Advancing the Competitiveness and Efficiency of the U.S. Construction Industry (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

D Technical Change and Its Impact on Construction Productivity Paul M. Goodrum, P.E., Ph.D. Associate Professor, Department of Civil Engineering University of Kentucky, Lexington Abstract Over time, technology has changed many construction processes. It may be debatable whether the construction industry has leveraged technology to its fullest, but there is little doubt that where technology has had an impact, there has also been significant improvement in construction productivity. This paper examines characteristics of technical change among construction equipment, materials, and information systems and among construction activities and processes. Understanding how distinct characteristics among construction equipment, materials, and information systems are related to improvements in construction labor productivity may help aid the development of future innovations. Much research is needed, however, to lead to an understanding of how technical change has improved the quality characteristics of the construction industry’s output and the potential impact that this has on the industry’s productivity measures. INTRODUCTION From the perspective of a casual observer, the U.S. construction industry can appear to be less technically progressive than other U.S. industrial sectors. The industry is perceived to involve primarily tedious, dirty, and physically exhausting work. As a testament to this perception, a popular career guide recently ranked the occupation of a construction worker (laborer) as 244th out of a possible 250 career choices (Krantz, 2002). Technological improvements have dramatically changed the process of construction over the past couple of decades as well as the quality of construction output. Unfortunately, the industry measures both outcomes poorly. The perceived lack of technological change is a primary argument supporting the belief that construction productivity has been declining since the 1960s (Rosefielde and Mills, 1979), which influences workforce strategies, research programs, and industry perceptions, and it is based on a number of productivity studies using industrial, macroeconomic data (Stokes, 1981; BRT, 1983; Allen, 1985). As a whole, the United States has enjoyed almost continuous productivity growth for the past several decades and especially strong growth in the past decade. In a relatively recent research effort, Triplett and Bosworth (2004) identified that much of the nation’s productivity growth could be attributed to improved production of information technology (IT), increased use of IT, increased competition due to globalization, and changes in workplace practices and firm organizations. However, Triplett and Bosworth (2004) also point out that construction bucked this trend by experiencing negative productivity growth during the time period of their analyses, 1995 to 2001. However, other studies have produced contradictory data. Research conducted through the Sloan Center for Construction Industry Studies at the University of Texas at Austin examined labor and partial factor productivity trends using microeconomic data for 200 construction activities as part of a larger effort to analyze the relationship between equipment technology and construction productivity (Goodrum et al., 2002). The results indicated widespread improvement in construction 76 

APPENDIX D 77 labor productivity across multiple construction divisions ranging from 0.2 percent to 2.8 percent per year between 1976 and 1998, especially in machinery-dominated divisions such as site work. Similar improvement was observed in partial factor productivity among the 200 construction activities. In a more recent effort to examine the relationship between material technology and construction productivity, the average percentage change in the labor productivity of an additionally sampled 100 activities was found to have an annual improvement compound rate of 0.47 percent between 1977 and 2004 (Goodrum et al., 2009). In addition to these measured improvements, there is anecdotal evidence shared among some industry practitioners that construction productivity has actually improved (Bernstein, 2003; Tuchman, 2004; Harrison, 2007). The potential reasons explaining the discrepancy between macro and micro measures of construction productivity are numerous, with most of the focus on issues relating to the accuracy of industry measures, particularly on the inflation indexes used to measure industry real output. The concerns range from overreliance on the use of proxy inflation indexes to deflate construction expenditures (Pieper, 1990), to the use of input cost inflation indexes instead of the preferred output price indexes (Dacy, 1965; Gordon, 1968; and Pieper, 1990), and the challenge of measuring the change in the quality of industry output (Rosefielde and Mills, 1979; Pieper, 1990; Gullickson and Harper, 2002). One alternative to using industry data to measure construction productivity is to use micro productivity data, which are typically reported in units of physical output per unit of input among construction activities. However, activity productivity data suffer their own measurement issues. In previous studies in which the writer has used activity productivity data (Goodrum et al., 2002; Goodrum and Haas, 2002, 2004; and Goodrum et al., 2009), the primary source of activity data has been commercial estimation manuals, which are often used by construction industry professionals for estimating the cost of a project. Previous studies have sampled 100 to 200 construction activities for the purposes of analyses, but the assumption that the measured changes in productivity among the observed activities actually reflect change throughout all of the construction industry has largely rested on sample size only. Weighting discrete activities to reflect their frequency of occurrence in the industry could help resolve this in part, but previous efforts by the writer have not attempted this. In addition, estimation manuals typically collect their data from contractors in multiple cities throughout the United States, but the methodology—precisely how the data are collected, the survey forms used, and the frequency at which both output and cost data are updated for every activity—is not documented in the public domain. Finally, contractors who submit information for the estimation manuals know that they are not required to construct a project using their own estimations, and this tends to create inflated estimates of construction costs (Pieper, 1989). Regardless, the annually published manuals are sold in volume for commercial use to a multitude of construction contractors, owner companies, and governmental agencies, all of which use the estimation manuals to predict project performance. Outside of commercial estimation manuals, there are relatively few sources of other micro productivity measures in the construction industry. One source is the Construction Industry Institute (CII) through its Benchmarking and Metrics (BM&M) program. Primarily focused on the industrial construction sector, the BM&M program aims to measure and assess capital project performance and find the best practices among similar projects. The BM&M data set is intended to allow participating companies to compare performance on their projects with similar projects and to help companies identify practices that may improve their respective projects. The database currently includes 86 projects, providing information about field practices and labor productivity. The field practices include different aspects of job-site management systems, such as materials management, constructability, and automation and integration of project systems, among others. The database collected activity productivity data on a variety of construction tasks among seven trades. The BM&M productivity metrics were identified through the use of literature reviews, documentation from owner and contractor organizations, and a series of workshops with industry experts. Details on its methods of data collection and standard accounts have been well documented elsewhere (Park et al., 2005). The combination of the commercial estimation manuals and the CII BM&M database affords an opportunity to examine the relationship between construction productivity and technical change at a 

