( I guess Adrian Orr will now have to add to the “two bloggers” –  the only people he concedes were critical of his tree god shtick – a pretty well-respected columnist in the nation’s largest-circulation newspaper?   When you masquerade as a tree god –  weird metaphors, worse history and all –  it might risk someone suggesting you were away with the fairies? It is a bad metaphor, badly used and shedding little light, instead serving mostly to misrepresent and exaggerate the contribution and place of the Reserve Bank itself. )

The Governor’s proposals to increase massively the amount of capital banks have to have are back in the spotlight.  There was apparently a briefing on Friday for favoured media/commentators, but they still won’t lay out some of the basics that (they say) support their claims.  One journalist emerged from the Friday briefing and rang me seeking a bit of information (that I didn’t have at my fingertips) that you’d have thought the Reserve Bank would have been proactive about having out there.   We’ve seen nothing from them on the expected transition (eg a range of scenarios as to how banks and markets will respond, and what the output and efficiency implications might be) and we still don’t have any analysis supporting their claims that (a) the new proposed capital requirements would be “in the pack” internationally, or (b) that there is a free lunch on offer (more stability and higher GDP).  Remember this graph from the Bank’s consultative document.

They never spelled out the analysis in support of it –  indeed, they largely eschew modelling, suggesting that they really have no idea where that “optimal” point might be (or, thus, whether we might not already be near it, or even above it (in capital ratio terms)).  Perhaps in tomorrow’s speech the Deputy Governor might finally set out the case in a more convincing terms –  he might even consider discussing why it is that if his boss really thinks big bank failures are what we have to worry about, they plan to still insist on tiny banks having almost as high capital ratios as, say, the ANZ or BNZ.  Or why, for example, the Reserve Bank has refused until now to allow Kiwibank to compete on the same (capital) terms as the big banks.  Or to carefully recognise that “levelling the playing field” (perhaps a worthy goal) is something quite different  – and needs to be evaluated separately – to raising the hurdle for everyone.  The case for something like the former might be easy enough to make.  The case for the latter really hasn’t yet been made at all.

The Bank attempts it with more folksy analogies.

Bascand said at a news briefing that the bank’s proposals were about “putting the roof on while the sun is shining”

That’s generally a good idea –  putting a roof on.  But when credible analysis –  an internal commentary on which was released by the Reserve Bank here – suggests that the actual capital ratios of New Zealand banks are already relatively high by international standards a better analogy might be putting on a whole new roof over the top of an existing one when the existing one was perfectly serviceable and about only five years old.   That would be worse than gold-plating, it would be just a rank waste of resources and a whole new layer of regulatory inefficiency.

Much of the Bank’s rhetoric about these proposals has been built around the notion of the probability of crisis.  This is from the Governor’s speech notes of 30 November

In making this assessment, our recent work makes the explicit assumption that New Zealand is not prepared to tolerate a system-wide banking crisis more than once every 200 years.  

That is an annual crisis probability of 0.5 per cent or less.

It has a ring of science around it, but when one digs into the background papers the Bank has (belatedly) released, it is striking how flimsy this number seems to be.   Just a few weeks prior to that speech of the Governor’s a decision paper was put up

36. We believe a reasonable interpretation of ‘soundness’ in the context of capital setting is to cap the probability of a crisis at 1% (or 0.5% if we wish to mirror approaches taken in insurance solvency modelling).

So staff actually thought that a one in 100 year crisis was the appropriate number, but included the 1 in 200 approach almost parenthetically, very late in a longrunning review.  In principle, that difference should, in turn, make a big difference to the appropriate minimum capital requirements. In practice, it looks as though the Bank had some capital ratio requirement numbers in mind, and then cast round for some props to support them.

As it is, the risk-appetite framework (the 1 in 200 year benchmark) is pretty questionable as a starting point.  In part, that is because any well-designed regulatory intervention needs to look at both the costs and benefits of a particular intervention. It is largely meaningless to put a stake in the ground (whether 1 in 100 year or 1 in 200 year) without some robust sense of what the efficiency implications of the resulting interventions might be.  There is little sign that any serious analysis of that sort was done before the Governor plumped, late in the piece, for his 1 in 200 standard.

