As a follow-on to my post the other day, here’s some further thoughts on risk and uncertainty, applied to revenue volatility:

• If long-term secular trends reflect low risk levels, fiscal volatility should revert to a mean where essential spending is almost certainly accomplished in the long run – that is, there is a low risk that the public sector will default on its fundamental obligations. This is true of most US state and local governments. But the path to get there is highly uncertain.

• Non-wealthy taxpayers and especially beneficiaries live in the short run, so for them a high level of uncertainty results in a high risk that something bad will happen. Contractual recourse for certain redress in future is not available to them – the ballot box is not at all precise or responsive. So allocating uncertainty to these parties is effectively allocating risk (in the sense of a “predictably bad outcome”).

• In contrast, infrastructure commitments are intrinsically long-term, which sets the stage for institutional investors to take a long-run view and incorporate the state’s low risk profile in their agreements. With detailed long-term contracts, the outcome of short term uncertainty (e.g. low payment this year) can be precisely and certainly addressed in future (e.g. a higher payment at some point). In effect, the state can allocate a high level of uncertainty – but not risk – to the investor.

• More specifically, cost-effective flexible debt capitalization can deal with the uncertainty of cash flow timing as long as the risk of ultimate repayment is low. Over the course of a 30-year infrastructure financing, a AAA-rated state is well-positioned to provide that assurance.

So overall, a P3 solution can allocate short-term fiscal uncertainty to the private sector (which is able to deal with this efficiently) but long-term risk would be retained by the state (which is well-positioned to keep it). This will reduce the harm of uncertainty when it results in a high risk of a bad outcome – directly for the infrastructure (planning, deferred maintenance) and indirectly for other “crowded out” spending programs (especially social services).

A minor additional point: one objective of a traditional-type P3 is “risk transfer” – it is useful to make it clear at outset that next-generation P3 is different, focused on “uncertainty transfer”

Something surfaced in a recent book I read by Mervyn King (ex-BoE governor) called End of Alchemy — the relevant parts are about about the limits of risk management in central banking.

One of King’s central points is that  “risk” is actually fundamentally different than “uncertainty”.  Risk, when strictly defined, includes more/less quantifiable probabilities that you can “optimize” around.  That does sounds familiar.

But “uncertainty” (“Knightian uncertainty” is the technical term, it seems) is by definition unquantifiable – and the best you can do is “cope” with it.

King then shows pretty convincingly (and consistently with my own experience) that central bankers, securitization experts, rating agencies, etc. all thought they were modelling “risk” to a high level of precision in the lead-up to 2008 FC — but they really weren’t.  The probabilities were not remotely correct.

In hindsight, that seems pretty obvious — even 20 years of data is meaningless in context of modelling something so complex as human financial markets.  Maybe a million years would do?

So for complex economic processes, it’s better to “know that you don’t know” and assume “radical” uncertainty — and then set yourself up, not to model precise outcomes, but ways to cope when the SHTF, to be blunt.

It dawned on me that a lot of this applies to P3s when we talk about “risk transfer”.

Maybe that’s valid for some specific physical properties of an infrastructure asset component.

But for something as complex as toll road usage?  Not likely.  If this risk gets transferred in a P3, it’s not “optimizing” risk allocation but naked speculation — one side will win big, and one side will lose.  Or neither. Or both.

The better question in a P3 risk transfer is not “who might win?” but “who can cope with this better when the SHTF?”

It strikes me that some of the lessons apply to practical modelling of P3 value – and that maybe the real role of P3s for complex assets like infrastructure is less about transferring insurable risk (because there’s not much of that actually) and more about the ability of the private sector to “cope” with uncertainty more efficiently (which seems true and might be actually measurable).

Draft of upcoming short opinion piece. Three WIFIA amendments are noted — 55-year loan term for dam safety projects, statutory interest rate reset, and a FCRA Criteria fix.

^{PDF here}

We don’t want any games from cost shares in federally involved projects sneakily hiding something from the budget, right? So, if there’s any doubt, include it, just like it says in the one sentence I read in some old report. And if the cost share ‘federal asset’ is included in the cash budget, the WIFIA loan paying for it ought to be as well. I mean, it’s the same federal asset, or sort of the same, or something. Anyway, it’s all just commonsense. Put the WIFIA loan in the nice and safe-sounding cash budget like all the other federal assets. No tricky accrual FCRA privileges for you, mister non-federal cost-share! Oh, that means the loan is effectively ineligible? Well, too bad — we must maintain the sacred integrity of the federal budget!