78 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY microlevel, which is the general focus of the discussion that follows. The paper examines the relationship between technical change and construction productivity in three sections based on the primary components of construction technology: equipment, material, and information technology. Each section examines previous research both by the writer and by other researchers. Using data from commercial estimation manuals, the sections addressing research on equipment and material technology examine the longitudinal relationship between the respective change in these technologies and corresponding changes in productivity. Considering the relatively new implementation of information systems in construction, the paper compares the use of automation and integration information technologies and reported levels of productivity across multiple projects using data from the CII BM&M database. Together, these sections present a comprehensive perspective regarding the relationship between technology and productivity in the construction industry. EQUIPMENT TECHNOLOGY Koch and Moavenzadeh (1979) completed one of the first research efforts that focused on the relationship between technology and construction productivity by examining the change in equipment and unit labor costs for various road construction activities over three time periods: the 1920s, 1950s, and 1970s. Controlling for inflation, Koch and Moavenzadeh noted how unit costs had continuously dropped over all three time periods. Figure D.1 shows an example of Koch and Moavenzadeh’s analyses involving the change in unit costs of excavation and hauling material over a 100-meter distance. They found that unit costs using the equipment technology from the 1970s were consistently the lowest in all three time periods, despite finding dramatic increases in labor and equipment costs through the 1930s, 1950s, to the 1970s. Koch and Moavenzadeh believed that these increases were offset by increased efficiency due to advancement in equipment technology. They found that increased usage and advancements in equipment technology were two primary causal agents. Furthermore, the capital costs of many activities increased while the relative labor costs declined reflecting an increase in the use of technology (Koch and Moavenzadeh, 1979). In addition, the rate of productivity improvements appeared to decline over time, with most improvements occurring from the 1930s to the 1950s when machine power innovations were introduced. Koch and Moavenzadeh (1979) did note that the greatest changes in technology and improvement in efficiency occurred from the 1930s to the 1950s. In the 1930s, “Small capacity, unpowered equipment operated largely by unskilled laborers with horses or mules as a source of power and a few skilled men acting in a supervisory role was most common” (Rossow, 1977). By the 1950s, equipment became powered and larger in capacity and was operated mostly by skilled laborers, with occasional help from unskilled assistants. The equipment technology transitions by the 1970s were not as great and can be seen in Figure D.1 in the smaller improvements in unit costs from the 1950s to the 1970s. By the 1970s, the changes primarily involved equipment that had become more powerful and larger in capacity, with only a few new types of equipment (Rossow, 1977). Although hydraulics had advanced controls of machinery, Rossow (1977) did not discuss advancements in machinery control as being significant. In a more recent research effort, Allmon et al. (2000) examined the impact of technology as well as real wages on productivity through a case study analysis of six construction activities. Using cost and output data from the R.S. Means Building Construction Cost Data, productivity was found to have increased in all six activities between 1976 and 1998. The study also found that some of the greatest increases in productivity were directly connected with the introduction of new, more technically advanced equipment. The study concluded that many of the improvements in productivity could be attributed to changes in technology. However, the research did not measure technology intensity, nor did it address the different types of changes in equipment technology and their relative importance with respect to productivity. It did, however, provide groundwork for a more thorough statistical analysis, which research by Goodrum and Haas (2002) addressed. 

APPENDIX D 79 FIGURE D.1 Unit costs of hauling of each technology period for excavation/hauling at 100 meters at prices of 1930, 1956, and 1974. NOTE: BCM, bank cubic meter. SOURCE: Rossow (1977). Goodrum and Haas (2002) examined the influence of equipment technology on construction productivity through a series of longitudinal studies that examined the changes that had occurred in the equipment technology and productivity of 200 construction activities between 1976 and 1998. Data were collected for years 1976 and 1998 on 200 construction activities from the R.S. Means Building Construction Cost Data (Means Company, Inc., various years), Richardson’s Process Plant Construction Estimating Standards (Richardson Engineering Services, various years), and F.W. Dodge Unit Cost Books (McGraw-Hill Inc., various years). One of the challenges in analyzing productivity statistically is that each construction activity has a different unit of measurement. For example, a concrete placement activity’s multifactor productivity may be measured in cubic yards of concrete placed per unit cost, while structural steel placement may be measured in linear feet of steel placed per unit cost. Using relative instead of absolute values is one way to solve this issue. Thus, the percentage change in productivity from 1976 and 1998 was used by Goodrum and Haas (2002). Expected physical output and crew formation data from the estimation manuals were used to calculate each activity’s labor productivity (Equation D.1). Expected physical output, labor input cost, and equipment input cost data from the estimation manuals were also used to calculate each activity’s partial factor productivity (Equation D.2). Expected Physical Output (Units) Labor Productivity, Year X = D.1 Workhour Requirements (Hrs) Expected Physical Output (Units) Partial Factor Productivity, Year X = Labor Cost (1990$) + Equipment Cost (1990$) D.2 

80 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY Input costs in partial factor productivity were deflated to 1990 dollars using the Construction Cost Index (CCI) from Engineering News Record. Note that Equations D.1 and D.2 assume that physical output measures do not change in quality. This is based on the assumption that changes in quality would be minimal, since the research examined construction activities that had not changed in scope between 1976 and 1998. Next, the percentage change in labor and partial factor productivity from 1976 to 1998 was measured for each activity using equations D.3 and D.4: % Change in Labor Productivity, ’76-’98 = ( Labor Productivity, ’98 – Labor Productivity, ’76 Labor Productivity, ’76 ) (2) · 100 D.3 % Change in Multifactor Productivity, ’76-’98 = ( Partial Factor Productivity, ’98 – Multifactor Productivity, ’76 Multifactor Productivity, ’76 ) · 100 D.4 Note, there is a weakness in using just two points in time to measure the change in productivity, since the results can be affected by the choice of the two years. Particularly, it is noted that 1976 was a year of stagflation and excess capacity in the United States. It is expected that fluctuations in the change in productivity would occur in a year-by-year analysis. However, by examining the changes in productivity over a 22-year time period, the research was designed to focus on the long-term trends in construction productivity. The average change in labor and partial factor productivity of the activities overall for each data source is shown in Table D.1. All three manuals indicate that labor and partial factor productivity increased from 1976 to 1998. The R.S. Means manual reveals a 0.8 percent compounded annual rate of improvement in labor productivity, Richardson a 1.2 percent increase, and Dodge a 1.8 percent increase. For partial factor productivity, Means shows a 0.7 percent increase, Richardson a 0.7 percent increase, and Dodge a 2.9 percent increase. The different estimates of productivity improvement are partially a reflection of the different distribution of types of construction activities in different divisions for each manual. Next, activities were grouped by Construction Specification Institute (CSI) Masterformat construction division, and the compounded annual rate of change in labor and partial factor productivity was calculated for each division (see Table D.2). It is clear from this sample that different sectors of the construction industry experienced varying degrees of change in productivity. On average, site-work activities experienced the greatest improvement in labor and partial factor productivity. Electrical activities, moisture-thermal protection, and woods and plastic activities experienced the smallest improvements in labor and partial factor productivity. Further research is required to determine the reasons for the differences in productivity change by division. TABLE D.1 Estimates of Labor and Partial Factor Productivity Trends (1976 to 1998), by Data Source Activity Labor Productivity: Partial Factor Productivity: Sample Compounded Annual Rate Compounded Annual Rate Data Source Size (total change) (total change) Means Building Construction Cost 100 +0.8% (19.1%) +0.7% (17.4%) Data Richardson Process Plant 50 +1.2% (30.2%) +0.7% (18.1%) Construction Estimating Standards Dodge Unit Cost Books 50 +1.8% (48.6%) +2.9% (88.5%) 

