EN
Back to the archive

The encyclopedia · Finance & Accounting · Financial decision · 2003–2005

GEAM's sequential LP optimized $30B of insurance portfolios

A patented sequential-linear-programming algorithm handled nonlinear risk in asset-liability management, optimizing 30+ portfolios worth over $30B.

GE Asset Management

The solution

GE Asset Management manages billions of dollars in cash flows from insurance, reinsurance and financial products, mostly invested in corporate and government bonds under risk and regulatory constraints; matching assets to liabilities is an asset-liability management problem.

Risk is naturally nonlinear, but portfolio managers traditionally needed linear models for tractability. GEAM's team developed a patented sequential-linear-programming algorithm that handles the nonlinearity iteratively and efficiently.

Implemented since 2003, it optimized more than 30 portfolios worth over $30 billion across GEAM, GE Insurance and Genworth. On a hypothetical $100 billion book, the present value of potential benefits could approach $75 million over five years.

Why it worked

  • Iterative LP captures nonlinear risk without nonlinear solvers.
  • It generalized across many GE entities' portfolios.
  • The patent protected the algorithm as a proprietary edge.
  • Linear speed kept the method usable in day-to-day portfolio work.
What it achievedIterate a linear model to capture nonlinear riskclever

What can be applied

Don't abandon LP when the problem is nonlinear; solve the linear model repeatedly, refining the approximation each pass, and you often get nonlinear accuracy with linear speed.

Aftermath

The algorithm moved into broader use at GEAM, GE Insurance and Genworth, and the case was published in Interfaces (2005).

Sources

spotted an error? The archive wants to know.

Related cases