1987).

Additional Factors

Factors such as low morale, poor supervision, poor training, and unsafe working conditions are generally related to worker motivation. A great deal of research has been carried out on the factors that motivate construction workers (Borcherding and Oglesby 1974; Borcherding et al. 1980; Borcherding and Garner 1981; Maloney 1983; Maloney and McFillen 1985, 1986). Summaries of these factors are also available (Warren 1989). Other studies have shown the effect that management (starting with the foreman) can have on crew performance. For example, a survey of 703 construction workers showed that foremen have "a strong impact on worker motivation, performance, and satisfaction" (Maloney and McFillen 1987). The relationship between productivity and a foremen's management style has also been recognized (Hinze and Kuechenmeister 1981; E6mna et al. 1986). Another study found "poor supervision. . . . poor planning, and generally poor management" to be major causes of absenteeism and turnover ("Absenteeism" 1982).

The organization and management of the site can have a great impact on productivity. In a recent study, ineffective material management practices were estimated to have caused an 18% work-hour overrun on a commercial project (Thomas et al. 1989). Motivation and productivity can be further reduced by unsafe working conditions. It has been estimated that poor safety practices increase construction costs by roughly 6.5% (More 1983). Further discussions of construction safety and productivity are available (Hinze and Parker 1978; Levitt and Samelson 1987).

Although they have not been the subject of detailed study, other factors including out-of-sequence work, congestion, dilution of supervision, and unavailability of manpower are frequently cited as potential causes of productivity loss (Neil 1982; "Change" 1987). In some instances, acceleration, change orders, and delays are cited as causes of productivity loss. However, these conditions do not themselves cause a loss of productivity. Rather, they may in turn lead to any of the factors discussed in this section.

Summary of Causes of Productivity Loss

A variety of factors may lead to a loss of productivity. In many instances, more than one factor may be at work. No review or study could address all possible causes or combinations of causes. However, the factors discussed in the previous sections are those most commonly cited, and those for which the most extensive documentation is available.

The research published to date on the causes of loss of productivity has focused on establishing whether or not certain factors, events, or conditions have the potential to cause a significant loss of productivity. These studies indicate the kinds of losses that perhaps in general or on average might be experienced. Such results are useful for estimating or forecasting. However, these same results are typically of limited value in determining the cause of a loss of productivity for any given project.

There are many factors that may lead to a loss of productivity, and many of these factors may interact to cause a loss. In addition, the factors frequently cited as causing a loss of productivity do not always or necessarily do so. When such factors occur, their impact varies from project to project, from activity to activity, and from crew to crew. Additional factors and processes must therefore be involved. The action-response model describes these additional factors and processes.

ACTION-RESPONSE MODEL OF PRODUCTIVITY LOSS

The action-response model of productivity loss graphically depicts how a variety of factors may interact to cause a loss of productivity. The model also shows how a crew and hence productivity is influenced by these factors. Additionally, the model portrays how the management of crews can mitigate, eliminate, initiate, or exacerbate any particular loss of productivity. As a result, the action-response model is able to represent why the factors frequently cited as causing a loss of productivity do not always do so. The model also provides an explanation of why such losses, when they are observed, can be expected to vary widely.

Basic Relationships for Model of Productivity Loss

The starting point for the action-response model is output or production. Physical production and thus productivity and productivity loss occur at the crew level. Modeling construction productivity at the crew level has been strongly advocated by others (Thomas et al. 1990). Productivity, or the lack of it, is a crew's response to a variety of factors, including actions of the contractor and the owner. The model developed here is therefore referred to as the action-response model.