APPENDIX D 81 TABLE D.2 Compounded Annual Rate of Change in Labor and Partial Factor Productivity for Activities, by Construction Division, from 1976 to 1998 Change in Labor Productivity Change in Multifactor Productivity 1976-1998 1976-1998 Compound Annual Rate Compound Annual Rate Construction Division (total change) (total change) Sitework +2.8% (83.5%) +2.4% (66.9%) Doors and windows +1.6% (43.1%) +1.8% (47.8%) Metals +1.5% (25.6%) +1.0% (25.1%) Finishes +1.2% (29.1%) +1.6% (37.5%) Masonry +1.2% (28.8%) +0.8% (25.0%) Concrete +1.1% (26.3%) +1.4% (34.4%) Mechanical +1.0% (25.3%) +1.4% (35.1%) Wood and plastic +0.3% (7.7%) +0.4% (17.5%) Moisture and thermal protection +0.2% (4.7%) +0.6% 14.9%) Electrical +0.0% (0.2%) +0.8% (18.6%) SOURCE: R.S. Means, Richardson, and F.W. Dodge estimation manuals. To begin examining how the changes shown in Tables D.1 and D.2 are related to simultaneous improvements in equipment technology, five factors were identified (defined below) that characterize significant changes in equipment technology related to improvement of the productivity of construction activities: • Amplification of human energy—Amplification of human energy involves technology designed to make an activity physically easier to perform. In its simplest terms, this amplification can be regarded as the shift in energy requirements from human to machine and causing an increase in machine output (e.g., revolutions per minute, horsepower). As examples, welding machines increased wattage output, and powder-actuated systems offered greater depth penetration for installing studs in metal decking. Further, most site-work machinery offered increased horsepower output (e.g., front-end loaders, dump trucks, backhoes, bulldozers, graders, asphalt pavers, and scrapers). • Level of control—Level of control relates to advances in machinery and hand tools that transfer control from the human to the machine. Welding machines in the metals division, for instance, are now equipped with remote-controlled amperage adjusters, and powder-actuated systems have semiautomatic loading capabilities. The pneumatic nail gun has replaced the handheld hammer in the wood and plastic division as well as in formwork installation in the concrete division. • Functional range—Changes in equipment’s functional range expand a tool’s or machine’s range of capabilities. Through advances in hydraulic controls and microprocessors, machinery for site work now offers more control precision and a greater reach with booms and buckets. • Information processing—Over time, construction equipment has been designed to provide greater and more accurate information regarding internal and external processes. Almost all of the advances in information processing occurred in heavy machinery with the development and improvement in engine performance monitoring and self-diagnosis systems. • Ergonomics—Ergonomics involves technology that helps the human operator to best cope with the work environment (Oborne, 1987). Construction workers are exposed to high noise levels, dust, weather, and other external factors, which can cause worker fatigue and thereby reduce efficiency. 

82 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY Advances in construction equipment technology have addressed these concerns. For example, operator stations on heavy machinery have been designed to provide a quieter environment with less vibration. Hand tools have been designed with molded grips to better fit in a worker’s hand for greater comfort. The activities that experienced improvement in the above equipment technology traits experienced more improvements in labor and partial factor productivity than those activities that did not, and this finding was statistically significant in a series of analysis of variance (ANOVA) results (Table D.3). Activities experiencing an improvement in energy, control, functional range, and information processing had at least twice as great an improvement in labor and partial factor productivity as the improvement of activities experiencing no improvement in the technology factors. Ergonomics was not statistically significant in any of the ANOVA results. Although it is widely believed that alleviating the physical stresses of the workplace would allow operators to be more productive, this relationship was not seen in the quantitative analyses. Perhaps ergonomic changes reduce insurance costs through a reduction in workers’ compensation and health insurance claims, but this study did not measure the insurance costs by activity. Previous research included regression models of the equipment technology characteristics on changes in both labor and partial factor productivity (Goodrum and Haas, 2002, 2004). The models used a series of technology measures, including changes in capital-to-labor costs for each activity. In addition, the research developed an equipment technology index, which is detailed in Goodrum and Haas (2004) but can be briefly described as a scoring system based on the five equipment technology characteristics considering each activity’s hand tools and machinery. Furthermore, the regression models examined a series of dichotomous variables to estimate the influence of the five equipment technology factors (energy, control, functional range, information processing, and ergonomics); however, only equipment technology control proved to be statistically significant in both regressions, with function being only marginally significant (p-value = 0.12) in the regression on labor productivity. The remaining factors of energy, information processing, and ergonomics were found to be statistically insignificant. Several lessons are gained from the regressions in Table D.4. First, an increase in the change in the capital-to-labor ratio was observed, with significant increases in labor and partial factor productivity, although the relation was weaker between the capital-to-labor ratio and partial factor productivity. Second, the technology regressions explain significantly more of the variability of the change in labor productivity versus the change in factor productivity as seen by the difference in the R-squared values for both models. Both occurrences demonstrate an expected labor-saving bias of technical change. Third, the regression models indicate that changes in the level of control have the strongest relation, which changes both partial factor and labor productivity among the five identified equipment technology characteristics. Finally, the R-squared values might be considered low in comparison to those factors from other statistical studies. However, when considering the multitude of other factors that influence job-site productivity performance (e.g., skilled labor, quality of design documents, and weather), the ability of equipment technology alone to explain as much of the variability in productivity as shown in Table D.4 (especially in labor productivity) should be considered significant. 

