Years ago, a VP of projects for one of the big oil companies complained to me, “Why do these darned [he used a different word] contractors take on risks they can't manage?” If it had been a real question instead of an outraged whine, I would have answered, “Because you make them, and they want to eat.” But thinking about that incident led me to some insight into what contracting is really about: it's about risk allocation and assignment. At some level everybody on both sides of the table knows that, but we often do not behave as if we know it. During the awkward dance of contract negotiations, a forthright give and take around who is taking responsibility for which risks and why rarely happens.

In this chapter, I want to discuss what “risk” really means in projects. We often don't use the word very carefully or even in a manner that some would deem correctly. Second, I want to explore the principles of risk averseness and risk pricing and how those principles apply to owners and contractors. Then I will discuss the big areas of risk that need a careful look in every contract negotiation.

The Meaning of Risk

In its simplest form, risk refers to anything that is probabilistic rather than fully determined. By that definition almost everything is risky. When we speak of project risks, we are usually talking about things that could turn out worse than we expect. That doesn't narrow the field very much!

Strictly speaking, most of the things we call risks in projects are actually uncertainties. A true risk has a known probability distribution within a known range of possible outcomes.¹ Some refer to this as an actuarial risk. The distinction between risk and uncertainty is important because true risks can usually be at least mitigated by buying insurance of some form.

For example, there is a risk that a piece of equipment will be lost or damaged in transport. We can obtain insurance that will at least cover the cost of the equipment if not the consequential damages.² We can insure the piece of equipment because an insurance provider can figure out how likely damage in transport is and, therefore, what the insurance premium should be as a function of equipment cost. But note that we usually cannot insure against the damage that the now late piece of equipment did to our project. That is because the distribution of damages caused to projects from a late piece equipment is not known by an insurance company and varies greatly. Sometimes the damage to the project is essentially zero because the equipment was not on the critical path and the need to replace it doesn't put it on the critical path. Sometimes, a single piece of equipment can cause a project to be a year late and create large damage claims from contractors whose work is stopped.

There are relatively few true risks in projects compared with the number of uncertainties. Uncertainties come in different forms and sizes. In one form, the uncertainty is identified, the range of outcomes known, but the distribution of outcomes is unknown. For example, we might be trying a new piece of major equipment that has never been applied at scale in our intended use. The range of outcomes might be reasonably bounded: it will work as expected all the way to it doesn't work at all. Because the uncertainty is bounded, we may be able to hedge it by leaving space to replace the new equipment with the two pieces of equipment required in the past version of the process. Bounded uncertainties are more manageable than unbounded uncertainties, where we cannot define the range of possible outcomes. For example, engineering productivity is an unbounded uncertainty. While we can imagine how poor it could be, there is no natural limit. Finally, there are unknown uncertainties, those that never occurred to us as possible. They are fundamentally unmanageable.

In my experience with capital projects,

• True risks are few in number, which is unfortunate because they are easily managed via some form of insurance.
• Bounded uncertainties are plentiful but usually not too harmful. Much of what goes on a typical risk register are bounded uncertainties.
• Unbounded uncertainties are fewer in number but much more likely to cause significant damage.
• The unidentified uncertainties—the “unknown unknowns” as they are sometimes called—are either of the force majeure type³ or are cascade failures that we didn't see coming. Force majeure events are simply a fact of life in projects and in life in general and are usually dealt with in contracts in whatever manner the parties negotiated, even if it is often an area of intense negotiation. The ugliest and most important uncertainties are the cascade failures.

The cascade failures are amplifying knock‐on effects. These get more common as projects get larger and more complex but can occur on a project of any size.⁴ For example, the owner team skipped preliminary HAZOP in front‐end loading despite completing the P&IDs. When the HAZOP was done at about 60 percent design complete, major changes were indicated that set engineering and procurement back substantially, which caused materials to not arrive for the already mobilized mechanical contractor, which cratered piping productivity, which ended up with the E&I contractor invoking the cancellation clause in its contract due to the delay, and on and on until the project becomes an utter disaster in the field.

