 |
About The Author: Jeffrey Frost, a PMTC Senior Partner, has years of experience as a banking treasury executive, trading room technology innovator, and Internet electronic commerce pioneer.While Jeffrey's prior executive and entrepreneurial roles have demanded numerous skills, much of his career has revolved around one simple theme: The use of new computing technologies applied to existing information to create profitable new business alternatives. The Power Marketing Technology Consortium is an IT and electronic commerce power marketing consulting organization which integrates and supports technologies related to energy trading and marketing.
|
February 1999 by Jeffrey Frost -- Power Marketing Technology Consortium
The commercial software systems referred to here are those designed for Risk Management and Trade Processing (RMTP) for the merchant energy business. These are the systems used by merchant organizations dealing in physical and financial natural gas and power plus other energy commodities. The credit risk management portion of systems is of interest here. There are over forty vendors offering Risk Management and Trade Processing (RMTP) software applications in the energy market. Most end users are familiar with some of the vendors. Names like Altra, ARC IT, Axiom, Energy Imperium, Nucleus, Open Link Financial, Powertrade, Primo, Riskworks, and Transenergy are well known. This article starts by identifying what credit managers need from these systems. After summarizing the capabilities energy credit risk managers need from their systems, there follows a review of what those needs mean in terms of software issues. In other words, first about the business issues and then the technology issues are discussed. The final segment briefly discusses specific vendor applications. Just what does a credit manager need from an energy Risk Management and Trade Processing (RMTP) system? Let us first look at "Measurement of Credit Exposure" and then "Credit Exposure Monitoring and Control." A) Measurement of Credit Exposure An RMTP application must allow the user to implement his credit policy in a quantitative manner. The system must unambiguously allow Counterparty X's limit to be input as a number, whether $5,000 or $50 million. The system must measure (calculate) the exposure for any given Counterparty X at all times. Measurement of credit exposure is a complex and evolving issue. Historically credit risk exposure measurement has included CURRENT exposure, but not POTENTIAL exposure. Lets look at each of these in turn. 1) Current Exposure Measurement CURRENT exposure is the exposure you face if your counterparty defaults today. It includes A/R plus open positions marked to market. Marking to market shows what it will take to replace the transaction with a new counterparty, at today's market prices, in the event of a default. CURRENT exposure is thus the sum of A/R and MTM. 2) Potential Exposure Measurement What about POTENTIAL exposure measurement? It was not considered by many organizations until quite recently. In fact, Fitch Investors Service wrote the following in their June of 1997 Special Report titled, Managing Credit Risk in the Electricity Market: "Fitch predicts that the failure and default of some poorly capitalized entities will result in the increased sensitivity to counterparty credit in the electric commodity market." Well they got that one right! Just one year later, June 1998 proved them right. The majority of losses suffered in June of 1998 were counterparty default losses, losses that could have been anticipated to a degree by a good POTENTIAL credit exposure measurement system. So what is meant by POTENTIAL credit exposure? Potential exposure refers primarily to the exposure that will result if market prices change. CVAR, or Credit Value At Risk, is the most widely used POTENTIAL exposure measurement methodology. CVAR as the name CREDIT VALUE AT RISK implies is a measure of POTENTIAL exposure for credit risk. Credit Value At Risk, CVAR, is perhaps best understood by looking first at the more familiar regular VAR, regular Value At Risk VAR measures the exposure that will result if market prices change. More specifically, it measures the expected change in the value of your energy portfolio in the face of possible market price changes. VAR, you will recall, is an estimate of the maximum expected loss, over a given period, for a given percentage of time. For example, VAR would allow you to state that you are 95% confident that in the next month you would not lose more than $3.7 million. There are three major variants of VAR derivation widely in use in the energy industry today: One is analytic VAR, a.k.a. the variance/covariance approach; the second is the Monte Carlo or simulation VAR methodology, and the third is historical VAR. While the experts do widely debate various energy industry VAR refinements and limitations for each of these two major methodologies, they are nearly universal in their agreement that VAR is one of the best risk management tools available. CVAR or credit value at risk is simply the VAR methodology applied to counterparties for purposes of credit measurement. Note that CVAR requires a separate set of calculations since CVAR and VAR ARE NOT identical for a given counterparty. You can visualize the basic difference between VAR and CVAR for yourself. EXAMPLE. Think about a fixed price, fixed quantity, open purchase position, due for future delivery. For example, assume that you enter into a contract to purchase 100 MW of power on a 5x16 basis for the month of March at a price of $25. Assume further that you have purchased this power without an offsetting sale. You plan to sell the power into the market at prevailing prices in March. You are long power. Entering into this single position results in an increase in both CVAR and VAR. Your VAR modeling will reveal that DECREASING prices increase your VAR risk. Think about it: You have purchased power at a fixed price (long) and if prices decrease you will sell it at a loss. Your VAR calculations will reveal this potential outcome as an increase in VAR resulting from this transaction. Note also that the measurement is of a potential actual $ dollar loss. Your CVAR modeling will reveal, in contrast, that INCREASING prices increase your CVAR risk. Again, think about it: Your counterparty will be selling you power at $25. Should they default when market prices have risen, the replacement cost will be larger. When your counterparty replacement cost is larger, you have an increase in risk. Your CVAR calculations will reveal this potential outcome as an increase in CVAR resulting from this transaction. Note that CVAR measures your potential counterparty credit exposure, NOT a potential actual $ dollar loss. Although the statistical methodology for calculating CVAR comes directly from VAR, the use of the term Credit Value At Risk can be misleading because of this difference. SUMMARIZE. What are the key points about CVAR?
