Stressed Out Investor

Judge’s comment: “A very detailed and well-thought out approach and clear commitment to develop the portfolio further in terms of diversification and risk.”

FRR, the €37.2bn French national pensions reserve fund, was created in 2001 to help balance the first pillar of the pension system from 2020. Until 2010, with no explicit liabilities and an investment horizon expected to run from 2020 to 2040, FRR’s allocation was dominated by performance-seeking assets. In 2008, however, FRR began to factor in inflation-linked hypothetical liabilities. This was the precursor for a pensions reform in 2010 to address the huge deficit in the French pubic pension system that was caused by the recession of the crisis years. This saw FRR’s mission reviewed with the result that it is required to meet an annual liability of €2.1bn per year between 2011 and 2024.

As a consequence, the appropriate investment allocation needed to change dramatically. FRR thus adopted a liability-driven investment framework based around a liability-hedging portfolio and a performance-seeking portfolio. Moreover, its asset allocation aims to diversify risk, while allowing for flexibility through tactical allocation taking into account the current market environment.

To formulate its strategic allocation across assets and build its portflios, FRR performs Monte-Carlo simulations, enabling the fund to simultaneously calibrate the composition of both its liability-hedging and its performance-seeking portfolios. But this can be tricky as even if it decides to allocate more to the performance-seeking portfolio, it cannot obtain the level of diversification it would prefer since it is not permitted to invest in long/short risk premia, hedge funds or catastrophe bonds.

FRR first forecasts French and US yield curves, using many different spreads, the euro/dollar exchange rate and returns on equities, real estate and commodities. It derives these from macroeconomic analyses and in-house models that consider different market situations. Levels of correlation and volatility are estimated historically with a short-term tilt.

The euro and US yield curves in FRR’s Monte-Carlo model rely on three Nelson-Siegel parameters, the evolution of which depends on auto-regressive models calibrated to reflect, on average, central yield-curve hypotheses. It uses a similar approach to spreads using a Vasicek model to determine the evolution of interest rates. Based on the historical relationship between index returns, spreads and the Nelson-Siegel parameters, FRR can then simulate the performance of all kinds of fixed income investments. In the scheme’s Monte-Carlo simulations, random correlated variables are based either on one set of regulations or a regime-switching model. One way to choose the optimal portfolio with respect to the efficient frontier is utility, which is based on past decisions and asset management surveys. This enables FRR to be rather contrarian if need be, as its return requirements will decrease if its performance is better than expected.

However, FRR admits it is difficult to predict future portfolio changes. So it uses another optimisation method based on Infanger academic papers on dynamic asset allocation strategies that use a stochastic dynamic programming approach. It starts by predicting the utility in 2024 and then taking each year’s projected final assets under management, turnover and any expected loss into account work back until 2015 to predict the fund’s optimal asset allocation. The final asset allocation is chosen after conducting stress and robustness tests and qualitative discussion and analysis. As most of the risk on the surplus comes from equities, in 2015 FRR introduced option-hedging strategies using put spread collars and it aims to substantially increase its investments in illiquid assets such as real estate, private debt, private equity and infrastructures in the future.

Finally, FRR adds tactical decisions on all asset classes, including foreign exchange, using an overlay manager. The tactical decisions are taken based on one-year time horizons and depend on stretched valuations or risk considerations.

2014 Essentials



Founded in 2004

Defined benefit sovereign reserve fund

Assets: €37.2bn


  • one year: 8.7%
  • three years: 8.1%
  • five years: 3.9%
  • ten years: 3.9%

Quick facts

  • €37.2bn reserve fund supporting the French pay-as-you-go system
  • Annual liability bill of €2.1bn caused by crisis years from 2008
  • Extensive stress testing models to determine optimal portfolio between 2015 and 2024


  • Deutsche Telekom Pensionswerke Germany
  • Pension Fund SBB Switzerland
  • Royal Mail Pension Plan United Kingdom
  • SPK Sweden
  • Trafalgar House Pension Trust United Kingdom


  • Hermann Aukamp
  • Jeroen De Soete
  • Paul Kelly
  • Bob Swarup