Revisiting an incredibly expensive insurance policy

The Reserve Bank’s radical bank capital proposals –  markedly increasing required capital for locally-incorporated banks, in a country with (a) a low demonstrated risk of financial crisis and (b) high effective capital ratios by international standards anyway –  hasn’t been much in the news lately.  The Governor –  unelected, but sole decisionmaker on this –  his own –  proposal has presumably retreated to his high tower to contemplate.   He hired some carefully selected overseas academics to review bits of the Bank’s analysis, and we might expect to see their reports shortly (but recall the tight constraints on what they were allowed to look at, who they were allowed to talk to etc).

I was doing an interview yesterday on various aspects of the proposal, including making the point that what the Governor is proposing can be seen as –  on the Bank’s own numbers – an incredibly expensive insurance policy, paid not by the Governor and his colleagues of course, but foisted on the people of New Zealand.      But I used a number in the course of the interview and when I got home I realised it didn’t sound quite right, so thought I should review the estimates.   I’d written a post on this some months ago.

There are various estimates around as to how much difference the proposed new capital requirements might make to real GDP.  I gather the ANZ has suggested anything up to a steady-state levels loss of 1 per cent (ie each and every year GDP is lower than otherwise by that amount).    Channels could include higher costs of credit and possible reductions in the availability of credit.

I wouldn’t completely rule out such numbers, but for my purposes I’m content to use the Reserve Bank’s own estimates.  In a speech in February the Deputy Governor told us the Bank thought the cost could be “up to 0.3 per cent per annum”.   So lets use 0.25 per cent  (if they really thought the effect would be no higher than 0.2 per cent, they’d have said “up to 0.2 per cent per annum”).

That might sound like quite a small number.  In fact is something like $750 million this year, and that cost is repeated each year.  And since the real economy is growing over time, the dollar value of the cost –  the annual insurance premium  –   only increases with time.

We can discount back to the present the value of all those annual insurance premia.  The answer, of course, depends in part on other assumptions.   Over the longer-term, real GDP growth will be largely a reflection of population growth and productivity growth.  In a New Zealand context, 1 per cent per annum each might seem a reasonable central estimate.   And then there is the discount rate: current Treasury guidance suggests using a real discount rate of 6 per cent for regulatory proposals, but of course long-term rates have fallen quite a long way since that guidance was issued, so I’ve also shown the numbers for a 5 per cent per annum real discount rate.

I’ve done the calculations for 150 years (recall that the Bank’s proposal is about having resilience to cope with a 1 in 200 year shock), but of course the bulk of the present value costs are borne in the first few decades.

Present value cost over 150 years ($bn)
Real GDP growth
1.5 2.0 2.5
Real discount rate 5% 21.60 25.20 29.90
6% 16.90 19.10 21.80

Recall that the output cost was the Bank’s own number, the discount rates are based off Treasury numbers, and that these are probably low-end estimates since (a) they take no account of transitional disruptions, which are front-loaded, and (b) being GDP focused they take no account of the additional income transferred to foreign shareholders in domestic banks (a very substantial sum on the estimates done by my former colleague Ian Harrison).

And what are we getting for our insurance premia?

You might recall that, waving his finger in the air, the Governor decreed that his proposals were conceived with the goal of ensuring that New Zealand didn’t have a financial crisis (probably, a succession of bank failures) more than once in 200 years.

So lets assume, just for the sake of argument, that the Bank’s proposals are implemented and they are sufficient to prevent one serious financial crisis that would otherwise have occurred every 150 years.  (Existing, quite high, capital ratios are assumed to prevent other smaller, more frequent crises).

We don’t know when in the 150 years that crisis might occur, so lets just assume it happens 75 years from now.

But then the key variable is what scale of output losses is saved.    The Bank likes to assumes very large losses, sometimes permanent ones (ie a financial crisis now itself reduces the level of GDP even 100 years hence).     I’ve argued that much of the analysis and discussion in this area is wrong and (when undertaken by regulatory agencies) self-serving.      Many of the output losses associated with (word chosen carefully) financial crises do not arise from the crisis itself (bank failure etc) but from the poor quality lending and investment decisions that happened in the years prior to the crisis, and which –  most likely –  would have happened anyway, regardless of the level of capital.   The segment of any apparent output losses than might be saved by higher capital requirements is some fraction –  probably a fairly small fraction –  of the total economic underperformance in the wake of the crisis.   As I’ve noted before, in my own analysis and that undertaken by William Cline at the Peterson Institute, actual differences in output growth between crisis and non-crisis countries (eg post 2007) are much smaller than the numbers the Reserve Bank and foreign regulators like to wave around.    And as I noted in my own submission to the Reserve Bank

Note also that the Cline methodology still overstates the amount that higher capital ratios alone might save, since his output path comparisons include (for the crisis countries) both kinds of losses – from the initial misallocation of resources, and the pure crises effects.   Only the latter should be relevant in assessing the costs and benefits of higher minimum capital ratios.

