The Subsidy to Infrastructure as an Asset Class - PowerPoint Presentation

Slide1

The Subsidy to Infrastructure as an Asset Class

by Aleksandar Andonov, Roman Kräussl, and Joshua Rauh

Discussant:Vesa PursiainenUniversity of Hong Kong & Imperial College London

This paper

Return characteristics of infrastructure investmentsDifferences in the performance of different LPsInternational sample of infrastructure investments from Preqin:640 institutional investors from 38 countries

425 funds, 206 GPs, 3,687 infra assets3,081 investor-fund observations, 1,096 direct investments38,676 investor-deal observations

Main findings

Average cash flow profiles of infra funds look similar to those of PE fundsSeems to contradict the usual sales pitch of infra funds, i.e. more stable, longer-term and less cyclical returnsUS public pension funds have lower average returns than other LPsPublic pension funds may be subsidizing infrastructure at the expense of plan members/taxpayers

Contribution

Shedding light on an important topic

Comments on methodologyRisk profile

Average cash flow profile of a fund does not directly measure risk, or even the stability of cash flowsA large part of the cash flow profile may be driven just by the investment cycle of the fund (especially if cash generated at the asset level is used to pay down debt, which is common in buyouts)Even if the cash flows of the underlying assets are different, the aggregate fund-level flows may look quite similar

Comments on methodologyRisk profile (cont’d)

Would be nice to see:The distribution and standard deviations of fund return metrics and cash flowsInfra vs. PE/VCBy project type (greenfield/brownfield/secondary)

By investor typeFund returns by vintage for different fund and investor typesAre infra fund returns more stable (or less cyclical) over time?Does the composition of infra funds (and hence the return distribution) change over time

Comments on methodologyInvestment timing

All of the main the analysis lacks controls for vintage Does the public pension fund dummy just capture the effect of timing?Public pension funds start investing in infra later than other investors (Table 1):

Comments on methodologyInvestment timing (cont’d)

PE fund PME by vintage (Ang et al., 2018, JF):

Avg. first infra investment for private pension funds

Avg. first infra investment for public pension funds

Comments on methodologyInvestment timing (cont’d)

BCG analysis of infra deal MoM by year:

Comments on interpretationWho is being subsidized? (cont’d)

If public pension funds (semi-)deliberately subsidize infra assets:The investments probably have to be local (no political incentive to subsidize far-away infrastructure)This can be directly tested by regressing deal/fund performance including an interaction of pension fund dummy and an indicator of proximity (same state/distance)

Are public pension funds more likely to invest locally than others (i.e., stronger home bias)?Why do non-US public pension funds not subsidize local infra?Explore the effect of LP board composition (data from Andonov et al., 2018, JF)?

Comments on interpretationWho is being subsidized?

US public pension funds have lower average returns than other investors as a wholeIs this because of:Conflicts of interest – political considerations vs. maximizing returns (Andonov et al., 2018, JF)?

-> Subsidy to (presumably local) infrastructureLack of skill relative to other investors?-> Subsidy to poor fund managersDifferent risk profile?

Comments on interpretationWho is being subsidized? (cont’d)

If it is lack of skill:Maybe there is a difference in how different LPs get into infra investments, and hence the funds are differentInfra-arms launched by traditional PE funds?Pension funds might be more likely to invest money in funds launched by GPs with whom they already have PE investments

Classify GPs as infra-specialists (Macquarie, Global Infrastructure Partners etc.) vs. traditional buyout funds (KKR, EQT etc.)?Is there a difference in asset profiles and returns?Do traditional PE managers’ brand extensions underperform?Are the fees different?Is the sensitivity of fund participation to past performance different for public pension funds?

Conclusion

Very interesting paper and an important topicStill room to:Expand the descriptive part on infra funds, especially wrt. risk profileFurther clarify the channels for the subsidy storyGood luck!

Appendix

Smaller comments & random thoughts

The regression control variables differ a lot between different analyses – they should probably be similar across different tablesMuch of the analysis is only testing US public PF vs. other LPsShould perhaps include indicators for other fund types as well to see if the other LPs are similar to each other, or if there is more variation across types than just public PF vs. others (Table 3 sort of has this, other tables don’t)

Maybe it would be fairer to just compare them with US private pension funds, given they have a more similar liability structure and investment goals?Non-US public PFs actually have lower exit rates than US ones, based on the estimated coefficient in Table 3. Similarly, the estimates for gov’t agencies and SWFs are actually larger than public PFs, although not significantTable 5 includes non-US public PF dummy, but Table 6 doesn’t – not clear why?

Smaller comments & random thoughts

Would be nice to have average returns by investor type (Table 1) and their standard deviationsInvolvement in direct deals as well might signal skill of the LP – this might be worth controlling for in the regressionsWhat if you show the cash flow profile of the funds most and least public pension fund exposure? Are they still the same?Might be worth controlling for some GP characteristics as well in Table 3 (like

AuM)From Table 5 it seems that the sample includes funds of funds and debt funds – might want to exclude those as they are fundamentally very different from equity funds

Smaller comments & random thoughts

Are individual deal investment periods longer in infra?