The basic relationships for an action-response model of productivity loss are shown in Fig. 3. Fig. 3 has three components: external conditions, contractor actions, and crew responses. External conditions are defined as events, situations, and decisions beyond the control of both the contractor and its crews. Fig. 3 shows that a crew responds to both external conditions and to the actions of the contractor. Similarly, the contractor takes action in response to both external conditions and to the performance of its crews.

plfig3.tif (71392 bytes)

The basic relationships shown in Fig. 3 are expanded into the more detailed model shown in Fig. 4. External conditions are divided into two groups. The first group consists of those conditions that should come to the attention of a contractor's management before the crew is affected. Examples of this first group include owner actions such as design changes, third-party actions such as regulatory changes, and force-majeure events. The second group consists of those conditions that may directly affect crew performance before management is alerted to a potential problem. Weather and environmental conditions are typical examples of this type of external condition.

In Fig. 4, the contractor's actions are also divided into two groups. The first group includes those actions taken as initial steps in managing a job. Planning, supervisor selection, crew training, and site management fail into this group. Poor performance of any of these functions can ultimately cause poor productivity. The second group of contractor actions are the management actions taken during the job in response to either external conditions or to crew performance differing from that expected. The extent to which a contractor is aware of crew performance (and of deviations from anticipated crew performance) depends on the extent to which productivity, cost, and schedule are monitored at regular intervals throughout the job,

plfig4.tif (159854 bytes)

Action-Response Model

A model of productivity loss is now built by expanding the relationships and components shown in Fig. 4 to include those factors frequently cited in the literature as causing productivity loss. The result is the action-response model shown in Fig. 5.

The action-response model has six components: initiating events, management-level constraints, crew-level constraints, contractor's management actions, consequences of management actions, and crew responses. The arrows between the components indicate a "may lead to" relationship. For example, design changes may cause a delay, but will not necessarily do so; a change in schedule may lead to out-of-sequence work, but will not necessarily do so.

By including those factors frequently cited as causing a loss of productivity, Fig. 5 extends Fig. 4 in several important ways. For convenience, external conditions and the contractor's initial actions now appear together as a group called initiating events. In addition, Fig. 5 contains three new components. These are management-level constraints, crew-level constraints, and consequences of management actions. These three components incorporate into the model the factors frequently cited in loss of productivity claims. The following section describes how the action-response model represents the origins of productivity loss.

plfig5.tif (228059 bytes)

Initiating Events

Initiating events are actions that initiate a sequence of events that may ultimately lead to reduced productivity. These actions can be within the owner's or contractor's control or beyond the control of either party. Here, there are four subcategories of initiating events: owner actions, force majeure and third-party actions, environmental conditions, and contractor's initial actions.

Owner Actions

Examples of owner-caused initiating events include untimely design changes, slow response to requests for information (RFIS) and change order requests (CORs). These events in turn may result in delays, disruptions, changes, or acceleration to the project.

Force Majeure and Third Party Actions

Events beyond the control of either the owner or the contractor are usually either force-majeure events or third-party actions. Examples include extreme weather conditions that could not be reasonably anticipated, tornadoes, floods, strikes, and changes in regulatory requirements. Such events may result in delays, disruptions, changes, or acceleration to the project.

Environmental Conditions

Environmental conditions such as extreme temperatures, humidity, precipitation, and dust can result in difficult working conditions and can cause workers to fatigue more easily. In the existing literature, weather, the major subset of environmental conditions, is typically treated as a direct cause of productivity loss. However, Fig. 5 shows instead that weather may lead to difficult working conditions, which in turn may lead to crew responses such as fatigue, slowed pace of work, idle time, or poor-quality work.

Contractor Initial Actions

The contractor can initiate a sequence of events that leads to reduced productivity. For example, poor site management or poor coordination of subcontractors may lead to difficult working conditions or an unavailability of resources.

Management-Level and Crew-Level Constraints

Project constraints fall into two categories. Management-level constraints include delays, disruptions, changes, and acceleration and are typically caused by conditions external to the contractor. The contractor is usually made aware of management-level constraints directly, for example, through a directive or change order. Crew-level constraints include difficult working conditions, unavailability of resources, and an unsuitable workforce. These constraints are typically caused either by the contractor's actions or by environmental conditions. The contractor may not become aware of crew level constraints unless crew productivity is observed or measured.

