image missing on blob storage!!!

After 4%: what should replace it?

01 March 2019
Richard Kemmish

One of the features holding back the Singaporean covered bond market’s development is a prudent one – a limit on covered bonds at 4% of an issuer’s balance sheet. Ahead of the Asian covered bond forum, a couple of pieces on this limit. Secondly, if 4% is to be dropped, what next?

In my previous article I discussed why Singapore felt an urge to impose a cap of 4% on covered bond issuance and decided that most of the possible reasons depended on the institution in question rather than the market as a whole. But whether the 4% limit should be relaxed for all Singaporean banks or just for a select few, the next question is, what should replace it?

The easiest (lamest?) replacement for a too strict and arbitrary cap is a less strict but still arbitrary cap. Let’s say 4% is replaced by 8%. There has to be a better alternative.

One possibility, which follows on from my previous article, is that issuance caps should be issuer specific and should be a function of the credit worthiness of the bank in question and their current levels of encumbrance. If a bank has a tier 1 capital of at least X% and can demonstrate unencumbered assets that are eligible for repo of at least Y% of it’s balance sheet then it can issue covered bonds equal to Z% of it’s balance sheet.

Couple of problems with this. Firstly, the limits all feel a bit arbitrary until someone (smarter and less lazy than I) creates a model to calibrate X, Y and Z. If such a model can be devised, is it and therefore the calibrations of X, Y and Z published? Or are banks notified of their issue limits privately?

Secondly, what happens when a bank falls from one category to another? If an ability to issue covered bonds is inversely related to creditworthiness there is a risk that such a rule becomes pro-cyclical. No regulator wants to introduce a rule like that.

Another alternative, that achieves a similar effect, is that issuance should be unlimited, but that issuance above a given level should be reflected by higher pillar 2 capital requirements. We avoid the procyclicality in the previous case but create an incentive for banks with more covered bond funding to become safer banks. Then it’s a commercial decision for each bank what their optimum level of covered bond funding / incremental capital charge is. This reflects their cost of capital and the saving from covered bond issuance. To my knowledge this hasn’t been tried anywhere but it feels like a rational answer.  

An alternative to changing the level of the cap is changing the way that it is calculated. Outstanding covered bonds as a percentage of total balance sheet is a simple measure. But it is also a measure that ignores the relative level of over-collateralisation in the programme which – arguably – should be the real concern for other creditors. If bank A borrows 100 using 120 of mortgages and Bank B borrows the same but with 110 of mortgages, the current method of calculation would treat them in the same way. But Bank A’s creditors are clearly in a far worse position in insolvency.

Could total issuance capped at 4% of balance sheet be replaced by total over-encumbrance of 1% of balance sheet? Again, hasn’t been tried bit feels like it should be.

I have no answers. But I’m interested to hear the thoughts of market participants in Singapore at the Euromoney Asian Covered Bond Forum on 12th March.

Event, Articles and Videos that might interest you

America should not have a covered bond law

10 June 2019 |

At the recent Euromoney North American covered bond conference we held an Oxford Union style debate on the motion that “This house believes that America should have a covered bond law”. To many people’s surprise I opposed the motion. Here’s why.

Tags: Covered Bond

4%. Again?

17 May 2019 | Richard Kemmish

Ahead of the Asian covered bond conference earlier this year I wrote a couple of pieces on Singapore’s 4% cap on covered bond issuance. Do the same factors apply to Canada’s 4% limit?

Tags: Covered Bond