REVISITING PROJECT DELIVERY PERFORMANCE 1998–2018

REVISITINGPROJECTDELIVERYPERFORMANCE1998–2018New benchmarks for unit cost, delivery speed, costand schedule reliability.University of Colorado Boulder

Revisiting Project Delivery PerformanceDoes the design-build delivery system still outperform the alternatives?SUMMARYOur study updated the median performance benchmarks for project delivery systems using a new sampleof 212 contemporary projects. A comparison ofthese results to the 351 projects used in the 1998 CIIbenchmarks is shown in Figure 1. After 20 years, DBprojects are still delivered faster and with greaterreliability in cost and schedule performance. Thesemedian results are highly consistent with the averageperformance of each project delivery system that wecalculated through regression modeling.Authors:Keith Molenaar, PhD, DBIAK. Stanton Lewis Professor and Associate DeanUniversity of Colorado Boulderkeith.molenaar@colorado.eduBryan Franz, PhDAssistant ProfessorUniversity of Floridabfranz@ufl.edu2 Revisiting Project Delivery PerformanceCost Growth (%)4.83.42.21.90.9DBB CMRDBDBB CMR1998Schedule Growth (%)Twenty years ago, the Construction Industry Institute(CII) published a report titled, “A Comparison of U.S.Project Delivery Systems,” which benchmarked theperformance of design-bid-build (DBB), constructionmanager at risk (CMR) and design-build (DB) projects.The report examined data from over 350 projects ofvarying size, sector, complexity and location that werecompleted in the mid-1990s. The analysis revealedthat DB projects outperformed both CMR and DBB interms of unit cost, cost and schedule growth, and allmetrics relating to the speed of delivery. These resultshad a profound impact on how projects were deliveredin the construction industry. In the decades since thisseminal report, our industry has changed and therehas been considerable interest in updating the benchmarks for contemporary projects.Figure 1: Median Performance Comparisonsfor 1998 CII and 2018 CII/Pankow Projects0.9DB20184.42.50.0DBB CMR0.20.0DBDBB CMR1998Delivery Speed 01721DBB CMR1998DBDBB CMRDB2018Updated January 2019

METHODSDetailed project information was collected using asurvey questionnaire. A total of 212 projects werereturned and verified for accuracy: 53 of which wereDBB, 79 were CMR and the remaining 80 wereDB. Sixty two percent (62%) of these projects werepublicly funded and thirty-eight percent (38%) wereprivately funded. All projects were completed between2008 and 2013 and represented a variety of buildinguses, including:— Light industrial— Simple office— Heavy industrial— Complex office— Multi-story dwelling— High technologyBuildings ranged in size from 5,000 square feet toover 1 million square feet, with approximately 60%being less than 200,000 square feet. The unit cost todesign and construct these projects ranged from 50per square foot to over 1,200, with 55% reportingless than 400 per square foot. The projects weregeographically distributed across the U.S. and notconcentrated in any specific region (See Figure 2).In addition to the median performance analysis, asubset regression analysis was performed for eachmeasure of project performance: unit cost, costgrowth, schedule growth, construction speed anddelivery speed.This analysis identified, in equation form, the set of keyproject variables that explained the greatest amountof variation in each measure of project performance.These equations were then used to calculate theaverage performance expected for projects deliveredunder each project delivery system.To provide additional certainty in our results, casestudies were conducted on 9 of the best performingprojects and 7 of the worst performing projects. Eachcase study consisted of semi-structured interviewswith the survey respondent to understand how keyproject variables contributed to each project’s performance. We also documented lessons learned formeach project to inform a set of recommendations forfuture projects.Figure 2: Distribution of Projects Across the U.S.NUMBER OF PROJECTS133CII/Pankow 3