Contractor's Management Actions

Whether the contractor is made aware of an initiating event through management-level constraints or by observing or measuring the crew's productivity, a variety of actions can be taken. These actions include adding or changing resources (for example, hiring additional workers), changing the schedule (for example, scheduling overtime or resequencing activities), and modifying the work method. In addition, the contractor can respond by taking no action, that is, by making no change in resources, schedule, or work method. Any of these actions, if appropriate and implemented properly, may mitigate or eliminate loss of productivity. However, if the selected action is inappropriate or is implemented poorly, additional constraints may be placed on the project. For example, a decision to increase manhours by adding extra crews without a corresponding increase in tools may result in a lack of resources (tools) which in turn may result in idle time. (This example arose from a discussion with Gregory A. Howell on the importance of providing appropriate support to a crew in order to ensure good productivity.)

Consequences of Management Actions

Depending on the specifics of the job and on the way the contractor chooses to implement a management action, a variety of consequences are possible. These consequences include increased workload, crowding of workers, stacking of trades, dilution of supervision, performing work out of-sequence or performing rework. Each of these consequences may in turn have a negative effect on the crew's ability to work productively. In the existing literature, management actions and their consequences are commonly treated as direct causes of loss of productivity. They are not. Rather, each action may in turn result in difficult working conditions, unavailability of resources, or an unsuitable work force. It is these conditions that directly affect the crew and cause loss of productivity. In contrast, when properly conceived and executed, management actions may actually increase productivity. Much of the literature on productivity loss often fails to mention that a possible consequence of management action can be an improvement in productivity.

Crew Responses

Loss of productivity occurs at the crew level. It is at the crew level that physical work is performed. Difficult working conditions, unavailable resources, and an unsuitable workforce can each cause slowed pace of work, absenteeism, worker turnover, idle time, fatigue, loss of motivation, or poor quality work. Such responses by a crew may be directly observed by the contractor or identified through an appropriate system of measuring and tracking productivity. The contractor, having observed a change in productivity, may then elect to take management action. This action may improve the situation or cause further problems with working conditions, leading to further reductions in the crew's performance. In such cases, spiraling loss of productivity may result.

PRACTICAL APPLICATIONS

The action-response model shows possible avenues of cause and effect in productivity loss. As a result, the model can be used to identify the causes of a particular loss of productivity. The following three case studies illustrate how the action-response model has been used as a practical tool to evaluate a loss of productivity.

Case Study One

A general contractor spent $5,000,000 constructing a bridge bid at $2,000,000. The contractor believed that $1,000,000 of the total $3,000,000 overrun, and a corresponding nine-month delay in project completion, was due to design revisions and the associated change orders. Most importantly, the contractor asserted a slowed pace of work and idle time on unchanged work due to disruption caused by untimely design changes and change orders.

There was no question that labor hours exceeded the estimate and that productivity was below that anticipated by the contractor. The question was why. An examination of Fig. 5 helped lead to the correct explanation. Design changes were the owner action asserted as the cause of the observed productivity loss. These design changes caused delays and changes to the work, and in response the contractor's management action was to change the schedule. However, none of the consequences of the contractor's management actions that could lead to a loss of productivity were evident. In particular, there was virtually no increased workload due to overtime, no crowding of workers or stacking of trades, no dilution of supervision or loss of experienced personnel. There was no out-of-sequence work and no significant rework performed. It was concluded that it was not the changes in schedule that led to poor crew performance.

The action-response model shows that if there were no consequences from the contractor's change of schedule, then lost productivity must be due to either environmental conditions, to the contractor's initial actions, or to contractor actions taken in response to poor crew performance. After interviewing individuals familiar with the project, it became clear that the low productivity on the job was due to poor planning, poor management, and poor training of crews by the contractor. The contractor's job superintendent had no experience in most of the skills and trades involved This inexperienced employee gave inadequate directions to the workforce, which led to a slowed pace of work. Additionally, the contractor had limited experience in coordinating subcontractors, which led to idle time for crews. Finally, the contractor did not properly schedule vendor submittal and shop drawing reviews, which led to late procurement of critical items.