APPENDIX D 83 TABLE D.3 Analysis of Variance (ANOVA) of Change in Construction Productivity and Change in Equipment Technology Characteristics Change in Labor Productivity Change in Partial Factor Productivity No Change in Change in No Change in Change in Equipment Equipment Equipment Equipment Equipment Technology Technology Technology Technology Technology Characteristic Characteristic Characteristic F-value Characteristic Characteristic F-value Energy 3.6% (49) 39.8% (151) 11.84* −4.8% (49) 18.9% (151) 5.38* Control 14.9% (101) 46.6% (99) 10.45* −1.1% (101) 27.6%(99) 10.94* Functional range 13.5% (106) 51.8% (94) 18.21* 3.1% (106) 24.5% (94) 5.90* Information processing 21.0% (144) 56.4% (56) 12.31* 6.8% (144) 29.3% (56) 5.25* Ergonomics 26.4% (91) 34.8% (109) 0.81 8.0% (91) 17.4% (109) 1.09 NOTE: Numbers in parentheses represent sample size. * Denotes significance at 0.05. TABLE D.4 Regression of the Equipment Technology Index (ETI), Capital-to-Labor Ratio (K/L), and Dichotomous Variables on Percent Change in Labor Productivity Dependent Variable: Percent Change in Labor Productivity; Independent Variable as Indicated by Column Heading Eqn. Constant K/L (K/L)2 ETI ETI2 Control Function F R2 Adj. R2 A 5.23 131.74 110.39 −18.53 6.60 19.43 14.51 19.08 0.37 0.35 (0.66) (6.88) (3.09) (−2.03) (3.08) (2.21) (1.55) Dependent Variable: Percent Change in Partial Factor Productivity; Independent Variable as Indicated by Column Heading Eqn. Constant K/L (K/L)2 ETI ETI2 Control F R2 Adj. R2 B 6.36 66.76 86.63 −19.23 6.18 26.08 7.83 0.17 0.15 (0.67) (2.79) (2.04) (−1.80) (2.32) (2.47) NOTE: t-values shown in parenthesis; N= 200 activities. MATERIAL TECHNOLOGY Other research has examined the relation between changes in material technology and construction productivity. These analyses examined how changes in material technology have influenced labor and partial factor productivity in the U.S. construction industry between 1977 and 2004, using methods of longitudinal analyses at the construction activity level similar to those used for the examination of equipment technology. However, this time, the analyses involved a smaller sample size of 100 construction activities (Goodrum et al., 2009). For these sets of analyses, labor productivity is defined as before (Equation D.1), but partial factor is redefined by Equation D.5 by replacing the equipment with the material input cost in the denominator. Expected physical output (units) D.5 Partial factor productivity, year X= Labor cost (1990$) + material cost (1990$) 

84 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY To achieve a critical understanding of the impact of material technology on construction performance, the researchers developed a metric based on a combination of literature review and industry interviews that were used to quantify the changes in material technology between 1977 and 2004. The metric was composed of five material factors that are briefly described below but are detailed elsewhere (Goodrum et al., 2009). • Reduction in unit weight—The obvious productivity benefits of reduced material weight include ease of handling and transporting by craft labor, although lighter materials have other benefits related to structural design and space requirements. • Strength—Technological advancements, especially with new admixtures and design of concrete mixtures, have increased unit strength of materials. • Curability—Several material advancements have reduced the amount of time required for a material to cure and reach its desired strength (e.g., concrete) and/or dryness (e.g., paint). • Installation flexibility—Installation flexibility refers to the environmental conditions under which a material can be installed. For example, extreme temperatures or moisture can have significant impacts on the installation of material. Technological advancements, such as epoxy coating, waterproofing, and cold-weather admixtures, have improved the durability of materials and allowed their installation in extremely moist and cold conditions. • Modularization—Modularization relates to the amount of material customization performed on-site prior to installation. Prefabrication of individual components is included in this factor. The purpose of including this factor is to measure the benefits of “customizing” materials in a controlled environment under ideal conditions before actual installation. Similar to the finding from the equipment technology analyses, the activities that experienced improvement in the above material technology traits experienced more improvements in labor and partial factor productivity than those activities that did not (Table D.5). In particular, activities experiencing an improvement in curability, installation, and modularity consistently experienced substantially greater improvement in labor and partial factor productivity. The differences among these three factors were also statistically significant above the 95 percent confidence level. In regard to the change in the unit weight and strength of the materials, the analyses found that a statistically significant relationship did exist among reduction in the unit weight of material and improvement in labor productivity, but the relationship was below the 95 percent confidence level in regard to partial factor productivity. Improvement in the unit strength of material was not found to be significant with either labor or partial factor productivity. One reason for the lack of statistical significance involving change in the strength of materials is that this change was more likely intended to allow structures to withstand higher loads rather than to expedite the process of construction. Paralleling the analyses on equipment technology, previous research also developed a series of regression models to examine the relationship between changes in material technology and construction productivity (Goodrum et al., 2009). Due to the smaller sample size (100 activities), the regression models were simplified by not including the capital-to-labor ratio as was done for the equipment technology regressions. A material technology index was developed based on a scoring system of the five material technology characteristics and is described in detail elsewhere (Goodrum et al., 2009). The regression models examined a series of dichotomous variables for the five material technology factors of strength, weight, curability, installation flexibility, and modularization. Not all of the material technology factors were found to be significant; thus only three of the factors are included in the regression models as shown in Table D.6, which shows separate regressions for labor and partial factor productivity. As shown in the regression equation for labor productivity (Table D.6, Equation A), the material technology factor weight produced statistically significant effects on labor productivity above the 95 percent confidence level. This factor, along with the material technology index (MTI), explained 17 percent of the total variation in labor productivity according to the adjusted coefficient of determination. Activities with a 

APPENDIX D 85 decrease in the unit weight of construction materials experienced a 31.0 percent increase in labor productivity compared to other activities that did not experience a change in unit weight. The regression models for partial factor productivity are shown in Table D. 6, Equation B. The material technology variables of installation flexibility and modularization produced statistically significant effects, above the 95 percent confidence level, and these variables, along with the MTI, explained 48 percent of the total variation in partial factor productivity, according to the adjusted R-squared value. It is noted that the material technology regressions had both a stronger substantial and statistical relationship with partial factor productivity than with labor productivity, which is surprising considering that technology overall typically has a labor-saving bias on production (Salter, 1966). Although the research did find evidence of declining material-to-labor cost ratios using industry data from the Engineering News Record, the exact cause for this decline remained inconclusive and warrants future research. TABLE D.5 Analysis of Variance (ANOVA) of Change in Construction Productivity and Change in Material Technology Characteristics Change in Labor Productivity Change in Factor Productivity No Change in Change in No Change in Change in Material Material Material Material Material Technology Technology Technology Technology Technology Characteristic Characteristic Characteristic F-value Characteristic Characteristic F-value Unit weight 10.7% (92) 48.6% (8) 12.93* 13.3% (92) 37.1% (8) 3.03 Strength 13.8% (93) 39.0% (7) 0.63 14.7% (93) 31.1% (7) 1.75 * Curability 8.9% (71) 24.4% (29) 5.34 1.8% (71) 30.1% (29) 17.6* Installation flexibility 8.7% (67) 23.1% (33) 4.95* 3.1% (67) 52.6% (33) 60.3* Modularization 8.1% (71) 24.2% (29) 9.31* 1.7% (71) 42.6% (29) 34.9* NOTE: Numbers in parentheses represent sample size. * Denotes significance at 0.05. TABLE D.6 Regression of the Material Technology Index (MTI) and Dichotomous Variables on Percent Change in Labor and Partial Factor Productivity Dependent Variable: Percent Change in Labor Productivity; Independent Variable as Indicated by Column Heading Eq. Constant MTI Weight F R2 Adj. R2 A 3.39 39.72 30.96 11.64* 0.19 0.17 (0.91) (−2.03) (2.97) Dependent Variable: Percent Change in Partial Factor Productivity; Independent Variable as Indicated by Column Heading Eq. Constant MTI Installation Modularity F R2 Adj. R2 B −10.65 42.15 23.57 15.79 31.58* 0.50 0.48 (−3.28) (2.40) (3.30) (2.44) * t-values shown in parenthesis; N= 100 activities. 

