Tuesday, 12 April 2016

Procurement case study: Holyrood 1997

 The Holyrood Building
The story of the building of the new Scottish parliament house, The Holyrood Building, is another instructive case study. In 1997 it was announced that the new parliament building was to be constructed and the cost estimate was £40 million. A design competition was held, one year later in 1998, and a construction management contract was awarded to Bovis Lend Lease at the beginning of 1999. 

The initial client was the Scottish Office, then after work started this was transferred to the Parliamentary Corporate Body, and later the Presiding Officer and an architectural advisor were added. A report in 2000 identified poor communication between the client and contractors as increasingly costly.

Contracts were based on concepts, not on a detailed finished design. The concept design also changed during construction, with floor space first increasing from 11,000 m² to 18,000 m² and ending up at 33,000 m². There were also increased anti-terror security measures including toughened glazing, which added another 10% to the final cost.

This all led to very many variations, in a six-month period between October 2002 and May 2003 there were 1,825 architect instructions which led to 4,600 instructions to trade contractors, with nearly 5000 variations in six months. Despite a design freeze that began in April, in May 2003 there were another 545 architect’s instructions. The time and cost implications are obvious. A report by the Auditor General in 2004 concluded that over 2000 design changes to the project were a major factor in the cost overrun.

During construction the lead architect died and the project director resigned. There were conflicts between the two architectural practices involved, and between them and the contractors. The Fraser Inquiry was critical of their dysfunctional relationship, poor communication and their claim, “without basis” that the building could be completed for £50 million.

By June 2001 the costs had escalated to £230 million. In 2002 the cladding contractor went into liquidation and the price increased to £300 million. By the time the building opened in September 2004 it was three years late and had cost a great deal more than the initial estimates. The eventual cost reported in 2007 was £414 million, 10 times the original estimate.

An inquiry was set up and the report was published in 2004. A major focus of the Fraser Inquiry was whether the procurement method or the client had caused the problems. A construction management contract is typically used on integrated design and construction projects, which is its strength, but is not particularly good at managing budgets. The original budget was obviously much too low, and this led to a lot of unwanted and probably unnecessary publicity about cost overruns. The main findings of the inquiry were the unrealistic nature of the initial cost estimates, use of a procurement model that passed risk wholly onto the state client, conduct of the tendering process and choosing the contractor, and security concerns that added to the cost but could have been anticipated. The conduct of civil servants was questioned.

The Fraser Inquiry identified two fundamentally flawed decisions. The first was procurement using a construction management contract instead of a Private Finance Initiative contract. Second was the insistence on a rigid program. The Scottish Office decided to use a construction management contract to speed construction, but without evaluating the financial risks of doing so, and without asking Ministers to approve it. Officials decided that rapid delivery of the new building was to be the priority, but that quality should be maintained, so cost blowouts were inevitable.

The client was obsessed with early completion and failed to understand the impact on cost and the completion date if high-quality work and a complex building were required. In attempting to achieve early completion, the management contractor produced optimistic programs, to which the architects were unwise to commit. The main causes of the slippage were delays in designing a challenging project that was to be delivered against a tight timetable.

Perhaps most damning in the inquiry’s report was the finding that senior civil servants withheld information on the problems between contractors and architects and the rising cost of the project from ministers. Ministers were not informed of concerns within the Scottish Office over the cost of the project and officials did not take the advice of the cost consultants, a serious failure of accountability.

In the final indignity the completed project suffered from flooding, as the original site had been a brewery and there were hidden underground springs. Then in 2006 one of the beams in the debating chamber swung loose due to missing and damaged bolts and poor glue, and the MPs were evacuated.

The lesson that is usually drawn from the Scottish Parliament building is that the client did not know what they wanted and proceeded with a brief that was poorly developed. However, the conduct of the civil servants involved was, in my view, deplorable. After the Fraser Inquiry’s report was handed down there were further investigations by the Scottish Parliament into their conduct, however no action was taken against any of the individuals involved.

Clearly, when civil servants get involved in these large complex projects, guidelines on governance need to be established and rigorous standards need to be enforced. One of the recommendations of the report was that independent advisers should be employed and those advisers need to have direct access to ministers, without their advice being filtered by public officials.

Holyrood is an extensively documented project. The link to the Fraser Inquiry report is below, the Conclusions and Findings are the most relevant. There is a Wikipedia page that is comprehensive, including a timeline of cost increases and many links.

This is the second parliament procurement case study. The first was on the building of Westminster in 1837.

Saturday, 2 April 2016

The Ratchet Effect in Construction

An Economic Perspective on Construction Procurement 

The incentive problem in short-term contracting is the main issue addressed here. This problem in construction has often been seen from the perspective of the principal-agent problem, where the focus is on motivating the agent and monitoring outcomes. By taking the incentive problem as the focus of the discussion the emphasis shifts away from the relationship between the principal and the agent, which is well understood, to the effort a contractor will make to minimise costs or improve performance. This aspect of principal-agent relations has been a major theme in labour economics, the economics of regulation and elsewhere.

