The Gothenburg-based Andra AP-fonden (AP2), one of the national ‘buffer’ funds of Sweden’s pension system, took the unusual step last year of terminating 16 of its domestic and European equity mandates at one stroke.
This bold move was part of a strategic decision to draw a clearer distinction between the active and passive management of risk. In practice, this meant AP2 bringing the management of Swedish and European equities in-house and moving more of its externally managed assets into global equities portfolios.
Two separate but inter-related issues were behind the move. The first was AP2’s realisation that some of its equity managers were not taking enough active risk. The second was the realisation that uncertainties about the pension fund’s long-term liabilities would mean that it would need every percentage point of returns it could squeeze from its assets.
Since AP2 began to manage pension assets in 2001, it has maintained a relatively high exposure to equities of around 60%. Yet the performance of its equity mandates has been mixed, says Petter Odhnoff, chief investment officer of AP 2. “We have been quite satisfied with our absolute returns last year, but relative returns were not satisfactory. Given the benchmarks and the remit we had chosen, we did not succeed in generating an alpha relative to those benchmarks,” he says.
Externally appointed managers of domestic equities, in general, performed poorly, he says. “Everyone says the Swedish market was very difficult last year, but there are always excuses. So we thought that we ourselves should do something to get a better return.”
AP2 has now moved the management of most of the mid- and large-cap Swedish equity mandates in-house. It has also built three ‘alpha processes’ or portfolios for these assets.
The first is a traditional, capital-weighted portfolio. “Here we try to achieve outperformance by underweighting or overweighting in certain stocks and sectors,” says Odhnoff.
The second is a long-term strategic portfolio, where performance is benchmarked against Sweden’s 50 biggest companies, equally weighted. “The focus here is never to look at the benchmark and only to invest in companies that will generate a return higher than our expected return from the asset class.
“We focus on two kinds of companies. We like companies that, over time, have been very good in generating shareholder value. We also have a value bias. We look at companies that we think are undervalued. We prefer companies that have strong cash flows and that pay dividends, so that we get paid for waiting.”
The third process is an enhanced portfolio run by a new quantitative analysis group that AP2 set up last year. “The group built up an index portfolio last year, and this year they are implementing alpha strategies that are developed from a more quantitative process,” says Odhnoff. “But it is no black box. Strategies have to be comprehensible and financially logical.”
AP2 has also has also re-distributed its equity mandates outside Sweden, which were originally allocated to European regional and sector managers, as well global equity managers.
“Our experience is that our global managers have done quite well. Our regional managers had mixed results, and most of our sector managers were struggling.
“The possibility for alpha depends on breadth and skill. We found that the breadth in the sector mandates was very limited. With different managers in each sector we had very few bets between sectors. And correlations being very high in the market, we saw that it was the wrong strategy for us.
“We think the skill was probably there but the fact that each manager tried to control risk in his or her own portfolio meant that the overall risk-taking was very diluted. We realised that what we were getting was an enhanced manager for an active fee with a negative alpha.
“So to get the better breadth we decided to close down these mandates and move the money to managers with global mandates. Our numbers show that you will get higher information ratios from enhanced and global managers.”
This has meant separating the wheat from the chaff. Odhnoff points out that managers with a tracking error of between 2% and 4% tend to operate mid-way between passive and active management. Managers with a tracking error of more than 4% behave differently. “They very seldom look at the index and will focus on their investments instead.”
Odhnoff prefers the perspective of these managers. “When you start your investment process, you don’t start with the index weight of the company, you start with the question ‘what kind of return will this company generate?’ We think there is more capacity for sustainable high returns from that approach.
“The managers that we’re going to choose are the ones who look more than six months ahead and who focus on the return of an investment rather than the extent to which it deviates form the index.”
These managers will be needed if AP2 is to do the job it was set up to do, says Odhnoff. One of the features of the new Swedish pension system is an automatic balancing mechanism. This is activated if long-term deficits appear in the system. The effect is to reduce the indexing of pensions, which is linked to growth in average incomes.
The mandate of the buffer funds, which collectively represent about 11% of pension system liabilities, is to manage their assets to minimise the risk that the balancing mechanism will be activated.

Calculating the size of the future liabilities is becoming increasingly difficult, however. Currently the Försäkringskassan, Sweden’s National Social Insurance Board, produces projections of future liabilities based on demographic assumptions, which are used by each AP fund as the basis for their asset liability modelling (ALM) studies.
AP2 does not believe that these projections are sufficiently comprehensive. The projections do not take full account of uncertainty, and in particular, the uncertainty that a projected ‘bad’ scenario could be worse than expected.
Last year, the pension fund developed an ALM model for its strategic portfolio, with the Frauenhofer Chalmers Research Centre for Industrial Mathematics, which incorporates this uncertainty.
The new ALM model introduces uncertainty and volatility not only in assets but also in the income index, inflation and the employment rate.
These new factors are described by an average growth rate and an uncertainty/volatility around the growth rate. Both the average rate and the uncertainty influence the key numbers of the pension system, says Odhnoff.
“The numbers show us quite clearly that over the next 10 years we could probably live with 70 to 80% fixed income. Between 10 and 20 years from now we would probably need 50:50 fixed income and equities.
“But between 20 and 30 years from now, for us to have any chance of living up to what we are supposed to do we need all the returns we can get and we need to take all the risk it is possible to take to be able to support the pension system.”
These returns will not be easy to find, and may mean looking for them outside the traditional asset classes, says Odhnoff. “We are now trying to get a better understanding of certain sub sectors of the equity and credit market to see if they improve the outcome of the model. We already have exposure to emerging market equities, for example, but the model is telling us we should have a lot more.”
One thing is certain – AP2 will be looking to managers for an active management of risk that is genuinely active.