I reckon the most one might allow for a saving might be 10 per cent of GDP (eg annual GDP 2 per cent lower than otherwise for five years, purely as a result of the crisis, and despite best stabilisation estimates of macro policy).   But, for illustrative purposes, lets say the number is 20 per cent (something like the Cline estimates) or even –  heroically 30 per cent.    Remember that the crisis is happening 75 years from now, so that even though GDP then will be much bigger than it is now, we need to discount that saving back to today’s dollars, to compare against the present value of the insurance premium.

In this table I’m assuming real (potential) per capita GDP grows by 2 per cent per annum (the middle scenario above)

Present value of GDP benefits ($bn) of averting a crisis 75 years hence
Size of saving (% of GDP)
10 20 30
Real discount rate 5% 3.2 7.2 9.5
6% 1.5 3.5 4.5

My view of the likely savings would be represented by the first column: it doesn’t really matter which discount rate one uses and the present value of the benefits is derisory relative to the present value of the annual insurance premia  (in fact the benefit/cost ratio would look almost as bad as for –  say –  light rail in Wellington).   But even if you use the Cline numbers, even if you go beyond that and assume really huge savings 75 years hence from this one regulatory intervention, in no scenario do the benefits (this table) come even close to matching the cost (first table).

And all this on the Bank’s own view of the annual insurance premium.

To all of which one could add the reminder that while the Governor talks today of implementing policies which made generate a real saving decades hence, he has no commitment mechanism.  His own term is only five years.  He’ll probably get reappointed for another term, but even then the government is consulting on proposals that would mean he would no longer be the sole decisionmaker.  Regulatory tastes, fashions, and judgements change and there is almost no chance that a regime inaugurated today would last 50 or 100 years (long enough to generate real benefits, since no one thinks the New Zealand banking system is at serious risk of crisis right now).   If the regime only lasts 10 years, until some other decisionmakers change tack, we’ll have paid a present value of perhaps $6 billion (plus all the transitional costs and disruption) for no benefits at all.  Another way of looking at it is to put a 50 per cent chance on any regime being persisted with for multiple decades: the expected present value of benefits halve, but the present value of the annual costs is frontloaded.  The equation looks even worse for the Bank’s case.

Not all risks are worth insuring against (try introspection on the risks to your own life or finances, or that of any business).  On the Bank’s own estimates of the premium costs, the policy they are offering (well, planning to impose) simply isn’t worth it.

Of course, if we had a decent policy process in the first place, we’d have had cost-benefit analyses –  perhaps various approaches –  laid out by the Bank months and months ago.   Don’t get me wrong.  No cost-benefit analysis is ever perfect, or the definitive “right” answer, but laying out the assumptions and the sensitivites helps enable more critical scrutiny of what the regulator is proposing, and (if done well) may even enhance confidence in what they are proposing.  The Governor’s approach remains one of not showing us any serious cost-benefit analysis –  and not engaging with the perspectives others offer –  until he has made his final decision.  At that point, any cost-benefit analysis –  done by his own staff to support his own decision –  serves only decorative (or PR) purposes, not functional ones.

It isn’t good enough.

UPDATE: A commenter points out, correctly, that non-linearities mean that the simplification of just focusing on a shock halfway through the 150 year horizon isn’t really valid (skews the results against the proposal).    I extended the analysis, and checked that the conclusion didn’t change, in a separate post here.


16 thoughts on “Revisiting an incredibly expensive insurance policy

    • It is not clear to me if the increased Capital equates to increased capital buffers ie more cash held or just the mere recording of increased capital and the additional cash buffers readily lent out.

      The banks lends out in order to make a margin on either the capital or the savers deposits or borrows from the bond market. In order to reduce the dependence on savers deposits and on the bond market, the RBNZ requires additional capital. End result is that the lending business is unaffected just the source of the funds has changed to from Savers deposits to Shareholders capital.

      Given that the big Australian banks dominate savers deposits then increased Shareholders capital allows for more competition as the smaller banks are wholly reliant of its shareholders capital in order to sustain its lending business. This is not a bad result if we achieve lower interest rates as a result of competition. Although the banks have indicated increased costs and increased interest rates, the current trend appears to be higher competition perhaps in anticipation that the smaller banks will get more competitive with a higher capital requirement regime.