This case study illustrates the following points. First, loss of productivity is a multistep process in which each step must be evaluated before the cause of any particular loss can be identified. Second, the existence of a cause and an effect does not imply a causal relationship-other factors may be at work. Finally, this case study illustrates how a contractor's own planning and management actions can significantly reduce productivity.

Case Study Two

A large prefabricated facility was to be shipped and erected by modules in a remote location. The construction season in this location was short, and work not completed in one season would have to be delayed eight months until the next season. The original plan had called for module erection to occur over two seasons, but the owner subsequently accelerated the schedule in order to have the work performed in one season. To accomplish the work under the changed schedule, the contractor added a second shift and added workers to both the first and second shifts. A change order was formulated based on the added labor hours necessary. These added hours were calculated using the estimated productivity for the smaller single shift originally planned.

The contractor's labor hours greatly exceeded those anticipated in the change order. This project was based on a cost-plus-fixed-fee agreement, and when labor hours exceeded -those anticipated in the change order, the question was raised as to whether or not these additional hours were reimbursable. The contractor argued that the additional hours were reimbursable, believing that adding additional workers to each shift led to an unavoidable crowding of workers and a resultant loss of productivity. By using the action-response model it was determined that the contractor was correct.

The contractor was experienced in the work involved, and the original estimate was evaluated and deemed adequate. An examination of the environmental conditions indicated no unusual conditions. Both the contractor and its workforce had worked in the area for a long period of time. The work was similar to other jobs successfully performed by the contractor, and the job was properly organized. Additionally, the contractor added the appropriate supervisory staff, resources, and office support to provide for its increased workforce, Investigation of the contractor's daily reports and time cards indicated that there was, however, a crowding of workers, which resulted in idle time and a slowed pace of work. By using the action-response model, all possible causes of productivity loss were eliminated except crowding of workers.

This case study illustrates another important aspect of analyzing loss of productivity. It is not sufficient to simply establish that a particular factor caused at least a portion of an observed loss of productivity. It may be equally important to show that other factors did not contribute to the loss.

Case Study Three

A structural steel detailer spent 18,000 hr detailing a job for which 10,000 hr had been estimated. During the job, the detailer encountered instances of missing and incomplete design information. The detailer submitted many RFIs to clarify design issues. Furthermore, the project was subject to a number of change orders (COs). The detailer believed that the 8,000-hr overrun in detailing hours was due to a loss of productivity caused by the disruption of RFIs and COs. Applying the action-response model of productivity loss to this situation resulted in the following conclusions.

The owner action was to issue design drawings which required RFIs and COs. The resulting management-level constraints were disruption and changes to the work to be detailed. The detailer's management action was to change its schedule, resulting in a resequencing of its work. The crew-level constraint thus created was believed to be difficult working conditions.

However, reduced productivity at the detailer's "crew" level was not to be found. There was no slowed pace of work, no unusual absenteeism or turnover, and little idle time. There was some poor quality work performed, but this had little, if any, impact on productivity. In fact, the detailer's actual productivity (drawings per hour), which had been measured on a weekly basis, was better than the estimated productivity throughout most of the job. This better than expected productivity was observed both in unimpacted periods and in periods impacted by RFIs and COs. It was only at the beginning of the job that productivity was below the original estimate. In fact, low productivity at the beginning of this job had been expected. The contractor correctly anticipated that lower than average productivity would continue until a certain percentage of the drawings had been produced. Only then would an "efficient" phase of the work be reached.

Examining the contractor's initial actions led to valuable insights into the true problems on the job. For this job the detailer had grossly underestimated the number of drawings to be produced. This underestimation lead to a longer than expected period of low production at the start of the job. Furthermore, once an efficient production level was reached, there was simply more work to be done than had been anticipated. However, there was no loss of productivity on the job.

