CHAPTER 11
The Effects of Scale
Large projects worry everyone involved because there is always a great deal at stake. They are very visible and can be career‐breakers and even company‐breakers. It is accepted wisdom in the projects literature and community that project outcomes tend to degrade as projects become larger and more complex.1 There is much less agreement with regard to why larger and more complex projects tend to fail. Flyvbjerg et al. suggest that the optimism bias2 and “strategic misrepresentation” are the primary causes.3 Without denying the reality of the optimism bias and the capacity of humans to lie, my colleagues and I have demonstrated empirically that the same problems that lead smaller projects to go poorly cause large complex projects to fail miserably. Those problems start with objectives that are unreasonable, with owner teams that are missing key functions, and with front‐end loading that is incomplete. Problems that would be damaging but manageable in smaller, simpler projects cause havoc in megaprojects because megaprojects are necessarily tightly woven projects and are therefore more susceptible to cascade failures.
Project performance degrades incrementally with size until projects approach about $300 million. Then degradation of project outcomes begins to accelerate. By the time project size approaches $1 billion, the average project performance is significantly worse for cost effectiveness, cost overruns, slip in the execution schedule, and operability. This degradation causes large losses of project value for owners and their shareholders.
Contracting strategy intersects with size and complexity in a number of ways. Understanding the difficulty that size and complexity cause for contracting makes it possible to side‐step the worst effects. First, as size increases, it becomes more likely that the project is contracted as a one‐off event. The only contracting strategy that is not one‐off that can accommodate megaprojects is repeat supply chain. However, RSC contracting tends to be limited to a particular region and a particular type of project. Partnering alliances could in principle be used for large complex projects but seldom were used even in the heyday of that strategy in the 1990s. A megaproject would exceed the capability of most partnering alliance contractors and would be disruptive to the workflow of all other projects within the partnering alliance purview. As discussed in several places already, a one‐off engagement contract is not self‐enforcing.
When contracts are not self‐enforcing, the principal‐agent problem at the heart of the contracting problem comes to the fore. The principal‐agent problem is exacerbated by the lack of transparency in the contractor's behavior on large and complex projects. Small projects provide line‐of‐sight management possibilities that are not available in large complex projects. This places much more burden on the contract to define the rules of engagement.
As project size increases, some elements of owner practices tend to improve such as clarity of objectives, assignment of roles and responsibilities, and owner control. Others, however, tend to degrade, and that degradation affects the effectiveness of some contracting strategies. The most important practice that degrades with project scale is front‐end loading. Poorer front‐end loading degrades the cost performance of all of the main EPC contract forms and Re/Re at about the same rate. However, Re/LS is only marginally affected by poorer FEL. The Re/LS form forces enough separation between engineering and construction that the effects of poorer FEL can be mitigated. All of this mitigation disappears, and the effects of FEL return when the Re/LS form is used on a schedule‐driven project. Schedule acceleration occurs, but it is traded for cost performance.
Finally, as size increases, the number of organizations involved in a project increases. It becomes necessary to assemble the scopes into multiple packages so that the size of any one contract does not become too large for a contractor to handle. The larger the contract size, the thinner the market that is in play. Contractors are limited in contract size by their bondability and balance sheets. In principle, contractors could form joint ventures to improve the size of contract they can handle. Although that route is sometimes taken, it is known to be quite risky for owners because it is very dependent on the ability of the joint venturers to work together. Therefore, multiple packages with multiple prime contractors is the usual solution. Multiple prime contractors, however, mean more interfaces for the owner to manage and more opportunities for contractors to interfere with each other, which creates the basis for claims.
Packaging Scopes
Packaging scopes is an art form. The following elements have to be weighed and balanced:
- The number of prequalified and truly interested contractors that are capable of executing one or more of the scope packages. Good and early prequalification is a must.
- How much work the interested contractors can take on.
- How many contenders there will be for each package. Each package must have at least three and preferably four willing bidders.
- The state of the market.
- The strength of the owner personnel to manage interfaces.
- Local content constraints.
- If the desire is to bid the packages EPC‐LS but one or more of the packages does not lend itself to EPC‐LS, how will the mixing of contract approaches at the single site affect the work?
- Project logistics—do the defined packages enable the contractor personnel to access the site close to their work area, or do they cross another contractor's area? Is it possible to organize the packages so that each package and contractor will have its own laydown area? Can fences be constructed to separate the contractors' work areas? If some of the packages are to be fabricated as large modules, how does the delivery and installation schedule interfere with other contractors' work?
The potential for interference is perhaps the biggest concern. In a multiprime arrangement, interference risk generally belongs to the owner with the contractors only required to be cooperative to the extent reasonable.
When contracting with multiple primes, the contracting strategy must center around a single contracting approach, especially with respect to compensation scheme of the constructors. Mixing lump‐sum and reimbursable construction risks labor migration from the lump‐sum contractors to the reimbursable contractors in the event of labor shortage, not an unusual situation for large projects. If the nature of the work requires some mixing of constructor compensation schemes, the contract terms and conditions must provide very clear rules about hiring labor away from other contractors onsite. The enforcement of those rules requires a functioning access control process at the gates.