86 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY INFORMATION TECHNOLOGY Finally, prior research examined the relationship between the use of information technology and construction productivity. While previously related productivity research examined longitudinal changes in equipment and material technology, research on the relation of IT and construction productivity has taken more of a latitudinal approach considering the relatively short history of construction projects’ use of IT. A number of research efforts have examined the impact of specific applications of IT and construction performance. Thomas et al. (2004) evaluated the relationship of design/information technology (D/IT) and construction project performance. The researchers measured the degree of D/IT usage specifically based on the use of four technologies: integrated database, electronic data interchange, three-dimensional computer-aided design modeling, and bar coding. Thomas et al. (2004) indicated that D/IT was positively related to project performance, especially cost and schedule. Grau et al. (2009) conducted an extensive field trial of an automated material tracking system for structural steel that integrated radio-frequency identification tags and the Global Positioning System and was able to tie the use of the system with improvement in steel labor productivity during the field trial efforts. O’Connor and Yang (2004) conducted one of the first studies that examined the comprehensive use of IT on construction job sites and its relationship to project performance. The researchers developed an integration and automation (IA) index ranging from 0 to 10 according to the IA use level on a series of project work functions. The statistical analysis of O’Connor and Yang (2004) indicated that the schedule success–technology relationship was stronger than that for cost. El-mashaleh et al. (2006) found a similar quantitative result when they also examined the impact of IT on construction firm performance (especially cost and schedule). The method that they used to develop an IT index was similar to that in the research of O’Connor and Yang (2004). Their analysis showed that for every one unit increase in their IT index, construction firms experienced an increase of 5 percent and 3 percent in schedule performance and cost performance, respectively. While these research efforts did identify a positive relationship between the use of IT and improvement in the cost and schedule performance in the construction industry, they did not link IT to actual productivity performance. As a result, recent research examined how the use of IT among different project work functions, such as supply management, communication systems, and cost and scheduling systems, is related to construction labor productivity (Zhai et al., 2009). For the purpose of the research, Zhai et al. (2009) adopted the following definitions of automation and integration, as developed by O’Connor and Yang (2004): • IT automation—The use of an electronic or computerized tool by a human being in order to manipulate or produce a product. Hard automation, such as robotics, is not included in this definition. • IT integration—The sharing of information between project participants or melding of information sourced from separate systems. The data used in this research came from the CII’s BM&M productivity database, described previously in this paper. Using data from the BM&M database that described the level of IT usage on specific project work functions as well as productivity measures on the same projects, Zhai et al. (2009) compared the level of IT usage on projects with the projects’ respective labor productivity among four common trades: concrete, structural steel, electrical, and piping. For the purposes of the study, the researchers measured labor productivity using the following equation: Actual Workhours Labor Productivity = D.6 Installed Quantity 

APPENDIX D 87 It is important to note that a lower productivity number per Equation D.6 is better. To ensure company confidentiality and allow comparisons across different tasks and trades, the raw productivities were normalized using the Min-Max method (Han and Kamber, 2000) based on the following equation: P −P P = raw raw min (P −P )+P D.7 −P norm norm max norm min norm min P raw max raw min In Equation D.7, Pnorm is the normalized productivity and, Praw is the raw productivity measure; Prawmin and Prawmax are the minimum and maximum raw productivity values in the construction task; and Pnormmin and Pnormmax are the minimum and maximum normalized productivity values, equal to 1 and 10, respectively. The normalized productivity (Equation D.7) is consistent with the Equation D.6 measure of labor productivity; a lower value indicates better productivity. Using methods developed by O’Connor and Yang (2004), automation and integration indexes were developed for each project based on the level of automation and integration achieved in 13 standard work functions. The range of each index is from 0 to 10. For purposes of the analysis, projects scoring 5 percent above the overall median among all sampled projects were classified as having a high level of automation or integration, and projects scoring 5 percent below the median were defined as having a low level of automation or integration. The projects falling within the 5 percent range were not used in the comparison between the two groups. In the automation-related analysis, four projects fell within this range, and in the integration-related analysis, nine projects fell within this range. The reason for using the median rather than the mean is that the automation and integration indexes did not have a perfectly normal distribution. The purposes of using such a 5 percent range below and above the median are to (1) create two groups with more distinct differences in automation and integration use levels, and (2) guarantee that the sample sizes are large enough to perform the statistical analyses. Next, Zhai et al. (2009) examined the productivity among the four trades as well as the productivity among all trades using the normalized productivity measure. All-trades productivity is a combination of the four trade-specific normalized productivity data sets, which includes all of the normalized activity-productivity available in this research combined into one data set. The results (Table D.7) indicate that automation usage is positively related to structural steel, electrical and all-trades productivity, and all of these relationships are significant at the 0.05 level. The results for the concrete and piping trades lack statistical significance although the relationships are positive. As indicated in Table D.8, integration usage was positively related to concrete, structural steel, and all-trades productivity at a statistical significance level of 0.05. The relationship in the electrical trade was significant at the 0.15 level. Again, no statistically significant result was observed in the piping trade, although the relationship was positive. While both integration and automation are related with better productivity performance, these analyses suggest that integration has a relatively stronger impact with better labor productivity. 

