Awarded a GARP Fellowship, Felipe Silva found flaws in a widely-used measure of risk

Silva's work shows that Value-at-Risk may decrease during a bubble.

Last year, the Global Association of Risk Professionals (GARP) awarded Felipe Silva M. Eng. ’12 a Risk Education Research Fellowship to study “The Impact of Asset Price Bubbles on the Computation of Value-at-Risk (VaR).”  The results showed that the widely used measure is “unable to assess the negative effect of market bubbles,” according to Silva.

Silva, who came to ORIE’s  Master of Engineering (M.Eng.) program from a career in Brazil, was one of five recipients of GARP Research Fellowships in the inaugural round of the program. Under the guidance of Robert Jarrow, the Ronald P. and Susan E. Lynch Professor of Investment Management at Cornell’s Johnson Graduate School of Management (JGSM), Silva “was able to read and apply cutting edge research on asset price bubbles to a practical problem,” according to Jarrow.

Choosing between academia and industry

“I was already in doubt between academia and industry when I applied to masters programs,” said Silva.  He was accepted into the Financial Engineering concentration of ORIE’s M.Eng. program and accepted admission.  “When I got to Cornell I could talk to some Professors and I realized that following an academic career is much more suitable to my profile,” he said.   Professor Jarrow, Professor George Gao of JGSM, and ORIE Professor Peter Frazier were “the great influences for my decision,” he added.

Aware of Silva's new-found aspirations to pursue an academic career, Victoria Averbukh, Director of Cornell Financial Engineering Manhattan (CFEM), encouraged him to pursue the Fellowship and suggested that he approach Jarrow to help with the proposal to GARP and to advise the project.  Silva spent the third semester of his FE program at CFEM and received his degree there at the end of 2012. 

Silva has now been accepted into the Ph.D. program in JGSM and will begin his studies there in the fall.  In the meantime he is working as a Research Assistant to Assistant Professor of Finance George Gao.  In the fall of 2012 he managed to carry out his GARP Fellowship research while at the same time completing course work as well as working on a completely different problem as a member of a 6 student project team supervised by ORIE Visiting Assistant Professor Tibor Janosi.

Cutting edge research to detect bubbles

With Younes Kchia and Philip Protter, Jarrow developed an analytical methodology to detect asset bubbles as they occur, based on real time measurements of the prices of the assets.  Jarrow is a member of the Field of Operations Research, Protter is a former ORIE professor, and Kchia spent several months in ORIE while he was a Ph.D. student at Paris-based Ecole Polytechnique.  

Market price bubbles, such as the recent bubble in housing prices, occur when there is a steady increase in the price of such assets, “with market players generally assuming there are new and promising opportunities for abnormal returns, followed by a crash as the bubble bursts,” said Silva.

A practical problem

Historically, bubbles create the risk of extreme losses; financial institutions and regulators are constantly on guard for risk.  A widely used measure of risk, called Value-at-Risk, or VaR , is presumed to be useful in determining the extent to which a portfolio of securities, such as stocks and bonds, is exposed to the possibility of extreme losses in its market value.  The calculation of VaR is based on statistical properties of the securities. 

Given recent events, it is natural to question the extent to which such a widely adopted measure of risk is sensitive to the presence of a market price bubble, i.e. whether VaR increases to reflect the increase in risk that accompanies a bubble.  Intuitively, VaR should increase in the presence of a bubble, signaling the holder of the portfolio to take extra measures, such as increasing capital reserves, to avoid the possibility of loss.  Jarrow proposed that Silva investigate how VaR responds to a bubble. 

How VaR responds to a bubble

Image removed.Silva applied the modeling approach of Jarrow, Kchia and Protter to this question, and used Monte Carlo simulation to test whether the risk measure is sensitive to the presence of a bubble.  He developed models of the evolution of prices under bubble and non-bubble conditions, and repeatedly simulated the dynamics of a portfolio of some 19 securities representing different market sectors (shown at right - each color represents a different run of the simulation) to see whether VaR would rise under bubble conditions. 

Remarkably, VaR calculated from Silva’s simulations showed that the most commonly used version of the VaR calculation, called unconditional VaR, can actually decrease in the presence of a bubble.  However similar VaR computations that measure the extent of loss under the condition that a loss actually occurs do increase in the presence of a bubble, making these computations more useful in assessing risk.  

Silva therefore recommends that incorporating such conditional risk measures along with unconditional VaR in defining the rules governing capital reserves.  “His results are useful for risk managers worrying about asset price bubbles,” said Jarrow, who regards Silva as “one of our finest M.Eng. students over the past ten years.” 

From aeronautical engineer to finance Ph.D. candidate, via ORIE’s M.Eng. program

In 2006, Silva graduated from Instituto Tecnológico de Aeronáutica in Brazil.  Before joining the Financial Engineering Concentration M.Eng. program  in ORIE he worked as a structural analysis engineer for aerospace company Embraer and as a market risk modeling analyst and a proprietary trading quantitative analyst for banks in Brazil. As an undergraduate he worked on the use of simulation software in analyzing and controlling chaotic, dynamic systems.

“Although my interests turned out to be different from the usual objective of a person who undertakes a professional program like the M.Eng., I do recognize the academic  excellence of our program as a leveraging factor to pursue my Ph.D. goal,” he said.  “One of the greatest advantages of the M.Eng. is the multidisciplinary environment that Cornell provides to its students, which allows each person to pursue his or her objectives with the different departments and courses.” 

GARP Risk Education Research Fellowships

The 2012 GARP Risk Education Research Fellowships were awarded to students at Rutgers University, the University of Macau, and Tongji University in China in addition to Cornell’s Silva.  According to Michael Nasatka ’04, Vice President of GARP’s Financial Risk Manager certification program, Silva was selected on the basis of his involvement with CFEM, the mentorship of Professor Jarrow, and his research direction.  “The performance of financial risk measures, with VaR being one of the classic ones, during crises is a hot topic in financial risk management today, and therefore one of particular focus for GARP’s educational programs,” said Nasatka.

Silva presented his results to the GARP research team in April, according to Nasatka, who said “his research was received quite positively.”  He added that  “GARP expects that this and other related research will help establish the application of a variety of risk metrics in bubble situations, and we recognize and commend [CFEM] for its efforts in research to this end.”  

Cornell’s GARP chapter is directed by CFEM Director Victoria Averbukh.  Nasatka expressed appreciation for the strides that the Cornell chapter has made in building a strong financial risk management community in New York City and beyond.  “Through our various relationships with Cornell University, we will certainly further our mission of creating a strong culture of risk awareness,” he said. 

Other Articles of Interest