This last case study illustrates two of the most important aspects of analyzing loss of productivity. First, the performance of the crew must be examined to determine whether or not a loss actually occurred. While this point may seem self-evident, it is often overlooked. Second, the performance of the crew can only be properly evaluated if productivity is measured at frequent intervals (in this case, weekly) throughout the project.

CONCLUSIONS

Research conducted to date on the causes of loss of productivity in construction has focused on showing whether or not certain factors, events, or conditions have the potential to cause a loss of productivity, These studies indicate the kinds of losses that in general or on average might be expected. Such results are useful for estimating and forecasting, and for identifying issues worthy of further attention. However, these results are typically of limited value in determining the cause of a loss of productivity for any given project.

The action-response model defines a new approach to evaluating the causes of productivity loss. This new approach is an improvement over current practice in the way it addresses the following six issues. First, the new model recognizes the importance of focusing on the crew in any discussion of productivity. Second, the model shows the extent to which productivity loss at the crew level may be removed from initiating events. Third, the model represents the existence of the may lead to relationships that exist in productivity loss. Fourth, the model shows the contractor's active role in influencing productivity through management decisions. Fifth, the model makes clear the two ways in which a contractor becomes aware of the need for management decisions: either specifically in response to constraints resulting from external conditions or in response to an observed loss of productivity at the crew level. Finally, the model shows how productivity can go into a downward spiral if inappropriate management actions are taken.

The action-response model of productivity loss in construction provides a framework for evaluating the causes of productivity loss on individual projects. This capability is important for two reasons. First, during the course of a project, it is important that appropriate management actions be taken to mitigate or eliminate the occurrence of a loss of productivity. If the cause of the loss is unknown, any remedial actions taken may be ineffective or, even worse, may exacerbate the situation. Second, if a loss cannot be eliminated, it is only fair that the party responsible bear the cost. To identify the responsible party, the cause of the loss must first be understood.

The techniques commonly used to identify particular cause-and-effect relationships in loss of productivity, such as productivity-time plots and reference to published studies, fail to account for the complex nature of these relationships. In fact, any number of factors may or may not interact to cause any particular observed loss of productivity. The action-response model takes into account this complexity, and thereby provides a practical tool for evaluating the loss of productivity that can accompany unanticipated conditions in construction.

ACKNOWLEDGMENTS

The writers acknowledge the contributions of the late Weston T. Hester, Professor of Civil Engineering at the University of California, Berkeley. Weston played a crucial role in developing the research that led to this paper, and it was with great sadness that we had to complete our work without him. Those who knew Weston will no doubt detect his influence. In recognition of Weston as a teacher, engineer, colleague and friend, we dedicate our efforts here to him.

The writers thank O'Brien-Kreitzberg & Associates, Inc., for its support of the research that led to this paper.

APPENDIX. REFERENCES

"Absenteeism and turnover." (1982). Rep. C-6, Business Roundtable, New York, N,Y,

Borcherding, J. D,, and Alarcon, L. F. (1991). "Quantitative effects on construction productivity." The Constr. Lawyer, 11(t), 1-48.

Borcherding, 1. D., and Garner, D. F. (1981). "Workforce motivation and productivity on large jobs." J. Constr. Div., ASCE, 107(3), 443-453.

Borcherding, J. D., and Oglesby, C. H. (1974). "Construction productivity and job satisfaction." J. Constr. Div., ASCE, 100(3), 413-431.

Borcherding, J. D., Sebastian, S. J., and Samelson, N. M. (1980). "Improving motivation and productivity on large projects." J. Constr. Div., ASCE, 106(l), 73-89.

"Change orders." (1987). Bull. CO-], Mechanical Contractors Association of America, Rockville, Md.

Dallavia, L. (1954). Estimating production and construction costs. Dallavia Co.,Houston, Tex.

Doyle, M. (1979). "Management's role in productivity improvement." Proc., Forum on Constr. Efficiency and Productivity, Revay and Associates Ltd., Calgary, Canada.