88 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY TABLE D.7 Results of t-test on Automation Index by Trade Levene’s Test Equal for Equality of Variances Equal Variances Normalized Productivity Variances Assumed Not Assumed High-level Low-level Trade Automation Automation Difference F Sig. t Sig. t Sig. Concrete 3.48 (33) 3.89 (37) −0.40 4.98 0.03 −0.69 0.49 −0.70 0.49 a Structural steel 3.74 (40) 5.24 (24) −1.50 16.91 0.00 −2.42 0.02 −2.14 0.04 Electricala 3.65 (52) 5.21 (19) −1.55 1.51 0.22 −2.04 0.05 −1.91 0.07 Piping 3.96 (53) 4.40 (37) −0.45 3.97 0.05 −0.71 0.48 −0.69 0.50 a All trades 3.68 (178) 4.54 (117) −0.86 20.62 0.00 −2.72 0.01 −2.58 0.01 NOTE: Numbers in parentheses are the sample sizes (activity productivities). a Denotes significance at 0.05. TABLE D.8 Results of t-test on Integration Index by Trade Levene’s Test for Equal Equal Equality of Variances Variances Not Normalized Productivity Variances Assumed Assumed High-level Low-level Trade Integration Integration Difference F Sig. t Sig. t Sig. a Concrete 2.91 (33) 4.71 (19) −1.81 19.90 0.00 −3.12 0.00 −2.61 0.02 a Structural steel 3.48 (39) 5.30 (10) −1.82 3.28 0.08 −2.58 0.01 −2.58 0.01 b Electrical 3.28 (48) 5.66 (8) −2.38 8.15 0.01 −2.36 0.02 −1.73 0.12 Piping 3.82 (52) 5.02 (15) −1.20 10.59 0.00 −1.39 0.17 −1.12 0.28 a All trades 3.37 (172) 5.06 (52) −1.69 28.89 0.00 −4.41 0.00 −3.57 0.00 NOTE: Numbers in parentheses are the sample sizes (activity productivities). a Denotes significance at 0.05. b Denotes significance at 0.15. The described t-test results were based on normalized productivity measures in order to preserve the confidentiality of the CII BM&M data and also to allow analysis across different tasks and trades, since the normalized productivity measures are dimensionless (Zhai et al., 2009). However, reporting the analyses using normalized productivity obscures the actual effects. To help clarify the results, the researchers calculated the means of raw productivity for the projects with high- and low-level technology use and then calculated the percentage difference using the following equation: Percentage difference of productivity (Mean P - Mean P ) D.8 = RawH RawL ×100 Mean P L where PRawH denotes the raw productivity with high-level automation (or integration) index. Similarly, PL denotes the raw productivity with low-level automation (or integration) index. As a reminder, raw productivity was measured on the basis of actual work hours per installed quantity, so the percentage difference of productivity indicates the approximate percentage of time saving per installed quantity when using a high versus a low level of technology usage (Table D.9). 

APPENDIX D 89 TABLE D.9 Percentage Improvement in Raw Labor Productivity Measurements Considering Automation and Integration of Construction Industry Institute Work Functions Percent Improvement in Labor Productivity Automation Integration All trades 30.9% 45.0% Concrete 23.3% 56.4% Structural steel 33.9% 41.5% Electrical 30.3% 38.4% Piping 36.4% 45.9% Overall, the analyses in information technology show that construction labor productivity is positively correlated with the usage of automation and integration information systems on the sampled construction projects. The average time savings per installed quantity were observed to be 30.0 percent and 45.0 percent when using a high versus a low level of automation and integration, respectively. Another important finding in the research by Zhai et al. (2009) is that automation and integration uses have different significance in various trades. It is intriguing that piping was the one trade that showed no significant correlation between automation and integration technologies on a project and productivity basis. Further research is needed to examine this occurrence. Although it is possible that the results lack significance owing to sample size, it is also possible that current automation and integration technologies are indeed not helping the piping trade become more productive. In the case of the latter explanation, attempting to understand why current automation and integration technologies are not helping is warranted. Meanwhile, O’Connor and Yang (2004) found similar results in their effort using similar automation and integration indexes described herein. In particular, O’Connor and Yang (2004) found that integration information systems had a more significant impact on project performance compared to automation information systems, which mirrors the results presented herein. From the definition of the automation and integration use levels, it can be seen that automation is a prerequisite to integration, and integration is an enhancement of automation. Therefore, in hindsight, it was not unexpected to observe that integration has a more significant impact on labor productivity. TECHNOLOGY AND ITS INFLUENCE ON CONSTRUCTION PRODUCTIVITY MEASURES Thus far, the discussion has focused on the observed relationships between technical change and related changes in construction productivity. However, it is unlikely that the influence of technical change is restricted just to productivity performance. There is evidence that technical change is likely influencing the industry measures of construction output, and it is the writer’s opinion that this influence needs to be considered in construction inflation indexes to help develop reliable industry measures of construction productivity. As mentioned previously, other researchers have expressed concerns regarding the need to understand how changes in the quality of construction influence the measure of the industry’s real output (Rosefielde and Mills, 1979: Pieper 1990; Gullickson and Harper, 2002). Although quality has many different meanings—such as reduction in process defects and improvement in customer satisfaction— quality changes in the context of this discussion are changes in the features of the built project. Two construction sectors are addressed below in this section along these lines: the residential and the industrial sectors. 

