Del 2
Topic article
6 Simulation of the real value of the Government Pension Fund 20 years into the future
In Report No. 10 (2009-2010) to the Storting, the Ministry presented updated estimates as to long-term real returns, risk (volatility) and correlations for global portfolios of equities, fixed income and real estate. These estimates were used to simulate the real value of the Government Pension Fund Global (GPFG) over a 15-year period. The simulation focused on expected developments in value, as well as the uncertainty associated therewith in the form of confidence intervals.
However, the simulation did not take into account the net inflow to the Fund from the surplus on the Fiscal Budget. These inflows are expected to be significant also in the coming years, although they will decline as a percentage of the value of the Fund. In the National Budget 2011, for example, the net inflow to the Fund was estimated at NOK 153 billion, or about 5 per cent of the value of the Fund. The ongoing inflows will therefore result in a stronger growth in the value of the Fund than suggested by the simulation without inflows.
This topic article presents simulations of the value of the Fund that also include the expected inflows to the Fund. Two issues are examined: a) How large can the Fund be expected to become during the years up to 2030, and how much uncertainty is associated with the projection? b) How large will the net inflow to the Fund be as a proportion of the value of the Fund, and how uncertain is this projection? Will the Fund go from net inflow to net outflow, and when?
The simulations assume that the actual use of petroleum revenues over the fiscal budget is 4 per cent of the value of the Fund each year. This is consistent with a spending of petroleum revenues that follows the 4 per cent trajectory of the fiscal rule over time. However, it is not taken into account that the fiscal rule allows for the actual spending of petroleum revenues to deviate from the 4 per cent trajectory at times, which represents an important source of uncertainty in the analysis.
The simulations are based on the Ministry's long-term estimates as to real return, volatility and correlations, which remain unchanged from last year. It is emphasised that these estimates are intended to apply in the very long run, and therefore will not necessarily be representative with regard to shorter periods of time, such as for example the coming 20-year period. The Ministry has thus far not prepared specific estimates for this period, or corresponding estimates more generally, cf. the discussion in Chapter 8 of Report No. 10 (2009-2010) to the Storting. The use of the long-term estimates in the simulations is another source of simulation uncertainty.
Simulation tool
The model used in the simulations is discussed in more detail in previous Reports to the Storting on the Management of the Government Pension Fund.1 The model has been refined recently to provide a more realistic description of, inter alia, the risk of major losses in the financial markets (so-called stochastic or time-varying volatility), and ongoing inflows to the Fund. A description of the simulation model will be posted on the Ministry's website (www.government.no/gpf).
Briefly summarised, the real value of the Fund over time are simulated by way of a so-called Monte Carlo (stochastic) model, which «draws» a large number of potential paths that the value of the Fund may follow for a given number of years into the future. The outcomes, which are «measured» at the end of the period, provide information about the expected value and the uncertainty around this expectation. The model used by the Ministry is based on known pricing processes for financial assets that allow for deviations from the normal distribution (stochastic volatility and «mean reversion»). This enables a more realistic description of the returns.
The model simulates the stochastic evolution of the portfolio's underlying equity and fixed-income benchmarks and exchange rates in Europe, the Americas and Asia/Oceania. The simulations are made using nominal figures, which at the end of the investment horizon are deflated by the chosen (deterministic) price index for conversion into real values.
The simulations presented here use Norwegian kroner as the reference currency.
Inflow to the Fund is given by the value of the net cash flow to the State from petroleum activities (cf. forecasts until 2030 in the National Budget 2011), less 4 per cent of the value of the Fund as per the end of the previous year. The model assumes, furthermore, some uncertainty with regard to the future net cash flow (the level of the path follows a normal distribution, with the forecast in the National Budget 2011 as the expected value2).
The simulations are made on the basis of the current strategic asset allocation and regional distribution. Return contributions from active management are not included. The value of the Fund is simulated from year-end 2010 (value NOK 3,077 billion) until year-end 2030.