TABLE 11.1 How Contracting Strategies Respond to Increasing Scale
| Contract Strategy | Scale Effects | ||
|---|---|---|---|
| Cost | Overruns | Slip | |
| EPC‐LS | Moderate increase | No effect | No effect |
| EPC‐R | Sharp increase | Sharp increase | No effect |
| EPCM | Sharp increase | Sharp increase | No effect |
| Re/LS | No effect | No effect | Tendency to decline |
| Re/Re | No effect | Small increase | Tendency to decline |
| Functional Specification | Small increase | Small increase | Moderate increase |
| Design Competitions | No effect | No effect | No effect |
| Guaranteed Max | Small increase | No effect | No effect |
| Convert‐to‐LS | Uniformly high | Sharp increase | Moderate increase |
| Integrated Project Delivery | Sharp increase | Sharp increase | Moderate increase |
| Repeat Supply Chain | No effect | No effect | No effect |
Given all of the complexities of large projects, many owners are tempted to hire a project managing contractor (PMC) either to assist or to take over primary responsibility for managing the project. However, as we have discussed elsewhere,4 hiring a PMC often creates more problems than it solves. If the owner is hoping to follow an EPC‐LS strategy for most of the work, the hiring of a PMC will dissuade many contractors from bidding, particularly if the PMC will also be allowed to bid and execute any part of the scope. Today, very few contractors want to serve in the PMC role without access to an execution role. Increasingly we are seeing PMC contractors simply hire the PMC staff “from the street” and charge premium prices for doing so. Owners on large projects are better served hiring the same people as individual contractors and foregoing paying the large markup.
Referring to Table 11.1, of the 10 contracting strategies we have quantitatively evaluated in this analysis, only 4 show no tendency for performance to degrade their cost performance with increasing project size—Re/LS, Re/Re, design competitions, and repeat supply chain. For the most part, cost growth (cost overruns) follows the same pattern as cost effectiveness. The split strategies, Re/LS and Re/Re, remove the incentive of the contractor to start field or fabrication too early by changing contractors. The split form then allows the owner to effectively fix problems and re‐sequence work as needed between engineering and construction or fabrication.
Design competitions and repeat supply chain projects also display no performance degradation with larger size, despite being EPC‐LS contract forms. Design competitions tend to be extraordinarily well‐defined projects, and well‐defined projects tend to have no cost growth and limited schedule slip because all of the optimism that accompanies early estimates and schedules has been squeezed out by the good definition. Repeat supply chain projects tend to benefit from a learning curve for cost. The benefits of that learning curve accrue to the contractors rather than the owner, but given the cost effectiveness of the projects, the owner has no reason to be concerned about that.
Again referring to Table 11.1, three contracting strategies degrade cost performance very seriously as projects pass through about $400 million and are often disastrous megaprojects: EPC‐R, EPCM, and IPD. EPC‐R projects are often the most intrinsically risky. They are more likely to be remote, new technology, or very large brownfield projects. EPC‐R is also the form that owners usually end up with when their attempts to secure a traditional EPC‐LS falter due to lack of contractor interest. The surprise member of this sad group is EPCM, but the case narratives for these projects point straight to construction management problems. Very large EPCMs often end up with too many constructors for the EPCM to manage effectively. Inevitably when there are a large number of constructors, a few of them will not be very good and the EPCM will need to step in. Too often, they are not up to the challenge.
Convert‐to‐LS and IPD falter at large scale for different reasons. Convert to LS is a bad contracting form at all sizes, but scale makes it even worse. The moral hazards that it poses become more extreme as size increases. The “off‐ramps” might be workable at small and intermediate size but are altogether impossible at large size. IPD/alliancing is a very complex contractual structure. As the number of players involved increases (as it will with size), the chances of getting the multiple agreements needed in place in a timely way decline and the opportunities for opportunism by one or more of the contractors increase. In IPD, owners often believe that they do not have to actively control the project because the contractors are incentivized to manage the interfaces. This is a failure to economize on the need for trust.
Referring to the final column in Table 11.1, when we look at slip in execution schedule, the effects of scale are greatly muted, and two strategies, Re/LS and Re/Re, tend to improve slip as scale increases. I have controlled for the completeness of FEED in all cases. I believe the small response of slip to increasing scale is because slip is ubiquitous in all project sizes. Schedule slip is responsive to cost growth for all projects taken together (Pr.|t|<.0001), with slip increasing one‐third of 1 percent for every 1 percent increase in real cost. However, when I separate the data by contracting strategy, the effects of scale are small on schedule extension.
Notes
- 1 E. Merrow, Understanding the Outcomes of Megaprojects, Rand Corporation, 1988; E. Merrow, Industrial Megaprojects, Wiley, 2011; Bent Flyvbjerg, et al., Megaprojects and Risk, Cambridge University Press, 2001.
- 2 See Daniel Kahneman, Thinking Fast and Slow, New York: Farrar, Straus and Giroux, 2011; and James Prater, Konstantinos Kirytopoulos, and Tony Ma, 2017, “Optimism Bias Within the Project Management Context: A Systematic Quantitative Literature Review,” International Journal of Managing Projects in Business, Vol. 10, Issue: 2, pp. 370–385.
- 3 Perhaps because they tend to focus on government‐sponsored projects, those who believe optimism and misrepresentation drive project results do not appreciate that effective project system governance will neutralize the effects of the optimism bias and make misrepresentation very hard to sustain. I guess it is ironic that governments often run poorly governed project systems, but that is not an inherent feature of government project organizations.
- 4 See Merrow, Industrial Megaprojects, 2011 op. cit., pp. 299–300.