90 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY It is the writer’s opinion that technology is significantly improving the quality of new homes by improvements in energy efficiency, fire protection, building security, and high-performance windows, to name a few examples. Preferably when deflating an industry’s output in order to measure its productivity, output price indexes are used. However, in certain sectors of construction, input cost indices have been used instead, since output price indices do not exist for all sectors of construction (Dacy, 1965; Gordon, 1968; and Pieper, 1989). In the residential construction sector, which is also the industry’s largest sector by volume, an output price index produced by the U.S. Census Bureau, called the Single-Family Houses Index, is used, but there are still concerns regarding the ability of the U.S. Census Bureau’s index to capture changes in quality adequately (Pieper, 1990). At the root of the concern is the hedonic regression model used to estimate the price variables used in the Census Bureau’s price index. The hedonic regression models are primarily based on a 1970s-style ranch home; thus, it is plausible that other characteristics resulting from technical advances of modern home structures that are significantly related with new home prices, but not included in the hedonic regression, may inflate the price variables due to omitted variable bias. If the price variables are overestimated, this effect could contribute to overestimating the Census Bureau’s price index, which would underestimate the real output of the residential sector. While this effect is plausible, it has not been quantified. Further work in this area is justified in order to quantify the effects of omitted variable bias in the Census Bureau’s Single-Family Houses Index and to determine if there actually is any discernable bias and what impact this may be having on the measures of the overall construction industry’s output. Overestimation of the Census Bureau’s price index is especially true if there is extensive growth in the omitted quality characteristics. Preliminary data from current research by the writer as well as others (NAHBRC, 2001; Hassel et al., 2003) suggest that this may be the case. Research by Dyer and Goodrum (2008) is examining the effects of omitted variable bias in the Census Bureau’s price index within one geographic area. The researchers have been quantifying the frequency of omitted quality characteristics of new homes in Bowling Green, Kentucky, by using sales data between 2002 and 2007 from the Multiple Listing Service, which tracks the prices of new homes sold along with many quality characteristics of new homes that are not currently measured in the existing price index models of the Census Bureau. Examples of omitted quality characteristics include (1) thermal windows, (2) floor coverings, (3) kitchen appliances, (4) whirlpool tubs, (5) walk-in closets, (6) smoke alarms, (7) tray/vaulted ceilings, (8) landscaping, (9) exterior lighting, and (10) structured cable wiring. The research intends to use the sales data and the housing characteristics to estimate the local price index using the Census Bureau’s price index methods both with and without significant omitted significant quality characteristics in order to measure the effects of the omitted quality characteristics for this one locale. The models developed by Dyer and Goodrum may be applicable to other geographic regions in the United States in order to help develop a broader effort of understanding of the effects of omitted variable bias in the Census Bureau’s price index on a national level. While the research discussed above addresses concerns of the productivity measures in the residential sector, it does not address other concerns about how to improve productivity measures in other sectors of construction. Ultimately, different sectors will likely rely on varying methods owing to nuance differences in volume and heterogeneous output. For the sake of brevity, the writer offers one approach for measuring productivity specifically for the industrial sector. The U.S. industrial sector is characterized as having relatively fewer projects compared to other sectors, but the projects are also typically some of the largest in the United States with strategic implications for the nation’s economy. For this reason, the writer proposes that work on developing reliable productivity measures is urgently warranted and proposes the use of a model price index, also known as an estimate price index, for doing so in the industrial sector. A model price index avoids the challenge of controlling for the change in quality of structures by holding constant a detailed specification for either an entire structure or different components of a structure. Individuals with experience in estimating construction prices are then asked to estimate the selling price of the model. This way, the price change can be observed while holding quality constant. Individuals can be asked to price the entire structure, which is called an aggregate approach, or to price only specific components, which is called the disaggregated approach. The aggregate model price 

APPENDIX D 91 index approach is typically preferred for relatively simple structures, like a single house. For more complicated structures, like an industrial facility, the disaggregated approach is favored (Mohammadian and Seymour, 1997). For the industrial sector, the framework for a model price index exists through the Construction Industry Institute’s model plant. The model plant was initially developed by CII in 1985 to represent a generic petrochemical facility (CII, 1986) and has been modified and updated through a series of related efforts since then. Its mock scope of work includes construction activities in the areas of civil, structural, electrical, mechanical, and architectural finishes. Since its development, the CII model plant data have been used to benchmark industry productivity (CII, 1988); to analyze the impact of multifunctional equipment (Guo and Tucker, 1993); to examine the schedule and manning impacts of utilizing a multiskilled work force (Burleson et al., 1998; and Gomar et al., 2002); and to examine the impact of alternative training strategies for a project’s work force (Castaneda-Maza et al., 2005; Brandenburg, 2004; Pappas, 2004; and Srour et al., 2006). The CII model plant would be useful in developing a model price index, but it needs to be updated, since it is still based on 1980s technical characteristics of a petrochemical facility. If it could be updated by including modern instrumentation and current specifications, it could be used to develop a disaggregated price index model in which different construction firms throughout North America could be used to price specific components of the index. Considering expected strong growth in the industry sector owing to anticipated energy projects, work in this area deserves attention. CONCLUSION From the perspective of an outside observer, it is easy to understand why it appears that technology has had little influence on construction productivity. It is arguable that the basic methods of construction have largely remained unchanged over the past several decades from the excavation of soil by mechanical means, to the placement of concrete, to structural-steel erection. However, there have been several significant changes within the processes of these methods. Changes in the energy, level of control, and functionality have made the equipment more productive. Changes in the ease of installation, curability, and modularity are characteristics of material technology that are significantly related to construction productivity improvements. Finally, information systems that support a project’s functions and its work crews have become more integrated and automated, which have also improved construction productivity. Looking forward, research is needed to examine how technology has changed the characteristics of construction output. This paper identifies two lines of research in this area: (1) an examination of how changes in the characteristics of new housing has influenced the accuracy of the Census Price Index, a major deflator used in the measure of construction output; and (2) an updating of the characteristics of the CII model plant, a hypothetical typical industrial project, for the purpose of developing a disaggregated price index model for the industrial construction sector. Doing so will help researchers and industry leaders understand whether productivity of the construction industry has actually declined or improved. More importantly, improving the accuracy of industry productivity measures will help develop effective industry strategies for improving the performance of the U.S. construction industry. ACKNOWLEDGMENTS A number of colleagues and organizations significantly contributed to the research efforts described herein. Although not exhaustive, a list of individuals who must be acknowledged include the following: Carl Haas (University of Waterloo), John Borcherding (University of Texas at Austin), Richard Tucker (University of Texas at Austin), Dong Zhai (University of Kentucky), Robert Glover (University of Texas at Austin), Carlos Caldas (University of Texas at Austin), Robert Chapman 

92 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY (National Institute of Standards and Technology), and Mohammed Yasin (University of Kentucky. Organizations that have provided funding and unique opportunities to support the writer’s research in this area include the Construction Industry Institute and the Alfred P. Sloan Foundation. REFERENCES Allen, S.G. 1985. Why construction industry productivity is declining. Review of Economics and Statistics 117(4):661-665. Allmon, E., C. Haas, J. Borcherding, and P. Goodrum. 2000. U.S. construction labor productivity trends, 1970-1998. Journal of Construction Engineering and Management 126(2):97-104. Bernstein, H. 2003. Measuring productivity: An industry challenge. Civil Engineering 73(12):46-52. Brandenburg, S. 2004. The Tier I Workforce Management Strategy: Concept and Application. Doctoral Dissertation, University of Texas, Austin. May. BRT (Business Roundtable).1983. More Construction for the Money: Summary Report of the Construction Industry Cost Effectiveness Project. New York: BRT. Burleson, R., C. Haas, R. Tucker, and A. Stanley. 1998. Multiskilled labor utilization strategies in construction. Journal of Construction Engineering and Management 124(6):480-489. Castaneda-Maza, J., R. Tucker, and C. Haas. 2005. Workers’ skills and receptiveness to operate under the Tier II Construction Management Strategy. Journal of Construction Engineering and Management 131(7):799-807. CII (Construction Industry Institute). 1986. Construction Industry Institute Model Plant. Publication 2-1. University of Texas, Austin. CII. 1988. Construction Industry Institute Model Plant Update. Publication 2-2. University of Texas, Austin. Dacy, D. 1965. Productivity and price trends in construction since 1947. Review of Economics and Statistics 47(November). Dyer, B., and P. Goodrum. 2008. Construction Industry Productivity: Examining the Effects of Omitted Variable Bias in the Census’ Price Index Models. Paper accepted for the 2009 American Society of Civil Engineers’ Construction Congress, April 2009, Seattle, Washington. El-mashaleh, M., W.J. O’Brien, and R.E. Minchin. 2006. Firm performance and information technology utilization in the construction industry. Journal of Construction Engineering and Management 132(5):499-507. Gomar, J., C. Haas, and D. Morton. 2002. Assignment and allocation optimization of partially multiskilled workforce. Journal of Construction Engineering and Management 128(2):103-109. Goodrum, P., and C. Haas. 2002. Partial factor productivity and equipment technology change at the activity level in the U.S. construction industry. Journal of Construction Engineering and Management 128(6):463-472. Goodrum, P., and C. Haas. 2004. The long-term impact of equipment technology on labor productivity in the U.S. construction industry at the activity level. Journal of Construction Engineering and Management 131(1):124-133. Goodrum, P., C. Haas, and R. Glover. 2002. The divergence in aggregate and activity estimates of U.S. construction productivity. Journal of Construction Management and Economics 20(5):415-423. Goodrum, P., D. Zhai, and M. Yasin. 2009. The relationship between changes in material technology and construction productivity. Accepted in Journal of Construction Engineering and Management 135(4):278-287. Gordon, R. 1968. A new view of real investment in structures, 1919-1966. Review of Economics and Statistics 50(4):417-428. 