Contracts generally can do a decent job of assigning the bounded uncertainties but are less able to deal with the unbounded uncertainties. Where contracts really struggle is with the cascade failures. Going back to my homely HAZOP example: who would end up bearing the costs of the fiasco? On an EPC‐LS, I am betting the contractor pays despite the owner being mostly to blame. After all, any FEED errors and omissions were supposed to be found in the first 60 days and so forth and so on. On a Re/LS, the piping contractor loses money and makes claims against the engineer and the owner, some of which may result in some recovery. And I can guarantee arguments all the way up and down the line about who the culprit really is. The cascade failures are where the interdependent nature of project activities and the joint product nature of owners and contractors come together. The lines of responsibility that contracts draw tend to be fuzzy and often unfair in dealing with the cascade failures because they are not anticipated.

Principles of Risk Assignment and Pricing

Jeff Bezos walks into a casino and places a $100,000 bet on 00 at the roulette wheel. He is followed by a poor man who puts the deed to his house on exactly the same bet. Both lose the bet. How do we respond? To Mr. Bezos we say it must be nice to be that rich. To the poor man we urgently refer him to a mental health clinic to deal with his gambling addiction. Same bet, different meaning. The first principle of risk pricing is that the degree of risk averseness (if you are a rational person) is a function of one's wealth to the size of the bet. This is why a great many people will flip a coin with you for a dollar, but very few will flip the same coin with the same expected pay‐off (zero) for a million dollars. The downside of the bet has a very different meaning.

In the industrial projects world, the owners are Mr. Bezos, and the contractors are the poor bloke. That derives from owners and contractors being fundamentally different types of economic entities. Industrial owners are heavily capitalized, asset‐heavy firms. Contractors are thinly capitalized, asset‐light firms. Owners earn and are measured by earnings against capital assets. Contractors earn by selling the hours of their employees at a usually small markup.

Let's consider the different implications of a significant cost overrun on a project. If a contractor takes on a well‐defined EPC‐LS project and the owner is disciplined, any cost overrun is deducted from the contractor's balance sheet. Given that the balance sheet is asset light, a major overrun on a large project may put even a large contractor in financial peril. But if the project were done on an EPC‐R or Re/Re reimbursable basis and suffered the same cost overrun, that overrun is added to the owner's balance sheet as an asset. The owner doesn't like having to earn against a larger asset, but the overrun has very little effect on the company's financial health. Owners are from Mars; contractors are from Venus.

There are very practical implications of these differences for capital projects. Other things being equal, contractors will always want a larger premium (price) to take on a risk than the owner would. In my experience, owners often think that contractors gouge them with risk premiums on lump‐sum contract bids. What they fail to understand is that from the contractor's perspective, the implications of taking on a risk are different than they are for the owner.

Does this mean that it is therefore inefficient to assign risks to contractors? No, it doesn't mean that at all, as we shall discuss. But it does mean that owners should expect contractors to be very careful about taking on risks. Given the tendency of contractors to price assignment of a risk higher than an owner, owners need to carefully examine whether any particular risk transfer is cost effective. Owners also need to examine whether the project circumstances make it easy for the contractor to shift the risk back to the owner. For example, if there is a reasonably high probability that government actions will slow a project, attempts to transfer delay risk to the contractor will probably not work out because actions of government risk are rarely accepted by a contractor. So any premium the owner pays in the contractor's bid for delay risk is probably money wasted. Owners should expect contractors to seek to shift risks back to the owner whenever owner behavior or circumstances makes that feasible. This is one of the fundamental flaws in “collaborative contracts.” The asymmetry in risk‐carrying capability means that contractors will be sorely tempted to shift risk back to owners in risk‐sharing contracts, and they do. The net result is that even successful integrated project delivery projects are expensive for owners.

The second principle of risk assignment and pricing is that the party most capable of managing and controlling a risk should be assigned the risk and with it the downside of failing to manage the risk successfully. This is a normative principle that accords with common sense. An alternative formulation is that assigning a risk to the party with greater control of the risk lowers the price of the risk, and vice versa.

Who is more capable of managing a risk may be a matter of skills or a matter of circumstance or both. Whenever a risk is assigned to a party that does not have control of the risk, a potential moral hazard is created in the party that does.⁵ All contracting strategies create some moral hazards. EPC‐LS creates a moral hazard that the contractor will cut corners in design or construction because the contractor may not bear the full consequences for those problems. Depending on how the contract is written, a moral hazard may be created by the owner not caring about flaws in their FEED package because the contract stipulates a transfer of responsibility.⁶ Straight reimbursable or time and materials construction contracts may lead the constructor not to care about craft labor productivity because the risk is shifted to the owner.