But there is also no “state of the art” when it comes to either defining financial crises, the cost of those crises, or the contribution that bank capital requirements could plausibly make in reducing those probabilities/alleviating those costs.   That really is the guilty secret here.  And it would involve no shame if the people involved were upfront and honest about what we don’t know –  there is lots in many areas of life.  Instead, largely-imaginary castles in the air are being built and marketed, used by people whose incentives aren’t necessarily that well-aligned with the longer-term public interest.

It is worth appreciating just how limited the data really are.  Consider that 1 in 200 year standard the Governor articulates. In principle, one might like 10000 years of data –  in fact we don’t have much more than 100 years of data, and for not much more than a handful of (advanced) countries.     And even of that supposed 100 years of data, a huge number of the observations of actual crises were really a single observation of one (interconnected) crisis, in 2008/09.   But even then, these are not physical processes (like storms or floods or earthquakes) we are dealing with, but human ones, and humans change and learn (for good and ill).   And there is rarely any serious consideration of the countries that didn’t experience crises, or of what contribution (if any) differential capital requirements made to those outcomes.

There is no agreement in the literature on whether the output effects of financial crises are temporary or permanent, let alone how large those effects actually are.  It makes a huge difference what you assume.   Perhaps it is fine for an academic modeller to drop in some median estimate, but that number will be almost meaningless if plucked from a relatively small number of studies, producing a very wide range of different results.

Linked to this (and crucially), it is very rare to see any papers in this area (and the Reserve Bank’s latest consultative document is no exception –  although its 2013 cost-benefit analysis supporting the now-current capital requirements did note the point) distinguishing between any costs that result from the misallocation of capital during the boom phase and any costs that results from systemic bank failures themselves.   Higher bank capital requirements can probably do something about the latter, but they can do nothing at all about the former –  in fact, there is an argument that unreasonably high capital requirements could induce some more reckless lending/borrowing, as banks attempt to maintain rates of return.   I’m not aware of any paper that has seriously attempted to estimate separately the two effects (if anyone is, please let me know).

Ireland over the last decade is a good example of the significance of this point, as are New Zealand (or the Nordics) in the late 80s and early 1990s.  There was plenty of reckless lending/borrowing, on over-inflated assumptions about future asset values, economic growth etc etc.  Much of the (building in particular) activity that followed simply involved wasted resources.  It wasn’t apparent during the boom –  it never is –  but the bust is partly about those effects crystallising.  If no bank had failed in the aftermath, all of those particular wealth losses would still have occurred, banks and potential borrowers would still have had to reconsider and review their business models, identifying better just what were good credits in thos particular economies.  I’m not suggesting that bank failures themselves had no adverse aggregate economic effects –  quite possibly they did –  but even if you could safely identify all the output losses relative to a pre-crisis trend, it would still no be remotely safe to ascribe all those to the banking failures.  And bank capital requirements will only affect the probability of banking failures.

One could throw in more points.  For example, the Reserve Bank has periodically tried to claim that there is good reason for New Zealand to be more cautious (in setting capital requirements) than other countries because of some specific New Zealand vulnerabilities.  But they simply never seriously spell-out the nature of those (asserted) greater vulnerabilities.  One might, for example, be more cautious if bank balance sheets were chock-full of complex instruments.  Our banks aren’t.  Or were very heavily exposed to new and highly uncertain industries. Our banks aren’t.  Or if your big banks were co-ops (with little ability to raise new capital if times get tough).  But our banks aren’t.  Or even if the government’s own finances were severely impaired.  But our government’s aren’t.  Or if your banks and country were part of a monetary area such that you had no independent monetary policy and no floating exchange rate.  That is the situation for much of the OECD, but it isn’t for New Zealand.   These are plain vanilla banks, mostly with parents that are among the better-rated banks in the world, operating in a country with a floating exchange rate and robust government finances.    But you won’t here those lines from the Reserve Bank –  well, you will when they proclaim, in every FSR, that the New Zealand financial system is sound and robust, but not when they assert (as here) that the system is far less sound than it should be and (expensive) core capital should be almost doubled.

And then there is question of the appropriate discount rate.  If we are worrying 1 in 200 year crisis we are worrying about events that are (probabilistically) a very long way in the future.   As even the Reserve Bank acknowledges (page 9, they don’t do their own modelling but they report that of other central banks), studies to date all use discount rates below those required by the New Zealand Treasury when agencies are evaluating potential investment projects and regulatory interventions.    Treasury’s latest recommended default discount rate is 6 per cent real.