APPENDIX D 93 Grau, D., C. Caldas, C. Haas, P. Goodrum, and J. Gong. 2009. Assessing the impact of automated materials tracking technologies on construction craft productivity. Journal of Automation in Construction 18(7):903-911. Gullickson, W., and M. Harper. 2002. Bias in aggregate productivity trends revisited. Monthly Labor Review. March. Pp. 32-40. Guo, S., and R. Tucker. 1993. Automation needs determination using AHP approach. In Automation and Robotics in Construction X: Proceedings of the 10th International Symposium on Automation and Robotics in Construction (G.H. Watson, R.L. Tucker, J.K. Walters, eds.). Amsterdam; New York: Elsevier. Han, J., and M. Kamber. 2000. Data Mining: Concepts and Techniques. San Francisco, California: Morgan Kaufman Publishers. Harrison, P. 2007. Can Measurement Error Explain the Weakness of Productivity Growth in the Canadian Construction Industry. CSLS Research Report No. 2007-01. Ottawa, Ontario: Centre for the Study of Living Standards. Hassell, S., A. Wong, A. Houser, D. Knopman, and M. Bernstein. 2003. Building Better Homes: Government Strategies for Promoting Innovation in Housing. Report to the U.S. Department of Housing and Urban Development, Office of Policy Development and Research, and the Partnership for Advancing Technology in Housing. Arlington, Va: RAND Science and Technology Policy Institute. Koch, J., and F. Moavenzadeh. 1979. Productivity and technology in construction. Journal of the Construction Division 105(4):351-366. Koebel, C., M. Papadakis, E. Hudson, and M. Cavell. 2004. The Diffusion of Innovation in the Residential Building Industry. Center for Housing Research, Virginia Polytechnic Institute and State University, and National Associate of Home Builders Research Center. January. Krantz, L. 2002. Jobs Rated Almanac. 6th Edition. Fort Lee, N.J.: Barricade Books. McGraw-Hill, Inc. (various years). Dodge Cost Guides. New York. Means Company, Inc. (various years). Building Construction Cost Data. Philip Waier, sr. ed. Kingston, Mass. Mohammadian, R., and S. Seymour. 1997. An Analysis of Some Construction Price Index Methodologies. Prices Division Analytical Series, Cat. No. 62F0014MIB—No. 2. Statistics. Canada. NAHBRC (National Association of Home Builders Research Center). 2001. Commercialization of Innovation: Lessons Learned. Upper Marlboro, Md.: NAHB Research Center. August 28. Oborne, D. 1987. Ergonomics at Work. 2nd Edition. New York: Wiley. O’Connor, J., and L. Yang. 2004. Project performance versus use of technologies at project and phase levels. Journal of Construction Engineering and Management 130:(3):322-329. Pappas, M. 2004. An assessment of implementation requirements for the Tier II Construction Workforce Strategy. Doctoral dissertation. University of Texas, Austin. August. Park, H., S. Thomas, and R. Tucker. 2005. Benchmarking of construction productivity. Journal of Construction Engineering and Management 131(7):772-778. Pieper, P. 1989. Construction price statistics revisited. In Technology and Capital Formation (D. Jorgenson and R. Landau, eds.). Cambridge, Mass.: MIT Press. Pieper, P. 1990. The measurement of construction prices: Retrospect and prospect. In Fifty Years of Economic Measurement: The Jubilee of the Conference on Research in Income and Wealth (E. Berndt and J. Triplet, eds.). Chicago, Ill.: University of Chicago Press. Richardson Engineering Services (various years). Richardson Process Plant Construction Estimating Standards. Mesa, Ariz. Rosefielde, S., and D. Quinn Mills. 1979. Is construction technologically stagnant? In The Construction Industry: Balance Wheel of the Economy (J. Lange and D. Mills, eds.). Lexington, Mass: Lexington Books. 

94 ADVANCING THE COMPETITIVENESS AND EFFICIENCY OF THE U.S. CONSTRUCTION INDUSTRY Rossow (formerly Koch), J. 1977. The Role of Technology in the Productivity in Highway Construction in the United States. Dissertation, Massachusetts Institute of Technology. Salter, W. 1966. Productivity and Technical Change. New York: Cambridge University Press. Srour, I., C. Haas, and D. Morton. 2006. Linear programming approach to optimize strategic investment in the construction workforce. Journal of Construction Engineering and Management 132(11):1158-1166. Stokes, H. 1981. An examination of the productivity decline in the construction industry. Review of Economics and Statistics 63(4):495-502. Thomas, S., S. Lee, J. Spencer, R. Tucker, and R. Chapman. 2004. Impacts of design/information technology on project outcomes. Journal of Construction Engineering and Management 130(4):586-597. Triplett, J., and B. Bosworth. 2004. Productivity in the U.S. services sector: New sources of economic growth. Washington, D.C.: Brookings Institution Press. Tuchman, J. 2004. Study concludes productivity is up, just not enough. Engineering News Record 253(1):18. Zhai, D., P. Goodrum, C. Haas, and C. Caldas. 2009. The relationship between the automation and integration of construction information systems and productivity. Journal of Construction Engineering and Management 135(10):746-753.