      The assumption by the RBNZ is that the increased Capital creates a cost. But if there is increased competition it is rather difficult for the dominant Australian banks to pass the cost onto New Zealanders due to competition. But someone has to pay, there is no free lunch.

      I am leaning more towards a lowering of Shareholders dividend payments or a move towards cutting business costs ie freezing salaries, shutting down branches, renegotiating office rents etc or increased IT investment to reduce staffing costs.

      I would suggest taking the slow careful approach so that banks can adjust. Rapid changes can only lead to a disaster. It has been rather frightening when Adrian Orr took that stand that he was a God Mahutu or something and that attitude can frighten off bank savers and a run on savers deposits is death to a bank.


  1. If I did this analysis I would use slightly different assumptions. First, I would place much greater weight on the GDP lost in a 10% recession than the additional 1% in marginal conditions. This is because I think deep recessions cause great personal suffering, trauma and family damage. Second, I would assume less of a growth penalty for higher capital costs, because I happen to agree with others that “financialisation” itself has a growth penalty over time. I would also use a lower social discount rate for the timing difference.
    My own guess is that the capital increases will have minimal negative impact even in the short term, and can always be tweaked if they cause some unexpected problem.


    • Totally agree that recessions can be and often are dreadful things for many, altho the worst of recessions usually arrive from other causes (including the initial misallocation of lending/investment).

      But, yes, good to lay out the differing assumptions and see where they each seem robust (or not).

      On your final point, I guess the counterargument is that once in place it will be v hard to tell just how large any adverse effect has been, and both good and bad policies tend to linger for quite some time.


      • Lets get it right. The last recession was not caused by the banks or by the GFC. The last recession was engineered by the RBNZ driving interest rates to a level unsustainable to businesses and individuals and as a result created a domino effect collapse of the economy.


  2. Is what you raise the question or is the better question ‘what is the optimal capital reserve over time for each bank’. From your own argument shouldn’t the ratio be dependent on the quality of the individual bank book.


    • Mine, at least in this post, is just one way of framing questions around what the RB is proposing. For each individual bank the “optimal” capital ratio will certainly depend on its loan book etc – and that is so whether or not there is any banknig regulation at all. Of course, the risk-weighted assets framework tries to take account of those risks, and the correlations among them.

      In many overseas regime, the regulator will specify a minimum capital ratio for everyone, and then have overlays for each bank based on the regulator’s assessment of that bank’s individual circumstances.


  3. what about the benefits the financial system may derive due to more people having greater faith in the stability and robustness of that system because they know their banks are well capitalised?


    • It is one possible story. The Bank itself hasn’t invoked it in the estimates it has presented so far. My sense is that most of the time people don’t think of the financial system as being unstable – if anything, regulators often worry that the public and markets assume the system is more stable than it really is (an argument that might have some resonance in the US story pre 2007).


  4. if all this extra capital is built up, what assumptions do the economist make in what the bank do with this capital. Obviously it wont stay in their bank accounts.

    So what are the options banks can use these funds for and keep as Tier 1?


    • David

      Since it is the regulatory minima the RB is proposing to raise, the main effect (as intended) will be simply to replace debt funding with equity funding. The ongoing debate is about the extent, if any, to which the Aus banks will look to reduce their overal exposure to NZ (the returns here will no longer be so attractive, and shareholders may push them to pursue higher yiellding opportunities elsewhere). For big firms, the effect will be small – the Aus banks will service them thru the Aus parents, who arern’t directly affected by these proposals. For NZ farms and SMEs any reduction in risk appetite could be quite costly.


  5. Hi Michael

    A good read, but I’ve think you’ve oversimplified the maths.

    Because the discount factor is non-linear in time, the 75 year ‘midpoint’ approach isn’t right ( i.e E[f(t)] ≠ f[E(t)] ). It’ll make a big difference to the numbers, although I’m not sure whether or not it fully tips the scales in your example.

    You’d basically need to calculate the discounted avoided crisis cost for each possible year the crisis could have ben in (say, over the next 150 years) and then do a probability-weighted sum of the discounted values. The simplest approach would be to assume equal probability of crisis in any of the next 150 years (which is consistent with your midpoint date of 75 years). Or you could use a more complex assumption.

    I think it is important to take a look at this, so that you know how much of a difference the shortcut you’ve taken makes. As I mentioned, I suspect it will be big.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s