• Previous
  • Next

Value-at-Risk

We publish our annual Value-at-Risk1 (VaR) to give our stakeholders an estimate of how our portfolio value may change over a 12-month period. This is based on historical data and assumes the same prevailing market conditions with no changes to our portfolio composition.

Our annual VaR of S$22 billion2 as at 31 March 2011 is broadly comparable to last year. However, this represents a lower 12% of our portfolio value versus 14% a year ago, indicating a reduction in our portfolio volatility.

Our top 10 holdings represented 60% of our portfolio and 68% of our VaR. SingTel, our single largest holding, was 14% of our portfolio and 16% of our VaR.

Share of Value-at-Risk by Sector (as at 31 Mar)


Share of Value-at-Risk by sector comparing 2010 to 2011 for Asset mix in percentage share and Value-at-risk mix in percentage share

Download

  1. 1Weekly price movements over three years are used to compute our annual VaR based on the assumption that the most recent 3-year history would be more indicative of market behaviour over the following 12 months.
  2. 2A VaR of S$22 billion signals a one-in-six chance of our portfolio value falling by S$22 billion or more within the 12 months ending 31 March 2012. Conversely, this also means a five-in-six chance of a gain in value or an outcome better than a S$22 billion decline.

Monte Carlo Portfolio Returns Simulations

The Monte Carlo simulation method is sometimes used to generate the likelihood curve for various future portfolio returns. This assumes market conditions and portfolio mix remain unchanged during the period.

The chart below provides the 12-month forward returns likelihood curves covering 31 March 2007 to 31 March 2011, based on our portfolio mix for each year end. Their peaks represent the most likely portfolio returns at the end of the following financial years.

At the height of market uncertainty in March 2009, the returns likelihood curve was flatter, indicating more volatility. Against the projected five-in-six chance of our portfolio returns falling between -30% and +76% a year later, our actual 31 March 2010 portfolio return was over +42%.

For our 31 March 2011 portfolio mix, the returns likelihood curve indicates a simulated five-in-six chance that our portfolio will return between -17% and +26% by 31 March 2012.

Simulation of 12-month Forward Portfolio Returns (as at 31 Mar)



Simulation of 12-month forward portfolio returns

Download

  1. 3Based on Monte Carlo simulation for 12-month forward portfolio returns distribution, assuming no change in market conditions or portfolio mix.
  2. 4Actual TSR achieved one year later.

 

Illustrative 20-year Annual Return Likelihood



Illustrative 20-year annual return likelihood comparing global equity portfolio to thematic portfolio

Download

The Monte Carlo simulation method may also be used to simulate returns over a longer period such as 20 years.

Assuming no changes to market conditions or portfolio mix, the chart above compares the likelihood of annual returns over a 20‑year period between a generic global equity index portfolio (), and a thematic portfolio () which simulates Temasek’s portfolio equity mix.

The simulation for the likely returns of a generic global equity index portfolio is based on the long term outlook of the global equity market, factoring in volatility of prevailing conditions in the near term. The likelihood of annual returns is projected to be within the range of the likelihood curve as shown.

In comparison, the thematic portfolio, shifted towards Asia and the growth markets, with sectoral exposure re-weighted to simulate the Temasek portfolio equity mix, projected a likelihood curve which is flatter. This means a slightly more volatile portfolio. However, the likelihood of higher returns has also increased.

Back to top