Suppose we are worrying about preventing a shock 75 years hence (the Bank’s proposals envisage that we would still suffer than 1 in 200 year event, but would prevent, say, the 1 in 150 year event) that might cost 10 per cent of GDP then.  Discount that 10 per cent loss back at a 6 per cent real discount rate and the present value of what you are trying to prevent is tiny (about 0.5 per cent of GDP even if allow some reasonable productivity growth, such that 10 per cent of GDP 75 years hence is quite a bit more than 10 per cent of today’s GDP).  A 6 per cent discount rate is, itself, ludicrously low in this context: it is not like evaluating a known technology (recall all those highly uncertain points I discussed above).    There are reasons why private businesses typically use hurdle rates of return well above estimated firm weighted average cost of capital (essentially what the Treasury numbers are based on).

Apply a higher discount rate –  even just take it up to 10 per cent real (just 4 percentage points above the Treasury-estimated WACC) – and the estimated future GDP savings 75 years hence reduce to near-invisibility.   You do not then need many costs upfront (say in the proposed five to seven year transition period, almost inevitably spanning the next recession) for this proposed regulatory intervention to fail the simplest sort of cost-benefit assessment.

And this current proposal is the whim/preference of one Governor.  He will not be Governor is 75 years time.  Most likely he will not even be Governor 10 years from now.  So there is no pre-commitment mechanism.  We could end up paying all the transitional costs only to find that 10 years from now some new Governor, some new government, some new studies all end up concluding that –  actually- the sorts of risk-weighted capital ratios we have right now were really just fine after all.

On which note, my former colleague Ian Harrison (now of Tailrisk Economics), who built the model the Reserve Bank used to evaluate capital requirements in 2012/13 and who thus know whereof he speaks, has been beavering away on his own review and critique of the Reserve Bank’s consultative document, drawing on a detailed examination of the documents the Bank has published and those they cite.   He sent me a draft and suggested I might like to highlight a few of his points, a teaser for the full publication expected in the next couple of weeks.    There is a lengthy and serious assessment, done in considerable detail, and accompanying it is a set of Pinocchio awards, evaluating the Bank documents using an approach adapted from the Washington Post’s fact-checking methodology.   Thus

One Pinocchio:  Some shading of the facts. Selective telling of the truth. Some
omissions and exaggerations, but no outright falsehoods.  ……

Four Pinocchios: Whoppers

On Ian’s assessment, the Bank’s material scores, shall we say, poorly.

From the more substantive document

The costs of the policy receive little attention. It is admitted that the higher capital requirements could make it more expensive for New Zealanders to borrow, but the Bank claims that  the impact will be ‘minimal’ and that they have taken it into account.  However, even on the Bank’s own assessment of 8 basis points[1] for each percentage point increase in the capital ratio, the cost to New Zealand will not be minimal. It is likely to cost around $1.5 billion per year, and possibly more.

The present value of the cost of the policy could reasonably be assessed at $30 billion. ….  A homeowner with a $400,000 mortgage could be paying an additional $1000 or more a year.

[1] This is the number that appears in the decision document. A figure of 6 basis points appears in the consultation document but there is no explanation for the difference.

(Wider margins could affect depositors as much as borrowers, if there is some OCR offset, but either way the effects are not trivial.)

Or

The central question that is addressed in this paper is whether the benefits, (‘being more resilient to economic shocks’) are worth more than the $30 billion. Our assessment is that it is not.  New Zealand could secure nearly all of the benefits of higher capital by increasing tier two capital, as the Australians are proposing to do, at about one fifth of the cost of the Reserve Bank’s proposal.  The Reserve Bank has not seriously considered this option.

Or

The Bank’s assessment that the banking system is currently unsound [implicit in the proposal to require such huge increases in  capital] is at odds with rating  agency assessments and borders on the irresponsible. The rating agencies’ assessment of the four major banks is AA-, suggesting a failure rate of 1:1250.

Ian concludes that using credible inputs to established Basle models, the Governor’s 1 in 200 year target would be adequately met with lower minimum capital requirements than we have at present.

Perhaps all the answers will be in the speech from the Deputy Governor tomorrow. Whether they are or not, this far-reaching proposal needs much more robust analysis in support. Declarations from the oracle of the forest really are not enough.