Taking different approaches to the management of assets
The continuous decline in interest rates and the stock market correction have put pension funds under significant pressure in recent years. To improve the sector's financial position, contributions have been raised and pension promises have been cut down. There has also been a reorientation in investment strategies, with a growing emphasis on asset liability management (ALM).
In this article, we examine the behavior of 77 Dutch pension funds over the period 2002-2005, using data from a quarterly survey. Our sample largely covers an episode during which pension funds had to take significant measures in order to restore their funding ratios.
Furthermore, this panel dataset allows us to compare different types of pension funds, in particular industry-wide versus company-linked funds. This provides important value- added to other studies, which are mostly based on aggregate figures.
We use data from a quarterly survey. Our dataset includes 77 institutions, representing about three quarters of total pension assets in the Netherlands. This means that the smallest pension funds are under-represented, although the survey does include a wide range of size categories. The data are available from 2002 onwards. This is a short sample, but largely covers an interesting episode during which pension funds had to deal with dramatic financial market developments.
Unfortunately, our dataset does not include off-balance sheet positions. Pension funds have an incentive to use derivatives to reduce interest rate risk, exchange rate risk and their exposure to the stock market. Presumably, such instruments will become more important with the introduction of a new, more risk-based regulatory framework in 2007.
The table above summarises some basic characteristics of the survey data for different size classes, as well as industry-wide versus company-linked pension funds. Presented are averages per subcategory; medians and aggregate numbers are not presented but show a similar pattern. Size categories are quartiles, based on the institutions' total assets in the fourth quarter of 2003, our sample's middle observation.
A large proportion of total assets is invested abroad, which can be largely explained by risk diversification. Large pension funds hold more foreign assets than small ones. Presumably, scale economies make it easier for large pension funds to operate on a global scale and deal with exchange rate risk. Differences in the asset mix are substantial as well: large pension funds invest more in stocks and real estate, while smaller institutions hold a higher proportion of fixed income investments.
The differences between industry-wide and company-linked funds are less pronounced. Interestingly, company-linked pension funds are more internationally oriented than industry-wide funds, despite the fact that they are much smaller. In addition, industry-wide funds invest more in real estate and less in stocks, while the proportion of fixed-income is similar for both categories.
We carried out panel regressions to investigate pension funds' financial behaviour, using a fixed effects estimator. We examine two important tools: financial transactions and changes in pension contributions. Financial transactions may be used to adjust the asset mix and risk profile, while contributions can be raised to improve the funding ratio. For financial transactions, we focus on net purchases of the most important asset categories - bonds and equity - which can be separated from changes in their valuation. For the results, we restricted our sample to the period 2002-2004, that is, we excluded the observations for 2005. In that year, pension funds were anticipating two important regulatory changes, which probably affected their financial behaviour.
First, a new supervisory regime was planned to be introduced in 2006, postponed to 2007 in September 2005. Second, in 2006 the new act on early retirement and life course savings arrangements entered into force. The act included changes in the tax treatment of pensions schemes, aimed at discouraging early retirement.
The regression results show that pension funds do not reduce the risk of their portfolio after a deterioration in their financial position. Rather, a relatively high funding ratio is associated with a net purchase of bonds, while the response of equity transactions is insignificant.
An interesting question is to what extent this is consistent with rebalancing, that is net purchases that counteract price movements in order to restore a strategically fixed asset allocation scheme.
Rebalancing behaviour has been established before on the basis of aggregate data for all Dutch institutional investors. Indeed, the evidence is that bond and equity transactions are negatively related to corresponding price changes.
This is most pronounced for industry-wide funds, which respond instantaneously to both equity and bond prices. Company-linked funds react immediately to changes in bond prices, while the response to equity prices takes place with a one-period lag or not at all
Altogether, especially for industry-wide funds, our results are in line with rebalancing while the evidence for company-linked funds, and for all funds together, is less clear.
Pension funds may restore their financial position by raising their contributions. In our dataset, pension contributions include both regular premiums that are negotiated every year and ad hoc contributions by the sponsor.
In a preliminary analysis, we investigated the relationship between pension contributions - as a proportion of total liabilities - and several lags of funding ratio and changes in the valuations of bonds and equity. In all regressions, only lags of the funding ratio were significant, albeit not for industry-wide funds.
The difference between both types of funds is in line with a recent survey among the 25 largest Dutch pension funds, which shows that company-linked funds adjust contributions more rapidly than industry-wide funds. Presumably, they can do this more easily because there are fewer parties around the table.
Another reason may be that company-linked pension funds are on average more mature and therefore need to raise contributions more than industry-wide funds to produce the same effect
We carried out several robustness checks. First of all, we repeated all regressions using a market-based discount rate instead of the fixed interest rate that is included in our dataset. This is not a straightforward exercise, but with some assumptions we were able to calculate a proxy for market-based liabilities. However, including these instead of the original liabilities only has a limited effect on the estimation results and does not change our conclusions.
Second, we repeated the analysis using pooled regressions instead of a fixed-effects estimator. Third, we examined possible changes over our sample period, particularly regarding rebalancing behaviour.
Finally, we re-estimated all equations for the entire period 2002-2005, that is, including the last four quarters prior to the planned introduction of the new regulatory framework and the Act on Early Retirement and Life course savings. For the regression results for bond and equity transactions this only leads to minor changes and does not change our conclusions. However, the regression results for pension contributions change substantially.
A closer inspection of the data reveals that at the end of 2005, for several funds the amount of pension contributions increased markedly, despite an improvement of their funding ratios - measured using a fixed discount rate. However, under the new market-based rules, funding ratios were deteriorating because the ongoing decline in interest rates.
Anticipating this new regime, several sponsoring companies raised their pension contributions. In addition, a number of pension funds have indicated that extra contributions by the sponsor were related to changes in pension arrangements. The new act prompted many funds to terminate pre-pension schemes, which the sponsor often compensated by a once-only contribution.
For similar reasons, pensions plans that switched from DB to (collective) DC often received an ad hoc contribution.
We have established several interesting results.
Pension funds show rebalancing behaviour, which is most pronounced for the largest institutions.
More generally, large pension funds are more active on financial markets to manage their financial position, whereas smaller institutions and company- linked funds are more prone to adjust the level of pension contributions.
Many industry-wide funds have made DB schemes more sustainable by moving from final salary to career average systems, while several company-linked pension funds are shifting financial risks to their participants by introducing more DC elements. Altogether, pension funds are dealing differently with their exposures to financial risks.
What are the implications of our findings when future trends are taken into account? As the proportion of active members falls, it will become more difficult to prevent underfunding by raising contributions. Other trends in the economy are increasing job mobility, mergers and acquisitions, and sectoral shifts. As a consequence, pension funds increasingly will have to accommodate transfers of pension rights and deal with discontinuity issues.
Differences across pension funds may be a further complication to the portability of pensions. Given these challenges and the limited tools to adjust a fund's financial position, a growing focus on asset liability management will probably be important to protect the pension sector's resilience.
Jan Kakes is at the De Nederlandsche Bank , Financial Stability Division. Views expressed are those of the individual author and do not necessarily reflect official positions of DNB. This article is an edited version of DNB Working Paper 108: ‘Financial behaviour of Dutch pension funds: a disaggregated approach'