A predictive sourcing model for multi Export Credit Agency financed large industrial projects

Abstract

CB&I is experiencing an issue in a new project to be executed in Russia, named NKNK. Despite the rich experience CB&I has with projects, there is a continuous struggle with the sourcing process in projects which it involves financing by multiple export credit agencies. The issue at stake is, CB&I does not know beforehand in which countries it is most likely to source its equipment to achieve to lowest possible sourcing costs. However, budgets available in countries will be set in an inception phase of a project. A preliminary estimation method is needed to determine the amount of budget needed in multiple countries, in order to increase the probability of minimizing total sourcing costs. In order to accomplish this, a new cost estimation methodology is needed. This combines strategic sourcing theory, descriptive statistics on suppliers, cost differentials among countries of manufacturing, macroeconomic theory, the role of export credit agencies in trade finance, conventional cost estimation methods, linear optimization, and Monte Carlo simulations. The importance of strategic sourcing is underpinned in this thesis. Theoretical optimal sourcing strategies are suggested on the basis of the level of perceived competition. The perceived level of competition within different industries is acquired through questionnaires with industry experts. The suggested sourcing strategies are tested on their practical applicability in large industrial projects. It turns out that there are serious limitations in applying multiple sourcing strategies, due to the nature of the highly customized equipment needed in these projects. Predominantly, single sourcing strategies are used, in which a number of suppliers is inquired for a bid. It is shown, through a linear regression analysis, there is a significant positive correlation between the perceived level of competition and the number of suppliers inquired for a bid. Descriptive statistics on suppliers involve per equipment type (more formally known as purchase order category), the number of suppliers selected and their most likely country of manufacturing. It is discussed that there are multiple restrictions in selecting potential suppliers for a project. Firstly, suppliers can only be selected and inquired for a bid, if they are stated in an ‘Approved Vendor List’. Secondly, ECA involved financing limits the budget available in each country to a certain extent. Therefore, selecting suppliers in a country where probably no budget is available, is a waste of effort. Thirdly, the increasing administrative burden in selecting larger numbers of suppliers poses limitations. Through a comparison on descriptive statistics on suppliers in two very similar projects, but with different project contexts, the effects of these limitations are determined. It is hypothesized there are sourcing cost differences among countries for particular purchase order categories. Through a literature review, macroeconomic factors that could explain these cost differentials are determined. These are categorized in economic-, infrastructural-, labor, supply based, and political factors. For each macroeconomic category indicators are selected to represent these. A total of twelve indicators per country are reduced to two factor scores per country, through a dimension reduction technique (principal component analysis). Based on quotations submitted by suppliers for a completed project in the near past, significant cost differentials among countries are determined using categorical variables in a linear regression. A statistical refinement has been done to place countries in a cost category. Factor scores per country and descriptive statistics on suppliers are used to substantiate these cost rankings. Combining cost differentials, macroeconomic indicators, and descriptive statistics proved to be a valuable tool to determine in which country one is most likely to receive the least expensive quotations. The role of export credit agencies (ECAs) in project finance is explored through a literature review. ECAs cover political and commercial risks for exporters and credit providing entities. ECAs are heterogeneous and there is no definitive model for ECAs. For terms associated with project finance (medium- to long-term), the most widely used mechanism by ECAs is buyer credit. ECAs are involved by issuing insurance, for defaults, directly to the exporter’s bank. ECAs are also involved in buyer credit by offering a precompletion risk facility. A recourse agreement is included, meaning defaults caused by the exporter can be reclaimed from the exporter and disbursed to the lending bank. To quantitatively compare differences in terms and conditions of ECAs, a new methodology is developed in this thesis. This methodology involves a discounted ‘Interest Rate Coefficient’, which incorporates ECA premiums rolled over into the loan in the financing period, and terms and conditions involved in the repayment period. Through a questionnaire terms and conditions applicable to the NKNK project are acquired, which are mainly budgetary constraints, insurance premiums, and interest rates. Combining the results of the questionnaire and the interest rate coefficient, necessary inputs are obtained for linear optimization and Monte Carlo simulations. The basis of the newly developed preliminary sourcing cost estimation methodology is a ‘sourcing allocation table’, which can be used as a direct input in a linear optimization model developed in line with this thesis. The methodology starts with listing all purchase orders for a project in the sourcing allocation table. Next, it is evaluated which data is readily available, with respect to suppliers, supplier countries, quotation values, and purchase order value estimates. Data which is not readily available on suppliers and supplier countries are estimated per purchase order category, based on the descriptive statistics on number of potential suppliers and their distribution among countries. For purchase orders of which no quotations or estimates are available, conventional estimation techniques are used. The order of magnitude method is used on a reference project, which is indexed to accommodate the inflationary impact of time. Dummy quotations are generated to fill in the missing data on suppliers, their countries, and quotation values. These dummy quotations take significant cost differentials among countries per purchase order category into account. In these quotations, values are randomly generated according to the average spread of quotation values, using a uniform distribution. Trade finance estimates are also included in the sourcing allocation table. Now the sourcing allocation tables contains, based on live data and dummy quotations, for each purchase order a number of suppliers, their country in manufacturing, and quotation values. As there are numerous randomly generated parameters, there is no definitive optimized value. Rather there is a range of possible outcomes, determined by doing a Monte Carlo simulation with the linear optimization model. The output of these simulations are, a probability distribution of the total optimized value, a probability distribution of the expenditures within each country, and an average distribution of ECA budgetary flows towards sourcing countries. The new methodology for preliminary estimation of sourcing costs is seen by CB&I as a valuable tool to determine in an early phase of the project where budgets are most likely needed. This allows to set ECA budgets properly, to increase the probability of minimizing sourcing costs. The first results are already presented to the client, which was impressed with the result. It gives a clear graphical representation of the estimated total costs, budgets needed in which countries, and where the budgets are spent. Evenly important, it shows the uncertainty in all these estimates, through probability distribution. In addition, this tool allows easy identification of the cost impact of different scenarios, such as exploring the cost effect of excluding budget from a certain ECA country.