"The effects of scheduled overtime and shift schedule on construction craft productivity." (1988). Source Document 43. Construction Industry Institute (CH), Austin,Tex.

"The effect of temperature on productivity." (1974). Rep. No. 5072, National Electrical Contractors Assocation (NECA), Washington, D.C.

"Factors affecting productivity." (1976). Bull. 58 and Cover Letter, Mechanical Contractors Association of America, Inc., Rockville, Md.

Grimm, C. T., and Wagner, N. K. (1974). "Weather effects on mason productivity." J. Constr. Div., ASCE, 100(3), 319-335.

Guide to electrical contractors' claims management; Vol. III. (1980). National Electrical Contractors Association (NECA), Detroit, Mich.

Hester, W. T. (1987). "Productivity of insulation." Rep., Construction Industry Institute (CII), Austin, Tex.

Hinze, J., and Kuechenmeister, K. (1981). "Productive foremen characteristics." J. Constr. Div., 107(4), ASCE, 627-639.

Hinze, J., and Parker, H. W. (1978). "Safety: productivity and job pressures." J. Constr. Div., ASCE, 104(l), 27-34.

"Hours of work and output." (1947). Bull. No. 917, Bureau of Labor Statistics,Washington, D.C.

"How much does overtime really cost?" (1987). Bull. No. OT-1. Mechanical Contractors Association of America, Rockville, Md.

Koehn, E., Seling, R., Kuchar, J., and Young, R. (1978). "Cost of delays in construction." J. Constr. Div., ASCE, 104(3), 323-331.

Lemna, G. J., Borcherding, J. D., and Tucker, R. L. (1986). "Productive foremen in industrial construction." J. Constr. Engrg. and Mgmt., ASCE, 112(2), 192-210.

Levitt, R. E., and Samelson, N. M. (1987). Construction safety management. McGrawHill, New York, N.Y.

Maloney, W. F. (1983). "Productivity improvement: the influence of labor." J. Constr. Engrg. and Mgmt., ASCE, 109(3), 321-324.

Maloney, W. F., and McFillen, J. M. (1985). "Valence of and satisfaction with job outcomes." J. Constr. Engrg. and Mgmt., ASCE, 111(l), 53-73.

Maloney, W. F., and McFillen, J. M. (1986). "Motivational implications of construction work." J. Constr. Engrg. and Mgmt., 112(i), 137-151.

Maloney, W. F., and McFillen, J. M. (1987). "Influence of foremen on performance." J. Constr. Engrg. and Mgmt., ASCE, 113(3), 399-415.

"Modification impact evaluation guide." (1979). Rep. EP 415-1-3, Department of the Army, Washington, D.C.

"More construction for the money." (1983). Summary Rep. of the Construction Industry Cost Effectiveness Project. Business Roundtable, New York, N.Y.

Neil, J. M. (1982). Construction cost estimating for project control. Prentice-Hall Inc., Englewood Cliffs, N.J.

Neil, J. M., and Knack, L, E. (1984). "Predicting Productivity." AACE Trans.,Paper U-3, American Association of Cost Engineers (AACE), Morgantown, W.Va,

O'Connor, L. V. (1969). "Overcoming the problems of construction scheduling on large central station boilers." Proc., Am. Power Conf., Illinois Inst. of Tech., Chicago, Ill., 31, 518-528.

Oglesby, C. H., Parker, H. W., and Howell, G. A. (1989). Productivity improvement in construction. McGraw-Hill Publishing Inc., New York, N.Y. "Overtime and productivity in electrical construction." (1989). Rep. No. 5050, 2nd Ed., National Electrical Contractors Association (NECA), Washington, D.C.

"Scheduled overtime effect on construction projects." (1980). Rep. No. C-2, Business Roundtable, New York, N.Y.

Stewart, J. M. (1982). "Chapter 16: practical aspects of human heat stress." Environmental engineering in South African mines. The Mine Ventilation Society of South Africa, Cape and Transvaal Printer (Pty) Ltd., Cape Town, South Africa.