Christina Atanasova and Gilles Chemla have posted a discussion of pension plans,where holdings show a tilt toward private equity and real estate investments.
Atanasova is associate professor of finance, Beedie School of Business, Simon Fraser University in Burnaby, BC, Canada. Chemia is professor of financial at Imperial College Business School, London, UK.
Together, they make the following empirical points: defined benefits plans backed by firms with high research and development expenditures (R&D) tend to tilt toward a large private equity allocation in their portfolios; and that defined benefits plans backed by firms with high land and buildings holdings (L&B) tend to tilt toward a large real estate investment in their portfolios. Of course, PE is a common mechanism by which high-tech, and thus R&D-oriented, firms raise their money so in that situation, as in real estate, this is a case where institutional familiarity with an investing space creates comfort with it.
Anecdotes and Imagination
The article uses anecdotes to confirm the existence of such a tilt. In this line, it mentions Delta Airlines, which had between 3% and 5% of its pension assets invested in oil and gas throughout the period 1999 to 2014. Of course as the article says the purchase of fuel (and the creation of hedges against unexpected increases in fuel cost) is a key operational concern for airlines. That appears to translate into a “we know this stuff” confidence on the pension side. Pfizer, which is very much an R&D oriented pharma company, had a DB pension fund between 7% and 14% in PE and venture capital in the same period.
In more statistical terms, Atanasova and Chemla find that a one standard deviation increase in the ratio of sponsor R&D expenditures increases the plan’s PE investment by 0.74%. Similarly, a one standard deviation increase in the ratio of sponsor L&B expenditures increases the plan’s real estate investment by 1.27%.
This relationship, the authors say, “remains robust to a number of alternative specifications and when we address concerns of endogeneity and changes in pension regulation.”
Is this a good or a bad thing from the point of view of the soundness of the pension fund? Presumably if pension plans were benefiting from value-relevant information from the operational side of the sponsor this would improve performance. One can imagine a scenario in which someone who buys airplane fuel for Delta is calling up a pension fund manager to say, “I’ve just learned something big about fuel prices and how changes in the near future may help oil company XYZ” and the pension manager changes the portfolio accordingly. That sounds like a scenario that would be subject to prosecution. The authors speak more vaguely of how information might “trickle down from the corporate sponsor to the firm’s trustees” and benefit decisions. But neither of those scenarios describe what is happening.
Instead, the familiarity tilt is a non-rational, one might say a behavioral one.
The Tilt and Its Cost
The tilt comes with a cost. And, let us pause to observe, that is what efficient markets theory would lead one to expect. For as we’ve noted “tilt toward what you know” is a behavioral bias, not a rational decision, and if markets are rational, even more-or-less rational, than they will punish it.
Behold, they do. Atanasova and Chemla report that pension funds with this sort of a tilt underperform other plans by up to 200 basis points.
These authors draw on theoretical work by Philem Boyle et al. in explaining their empirical results. In 2012, Management Science carried a paper by Boyle (the Irish economist who, in the 1970s, pioneered the use of Monte Carlo methods in the study of options pricing) and three colleagues. This 2012 paper is called “Keynes Meets Markowitz: The Trade-off between Familiarity and Diversification,” and it proposes that as the correlation of other assets increases, i.e. as the benefits of diversification lessen, the pay-off from investing in unfamiliar assets also lessens. So higher correlation should lead to an increase in the size of the familiarity bias.
Atanasova and Chemla find support for Boyle’s model in their data.
Their results are also consistent with the hypothesis that “trustees have pessimistic beliefs about alternative assets as long as the parent firm characteristics are not related to these alternative classes.”