Placing the Production of Investment Returns: An Economic Geography of Asset Management in Public Pension Plans

Abstract

Public pension funds are engulfed in a severe funding crisis. At stake is the financial stability of state and local governments as well as the welfare of over thirty million public-sector employees. Although cutting back on external asset management expenses could help save billions in taxpayers’ money and improve public pension funding, recent research suggests that public pensions remain predominantly outsourced and keep paying high fees to private-sector asset managers. This article examines why public pension funds outsource their asset management functions. It relies on mixed methods, juxtaposing positivist and reflexive approaches. The study relies on an econometric analysis of a unique panel data set of twenty-one state pension plans. The model tests specific relationships between levels of outsourcing and the organizational, economic, and political context in which these plans are embedded. The results indicate that outsourcing is linked to plans’ (1) investment return targets, (2) allocation to nondomestic and private market investments, (3) local financial sector vibrancy, and (4) proximity to a leading financial center. These quantitative results are enriched with insights from thirty-seven semistructured interviews with investment professionals employed by a top-performing state pension plan. The interviews help shed further light on how distance, politics, and governance affect pension plans’ outsourcing strategies. This article contributes new insights on context to economic geographic literature on pension decision-making as well as new perspectives to financial geography literature on the role of place and distance in institutional asset management.

Key words:

• financial geography
• insourcing
• asset management
• international financial centers
• outsourcing
• state pension plans
• pension decision-making

JEL codes:

• G11
• G17
• G23
• H11
• H55
• H75

Acknowledgments

The author thanks Gordon Clark, Adam Dixon, Rajiv Sharma, Rebecca Sandell, and Jane Pollard, as well as four anonymous referees for providing valuable comments on previous drafts of this article. None of the above should be held responsible for the opinions expressed herein. This research was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement number 681337), by P&I Research Centre, Crain Communications, Inc., New York, NY, and the Smith School of Enterprise and the Environment, University of Oxford. None of the above played any role in the formulation of the article’s content and argument.

Appendix

This appendix presents the detailed rationales and expectations for each variable included in the regression analysis.

Micro-organizational:

1. Total membership: the variable allows testing the importance of economies of scale; generally, I expect larger plans to manage a larger proportion of their assets internally.

2. Demographics ratio: the variable allows controlling the importance of the demographic distribution of plans’ members. I expect plans with a larger proportion of retirees per active members to be less inclined to build the required infrastructure and manage employment contracts to invest assets internally since their assets are declining to pay out benefits.

3. Home bias of asset allocation: the variable allows testing for the effect of the geography of asset allocation. I expect pension plans to have a preference to manage domestic assets in public markets internally because they entail low to medium place-specific information content (Clark and O’Connor 1997).

4. Complexity of asset allocation: the variable allows testing the effect of plans’ asset allocation toward private markets and alternative investments. Because these types of investments entail high place-specific information content (Clark and O’Connor 1997), I expect public pension funds to outsource their management to external contractors.

Mesoeconomic:

5. GDP, metro: the variable allows controlling for local economic activity at the MSA level. I expect plans that find themselves in less vibrant economic environments to rely more extensively on external contractors.

6. Unemployment, metro: the variable allows controlling for the effect of unemployment at the MSA level. High unemployment levels may make the area less desirable for investment professionals and increase recruitment difficulties, thus pushing plans to outsource. On the other hand, local governments may be inclined to build internal capabilities to create local jobs.

7. Finance and insurance percent of GDP, metro: the variable allows testing for the effect of plans’ local financial sector at the MSA level. I expect plans located in areas with a developed financial sector to outsource their investment management.

8. Co-location with IFC, one hundred miles (dummy): the variable allows testing for the effect of being colocated with an IFC. Similar to variable 7, proximity may increase difficulties to recruit specialized labor. Since IFCs provide competitive institutional investment services, proximate plans may also be more inclined to outsource, since they have many contractors to choose from at their doorstep. Finally, proximity might improve the quality of relationship with contractors.

Macropolitical:

9. State government debt-to-GDP ratio: the variable allows testing for the effect of budgetary deficit on outsourcing. I expect an increase in the former to increase the latter. This is because expenses on external investment management are charged directly to plans’ assets whereas expenses on internal capabilities are charged to plans’ sponsors (Urban 2018c). Consequently, outsourcing may be a way to reduce annual government expenses (not long-term expenses).

10. Funding ratio, n-1: this variable allows controlling for the effect of funding performance. I expect poor funding to be associated with outsourcing as plans seek external help and may want to transfer responsibility onto external contractors.

11. Investment return five years: this variable allows controlling for the effect of investment performance. I expect poor performance to be associated with outsourcing, since plans seek external help and may want to transfer responsibility onto external contractors.

12. Assumed rate of return (ARR): this variable allows testing for the effect of investment return expectations. I expect plans with ambitious investment targets to seek external help, since star performers are too expensive for public pensions to hire but can be contracted out.

Notes

¹ This figure comprises 14.7 million active participants,10.3 million beneficiaries, and 6.2 million other members, including inactive vested and inactive nonvested participants.

² The period 2006–12 is the longest time series available. Close examination of temporal patterns showed that the GFC had little short- and mid-term effect on insourcing strategies.

³ These include plans with no data points at all as well as plans with two consecutive data points missing; isolated missing data points (excluding 2006 and 2012) were approximated using the average of the two closest years.

⁴ Total membership is used here as a proxy for AUM since AUM is already the denominator of the dependent variable.

⁵ MSA-level data have been used extensively in empirical research in economic geography and urban studies (see Florida 2002a, 2002b; Wójcik 2011; Dougal, Parsons, and Titman 2015).

⁶ Funding ratio is tested, since it includes both the effects of funding discipline as well as investment returns. The previous fiscal year’s funding ratio is used to capture the effect of available information.