Whatever moral hazard is created by a contract should become the focus of management by the other party. Smart owners using EPC‐LS focus on a strong corps of inspectors; smart contractors doing EPC‐LS focus on very deep evaluation of the FEED package in a short period of time; smart owners in reimbursable contracts focus on how they will manage productivity risks, and so forth.

The third principle of risk assignment and pricing is that assignment of a risk to a party with little or no control over the risk means that risk will go unmanaged. When an owner dumps a risk on a contractor that the contractor cannot manage, the risk doesn't go away; it goes unmanaged. If the contractor cannot find a way to transfer the risk back to the owner or to a third party and the risk materializes, the contractor loses money. But if that risk results in a poor project or a low‐quality asset, the risk all comes back to the owner and both the owner and the contractor lose money. Was that intelligent risk transfer? When working on contract terms and conditions with owners, I am sometimes quite amazed that owners believe that contractually transferring a risk actually solves the problem. Contracts do not change project reality. The reality is still there no matter what the contract says.

The fourth principle of risk assignment and pricing is there is no such thing as a shared risk. Risks must be assigned, not shared. That doesn't mean that the personnel of both owner and contractors may not be involved in the management of the risk, but the lead role in that management process will belong to the risk holder, and the consequences of failure to manage the risk successfully will accrue to the risk holder. Smart owners cooperate fully with contractors to assist in the management of risk that the contractor is assigned. Smart contractors cooperate fully with owners to manage risks the owner is assigned. But that is not a shared risk; that is good cooperative behavior and is normal.

One of the mistakes that owners and contractors sometimes make on EPC‐LS projects is to exclude personnel from the other party from helping with the management of a risk. Owners need to be ready and willing to support contractors who are struggling on EPC‐LS projects. This is why small execution owner teams are not positively indicated on EPC‐LS. And recall the example of the brilliant project director with the “bring me the bad news incentive.” That was an EPC‐LS project, but he knew that he had to be in a position to help the contractor resolve problems, and the only way to do that was to know about them in real time. Shared risks are your risks, not mine!

So, what do the principles tell us about how to transfer risk in capital projects between the owner and the contractors? First, the principles tell us that owners who attempt to do wholesale, complete risk transfer to contractors are going to pay a higher price for their attempts than they probably realize in both the short and long term. Long term, owners who push all risk onto contractors get a market reputation that affects the bids they receive. For example, we work with a national oil company that always contracts EPC‐LS and simply does not recognize as legitimate any project change. They require that dispute resolution be done locally and—mirabile dictu—they never seem to lose. What they do not seen to understand is that every contractor knows the game and bids 30 to 40 percent higher than they otherwise would. We see it in their benchmarks. Short term, owners who seek wholesale risk transfer will pay top dollar any time the market heats up. In the Middle East, for example, where most industrial projects are done EPC‐LS, the swings in market prices for projects can be as much as 100 percent from market low to market high.

Second, risk transfer decisions (and negotiations) should be guided by who has the greater control over a risk. For example, I would never advise an owner to take on construction labor productivity risk. Owners simply lack the construction management chops needed to organize and supervise field labor except for small projects, and some owners cannot even manage that. Sometimes owners get away with straight reimbursable or time and materials construction arrangements, but they are carrying a risk they do not know how to manage. Lump‐sum or unit rate construction is far more appropriate. Conversely, I would never advise a contractor to agree to some of the liability provisions that owners sometimes try to push into their contracts. For example, a provision that the contractor is fully liable for an event “in which owner is deemed to be primarily responsible” is a recipe for disaster unless the liability is capped at insurance, which many owners resist.

Third, contract negotiation should be done in the form of a “risk assignment workshop.” As I discuss in the next chapter, this is where a solid construction lawyer can do a world of good for both sides. A logic diagram similar to the one found in Figure 9.1 should be used in conjunction with a risk register. Starting with an identified and defined risk from the risk register, the first question is the controllability of the risk. If the risk is not controllable within the project, the risk goes into a separate bucket that will also include all force majeure type events. Those will all have to be negotiated, but those risks follow a different path.

If a risk is considered controllable, is the risk hedgeable? For example, potential losses due to foreign currency exchange are hedgeable. If a contractor is going to accept those risks, then responsibility for hedging needs to be explicitly addressed in the contract.⁷ If the owner is going to be responsible, the contract should stipulate that the owner will hedge or self‐insure. If the risk is not hedgeable, who has the greater ability to control?

What are the underlying assumptions on the ability of the party to control? For example, the lump‐sum construction contractor usually accepts responsibility for labor productivity, but there are a host of exceptions. For example, if engineered materials are incorrect, that is not the constructor's responsibility. If design has errors, the productivity loss normally ends up back with the owner or the engineer. If the mechanical contractor is working but the plot plan changes, all that comes back to the owner and/or the engineer. And what level of detail is assumed in the IFC design by the construction contractor? Rather than spelling out the conditions, many contracts leave it to later disputes to settle such issues.

In other words, contracts are risk‐assignment documents and should be treated that way. That is really their primary purpose. However, the risk assignment is often not thorough because no systematic accounting of project risks is used. When risks are explicitly assigned to a party, it is rare that there is any discussion in the contract, or even in the negotiations, about why the party was assigned the risk. As a result, there is not deep agreement about the assignment of the risk. Making the risk assignments clear and discussing them fully is not just a matter of writing a more solid contract. It is also a matter of making it more likely that the assigned risks will be fully understood and fully managed by the party to whom they have been assigned. Disputes most often occur because the party assigned a risk in the contract did not fully appreciate or accept the risk they were taking on. Clarity in risk assignment is one of the few areas in which the contract itself can actually improve the reality of the project.

In a well‐crafted contract, all of the true risks and all of known‐unknowns—those things identified in the risk register—will be assigned a risk owner or otherwise accounted for. The unknown unknowns, of course, still remain. When an owner has done a solid job of preparing the project in front‐end loading, most of the risks and known‐unknowns have been identified, and the potential for unidentified uncertainties has been driven down as much as practically possible. In such cases, it is much more likely that any unidentified uncertainties—those ugly unknown unknowns—can be successfully managed should they arise. Projects fail when there are so many moving parts due to unpleasant surprises that project management simply cannot cope.

Key Risk Areas

I look at areas of risk not as a lawyer, because I am not, but from the perspective of project management and control. A good construction lawyer looks at how the allocation of risks in the terms and conditions combined with the compensation scheme affect the probability of disputes between owners and contractors (i.e., which Ts and Cs generate claims that will result in dispute resolution at the end of the project)?

My focus, however, is on how risk allocation combined with the contracting strategy generate better or worse projects measured primarily by cost and schedule results. Like the good construction lawyer working on the front end of the project, I hate to see projects end up in arbitration or litigation, but I hate to see an expensive, poor‐quality asset produced even more.

In this regard, not all risks are equal. Inappropriate assignment of some risks tend to damage owners or contractors but have relatively little effect on the overall project. Other misassignments seriously damage projects and often take the contractor and its reputation as collateral damage.

For this discussion I am going to confine myself to a small subset of risk assignment areas that I see as especially problematic because they are associated with projects turning into complete disasters. As mentioned, these disasters are almost always due to the triggering of a cascade failure in the project. Often, the activity or failure to act that initiated the cascade had occurred before the contractor was even brought on board. Sometimes the five areas listed next are areas of contention and dispute between owners and contractors in the contracting process, but more often they are glossed over like the elephant in the room that no one wants to discuss for fear of getting the elephant upset.

• Schedule
• Site conditions
• Interface management responsibility
• Local content
• Labor shortages

When the contracts are lump‐sum, complaints about the previous items mostly come from the contractor and are addressed to the owner. When the contracts are reimbursable, the complaints flow in the opposite direction. Either way, these are perennial areas of conflict.

Schedule

Three of four industrial projects are authorized on a schedule that will not be achieved. The average project starts with a schedule that is 85 percent of the time that the industry spends on average to complete that scope of work. This industry‐wide tendency toward schedule aggressiveness is the trigger for more owner/contractor disputes and more project disasters than any other factor.

Schedule aggressiveness at FID means that delay will be the norm and the big issue becomes who pays for that delay. In my experience, who pays is often completely unhinged from who caused the delay. Therefore, the result is often unfair and comes down to who was more clever in negotiations on contract terms or who is better at playing “claimsmanship” games.⁸

Delay is never free, and in today's industrial projects world, delays in schedule are the sorest point in project performance. In the database underpinning this book, the average slip in the execution schedule from FID to mechanical completion was 18 percent. That contrasts with only 4 percent average cost growth. As is clear from Figure 9.2, incomplete front‐end engineering design is the key driver of engineering slip, which in turn drives construction and total project execution slip.⁹ Note that when FEED is complete, most of the slip in engineering can be accommodated, and the project ends up about 12 percent late. When FEED status is preliminary, the project slip jumps to more than 20 percent. Significant schedule slips are almost always cascade failures.¹⁰

The fact that FEED is the primary driver of slip is quite important for how we think about delays. FEED is always the owner's responsibility except in duty spec contracting. Even when the form is F‐EP or F‐EPC, the decision about whether to complete FEED prior to mobilizing detailed design is the owner's decision, not the contractor's. When FEED contractors who are going to continue into execution on a reimbursable basis encourage the owner to start detailed engineering with incomplete FEED, they are creating serious mischief, causing the whole project to slip, and likely enriching themselves with many extra hours in the process due to out‐of‐sequence work. When owners cut FEED short, they are engaging in moronic behavior that will do the same.

Failure to complete the FEED work harms more than project duration. It also means that the FID estimate is not a quantities‐based material take‐off estimate¹¹ and therefore may be quite unreliable. It means the quantities baseline for project controls and, tellingly, control of engineering cannot be established, which makes setting target hours for engineering difficult if not impossible. Finally, it means that the project schedule is unlikely to be resource loaded. When FEED was not complete, fewer than one project in five had completed the schedule and execution planning work and two in five did not even have the critical path defined.

When the owner does not complete FEED on EPC lump‐sums, they are passing a lot more risk to the contractor. Most of the time the contractors then price that risk into their bids. Sometimes, however, depending on the situation and their knowledge of the owner, the contractor will “bid their claims.” That means when they see holes in the FEED package, they know that significant changes will be forthcoming, which enables them to bid lower, thereby winning the job and then recover cost with expensive changes. The cost competitiveness of EPC‐LS with incomplete FEED is 13 percent poorer (higher cost) than the EPC‐LS with completed FEED (Pr.|t|<.0001). Almost all cascade failures start with incomplete FEED and execution plans. One has to look hard and long to find a disaster project with completed FEED.

The risk mitigation strategy for owners in dealing with schedule problems with contractors is clear: complete FEED and construct a fully resource‐loaded integrated schedule. The importance of FEED has been discussed, but the importance of the resource‐loaded schedule is almost as great. The definition of the critical path establishes the floor on a project's duration but doesn't ensure that the duration is actually feasible. The resource loading combined with a probabilistic risk assessment does establish feasibility. For example, when a schedule is properly resource‐loaded, it tells you how many engineers by discipline will be needed to achieve the planned engineering schedule. It tells you field labor requirements over time by trade. Those numbers can then be tested for reasonableness. If the schedule requires 400 pipe designers and the engineering firm has only 200 available, you are going to be late. We have seen projects in which, if they had resource‐loaded the schedule, the owner would have realized they would need more qualified alloy welders than existed in the entire region. Realistic schedules do not ensure that delay will not occur, but unrealistic schedules ensure that it will.

The failure to complete the owner front‐end work is perhaps the greatest confounding factor for contracting. I call it “the mother of failure.” Some argue that the line demarcating the boundary between FEED and execution is fuzzy at best and nonexistent at worst. The line is, of course, artificial. But it is not arbitrary. Until the P&IDs are completed and a schedule is complete, there is no possibility of a definitive baseline from which to assess and control execution. Contracting is made more difficult because in the absence of a completed front end, it will be difficult to assign responsibilities (risks), and it will be difficult to separate causes and effects when it comes to reckoning results.

When a reimbursable FEED contractor who will continue on into detailed engineering whispers in the ear of the owner PM or business sponsor that completing FEED will cost the project time, they are factually incorrect, and their argument is very self‐serving. When FEED is not complete, the control of engineering hours is all but impossible. The median (half above, half below) slip in detailed engineering is 27 percent for projects with incomplete FEED, and the mean is 52 percent. Many more total engineering hours are consumed than on completed FEED projects after controlling for size.

For EPC‐LS contracts, the definition of excusable delay, ownership of float, the aggressiveness of the schedule, and the completeness of front‐end loading must be considered together. If the contractor is going to be required to pay anything for delay through LDs, penalties, or through cost overruns to achieve schedule, it should not be for failing to achieve an unachievable schedule. Yet that is the situation in which we often find ourselves.

The owner's response is that the contractor agreed to a given schedule, so why should I provide relief? This response contains an important premise as well as a failure to acknowledge the asymmetry in the owner/contractor relationship. The unspoken premise is that the contractor actually knows how long the project will “normally” take. In fact, the contractor's view of how long the project will require is an estimate at best. Without a detailed integrated, networked, and resource‐loaded schedule, the contractor does not know how long the project should take, and neither does the owner. The poorer the quality of the schedule at FID, the more aggressive that schedule is (Pr.|t|<.02). The asymmetry in the relationship is that contractors feel they must bid the owner's schedule or forego any real chance of winning the contract. A similar dynamic often plays out in reimbursable contracts as well, where the contractor must profess the belief that the owner's schedule ambitions are reasonable or risk losing the work. The root cause is the owner business sponsor's drive for speed, but the critical enabler of the problem is the lack of complete FEL.

There are a great many other examples of the same basic dynamic, such as time extensions for events beyond the contractor's reasonable control. I cannot begin to cover all of these circumstances, but holding contractors responsible for things they do not control can easily ruin a project. Things that happen early in execution are especially damaging.

Site Conditions

When thinking about site conditions, I am referring to much more than soils, although that is part of the set. I also include site access and logistics, laydown, the local regulatory and approvals process, the site health and safety requirements, local materials availability, and the nature of the local construction labor market.¹² Site conditions are a third of the front‐end definition of projects.

Adverse site conditions that are not discovered as part of front‐end loading are ideally suited to triggering cascade failures. Discovering hazardous materials in the soils at the start of digging foundations will in some countries, such as the United States, often shut the project down completely while soils remediation is done. Why is anyone ever shocked to find hydrocarbon contamination at a 100‐year‐old oil refinery?

We have a significant number of projects, usually large projects, where poor access to the site for construction labor led to hours‐long delays. Workers either receive pay for the time or they walk. Sometimes site access requires the agreement of a third party, and when that is not forthcoming, the soils work cannot be completed. We have cases in which delayed site access meant that arriving equipment and materials had no laydown area and were dispatched to remote warehouses or left out to rust. Anything that disrupts construction just as it is starting usually triggers a cascade failure. But those things usually had their genesis in front‐end errors or omissions.

In EPC‐LS projects, the contract terms will determine who is charged with paying for the sorts of debacles cited earlier. The problem is that the consequences cascade failures are so large that the contractor cannot carry the risk, which means that one way or another the risk will return to the owner.

Site conditions are second only to schedule as a source of owner/contractor disputes. The effects of site conditions on cost growth are quite uniform across all contract forms. For each point on a 12‐point site definition scale, cost growth increases about 2 percent for the EPC forms from the FID estimate and about 1 percent on split forms (Pr.|t|<.001). The effect of site conditions on execution schedule are more sensitive to contract form. Each point on the scale is associated with nearly 3 percent of schedule slip for the EPC‐LS, EPC‐R and EPCM forms (Pr.|t|<.0001) but has no association with slip on Re/LS and Re/Re (Pr.|t|<.87).

These differences in response between the EPC and split forms provide some richer insight into how these contracting approaches differ in practice. Inadequacies in defining the site‐related issues are associated with finding more changes, and there is no reason to suppose the changes found are a function of contract type. However, in the EPC forms, the engineering and the field work are tied together in a single contractor, and the site definition deficiencies slow the project and construction in particular. The construction time on EPC forms increases at 3 percent per site definition point, right in line with overall execution. But in the split forms, the site issues are being caught and incorporated and have no effect on execution or construction time at all. On EPC forms, including EPCM, the EPC contractor, not the owner, decides when to go to the field, and they often decide to go too early. In the split forms, the changes translate into much less schedule slip than in the EPC forms because they are resolved before we move to the field or fabrication where they are more likely to be critical path. Split forms tend to be more robust against unpleasant surprises and reduce the possibilities of cascade failure.

Interface Management

There are interfaces to be managed on every project, large and small. The responsibilities for interface management are always divided in some fashion between the owner and the contractors. The assignment of interface responsibility should be part of the risk assignment process in the contract. Often, however, interfaces are left without clear assignment, or one of the parties doesn't want to live with the assignments made. And sometimes the assignments are joint, and that is akin to shared risk.

The contract terms should explicitly account for and assign the following interfaces:

• Dealing with government agencies at every level
• Community relations
• Site approvals
• Environment, health, and safety approvals
• Interface between project and site operations
• Partner investors and other external stakeholders
• Special transportation and logistics issues, especially those requiring extensive government interface, such as module transportation to site
• Equipment vendors and how owner frame agreements interact contractor vendor selection
• Subcontractors, especially owner‐directed subs
• Third‐party projects needed for successful startup, such as power and other infrastructure
• Specialty contractors such as scaffolding and heavy lift, etc.
• Disciplinary constructors on EPCM
• Prime‐to‐prime interfaces in multiprime arrangements

Regardless of contract form, every one of the previous interfaces needs to be assigned responsibility in the contract terms. Unassigned interfaces are likely to be unmanaged or managed poorly. The problem with interface management often starts with a failure to carefully and fully define the boundaries of the scope of work, which is where interfaces occur.

Obviously, the assignment should, like any risk, be based on the ability of the party to manage. Often, the assignments are assumed. For example, in EPCM, it is often assumed that interface with disciplinary constructors will be the responsibility of the EPCM. However, some owners believe the disciplinary subs should be responsive to their direction and the assumption proves wrong. Confusion follows.

There are some general rules of thumb in interface assignments. Government interfaces should be owner‐managed because government officials often balk at dealing with contractors instead of the ultimately responsible party. Stakeholder management should be with owners because contractors have no role in the allocation of project value, which is the essence of stakeholder alignment and management. Prime‐to‐prime interfaces should be owner managed and not self‐managed or PMC‐managed. The interface with site operations should be owner managed because contractors lack influence within the company. Most of the other interfaces should have contractor leadership. However, for every case there is a counter‐example.

The problems with interface management are usually created by owners. Sometimes, the owner team wants to minimize its involvement, even on reimbursable projects. The team is not adequately staffed to manage the usual owner‐managed interfaces, and they are turned over to the contractor by default. The contractor did not anticipate being responsible for the interface because it was not clearly assigned or doesn't actually have the requisite skills to manage the interface. It is essential to remember that no prime contractor has contracts with other prime contractors and therefore lacks the wherewithal to manage them. On EPC‐LS projects, some owners want essentially all interface management to be done by the contractor because they believe it helps maintain “the wrap” nature of the contract. That works only if the contractor knows he will be responsible and is able to marshal the skills to do it.

Owners can also go the other way—they dabble in everything. In one case I will never forget, the owner decided they would do management of the large multiconstructor laydown area for the project. Nobody could find anything in the laydown yard, and labor productivity plummeted. The owner generated an interface that didn't need to exist. Sometimes owner procurement dabbles in the contractor‐to‐vendor interfaces because they have an agreement with a vendor that supplies equipment and they want that agreement used. That can be done as long as it is clearly laid out at the beginning of the project, often in the ITB or even prequalification.¹³

The interface with third parties is a particularly difficult area. By definition, contractors have no control over third parties. Generally, the owner has more influence with a third party than the contractor and should therefore take the risk assignment. But if the third party fails to perform, that could open up major liability for the owner if the third party interferes with the contractor's work. In EPC‐LS, that's a risk that smart contractors will price handsomely if it is passed to them. But this is perfect case of passing a risk that the assigned party cannot control and then imagining that the problem is solved. It may be priced by the contractor, but if the third party does not perform, the damage to the project will likely dwarf the damage to the contractor.

I have two recommendations in interface management, one obvious, one maybe not. First, do not attempt to dissolve interfaces that are really there. Clarify those interfaces and plan and assign their management instead. Second, network all of the interface managers, both owner and contractor, to promote cooperation, and inform and improve interface management. Above all, never dabble; either take on responsibility for an interface or fully and completely assign it to a contractor and then formally require all contractors to cooperate in its management.

Local Content

Local content is recognized as an important risk issue in large international projects in the developing world. However, local content is an issue for most projects, large and small, everywhere. On the large international projects in which local content is mandatory, onerous, and prescriptive, that is, very hard to meet, contractors add substantial risk premiums to their bids for EPC‐LS work. In cool‐to‐normal markets, owners should expect all international bidders to add 10 to 20 percent to their bids when local content provisions are onerous. In hot markets, contractors will sometimes try to add enough to their bids for such projects that they will not be unlucky enough to win them. Onerous local content provisions on EPC‐LS contracts are viewed by the contractors as almost untenable. In some locales, contractors view onerous local content as a “corruption required” provision.

I am not suggesting that local content provisions are automatically bad or not legitimate exercises of sovereignty. What I am suggesting is that local content provisions where local markets cannot provide the required content competitively and in a timely manner are enormously risky for contractors. Owners in such circumstances need to understand they are paying handsomely for the local content requirements. Host governments need to understand it as well.

What is often overlooked are informal local content provisions in projects everywhere. Every petroleum refinery, chemicals complex, or mining site has a need to keep the community in which they reside happy. One element of that process is to funnel work to local contractors and suppliers. Almost every directed subcontract by an owner is local content at work. And like local content elsewhere, it carries risks for the prime. Of course, overtly directing work carries risk for the owner as well, which is why local content motivated subs are often “encouraged” or “introduced” rather than directed. My view is that prime contractors should always fully review what goods and services can be provided locally in a competent way and bid or otherwise employ those providers. When an owner points to a particular supplier, that situation needs to be viewed with the same suspicion as onerous local content requirements on large international projects.

The most insidious local content situation is where government approvals, at any level, become contingent on the use of a particular contractor or a local contractor, however local is defined. This is rarely done in a transparent manner, but as an informal and unstated requirement. This is far more common in North America and Europe than generally recognized and acknowledged.

Contractors that have worked in a local market for any length of time have already learned to navigate the local content requirements. The contractor who is new to the market, however, is in real danger of being blindsided by requirements that do not even seem to be there.

Labor Availability

For every on‐shore project, regardless of location, the availability of construction labor is a concern. Even for the fabrication of offshore platforms and modules, the availability of labor at the prequalified fab yards is an important issue. The labor situation for a project has a large effect on the choice of contracting strategy. When labor is tight, EPC‐LS becomes more expensive or simply unavailable. Owners usually then resort to EPC‐R or EPCM as their fallback. Rarely does the owner who was looking for an EPC‐LS turn to one of the split forms. The things that pushed them to EPC‐LS, primarily their own staffing weaknesses, push them to another EPC strategy with a different compensation scheme rather than to a split form. When an owner is forced to abandon EPC‐LS as a strategy late in the front end or even after no acceptable bids are received, the resulting project is usually problematic because the project team has not prepared for a reimbursable strategy.

Poor labor availability rarely tanks projects, although it can play a role when other factors are in play. A tight labor situation extends execution and increases cost some, but by itself is not the cause of disaster. Depending on how it is handled, poor labor availability may drive poorer productivity. If, rather than slowing execution to accommodate the supply of productive labor, the owner (or contractor) attempts to maintain schedule by bringing in less qualified labor, productivity suffers substantially. Mixing high skill craft with low skill craft generally drags down the productivity of all. If the owner is smart enough to extend the schedule to keep craft competence high, the schedule slips, but the effect on cost is minimal as it affects only time‐related costs. Planning a longer schedule to keep productivity up often leads to a faster project than staffing up with less productive labor.

The assessment of labor availability for a project is part of the owner's normal front‐end loading requirements in execution planning. Owners completed what we at IPA deem a definitive labor assessment as part of FEL in only 55 percent of the projects in our sample of 1,148 projects. A definitive assessment includes a full craft labor survey. Somewhat perversely, owners did a better job assessing labor availability on smaller projects than larger projects, probably because it is easier. Unfortunately, it is more important for the larger projects.

To state the obvious, assessing the labor availability situation for a project does not change labor availability. We would expect the failure to assess the labor situation to affect predictability of cost and schedule more than competitiveness. It turns out that understanding labor availability does significantly improve cost and schedule predictability—less cost growth and less slip. But it also is associated with significantly better cost competitiveness, which suggests that intelligent adjustments are being made when labor is known to be tight.

During hot labor market periods, owners are not making optimal adjustments in contract strategies. Yes, they are responding by reducing EPC‐LS, but that is undoubtedly more forced than by design. They run primarily to EPCM and actually reduce their use of Re/Re. But in hot market periods, Re/Re is far and away the most cost‐effective contracting strategy available, even if construction is not done with unit rates. It also provides better schedule performance than EPCM. Even in hot markets, owners are better off hiring a construction management firm to supervise reimbursable construction on a Re/Re form than resorting to EPCM. It requires very little change in their staff capabilities to go that route.

The other recommendation I would make in hot labor markets is to discuss with the business sponsor the relaxation of the execution schedule requirements. I am not recommending slower projects. I am urging recognition that the project will be slower in a hot labor market so that it need not be a lot more expensive as well. I recognize that this hope is probably in vain because the circumstances that create a hot labor market are usually created by hot product markets.