AVERAGE PERFORMANCEIn addition to median benchmarks, the 1998 CII study used regression analysis to calculate and compare theaverage performance of project delivery systems. Using the same method, our average performance results arehighly consistent with the median benchmarks and offer greater confidence in the comparison. Design-build wasthe best performing project delivery system in terms of both cost and schedule performance. On average, whencompared to other project delivery systems in our sample:Cost PerformanceSchedule Performance— DB projects are 1.9% less expensive than CMR ona cost per square foot basis and 0.3% less thanDBB.— DB projects see 3.9% less schedule growth thanCMR and 1.7% less than DBB.— DB projects see 2.4% less cost growth than CMRand 3.8% less than DBB.— DB projects are 13% faster than CMR during theconstruction phase and 36% faster than DBB.— From design through final completion, DB projectsare delivered 61% faster than CMR and 102%faster than DBB.COST RESULTSWith respect to unit cost and cost growth in oursample, DB has the best performance (See Table 1).These findings are consistent with the 1998 CII benchmark, however, the performance gap between projectdelivery systems has narrowed. On average, projectsusing DB are expected to cost 1.9% less per squarefoot when compared to CMR, and 0.3% less whencompared to DBB. Similarly, DB projects are expectedto average 2.4% less cost growth than a comparablyscoped project using CMR and 3.8% less cost growththan a project using DBB. The most surprising difference between the 1998 CII and current benchmarkswas in the cost performance of CMR. When comparedto DBB, CMR is now expected to cost 1.6% moreper square foot and have 1.4% less cost growth onaverage.Table 1: Cost Performance ComparisonPerformanceMeasureDB vs.CMRCMR vs.DBBDB vs.DBBR²Unit Cost1.9% less1.6% more0.3% less99Cost Growth2.4% less1.4% less3.8% less22Note: R² is the percentage of the variance in each performance measure predicted by variables in the regression model. Ahigher R², up to a maximum of 100%, provides greater certainty in the benchmark.For unit cost, the following conditions have the mostimpact in reducing the cost per square foot of thecompleted project:For cost growth, the following conditions are the mostinfluential in controlling design and construction coststhroughout the delivery process:— Higher team chemistry among the Owner, designerand builder (GC, CM or design-builder)— Use of a DB project delivery system— Open book contracting terms, such as a cost plusa fee with a guaranteed maximum price (GMP)— Lower initial contracted unit cost— Higher team chemistry among the Owner, designerand builder (GC, CM or design-builder)— Smaller gross square footage of project— Open book contracting terms, such as a GMP— Earlier involvement of the builder4 Revisiting Project Delivery Performance

SCHEDULE RESULTSDesign-build was the best performing project deliverysystem in terms of schedule growth, delivery speedand construction speed (See Table 2). Compared tothe 1998 CII benchmark, the differences in schedulegrowth across project delivery systems is tightening,while the gap in delivery and construction speeds iswidening. Projects using DB are expected to have3.9% less schedule growth than a comparable projectusing CMR and 1.7% less schedule growth than aproject using DBB. On average, DB projects are delivered 13% faster during construction and 61% fasterfrom design through final completion when comparedto CMR projects. Even more disparate, DB projectsare delivered 36% faster during construction than DBBand 102% faster over the entire project duration.Table 2: Schedule Performance ComparisonPerformanceMeasureDB vs.CMRCMR vs.DBBDB vs.DBBR²Schedule Growth3.9% less2.2% more1.7% less21Construction Speed13% faster20% faster36% faster88Delivery Speed61% faster25% faster102% faster89Note: R² is the percentage of the variance in each performance measure predicted by variables in the regression model. Ahigher R², up to a maximum of 100%, provides greater certainty in the benchmark.For schedule growth, the following conditions are themost influential in controlling the duration of the designand construction process:For construction speed, the following conditions aremost responsible for increasing the rate of completionof the construction scope, from notice to proceed tofinal completion:— Participation of the designer and builder (GC, CMor design-builder) in project goal-setting— Use of a DB or CMR project delivery system— Earlier involvement of the builder— Larger gross square footage of the project— Lower project complexity— Higher initial contracted unit cost— Private funding source— Simpler foundation systems, such as slab-on-gradeFor delivery speed, the following conditions are mostresponsible for increasing the rate of completion of theentire project, from design initiation to final completion:— Use of a DB or CMR project delivery system— Larger gross square footage of the project— Higher initial contracted unit costCII/Pankow 5

BEST PERFORMERSWORST PERFORMERSAcross the case studies of the most successfulprojects, there were two recurring themes:Across the case studies of the least successful projects, three themes emerged:— The Owner placed a high emphasis on creating arelational project culture— Lack of experience with the project delivery systemor project management in general— Repeated use of the same designer and/or builder(GC, CM or design-builder)— Poor communication between the Owner and thebuilderEach Owner had a different approach to shaping theproject culture. One Owner, in particular, made theirexpectations clear during the first meeting. They challenged each member of the design and contractor’steams to not tolerate arguments or unprofessionalism,be willing to express their opinion and to treat eachother fairly. Another Owner described a constant drivefor greater accountability and one project successfullyimplemented periods of co-location that began duringthe schematic design phase.The majority of Owners on the best performing projects had worked alongside either the designer or thebuilder, or both, on prior projects. Because of theseprior working relationships, they were more comfortable communicating and, specifically, more willingto share challenges or problems encountered on thejobsite with other team members.— Understaffing or turnover within the Owner, designer or builder’s organizationInexperienced project managers were frequently citedas the underlying cause of poor planning and inconsistent quality of installed work. In one case, the projectmanager for the contractor was new to the companyand assigned to a project with tight deadlines. He hadlimited experience and, as a result, managed reactivelyto current problems, losing sight of the overall schedule. Staffing was also a challenge on some of theworst performing projects. On one case, understaffingplaced a heavy workload on members of the projectteam and forced long hours. On another, frequentturnover meant that few team members retained acomplete understanding of the project. The result inboth cases was increased stress and animosity amongthe Owner, designer and builder.Figure 3: Project Delivery SystemsDesign-Bid-BuildConstruction Managerat SDESIGNCONSULTANTSSPECIALTYTRADESCONTRACT6 Revisiting Project Delivery TANTSSPECIALTYTRADES