Long Run Technical Document

Apr 1, 1999

Preface

This technical document details the long term forecasting and scenario generation methodologies in LongRun. It contains two sets of techniques for computing forecast values and confidence intervals for asset prices and a procedure for generating scenarios for use in Monte Carlo. In some circles of the economics and finance professions, forecasting is not a highly regarded activity.

For some, it evokes images of speculators, chart analysts and questionable investor newsletters; for others, memories of the grandiose econometric forecasting failures of the 1970's. There is nonetheless a need for forecasting in risk management. A prudent corporate treasurer or fund manager must have some way of measuring the risk to earnings, cash flows or returns. Any measure of risk must incorporate some estimate of the probability distribution of the future asset prices on which financial performance depends. Forecasting is an indispensable element of prudent financial management.

How should corporate treasurers and fund managers approach forecasting? Forecasting accuracy per se is not the object of the exercise: every currently known forecasting tool often falls wide of the mark. In a risk management context, the forecasts should rather be practical, based on objective techniques, it must be possible to examine how the methodologies would have performed had they been applied in the past, and it should be possible to articulate the techniques to shareholders, investors, and regulators. It is also desirable to have available different, methodologically independent forecasting techniques. The risk manager can then compare the results with one another and with his own judgements about future asset prices. We believe that LongRun meets these criteria for forecasting techniques.

The RiskMetrics Group's policy is to make public its risk management methodologies. In doing so we aim to foster pubic discussion of our approach, to help our clients grasp the methodologies which underline our products, and more generally to promote public understanding of risk management issues. We hope that interested practitioners and scholars will examine the LongRun methodology and look forward to studying their criticisms, alternative approaches, and suggested applications.

The authors have enjoyed support and constructive comments from a number of colleagues. We would particularly like to thank Ethan Berman, Alvin Lee and Jim Ye of the RiskMetrics Group, Mark Everson of Ford Motor Company, and John Byma of the Procter & Gamble Company for their detailed comments on several drafts of this document. Christopher Finger of the RiskMetrics Group helped formulate LongRun's simulation procedures and had many useful suggestions throughout. Peter Zangari was instrumental in the early stages of this project. We would also like to thank Tatiana Kolubayev for editing and producing this document.