The process of procurement has a number of side effects. While the intention is to purchase, the method determines outcomes. For building and construction projects the method generally used is a form of auction, typically a common-values low bid auction, where bidder costs are the same or similar and the project is awarded to the lowest bidder. This process has all the characteristics of economic models of one period contracts in short-term contracting under information asymmetry. It is also the case that the major public and private sector clients are repeat clients as they regularly bring projects to the market, and this is equivalent to the two periods in models of regulation

Here some of the insights into the behaviour of suppliers or contractors from the economics of regulation is applied to construction contracting. The approach is interesting because it is now generally accepted that procurement can be treated as a subset of regulation, following the model developed by Laffont and Tirole, which treats regulation as a principal-agent problem, with the government as regulator the principal, and the regulated firm (in fact its manager) as the agent. The regulator can observe realized production costs, but not how much effort the firm puts into cost-reduction (a post-contractual hidden effort problem). Importantly, the firm knows more about cost-reducing technology than the regulator (a pre-contractual hidden information problem).

In their model there are two types of firms: low effort firms will not try very hard to reduce their production costs, while high effort firms will be very responsive to cost reduction incentives. Therefore the problem is modeled as one of information asymmetry, with the focus on discovering the manager’s type, whether they are a high-effort Type 1 or a low-effort Type 2.

The first type responds to contractual incentives while the second does not, so the principal can use incentives to induce more information revelation from the agent, i.e. to get the agent to disclose whether they are a Type 1 or Type 2 manager, and the regulator can make transfers to the firm. Such transfers are clearly necessary in the case of procurement, where the principal/client pays the firm/contractor for work performed under a contract to supply goods and/or services.

In this context what is called the commitment problem arises, because the optimum outcome possible in the first period, or round of tenders, cannot be repeated twice. The problem turns on the existence of asymmetric information. In each of the two periods the government/regulator wants to procure a public good, and if they could credibly commit to a long-run (two-period) contract the optimal two period outcome would be the same as the one-period optimum twice. They call this the perfect commitment outcome. The perfect commitment outcome requires credible commitment to a long-run contract.

If the regulator cannot make credible long-run commitments, long-run contracts are ruled out. With the regulator unable to write a long-run contract with the regulated firm, it has instead to govern the relationship by a sequence of short-run (one-period) contracts.

This gives rise to what is known as the ratchet effect, an outcome of the regulated firm’s unwillingness to reveal whether it is a Type 1 or Type 2 firm in the first period, because that would mean the regulator no longer faces asymmetric information, and allow the regulator to take any gains by the firm from, for example, cost reductions that might be the result of the firm’s efforts or use of new technology.

Laffont and Tirole proved that after period 1 the regulator will in general not know the firm’s true type. Intuitively, the ratchet effect implies that information unfolds slowly, as the manager tries to protect his information rents by not revealing his true type. Thus the ratchet effect happens when an agent works hard and shows a good result, but the principal then may demand an even better result in the future. Anticipating this, the rational agent has little incentive to work hard in the first place, and this tendency for performance standards to increase after a period of good performance is called the ratchet effect.

Early formal models of ratchet effects emerged in the 1980s, and ratchet effects were predicted in specific informational and contractual environments where hidden action and hidden information must be present, and the parties must be in a repeated relationship yielding some quasi-rents to both where binding multi-period agreements are not feasible.

How does this apply to building and construction?

Is it likely that construction contractors respond to clients’ requests for bids by attempting to preserve hidden information? Is it possible they will not want to reveal themselves as a Type 1 cost-minimising firm? There seems to be three reasons.

Firstly, it reduces competition to a straightforward shootout on price, but because all tenderers have similar costs this is just a decision on margin, based on current and expected workload. Therefore the competition in any given tender is likely to be driven as much by contractors’ workload considerations as their estimated cost of the project. Even without cartel arrangements this is a form of managed competition, whereby the tenderers will not deviate too far from the client’s expected cost for the project, which will also be similar to industry estimates, thus avoiding revelation of a significant cost advantage on one project that might jeopardise margins on future projects.

Secondly, it allows for gradual improvements in productivity and efficiency, which are neither disruptive nor expensive to contractors, but will deliver a windfall gain to the contractor if a project comes in well under budget or schedule, which may be the result of some innovation by the contractor. This gain will, of course, be hidden from the client and from competitors as much as possible.

This suggests that there might be many cost reducing innovations available to contractors at any time, but the pressure to apply them will be muted by market conditions and a contractor’s appreciation of competitors’ likelihood of using them. Innovation is used here as a broad term that covers any and all product and process developments that can reduce final construction costs. There are costs and risks associated with innovation, so it is in the interests of all bidders to minimise these costs to themselves.

Thirdly, the winning bidder will always have the option of revealing themselves to be a Type 1 firm, if for some reason they want to. There will usually be some innovation available that will reduce project costs, but will be costly (i.e. require upfront investment) to the contractor. Thus the success of many major contractors in winning repeat work through negotiation rather than tendering is explained. By pushing the innovation boundary to reduce costs on the period 1 project the contractor gets the period two project without tendering costs at the new level of the client’s price expectation. (This does not exclude more traditional methods of cost reduction such as cash farming or subcontractor oppression of course).

The general argument made here is that the ratchet effect in the procurement process used in building and construction (typically auctions of single projects) will limit cost reductions from productivity and efficiency gains by contractors and subcontractors. This is an outcome of the unwillingness of bidders to reveal their hidden knowledge to clients, who will then expect future performance at the improved level. This is because clients typically only offer a single project at a time, or sometimes a bundle of projects, instead of sequences of projects. Thus short term contracting under information asymmetry.