Findings
Figure 6.1 shows the simulated real value of the GPFG for the next 20 years (in NOK billions at 2011 prices). The expected path (the median) is represented by a solid black line. The real value of the Fund is expected to grow to somewhat in excess of NOK 7,400 billion in 2030. The orange and brown fans show the 68 per cent and 95 per cent confidence intervals, respectively, which under the adopted assumptions represent the probability that the real value will fall within these intervals. The figure shows that there is considerable uncertainty associated with the size of the Fund 20 years from now. However, there is a very high probability that the real value of the Fund will increase. The probability that the real value of the Fund in 2030 will be lower than its value as per year-end 2010 is estimated to be less than 1 per cent.
Figure 2.2 (cf. Chapter 2) shows that the net inflow to the Fund is expected to decline as a proportion of the value of the Fund.3 The expected path (the median) is again represented as a solid black line, whilst the coloured fans show 68 and 95 per cent confidence intervals. Whilst the proportion in 2010 was about 5 per cent, it is expected to decline below zero around 2020, and to be about minus 2.4 per cent in 2030 (net outflow). The figure shows that this expected path is, under the adopted assumptions, associated with an uncertainty of 1 – 2 percentage points. Given these assumptions it is, nevertheless, almost certain (probability of more than 99 per cent) that the net inflow, as a proportion of the value of the Fund, will be negative in 2030. In practice, this means that capital will need to be retrieved from the Fund, which implies a certain liquidity need. However, the expected net outflow is lower than the expected real return (which is somewhat in excess of 4 per cent), thus implying that the real value of the Fund is expected to increase also 20 years from now, as shown in Figure 6.1.
A possible measure of the risk associated with the Fund may be the risk of large net outflows as a proportion of the value of the Fund. In the simulations, a large net outflow in a single year can be triggered by a large decline in the value of the Fund through the fiscal year, while the net cash flow as a proportion of the value of the Fund as per the end of the previous year will be low (less than 4 per cent). Under the stated assumptions, the simulation shows with a 95 per cent probability that the largest net outflow as a proportion of the value of the Fund until 2030 will fall within the interval 1.4 – 3.8 per cent (median 2.8 per cent). There is a very low probability (less than 0.1 per cent) of an annual net outflow as a proportion of the Fund in excess of 5 per cent over the coming 20-year period.4
There will always be uncertainty associated with this type of simulations. The findings presented here are influenced by, inter alia, the assumption with regard to the spending of petroleum revenues. Other elements of uncertainty are the assumptions with regard to future real returns, risks and correlations and their applicability to the coming 20-year period, and whether the simulation model to a sufficient extent absorbs the probability of very large losses in the asset markets. It may therefore be of interest to compare these simulations with findings from so-called historical simulations, in which historical return figures are used instead of model-simulated ones, but to retain the assumptions with regard to portfolio composition and inflows. Historical real return data for equities and government bonds going back to 19005 provide 92 overlapping 20-year periods that can be used to simulate the annual real return on the GPFG into the future.
According to this historical simulation, the real value of the GPFG will in 2030 be between NOK 2,667 and 19,556 billion at 2011 prices, with a median of NOK 8,521 billion. Furthermore, the simulation shows that the net outflow in 2030 will be between 0 and 3.7 per cent of the value of the Fund. The median net outflow is 2.5 per cent. Over the period until 2030, the largest net outflow will be between 0.8 and 3.7 per cent of the value of the Fund, with a median of 2.7 per cent. In particular, historical developments over the 20-year periods until 2000-2002 result in relatively large simulated net outflows (3.5 – 3.7 per cent). These periods are characterised by many years of high returns, followed by some years of major contractions in the stock market.
The findings from the historical simulation are reasonably consistent with the findings from the simulation model.
Footnotes
Credit Suisse Global Investment Returns Sourcebook (2011), and Elroy Dimson, Paul Marsh and Mike Staunton (2002) Triumph of the Optimists: 101 Years of Global Investment Returns, Princeton University Press, and Morningstar.