Monday, October 17, 2011

Public Megaproject Overruns: Error or Lie?

As I've disclosed previously, in my real life, part of my job as an (un-)civil engineer revolves around estimating construction costs for heavy civil/transportation/marine projects.  These projects include small projects that I design, and sometimes work on the construction side, and $100M+ projects of which I am a team member.  One of the professional requirements for such work is the identification and mitigation of risks.

Risk based estimating, as opposed to traditional estimating approaches, attempts to proactively recognize threats and opportunities to a project's scope, schedule and/or budget. In this process, known and unknown risks are assigned relative weights as some will impact the project to a greater extent than others. Identifying risks (and opportunities) will also keep management and executives from being surprised as the project progresses. Certain risks are ongoing, whereas others pass with time. Although unlikely, a major earthquake could occur at any time, whereas a high bid cannot occur at the end of construction.

Risk assessment and management is something we constantly do everyday. Nobody would cross a street without looking for traffic, and managing projects should be no different. Regardless of the size of the project, risk may be both quantitative and qualitative. Market conditions pose a risk to the cost of a project. The loss of key personal is more difficult to assess numerically. Risk is something that should be assessed continuously (or at least daily). Once a risk is identified, it should be documented. Likewise, if a risk is retired, it should be removed and the reason(s) for doing so noted. Therefore, to be effective, a risk registry is a living document that must be utilized throughout the life of a project. In order for it to be proactive, it must be acted upon. A risk is an issue that has not yet occurred; an issue is a risk with a 100% probability of occurrence; and an action item is the implementation of the mitigation plan. A transportation analogy is defensive driving.

One advantage of risk based estimating, as contrasted to traditional estimating approaches, is the estimate is given as a range of values rather than a singular number. As the design progresses, the range is narrowed. As with traditional estimating, it is important that risk based estimates are not skewed due to pressure to advance a project. Unfortunately, when it comes to public megaprojects, which I'll define as projects with over $100M in costs, there is a tremendous amount of political pressure to downplay the impacts of risk.  There is a great deal of research on under-reporting projects costs by Professor Bent Flyvberg.  For more background his seminal paper Underestimating Costs in Public Works Projects Error or Lie? is essential.

In my experience and as shown with extensive analysis by Professor Flyvberg, the riskiest projects, no matter the size, are those where something that was previously unknowable becomes exposed through the work process.  Examples are building remodels and underground work.   For example, during a building remodel, asbestos or mold may be discovered when a wall is removed, therefore leading to costly abatement work.  In the case of underground work, the geotechnical boring test pattern may have missed a large rock that is only removable by expensive means, or contaminated soil may be discovered.  (Change orders are always more expensive than work as it was originally bid.)  There are several means of mitigating these unknowns.  More preliminary work, during design, may help to identifying potential risks, such as performing more borings.  Another means is for the owner to take the risk from the contractor by a contracting tool such as force account, where the owner's representative directs the work methods.

There are also more general methods to mitigate risk, such as contract cost escalation clauses, which ensure that the contractor is protected from material and/or fuel cost hikes, therefore eliminating the need for hedging bids to account for future price shocks.

Another method for estimating the cost of risks is to create a risk model that uses a Monte Carlo simulation to develop a range of costs.  The probability of occurrence and the cost range of possible risks (and opportunities) are fed into the model and a random number generator computes the range of costs from best to worst case.  This is considered a cost-risk model.  For example, hypothetically, if the project was constructed 100 times, and the 60th percentile is chosen as a baseline this means the amount of funds budgeted for a certain project will correspond to the 60th highest cost per the model.  However, for public projects, this strategy is subject to manipulation as the percentile chosen will dramatically alter the cost estimate as it is released to the public.  A case in point is as follows.

In Seattle, as the residents of Washington State are aware, the earthquake damaged Alaskan Way Viaduct is being replaced, not with an elevated structure in-kind, but with a highly expensive deep bore tunnel.  The project has already started.  A cost-risk model, using the base estimate, has been developed.  The cost of the project, as reported to the public, is based on the results of the 60th percentile, which is the current policy used agency wide.  Although the 60th percentile might be a conservative estimate for rebuilding an interchange, it is rather low for an underground project with many known unknowns and unknown unknowns.  As Professor Flyvberg has convincingly demonstrated, tunnel projects worldwide are the projects that most susceptible to underestimating costs.  One reason, other than the nature of the projects, are these types of projects often allow politicians and others an opportunity to make their mark in a big way.  At the same time, the policy makers responsible are usually long gone by the time the final cost come due, so their ego is appeased and they do not have to deal with the fallout from trying to pay for it.  The deep bore tunnel is another such case.  Since tax receipts are falling, this project is taking funds that could be used to repair roads and other less glamorous, but necessary, transportation work.

It isn't only economists that use statistics to lie.

No comments: