Journal of Structured Finance AcuLead

To appear in Journal of Structured Finance, Winter 2016

1

A Framework for Understanding and Modeling Risk in Mega-projects and its Impact on the Markets for Project Finance

Atanu Mukherjee1 and Purnendu Chatterjee2 {[email protected]}, {[email protected]}

Abstract This paper discusses our approach to characterizing risk of large projects from the perspective of project finance markets. Unlike current statistical and correlation based mechanisms of risk characterization, we evaluate the dynamic nature of complexity in large projects based on causality and explore its relation to risk. We then suggest a framework to understand and model complexity and risk so as to distil it into components, their interactions and their impact on access to project finance and risk premiums. We then argue that adoption of such a framework can have a transformative impact on mega-projects and the markets for project finance and suggest how innovations in project finance can be accelerated using our framework. We believe that infrastructure projects, project finance industry, banks, institutional investors, mega-project sponsors, governments and multi-lateral agencies will benefit from such a framework by being able to model, measure and monitor mega-project dynamic risk characterization so as to structure, organize and finance projects effectively on an ongoing basis. This we think will eventually expand the market by increasing the number of participants, leading to enhanced market liquidity and opportunity expansion in functional project finance markets. Introduction The worldwide market for infrastructure projects is estimated at over 40 TT$ over the next twenty years [1]. Large multi-billion dollar industrial infrastructure projects in areas like energy, steel, petrochemicals, and transportation can yield superior long-run commercial and social returns but are prone to cost and time overruns. These can often be severe leading to eventual project suspensions and abandonments. Megaprojects are complex undertakings, with uncertainties and changes being the norm and the success of the project depends on the ability to characterize and control risks across the project lifecycle. This means that for these projects to be successful they need to be structured, organized, financed and managed in a way, which accommodates change, characterizes risk while minimizing late cost, functionality and schedule impacts. A key requirement for mega-projects is the need for project financing. Project financing is different from traditional corporate financing of projects in that the repayment of debts is tied to the cash flows generated from the project alone, and any recourse to debt recovery in the event of default is collateralized only to the assets of the project. The corporations liability as a sponsor of the project is limited to its equity or debt participation and the project finance agencies recourse to recoveries cannot be collateralized to the corporations other assets and cash flows.


2

This separation is necessary to encourage sponsors to undertake large and complex projects without jeopardizing their current operations while financing it through project finance markets at an appropriate cost of capital reflective of the risks of the project. Financing projects at this scale requires the participation of capital markets, the government and multilateral agencies. Unfortunately, financial markets have proven unequal to the task of funneling savings from places where incomes exceed consumption to places where infrastructure investment is needed. In a world with such huge infrastructure needs, the problem is not a surplus of savings or a deficiency of good investment opportunities. The problem is probably related more to the underdeveloped frameworks, instruments and institutions for infrastructure investments, which incentivize a financial system to intermediate savings and investment on a global scale [2]. In the aftermath of the 2008 financial crisis, most of the existing structuring mechanisms and instruments for large project financing disappeared dealing a severe blow to the capital markets for project finance in more mature markets of Europe and the United States [3] . In the past one of the biggest constraints to project finance in mature markets has been the perceived dangers surrounding the initial construction phase of a project. This had lead to the muted development of project bond markets. On the other hand markets for project finance in developing countries - which has the most need for infrastructure capital - are underdeveloped. Large transaction values, long tenors, ambiguity and uncertainty which results in inadequate characterization of value and the risk spreads, are the primary deterrents to markets for non-recourse or limited recourse project financing. Thus the incentives to participate by banks, institutional investors and other financial intermediaries in financing such large projects are significantly reduced. Because of the large transaction values, the liquidity requirements from the market are also high, and absence of scale in market participation makes the markets illiquid. While mature markets have functioning project finance and limited project bonds market [4], the avenues for project finance in developing countries today are very few. The options in these countries are limited to some traditional large liquid banks and institutions like Exim and IFC funding projects through some forms of syndication. This has resulted in few participants leading to high market concentration and artificial bucketing of all mega-projects into the generally undesirable below the investable BBB category. Inadequate characterization of risks thus engenders conditions in a funds-scarce market for higher project risk premiums resulting in depressed project returns and often derailing the commercial viability of the project. This situation is largely because of the ambiguity in the understanding of the dynamics of large-scale projects consequently resulting in a lack of clarity on the understanding of project value, characterization and assessment of risk and structuring the project finance. Traditional methods of statistical risk modeling techniques applied to mega-projects fall short of providing any meaningful value in risk assessment due to the highly inter-coupled, feedback driven and non-linear nature of interactions in a mega-project. Additionally, the unique and craft nature of each mega-project limits the extent of reasonable analysis based on correlations of comparative data.


3

If the causative factors driving the dynamic behavior of projects, its management and hence its associated risks can be characterized, quantified and monitored in a more accurate manner, the transparency and understanding will result in meaningful assessment of dynamic risk profiles resulting in better risk assessments and investments at appropriately priced project risk premiums. It will also expand the market by attracting many more participants by drawing in large institutional investors (e.g. insurance, pension funds) whose investment horizons match that of the project tenors thus making the markets sufficiently liquid. Inclusion of more institutional investors to complement existing banks, and enhanced market liquidity also leads the way towards greater financial innovation and opportunity expansion in fully functional project finance markets. This can have a transformative impact on the industry and society. The project finance industry and mega-project sponsors need a framework to be able to model, measure and monitor mega-project dynamic risk characterization so as to structure, organize and finance projects effectively on an ongoing basis. Approach to Characterization of Risks in Megaprojects In the context of mega-projects we view risk as the possibility that events, the resulting impacts, the associated actions; and the dynamic interactions among the three may turn out differently than anticipated [5]. Risks and uncertainty combine with indeterminacy to create ambiguous decision-making contexts. Decision analysis methodologies that have emerged over the last fifty years have been applied to mega-projects with a hope to understand, anticipate and mitigate a mega-projects turbulence. Some of the risks that can be identified through statistical analysis have been applied with some success. The heuristics and correlation based mechanisms used to establish causal relations for risk characterization traditionally are weak as they systematically ignore feedback effects, multiple interconnections, non-linearities, time delays, and the other elements of dynamic complexity seen in large projects. Various methods to infer causality used today is based on things like temporal and spatial proximity of cause and effect, temporal precedence of causes, covariation, and similarity of cause and effect. These methods lead to difficulty in complex systems like large projects where cause and effect are often distant in time and space, where actions have multiple effects, and where the delayed and distant consequences are different than proximate effects. Further, the assumptions in the traditional models are typically at higher levels of aggregation of the project and often underestimate the tail behavior and non-stationary effects that are typically encountered in mega-projects The multiple feedbacks in complex systems cause many variables to be correlated with one another, confounding the task of judging cause through aggregate level black-box empirical observations. This leads to inaccurate representation of risk because the causality is frequently misattributed and we often hear consequences attributed to fat-tail effects and the likes. Understanding and modeling causality based on system structure and interactions is thus a determining factor in understanding the underlying behavior of these projects. In mega-projects indeterminacy arising from possible external and internal events, nonlinear interactions, feedbacks along with randomness and uncertainty


4

thus calls for a more careful and rigorous causality based approach to model and characterize risk. Our approach to modeling causality in a dynamically complex system like large projects is based on the fact that system structure gives rise to behavior and hence is key to understanding cause-effect relationships in terms of attributions, feedbacks, delays and non-linear and higher order effects. Essentially, our claim is that if we do not understand how large projects are structured, how the elements interact with each other at the project, macro and industrial sector level, and how things like change, shocks and delay effects propagate through the project - one cannot characterize large project behavior and scenarios and hence one will erroneously attribute causality and largely mischaracterize risk. In other words, establishment of causality based on current models is largely based on aggregate level black-box empirical observation and statistical inference in an "open loop" system. Causality estimation based on facts, data and intuition around system structure and behavior in the context of the macro-industry dynamics allows us therefore to understand attributions, feedback effects and time delays in a iterative top-down, bottoms-up manner in a "closed loop" system based approach. We characterize risk based on this model of causality by framing dynamic models of project structure and interactions and infer patterns of behavior based on hybrid simulations using techniques from system dynamics [6] and traditional Monte-Carlo mechanisms. The Nature of Risk in Megaprojects Megaproject risks differ according to types of projects depending upon the intensity of technical, market, and institutional difficulties that they pose to the promoters. Integrated steel plants, for instance, are technically complex and challenging to execute, but they typically face fewer institutional risks, as they are socially desirable because of their employment generation capabilities and industry multiplier effects. The market risks are moderate though as they are particularly appealing in developing countries with a high appetite for gross capital formation. On the other hand power projects in developing countries, while technically less challenging can have significant institutional and supply side market hurdles. This is particularly true if the state owns the resources and institutional mechanisms for resource allocation is in its formative stages. Whereas, petrochemical complexes have higher completion risks, low institutional risks and lower market risks. Since the output can both be sold in domestic and international markets, the primary market risk is that of volatile prices than of capacity utilization. Megaproject risks can be largely put in the following categories 1. Market Risks Market forecasts for steel, transportation and power projects are based on assumptions about the structure and drivers of demand and supply. Demand projections for high-volume commodity outputs from mega-projects, often turn out to be widely off the mark. In some cases, errors result from shortfalls in


5

overall economic growth and in others, because of the unanticipated nature of changes in the structure of demand. Dynamic demand models, which model the underlying causality and interactions, supplemented with traditional methods is, in our opinion, a better indicator of demand behavior and associated variability. Similarly, supply risks also involve price and access uncertainties. Particularly, in a developing country supply risks can be very important as nascent state institutional mechanisms can misallocate raw material supply capacities [7]. Understanding supply options and alternatives therefore becomes very important in the context of the market risk characterization for mega-projects. 2. Execution Risks The Achilles heel of mega-projects is in technical, engineering, design and construction risks during the execution phase. While mega-projects usually have mature technology, arguably the principal risk in cost and time escalations is in the design, engineering and construction phases of these projects. The nature and scale of impact of risks in this phase can be severe both in terms of direct project costs and importantly in terms of opportunity costs lost due to schedule escalation. The resulting tardiness in time to capture capacity and escalation of product price due to increased project costs can drive many projects uncompetitive in the highly competitive global commodity markets. The Chevron Gorgon LNG project in Australia jumped from by over 15 BB$ to over 50 BB$ [8] and with a looming supply glut of LNG, the viability of the project remains to be seen. Similarly, the Kashagan oil project in Kazakhstan, currently tracking at 50BB$, is 35 BB$ over budget and nearly a decade behind schedule [9]. Apart from institution related external shocks - poor project shaping ( aka as Front End Loading ) , changes in scope, productivity, workforce availability, ordering delays, project supply chain disruptions, rework cycles and technology uncertainties. can escalate a projects time and cost baselines many fold. In our evaluation and simulation of large projects using integrated steel plants as some of the candidate models, we find that interactions of these factors form the basis of execution risk in large projects. It has been argued that overall risk exposure could be minimized if risk could be assigned, allocated and transferred based on the capacity of the party to bear and control the risk in a large project [10]. Thus, frequently attempts are made by mega-project sponsors to control the execution risk by transferring the engineering and construction risk to turnkey contractors. However, if the project execution risk is not well understood and its impacts not reasonably well comprehended, attempts to transfer engineering and construction risks wholesale to an EPC (Engineering Procurement & Construction) contractor may well increase the risk of the project as the incompleteness of information is likely to leave sufficient gaps for re-negotiations, escalations and claims in the contractual mechanisms. We think that dynamic modeling of project behavior along with reasonably accurate stochastic characterizations of the risk levers in the execution phase give us better ability to characterize the dynamic behavior of the project during execution. This gives us the ability for better qualitative and quantitative attribution of risks during this phase which can then be used as a basis for allocation, assignment and transfer on a more reasoned basis. 3. Institutional Risks Mega-projects depend on laws and regulations that govern the appropriability of returns, property rights, and contracts in a country. Institutional


6

risks are greatest in developing countries, where laws and regulations are in formative stages or are being reformed. Most common among these, are the regulatory risks associated with delays and difficulties in obtaining approvals for environmental, land, or social permits. Government agencies in many developing countries have labyrinthine bureaucracies, which can delay and effectively endanger projects simply by not granting permits. Rules on competition, pricing, entry, unbundling, regulated rates of returns and other elements have adverse impacts on mega-projects in most developing countries. Sovereign risks involve the likelihood that a government will decide to renegotiate contracts, concessions, or property rights. Changes in rules, property rights, and so forth, triggered by general economic or political shifts are sovereign risks, which can lead to expropriation risks. The seeming unpredictability of these events and the degree of adverse impact on a project can well lead to catastrophic results, like what happened with Enrons Dabhol Power Corporation in India [11]. In hindsight though, the seemingly unpredictable event of reneging on off-take agreements in the Dabhol project was an event waiting to happen. A more reasoned approach to understanding the potential market side risks in the shaping phase would perhaps have revealed flaws in the business model of generating power based on naphtha and LNG at four times the prevailing local power price that was clearly unsustainable. Although, the disruption manifested itself in terms of institutional behavior the underlying driver was probably due to inadequate diligence in the shaping phase to understand the market risk. The ability to enumerate and model causality based on scenarios of known-unknowns can provide useful insights into the nature of the impact of such events on a mega-project. This provides the project sponsors and participants the ability to design options that can be exercised in the event such risks unfold. The ability to inject external events in a mega-project simulator during the project lifetime allows the project sponsors and participants to take more real-time decisions on risk trade-offs and alternative strategies for coping with institutional risks. The Dynamic Interaction of Risks over the Megaproject Lifecycle The market, execution and institutional risks in megaprojects interact over time in a way where the risks emerge and characterize themselves from the point of decision to initiate a project, through execution to the operation and end-of-life. A typical mega-project lifecycle has three broad phases and the dynamic interaction of risks manifests themselves in varying magnitudes during each of these phases as illustrated in Fig 1 below. 1. Shaping Phase - This is the initiation and exploration phase which has the highest

risk and is typically financed by risk capital of the sponsoring parties [12] . Working details to ensure viability, front end loading and pre-feasibility, identification of known execution risks, sponsor identifications, financing alternatives and evaluating options are key activities. While from a project finance perspective there is little, if any, capital committed by the financing institutions it lays the groundwork for all future phases and key to understanding, shaping and mitigation of risks in the later phases. However, in our experience with many


7

large projects, we find that more often than not sponsoring parties

2. Execution PhaseThis phase entails the design, engineering, construction and commissioning of a mega-project and has the maximum investment intensity. For instance project finance requirements for large integrated steel plants can range from 3 BB$ to 20BB$ over a period of 3-5 years and many risks emerge and evolve during that time period. Similarly, a LNG plant can range 20 BB$ to 45 BB$ and if not shaped and managed pro-actively, the execution risks can escalate the project cost many times. Diligent shaping has far reaching impacts on risk characterization and reduction strategies during this phase. Characterization and mitigation strategies for execution risks during this phase largely shape the financing options and the risk premiums of financing.

3. Operations Phase Transition to successful and continued operation requires

that market risks are understood and performance of the plant reaches its quality and capacity goals. While it is not possible to completely anticipate the supply and demand side risks during operations, the deeper understanding of the industry dynamics at both a macro and micro level can lead to a model framework that can be made sufficiently robust to indicate ranges of variation on demand and supply side parameters to address the effects of external shocks and variations. Insights on potentially knowable external shocks are industry expertise and knowledge dependent and a project can yield useful information in framing the project investment strategy and mitigation mechanisms through potentially exercisable options.

Fig 1- Phases in mega-project execution

It might be now useful to look at how the different risks interact over time across the mega-project lifecycle. Institutional risks, for instance, usually diminish soon after permits are obtained. However, they may re-emerge and affect the supply side or demand side of market risks. For example a retroactive promulgation of ordinance


8

canceling the mining block allocations for a steel plant can negatively affect the cash flows in the operations phase and adversely affect the returns temporarily or permanently. Pricing de-regulation may emerge well after the initiation of the mega project and may manifest itself in competitive demand side and supply side pricing, which will positively affect the cash-flows in the operations phase and hence the cost of capital. Institutional risks may also affect the execution risks due to duties, taxes and labor laws, which may positively or negatively affect the project cash flows. Execution risks on the other hand can be characterized as engineering design is initiated during the shaping phase, and construction plans are firmed up. As the execution phase evolves in the early stages and the parties get better visibility into the parameters, which have higher likelihood of changes and impact on schedule, resources and costs, the execution risks can be better understood. This will be able to provide the sponsors and the project financing institutions, a more objective and dynamic risk trajectory of project execution. Generally speaking though, the execution risks will start dropping after the initial engineering and design phase completions. However, it is impossible to know the magnitude of all the risks and shocks that a mega-project will encounter particularly during its protracted execution phase and to a lesser extent in its operations phase. Known risk levers, whose magnitude may be unknowable in advance but demand appropriate actions when they manifest themselves, can be tested for their effect in a model simulator. Our model framework acts as a management flight-simulator for the reasoned assignment of knowable risks within which risks are discovered, imagined, and assigned to a coping strategy. The mitigation and coping strategies can be simulated using financial mechanisms, institutional shaping, and project execution trade-offs. These what-if simulations have the ameliorating effect of lowering the risk trajectories by testing the effect of mitigation mechanisms thus helping increase the incentives to invest through private placements, institutional investor participation and capital markets while appropriately marking the risk premium of mega-projects. A Framework for Modeling Causality Based Risk Characterization of Megaprojects In the preceding sections we outlined the nature of complexity, causality and risk in large projects due to non-linear interactions, feedback delays along with the random nature of variability, which leads to indeterminacy of outcomes in various forms. Our approach to modeling causality in megaprojects is based on iterative staged approach which decomposes the structure and interactions into two stages. The first stage lays out the project in the context of the industrial sector and the related macro-economic dynamics by using a more top down model. This pre-supposes a keen understanding of the working of the industry sector and the related macro-economic and policy environment. In the second stage we model the dynamics of the project using a more detailed bottoms-up approach and integrate its outcome to the first stage. Stage One Model


9

The top down Stage One model structure is built around three inter-related blocks of demand, supply and price along with representations of shocks on the demand and supply blocks for each sub-area of the model. We call this pattern an archetype as it is the building block for representing the sub-areas and interactions between them for the Stage One model. Fig. 2 below is our representation of the archetype.

Fig 2- Project investment causal model archetype In the representation the boxes represent the stock variables and the arrows with polarity are a simplified representation the causality. A (+) arrow means that an increase (decrease) in the causative variable increases (decreases) the affected variable. A (-) polarity means that an increase (decrease) in the causative variable decreases (increases) the affected variable. The cause-effect relationships of these variables may have time delays or propagation delays, in that the effect of a change of the causative variable on the affected variable is delayed based on the nature of the causative relationship. These causal variables along with time delays and feed-backs interact with each other in self-reinforcing or balancing manners to exhibit a pattern of outcomes over time which may otherwise appear counter-intuitive. We will briefly describe one of the instances of the archetype, the Project Investment archetype in the figure, which is primarily driven by Project Starts (Drivers). The Project Starts will use price of steel as the signal and can also be due to government investment initiatives, especially in a developing economy. As the in number of project starts increase so does the demand for investments (DemandFunds), however there will be lag for the investment demand to be realized because a project start may have significant front end engineering and shaping required e.g. with an average of 12 months before the actual demand for the investment is realized for application to the project execution. This lag is represented by the random variable Price Demand. Similarly, project starts will signal banks and financial institutions to prepare for supplying funds (SupplyFunds) and there could be a preparation lag by institutions (Price Supply). The interaction of demand for project investments and supply of funds from the market will determine fund gap or glut in the market (Funds

r). The increasing fund gap between demand and supply of funds will increase the price of funds (PriceFunds) with a propagation lag (Fundsr) as it gets applied to the project finance markets. Increasing price of funds will in turn increase the supply of funds with a time lag (Price Supply) which will decrease the gap, Funds

r. Similarly, increasing price of funds (PriceFunds) will

negatively affect the project starts thus reducing funds demand after a lag (Price


10

Demand) and so on. The demand and supply variables may be subject to shocks (ShockSupply, ShockDemand). A positive supply and demand side shock in the investment archetype could be sudden easing of monetary policy and a negative shock for projects in emerging markets could be throttling down quantitative easing in the US. Again these shocks manifest themselves with lags (Shock Demand ,Shock Supply). These interacting loops of demand, supply with lags will exhibit a stable or unstable dynamic behavior over time. We have used this primary archetype for the other building blocks in the Stage One model and then modeled their inter-relationships and an instance of the model for the steel industry is shown in Fig 3. It integrates the product market, raw-materials market, project investment, land, labor, and technology & equipment instances of the basic archetype and drives the Stage Two model through new project starts with the project execution model [Fig 5- Project Work-Rework Model] archetype. We use a hybrid continuous and discrete mechanism to simulate Stage One and the Stage Two models to understand the overall behavior of the project in the context of the macro-economic and industrial environment.

Fig 3 Stage One model Stage Two Model The Stage Two Model is embedded in the Stage One Model which is more of a bottom-up project execution dynamics model based on our analysis. Stage Two model uses the outputs of workforce, supplier and resources like land as stocks from Stage One model, which otherwise would have been defined exogenously. Similarly, Stage Two models outcomes further drive the supply and capacity addition


11

parameters in the Stage One model. In our analysis of mega projects, ( a mega project being defined by us as one with a minimum project cost of 1 BB$ and spanning over 36 months) we find that the degree and range of indeterminacy of outcomes can be made more tractable by understanding causality across the various interacting external and internal factors within the project lifecycle. In particular we think that and interacting nature that drive complexity and its effect on risk characterization in large engineering projects are particularly relevant in the execution phase of the project. This is especially germane from a project financing perspective as majority of the financing is in a large project is gated on understanding and mitigation of execution risks. We begin by laying out a framework for understanding complexity and critical areas of risk assessment and characterization, which addresses the execution phase of a large project through our Stage Two Model. We find that multistage project architectures, change, rework and information and physical delays are inherent characteristics of large projects. They interact in seemingly unpredictable ways to create impacts, which are delayed, non-linear, indirect and self-reinforcing. The nature of the interactions is such that it is difficult to perceive the full significance before or even after the occurrence. This results in misunderstood risk characterization in execution and an obvious inability to comprehend the interconnectedness of the risks. Current traditional models of risk for large projects do not have appropriate mechanisms for understanding the self-reinforcing degenerative dynamics of the risk factors, which often results in incorrect conclusions and ineffective coping strategies. These counter-intuitive effects then often prompt dysfunctional management actions that often further destabilize the project. Execution complexity and its effect on risk at that stage can be better understood by understanding the structural determinants of complexity in large projects. We find that the major determinants of complexity

a. Structure and Form Large projects have many interacting and overlapping stages across design, engineering, construction and commissioning. For example stages of design , also known as Front End Loading (FEL) in industry parlance, can be stage gated across three overlapping and interacting phases which can extend over 18 months and consume anywhere from 100,000 to 500,000 engineering hours [13] . The iterative and interactive nature of internal and external factors shapes the FEL stage. We find that the initial stages of FEL are largely impacted by market factors and institutional factors ( e.g. land and zoning regulations, feedstock policy, sourcing constraints, environment, competitive dynamics) . Similarly, successful engineering and construction demands that the project is partitioned and structured in a way which assigns functional coherence within partitions with clear interface definitions and dependencies while maintaining interface integrity across partitions. Structural integrity of these partitions is very important for successful project execution due to the sheer scale of engineering and construction. An adequately partitioned multi-staged engineering and construction phase may have well over 100 interacting and concurrently


12

executing partitions or packages spanning over 15-25 million engineering and construction hours of work [14 , 15] . As mentioned earlier, sponsors, especially exhorted by financing agencies, frequently outsource the partitioning complexity and the associated risk through a turnkey EPC contracting mechanism with the hope that the turnkey contractual mechanism will transfer the risk of partitioning complexity and execution. This is a big fallacy and rarely does a contracting mechanism solve the problem of partitioning complexity and its related execution. That is because it is highly unlikely that a sponsoring agency can design a bulletproof turnkey incentive-contract mechanism for risk transfer without clearly understanding the underlying drivers of execution risk which is closely related to the form, function and interaction of work within and across the partitions [16] . A large project may also have procurement across a global project supply chain ranging anywhere from 200 to 2000 suppliers. In our analysis using our models, procurement structuring and sequencing and supply chain interaction and delays have profound and counter-intuitive adverse impacts on project cost and timelines as it directly impacts the erection-sequence within and across partitions. Concurrent and stage-gated partitioning is a key determinant to the structure of the organization across the sponsors, contractors, consultants and the vendors from a perspective of project execution. The structure of the resultant organization is a key determinant in the speed and accuracy of decision making. The delay effects of decision making creates substantial cascading impacts in terms of time and cost across the project depending upon when and how they occur in the project lifecycle.

b. Structuring and partitioning large engineering projects is thus a key determinant in our framework for understanding project behavior and hence the nature of the risk that may evolve. In our projects we have found that Design Structure Matrices (DSM) [17] as a useful tool to structure projects from partitioning, decoupling, sequencing and interfacing perspective. Structuring thus forms one of the fundamental basis for understanding complexity and associated risk behavior in our framework. Fig 4 illustrates how we structure and partition mega-projects using DSM for large integrated steel project. It is not atypical to have over 40,000 tasks across partitions in the matrix.


13

Fig 4- Design structure matrix of partitioning mega-projects

c. Changes and Rework- One clear outcome from the analysis of our large projects is the importance of changes and the ability to manage the impact of changes and re-work on the risk profiles and the success of project. After plans are in place, during the shaping and FEL stage changes occur. They occur on most projects and throughout the life of many projects. They are the means by which projects products are refined and improved. Some changes are institutional mandated (e.g., labor laws, resource allocation policy, environmental regulations), some are market driven (e.g., changing structure of demand or feedstock capacity), and some reflect changing technical and performance requirements. The ability to progressively delineate change scenarios and understand the critical levers for managing and adapting to change is crucial in outlining possible project outcomes and their deviations from planned goals. This then provides the sponsors and project participants the ability to learn and pro-actively manage change in a manner which minimizes the deviations and risks while conforming to the dynamic market requirements. One of the important and seemingly indeterminate aspects of change is what we call as the second order cascading impact of changes on project outcomes. Traditional project risk analysis methods typically underestimate the impacts of the second-order effect of changes as they are frequently unforeseen and because they tend to appear unexpectedly later in projects. Frequently, second order impact of the effect of changes manifests itself as cost of reworking previously completed work. Cost of changes has a cascading multiplier effect when multiple changes occur across the lifecycle of the project. The degree of rework varies depending upon the project stage when the change is occurring. For example a late stage design change discovered during construction phase will have a much more pronounced impact than an early stage engineering change before the construction has started. Changes will have other multiplier


14

effects like productivity impacts, resource availability, coordination effects and quality depending upon when and how they occur. For example, our model simulations consistently corroborate the fact that sustained overtime causes direct productivity loss [18] and finds that knock-on productivity losses through attrition, new worker learning curves and morale effects can be substantial. Such productivity loss also occurs on unchanged workpeople get tired and are less productive on whatever they are doing. Overtime induced productivity loss thus in general means higher cost to perform the workhence, the second-order impact. Through this and several other paths, changes cause productivity impacts that can substantially increase the cost of executing the entire project. Although challenging to quantify and explain, this secondary impact may well be the single largest source of project performance problems.

Change impact quantification and causal attribution is thus a key element in identifying the possible impacts on the project in terms of cost and time escalations. From a risk characterization perspective as a function of change, we need to understand the scenarios and the project response to the following questions a) What would happen if b) When is it likely to happen c) What is the major cause of the change

The ability to model and manage the injection and mitigation levers for change and rework is thus the second important structural factor contributing to the impact on project outcomes and the risk behavior.

d. Feedback and Delay Second-order impacts in large engineering projects are

not only because of scope changes or visible direct changes like addition or modification of work. Delays occur due to changes as well as information - e.g. late arrival of data from equipment suppliers, late basic engineering development, approval cycles, delays in management response to changes and rework discovery time-constants. Some of the delays, what we refer to as time-constants and information delays are intrinsic to the nature of the work and organization. For example the rate at which rework is discovered follows a pattern which is dependent on the organization, project type, nature of work and the stage of the project execution and is difficult to change in a short span of time. The rapidly growing impact of many small delays especially the intrinsic time-constant based delays at various stages can snowball into a large and seemingly unforeseen project impact. Understanding the timing and patterns of material and information delays thus plays a significant role in terms of its effects on the possible project outcomes. Qualifying and quantifying the causal effect of delays and its mitigation strategies is thus important in understanding risk characterizations of large projects. This is the third building block for understanding the nature of risk in our framework.

Our Stage Two model models these three structural blocks and along with Stage Two model archetype (depicted in Appendix 1- Figure 10), we have found it to be a reasonably accurate representation of the project execution dynamics. The Stage Two model archetype is the work-rework-delay cycle which has been applied to


15

modeling execution dynamics of projects across design, engineering and construction phases as shown in Figure 10.

Fig 5 - Project work-rework cycle model

Using scenario simulations to determine the interactions and the outcomes, the causal models at this stage are essentially the embodiment of our mental models and along with variations and simulated shocks on project parameters and inputs from the Stage One model, the outputs typify the possible range of outcomes. Synthesis of Framework by Integrating Stage One and Stage Two Models As discussed earlier, our approach is both top-down and bottom up based on staged iterative evaluation of the causality at the macro and the micro level in the context of the mega-project in the industrial sector and the macro-economic environment. Specifically,

1. A top-down macro-level industrial and sectoral dynamic model which feeds into the project execution model. In the process many of the factors that would have been taken as exogenous, are endogenous to the execution model providing better fidelity on project behavior patterns in terms of outcomes, resiliency and sensitivity to these events.

2. It models the project as a set of interactions across project work items and

resources along with delays, feedbacks, behavior functions and policy response at an aggregate level. The level of aggregation is dependent on the expected granularity level of the output desired.

3. We also model the individual unit of project work from dynamic bottoms up

perspective through the work-rework cycle [19] as opposed to the static work unit modeled in typical project approaches. This unit of work-rework is then aggregated at different levels of the model [ Figure 10]


16

4. Our partitioning of the project also follows a bottoms-up approach to

aggregate and determine couplings across activities. This allows us to structure the project into partitions or packages. These packages and couplings then form the basis of work items for the causal model.

This then forms a more realistic basis for characterizing risk behavior of the project and provides us the basis for structuring, organizing, managing and potentially insuring mega-projects by integrating both micro and macro level effects in the context of the large project. Figure 6 below pictorially depicts our framework.

Fig 6- Framework to Model Large Project Dynamics

Risk Scenario Analysis Using the Framework

Our scenario analysis for understanding the dynamic response of the modeled project starts by trying to understand the major internal and external input levers which can change both in terms of extent/range and immediacy in terms of time. For example based on organizational knowledge and experience, we know the extent to which labor productivity changes over time if schedules are changed and over-time is increased. Similarly, based on the remoteness and location of the project we have reasonably good estimates of the effect of a supply chain disruption on the magnitude of introduced delay for an equipment supply. We can enumerate with a reasonable degree of accuracy all such important internal and external levers affecting the project. The model can evaluate how these changes, delays and disruptions propagate through the system and manifest in terms of schedule and cost impacts. Fig 7 represents an example screen on the projects sensitivities to disruptions and changes.


17

Fig 7- Example screen on project outcome sensitivities to changes

In essence our scenario evaluations attempt to recognize the major sources and ranges of overall execution risk factors and interconnectedness of key risks. Based on this understanding we evaluate various options for desensitizing the projects and assess the value of reducing the range of uncertainty surrounding the significant risk factors. For example, mitigation measures to limit the magnitude of change in engineering by limiting the design-approval cycles. We may find that shifting the deadlines gives us a window of change while minimally impacting the direct and opportunity cost. We may find that simulating a range of price shocks in feedstock may lead us to look at flexible design options in the project for a plant capable of handling multiple feed stocks e.g. a dual feed ethylene cracker plant option based on naphtha and natural gas or options for scrap, ore or pellet charge for iron manufacturing. Once we have reasonably desensitized the project along with mitigation and action plans, we will have a reasonably good estimate of the magnitude of variability of the risk factors. Here are some of the typical questions and what ifs the framework helps answer Market Risks 1 Is there an opportunity for investment in a project in the

market? 1. Potential return envelope 2. Capacity bounds 3. Market segments

2 How would the commodity price evolve?

1. Price evolution trajectory 2. Price bounds and sensitivities

3 How would raw material availability and pricing evolve?

1. Raw material supply trajectory 2. Raw material price envelope

4 Impact of availability and pricing of

Land Equipment & technology Labor & workforce

1. Impact of returns on project 2. Feasibility of new projects


18

Execution Risks 1 How should the project be structured? 1. Complete (EPC) and discrete

turnkey option 2. Number of packages 3. Difference in cost between

contracting options

2 Impact of Timing and magnitude of Scope changes Delays in decision making Delays in discovering rework/changes

1. Cost impact 2. Schedule impact 3. Workforce impact 4. Trade-off zone

3 Impact of supply chain disruptions

4 Impact of unavoidable events Labor strikes Forces majeure Inter-contractor dependencies

We can then embark on the next stage of characterizing these range constrained input risk factors using statistical functions. These we believe are much more accurate and controlled representation of uncertainties as they are steeper and tighter patterns and limits the fat tail effects encountered in other non-causal statistical modeling approaches at aggregate levels of work. We layer the Monte Carlo simulations based on these patterns on our causal models to characterize and pattern the behavior of the project over time. Shaping Markets for Project Finance through Financial Innovation We think that our project dynamics and risk framework fill an important need in the area of risk analysis, characterization, rating and monitoring of large projects. This in turn can help in attracting and structuring syndications and private placements as well as enable better functioning of project finance markets. While the applicability of our framework to syndication and private placements for project finance transactions is straightforward, we next discuss its applicability in exploiting the capital markets for project finance. As discussed earlier, in the current market large international project finance transactions are primarily served by commercial banks and some multilateral institutions like the EIB, ADB, Exim, Mitsubishi UFJ and IFC. The commercial banks by their charter and the nature of their business are not the best suited for finacing large scale, long tenor project transactions. The Basel III liquidity requirements further limits the ability of the banks to participate in such transactions from a liquidity requirements perspective [20]. The project finance markets for infrastructure financing fits more comfortably with natural long term institutional investors e.g. the pension funds and insurance companies, who seek a diversified portfolio of assets to match their long term liabilities. However, particpation by these institutions is gated by their pre-requisite for investment grade rating of projects at least at the A- level. In our opinion, there are currently no known mechanisms or institutions to rigorously and accurately analyze, rate and monitor risk for large industrial infrastructure projects, with a result that most of the projects are rated much lower than investment grade. This mismatch between need for liquidity and financing and the availability of liquidity can be perhaps be bridged


19

though our framework to enable an ecosystem of institutional investors, banks, risk arbitrageurs and financial products. This in turn can eventually help enable a vibrant capital market for project finance. Up until the financial crisis of 2008, participation of long term institutional investors in the project finance markets, especially in Europe, was enabled through monoline insurers [21] . These were AAA rated entities providing credit insurance (known as wrapping), typically for BBB-/BBB and below rated projects, thereby enhancing the rating to AAA and enabling debt to be sold to a bond market. The monolines provided risk analysis, credit structuring and monitoring skills that meant institutional investors did not have to invest in developing specialist in-house project evaluation and risk analysis teams. After the financial crisis, the ratings of these companies began to crumble and this business model no longer worked. As a result the project bonds almost completely disappeared as funding option in project finance markets.

Today the institutional investors (i.e. insurance companies and pension funds) face increasing investment pressure to find attractive investment opportunities. As a result, investors are re-exploring investment opportunities in industrial infrastructure. The common deal breakers of the past (e.g. construction risk; deferred draw-downs in accordance with construction progress rather than one single disbursement of the total amount at the start of construction) are now increasingly accepted. The most significant challenge though is in the risk assessment, monitoring and credit rating requirements of investors. Institutional investors do not have the resources, expertise, analysis frameworks and monitoring systems for them to be able to evaluate and participate in this market. Especially, the incomplete characterization of execution risks tends to restrict the ability of mega-projects to achieve a high credit rating and is typically bucketed at either default ratings of BBB or goes unrated. Significant bond market liquidity only really exists for credit ratings at BBB+/ A- and above. Under these circumstances, achieving this level of rating for a mega-project to attract financing will typically require significant market and construction/delivery risk mitigation in the form of corporate and third party credit support through parent company guarantees, letters of credit or surety bonding. This is both unattractive and expensive for contractors and sponsors and therefore inhibits the development of the project finance market. Alternative credit enhancement structures thus provide an opportunity for securitizing project finance transactions and making the investment opportunity attractive to different class of investors with various risk appetites. Some of these structures like the EIB Project Bond Initiative [22] and the Hadrians Wall Capital product [23] are useful innovations in this area. They are different than monoline insurance, wherein essentially, monolines swapped their AAA credit rating for guarantees on project finance risks for a fee. In the current products the senior level tranches are credit enhanced, in that the first loss tranch of the debt for the project through the subordinated notes is funded in the form of draw-down rights or a fund for mezzanine like debt. This credit enhancement of the senior notes draws in the participation of long tenor institutional investors increasing the liquidity of the market. Additionally, one can imagine the growth of complementary risk insurance products around all the risk categories, and this in effect can raise the ceiling on


20

project bond issues, improving their ratings and elevating them to investment-grade status.

However, for these credit structures to work on a consistent and long term basis, the risk characterization, analysis and monitoring needs to be reasonably bullet proof. We think that a delegated and objective institutional structure based on our framework is an appropriate mechanism to evolve such financial innovations for the project finance markets. Fig 8 below illustrates how this would work in conjunction with an EIB, NDB, AIIB and such multilateral institutions.

Fig 8- Risk framework applied to financial innovation in project finance

We think that our framework can help in eventually evolving the markets for project finance to a higher level of liquidity and access by supporting both the capital markets and the syndication and private placement markets. Its adoption along with the right institutional structures can advance the state of the project finance markets with more mature capital markets, while enabling the creation of competitive capital markets for project finance in developing markets. Fig 9 illustrates how we envision the markets for project finance to develop.


21

Fig 9- Development of markets for project finance

Conclusion Characterization of risks and understanding its interaction behavior is key to accurately structuring and financing large projects. The risk characterizations based on understanding and modeling of causality from inception to completion along with the layering of the stochastic nature of the related internal and external factors leads to transparent project assessments, more accurate risk characterizations and access to funds at appropriately marked risk premiums. Characterization of risk also allows matching risks with a financial structure that allocates risks to the appropriate parties who are best able to bear them through a combination of debt, equity, recourse and guarantees. We believe that such an approach to modeling mega-project risks will enable the creation of market for project finance along with financial devices that enable more widespread access to funds for mega projects.


22

Appendix -1

Fig 10 - System dynamic model for execution phase of a mega-project


23

References

1. Cohen and Steers, Global Infrastructure Report 2009 2. World Economic Forum Infrastructure Investment Blue Print, 2013 3. Investment and Investment Finance in Europe, European Investment Bank,

2013 4. Infrastructure Projects Face Funding Gap , Financial Times, May 2 , 2012 [

http://www.ft.com/intl/cms/s/0/0a3c51ee-9457-11e1-bb0d-00144feab49a.html#axzz3XTElxe6a ]

5. Mapping and Facing The Landscape of Risks, Donald Lessard and Roger Miller in The Strategic Management of Large Engineering Projects, 2001

6. Industrial Dynamics, Jay Forrestor, 1961, MIT Press 7. Coal gate The Times of India, [

http://timesofindia.indiatimes.com/specialcoverage/15739564.cms ] 8. Gorgon Project LNG Cost Jumps $15 Billion , Financial Times, December 6,

2012 [ http://www.ft.com/intl/cms/s/0/62e4f5e8-3f84-11e2-b0ce 00144feabdc0.html#axzz3XTElxe6a ]

9. Oil Firms in Kazakhstan: Cash All Gone, The Economist , Oct 11, 2014 [ http://www.economist.com/news/business/21623693-one-worlds-biggest-oil-projects-has-become-fiasco-cash-all-gone ]

10. The Financing of Large Engineering Projects, Richard Brealey et. al. in The Strategic Management of Large Engineering Projects, 2001

11. The Enrons Dabhol Project, Minority Staff ,Committee on Government Reform , U.S. House of Representatives February 22, 2002

12. Transformations in Arrangements for Shaping and Delivering Engineering Projects , Roger Miller and Serghei Floricel in The Strategic Management of Large Engineering Projects, 2001

13. Internal communications M. N. Dastur and Co. 14. Internal report M. N. Dastur and Co. 15. Chenierre Energy, Sabine Pass Presentation, Platts 13th Annual LNG

Conference 2014, Houston, TX 16. Industrial Megaprojects , Edward Merrow, John Wiley and Sons, 2011 17. A model based method for organizing tasks in product development, Steven

Eppinger et. al. , Journal of Research in Engineering Design, Vol6, 1-13, 1994. 18. Effect of Scheduled Overtime on Labor Productivity, H Randolph Thomas,

Journal of Construction Engineering and Management, Vol ' 118, No. 1, March, 1992

19. System Dynamics Modeling for Project Management, John Sterman, Working Paper MIT Sloan School of Management, 1992.

20. Basel III : The Liquidity Coverage Ratio and liquidity risk monitoring tools [ http://www.bis.org/publ/bcbs238.pdf ]

21. Monoline Revival Could Aid Infrastructure , Financial Times , 22nd July 2012 [ http://www.ft.com/intl/cms/s/0/9790c5c2-d27b-11e1-8700-00144feabdc0.html#axzz3XTElxe6a ]

22. The Europe 2020 Project Bond Initiative : Innovative Infrastructure Financing [ http://www.eib.org/products/blending/project-bonds/index.htm]

23. Hadrians Wall Capital, http://www.hadrianswallcapital.com/


24

Authors

1. Atanu Mukherjee Vice Chairman of Board, AcuLead Corporation, Houston, TX and President, M.N. Dastur and Co., Kolkata, India Atanu provides the strategic leadership at AcuLead Corporation and also leads Dastur, a consulting firm which focuses on the metals, mining, energy and infrastructure industry worldwide. He holds a joint graduate degree in Engineering and Management from the Massachusetts Institute of Technologys School of Engineering and the MIT Sloan School of Management. He can be reached at [email protected] / [email protected]

2. Dr. Purnendu Chatterjee Chairman of Aculead Corporation, Houston TX and The Chatterjee Group

(TCG), New York Purnendu leads TCG, a 2 BB$ private equity group investing in industrials, technology and infrastructure worldwide. He is also the Chairman of AcuLead, an energy engineering and technology consulting firm. He holds a Ph.D. in Engineering from the University of California, Berkeley. He can be reached at [email protected]

Journal of Structured Finance AcuLead

Documents

Transcript of Journal of Structured Finance AcuLead