b) Stress Testing and Scenario Analysis Stress Testing and Scenario Analysis are important alternative POTENTIAL exposure measurement methodologies. They model the change in a measure (e.g. the change in CVAR) based upon some random shock to a system or some change in underlying relationships. The VAR/CVAR interaction is one use of a scenario analysis capability. It is important to model the interplay between VAR and CVAR under various hedged and unhedged scenarios. Sometimes a reduction in VAR (market price risk) is achieved by effectively transferring the risk to increased credit risk, CVAR. POTENTIAL exposure refers primarily to the exposure that will result if market prices change. But there are other types of POTENTIAL exposure as well. Examples of these other types include: a change in the legal or regulatory environment or change in the credit worthiness and/or credit rating of a counterparty. Now lets write a formula for the Combined Measure of Credit Exposure. The result is A/R + MTM + CVAR (CURRENT plus POTENTIAL exposure). One should reduce this calculated exposure measure by the amount of any collateral held, yielding the amended: A/R + MTM + CVAR - Collateral equation. b) (A/R + MTM + CVAR) - Collateral This is a measure of exposure in the event a counterparty does default. For a full risk assessment, not just a worst case, each counterparty's Combined Measure of credit exposure should be adjusted for its probability of default, shown as Default Probability on the next line. c) ((A/R + MTM + CVAR) - Collateral) X (Default Probability) Similarly, if an organization estimates loss recovery rates that can be expected in a post default situation, the system must be able to apply a factor and adjust the credit risk exposure calculations. Loss recovery data in this industry is little used at this time, but at least one software vendor is ready. d) ((A/R + MTM + CVAR) - Collateral - Loss Recovery) X (Default Probability) That completes the Measurement of Credit Exposure discussion. The second part of Identifying The Need concerns Monitoring and Control issues and will require a much briefer description. B) Credit Exposure Monitoring and Control Exposure Monitoring and Control will be addressed nly from a systems standpoint, not from an organizational management standpoint. In software systems terms, Monitoring and Control, boils down to a few essential issues. For example:
These issues are partially policy and implementation issues for management, but each has systems implications as well. There will be more about these Monitoring and Control issues within the following section on Idealized Applications. The application's scope must cover all of them required energy commodities. This usually means both physical and financial natural gas and power. It may mean other energy commodities such as coal, natural gas liquids, emissions credits, transmission, and weather derivatives. Another scope issue is the tightness of integration across separate software modules for separate commodities. The various application modules must be integrated in a manner that allows a user to measure credit risk across all energy commodities for a given counterparty. This same integration issue often determines whether one can use a single client database including integrated management of limits and collateral tracking. Another scope issue is the systems breadth of coverage from front to middle to back office. Some systems e.g. do a poor job of tracking cash flow and receivables, a back office function. Consequently, their ability to measure the A/R portion of CURRENT exposure is severely limited. Other systems are weak on tracking customer documentation and are very clumsy about their handling of contractual issues like netting, setoffs, and collateral. Measurement primarily refers to the ability to calculate all portions of the Combined Measurement formula shown above. The system should have the flexibility to calculate, view, and react to both the combined totals as well as the individual elements of the equation. For example, if a trading manager comes to the credit manager and requests to do a deal with a counterparty already at its limit, the credit manager needs to know the total combined measure of exposure and how much is current exposure and how much is potential exposure. In fact, the system needs to be able to track separate limits for both types of exposure for each counterparty. Measurement must also be allowable at differing levels of aggregation. While credit decisions are usually made at an individual counterparty level, a risk manager also needs to know the entire portfolio view, a view which is not just the sum of the individual counterparties due to the mechanics of CVAR calculations. Similarly, the risk manager very much needs to know which geographic areas are the sources of the risk. Remember how regional transmission system operating factors played such a large role in the June 1998 debacle. Other levels of aggregation, which might be needed, include summaries by trading group or summaries by commodity or summaries by currency. Some users define counterparty limits in terms of the volume measures or even the lessor of separate volume and dollar criteria. This is a good example of how it is important for the measurement system to be flexibly configurable and highly customizable. 2) Other Technical Calculations The way in which applications provide the myriad technical details of measurement is also critical for:
Once the system has been chosen, installed, customized, and configured, it must provide the means for users to employ it effectively for Monitoring and Control of credit risk. Reporting must include a user programmable report writer. Reporting should have graphic options. Reporting should be both paper based and screen based. It should provide both regular and ad-hoc report requests. Exception reporting is mandatory. The best systems have the ability to notify affected parties when "user defined events" occur. "User defined events" are based upon customizable business rules about limits, types of trades, approvals required, price breaks, etc. Notification can be via email, screen alerts, or database updates. Event notification is best facilitated via the mechanism of middleware. Middleware has many flavors, but the type most germane to trading environments is called Message Oriented Middleware (MOM). Middleware is an essential system consideration. See the middleware articles within the Technology Corner archives for a fuller discussion of this issue. A good system should reject trades or transactions that violate user defined rules such as a counterparty credit limit. It is equally important for there to be system rules about who may force an override and under what circumstances they may do so. The are many other system issues that are not credit specific. These other issues include many important topics such as the navigability and configurability of the user interface. Equally important are issues such as system fault tolerance, stability, and scalability. Of course, pricing and contractual issues always play a role as well. In addition, in this industry at this time a company's financial strength, reputation, and personnel are major issues too. The list goes on and on. Consequently, you must be rigorous. "Rigorous" means that you should produce your own detailed list of needs and that you must spend the time immersed in the details. You are best served by employing a quantitative methodology for decision making. Use a methodology that allows you to assign both an importance ranking to key needs and assign a separate rating number to each issue for each vendor. The resulting rank-weighted number calculated for each vendor is very useful. An interesting discussion always develops around this process and the process forces a great deal of clarification about decision drivers. III) ACTUAL ENERGY APPLICATIONS Systems developers do not have an easy job. The wholesale energy industry is in a state of rapid change. No two users conduct business in an identical manner. There are major regional differences emanating from ISO's and regional markets such as the CA PX and PJM. Market events such as June of 98 change the way users want the systems to operate. The market and products are immature and in flux. In short, developers face many demands. The results found while investigating the credit risk capabilities of eight leading vendors were less than expected. Therefore, in keeping with a policy of only writing about vendors in a positive vein, just a couple of vendor systems will be mentioned. In spite of the difficulties faced by vendors, it was surprising how inadequately some vendors' treat credit risk measurement and management. For example, one market leader (not to be mentioned here) still calculates Current Credit Risk exposure, not Potential; they also still do not aggregate by counterparty across their power and natural gas modules. Rather than talk about the Ugly, let's talk about the good. A couple of vendors, Nucleus and ARC/IT in particular, seemed to offer a very solid approach to credit risk measurement and management. There are others such as Open Link, Primo, and Energy Imperium who appear strong although less time has been spent on credit risk exploration with these vendors. The most impressive credit risk management system reviewed for this article is Axiom. If you wanted to set up an institution wide risk management system, no one has a more refined and more powerful credit risk system. This vendor will do most if not all of the functions outlined in the Idealized Application. They are not an easy turn key solution for a smaller merchant energy function, but they are the best of breed found to date for a larger organization needing a comprehensive approach to credit risk. Building your own risk management system is always an option. Be very cautious about starting down this path. With all of the commercial alternatives existing today, few organizations have even a ghost of a chance of being able to promptly and cost effectively create a better in-house alternative. There can be an incredible naivete about the degree of difficulty in creating a quality Risk Management and Trade Processing system. This is a crowded market with some good answers. Unless you have some highly unusual conditions, your needs will best be met via the buy choice, not the build choice. Even having chosen to buy and not build, there are still plenty of tough decisions around the choice between selecting a unified suite or choosing separate best of breed modules and integrating them in-house. Most if not all of the vendors are working to rapidly improve their systems as we speak. Your challenge is to sort through the noise and mine the nuggets. It is hoped that this article can assist in clarifying the issues you need to be aware of for purposes of evaluating credit risk management applications.
|
