Article ID: | iaor20062152 |
Country: | United States |
Volume: | 35 |
Issue: | 5 |
Start Page Number: | 370 |
End Page Number: | 380 |
Publication Date: | Sep 2005 |
Journal: | Interfaces |
Authors: | Kiaer Lynn, Bollapragada Srinivas, Chalermkraivuth Kete Charles, Clark Michael C., Deaton John, Murdzek John P., Neeves Walter, Scholz Bernhard J., Toledano David |
Keywords: | programming: linear |
GE Asset Management Incorporated (GEAM), a wholly owned subsidiary of General Electric Company (GE), manages investment portfolios on behalf of various GE units and over 200 unaffiliated clients worldwide, including Genworth Financial (Genworth) and GE Insurance (GEI) portfolios worth billions of dollars. GEAM invests portfolios of assets – derived from cash flows for various insurance, reinsurance, and financial products – primarily in corporate and government bonds in accordance with risk and regulatory constraints. In asset–liability management (ALM) applications, portfolio managers try to maximize return or minimize risk and match the characteristics of asset portfolios with corresponding liabilities. While risk is widely represented by variance or volatility, it is usually a nonlinear measure; ALM portfolio managers traditionally need to use linear risk sensitivities for computational tractability. We developed a novel, sequential-linear-programming algorithm that handles the nonlinearity iteratively but efficiently. Patented and implemented on a limited basis since 2003, GE used it to optimize more than 30 portfolios valued at over $30 billion. It is now in broader use at GEAM, GEI, and Genworth. Hypothetically, based on $100 billion of assets, the present value of potential benefits could approximate $75 million over five years.