So I was told today, by a senior banker, that banks have been actively re-writing non-performing loans (since at least April this year) under new contracts with extended principal and interest holidays in covenants. These, in preparation for Nama, are priced at higher rates so they can get more on the loans once Nama discount applies.
This makes sense.
Do the math - assume:
- 20% cross-collatera
lized Euro100mln loan (see explanation of this below), written in 2006 - 3 years rolled up interest at 19.1% accumulated at 6% pa - which gives us loan face value at placement on the bank watch list of Euro119mln
- New covenants set in April 2009 at 9%pa, with no interest yield or principal repayment required for the next 3 years.
- On the date of Nama initiation, then, the loan is performing with expected yield of 9% on Euro119mln.
- Now, suppose the LTV ratio of the loan is 75% of principal (meaning the value of the underlying collateral was 133mln in 2006)
- Assume that collateral value has fallen 30% (an under-estimate to be palatable to all optimists out there), which means that with 20% cross-collateralizati
on writedown, plus 2% inflation annually since 2006 (cumulative inflation discount of 6.1%) collateral now is valued at 63mln, - By the time new covenants on the loan kick in in 2011, the rolled up interest on the loan and principal will mean total loan value will be roughly Euro 154mln.
- Now, to break even on this loan Nama will have to pay 1.5% interest charge on bonds, plus 0.5% management cost (including bank fees), implying that 3 year average mark-to-market writedown (at 2% pa or 6.1% cumulative) plus inflation at 1% pa on average (3% cumulative) is (1-63mln/154mln*0.91)*(100%)=62.7% (assuming no growth in the property market between now and 3 years from now).
Remember that figure in the Irish Times article signed by 46 economists, including myself? It stated that the real value of 90bn worth of distressed loans is around 30bn. That implied a mark-to-market writedown of just 67%! When published it caused Garret Fitz to go ballistic and the entire pro-Nama crowd to shout "Extremists are at the gates!" Not that far off from 62.7%.
Of course, this is an illustrative example. But notice that it assumed very modest decline in underlying assets value (30%) to date, plus a very generous (75%) LTV ratio. House prices alone are already down by more than 30% from the peak.
Challenge the rest of my assumptions?
Whether you do or not, one thing is clear - if you are a bank you had no incentive to manage your stressed loans since the very least this April. And you had a massive incentive to push up the face value of the loan without forcing it to become non-performing. The latter can be done by re-writing the loan with new roll up covenants.
Cross-collateralization:
Banks gave multiple loans on same properties in several forms -
Suppose a site was bought for 100mln at 75LTV, so that the developer borrowed 75mln for it. New zoning applied lifting the site value to 200mln, providing another 75mln loan facility at 75%LTV on 200-100mln capital gains. The building permission was then granted that, say lifted the site value to 300mln, and a new loan was taken out at 75LTV. Total value of the site was 300mln. Suppose each step in borrowing and capital gains took 1 year (a very short period of time), suppose interest rate was 5%. This means that:
Loan 1 now totaled Euro83mln
Loan 2 now totaled Euro78.8mln
Loan 3 now totals Euro75mln.
At loan 3 origination, LTV ratio on the entire site was 236.8/300=79%.
I assumed in my calculations on the blog that 20% of loans are written against sites that are cross-collateralized - so that other banks hold claims against the same site.
This assumption is based on a guess. It can be challenged if someone has any evidence on better numbers.
Now, that means in example above that some 20% of the site value was cross-collateralized with another bank. If it was the first loan that was cross-collateralized, LTV rises to (236.8+20% of 75)/300=252/300 or 84%. If all three loans were cross-collateralized at 20%, the resulting LTV is (236.8*1.2)/300=284/300 or 95%.
So here you have the maths on Nama - 75LTVs on each loan in reality can mask a 95% LTV of total loan package.
Of course, this is an illustrative example. But notice that it assumed very modest decline in underlying assets value (30%) to date, plus a very generous (75%) LTV ratio. House prices alone are already down by more than 30% from the peak.
Challenge the rest of my assumptions?
Whether you do or not, one thing is clear - if you are a bank you had no incentive to manage your stressed loans since the very least this April. And you had a massive incentive to push up the face value of the loan without forcing it to become non-performing. The latter can be done by re-writing the loan with new roll up covenants.
Cross-collateralization:
Banks gave multiple loans on same properties in several forms -
- most commonly, a property was valued several times consecutively and whatever capital gains accrued on the property, these gains were re-mortgaged under new loans;
- also commonly, capital gains were priced out of new building permits being extended to the properties. I am aware of several cases of mega deals (hundreds of millions borrowed) where a developer/investor bought a site with the site itself being collateralized for this first round of borrowing at the market value, then rezoned it, taking out a new loan against the site value after rezoning in excess of original loan, then obtained a planning application and re-collateralized the site again;
- less commonly, the banks simply did not check if a collateral property has already been pledged elsewhere.
Suppose a site was bought for 100mln at 75LTV, so that the developer borrowed 75mln for it. New zoning applied lifting the site value to 200mln, providing another 75mln loan facility at 75%LTV on 200-100mln capital gains. The building permission was then granted that, say lifted the site value to 300mln, and a new loan was taken out at 75LTV. Total value of the site was 300mln. Suppose each step in borrowing and capital gains took 1 year (a very short period of time), suppose interest rate was 5%. This means that:
Loan 1 now totaled Euro83mln
Loan 2 now totaled Euro78.8mln
Loan 3 now totals Euro75mln.
At loan 3 origination, LTV ratio on the entire site was 236.8/300=79%.
I assumed in my calculations on the blog that 20% of loans are written against sites that are cross-collateralized - so that other banks hold claims against the same site.
This assumption is based on a guess. It can be challenged if someone has any evidence on better numbers.
Now, that means in example above that some 20% of the site value was cross-collateralized with another bank. If it was the first loan that was cross-collateralized, LTV rises to (236.8+20% of 75)/300=252/300 or 84%. If all three loans were cross-collateralized at 20%, the resulting LTV is (236.8*1.2)/300=284/300 or 95%.
So here you have the maths on Nama - 75LTVs on each loan in reality can mask a 95% LTV of total loan package.
How does the cross collateralisation work? It all seems a bit sneaky (and therefore likely to be going on !) alright. Not sure how the bank benefits though. The loan evaluation questionnaire will establish that the loan has been non-performing for some years and has now moved to a moratorium. Not sure that it would be treated as performing on that basis? Would it? To an extent . . so what if it is? The valuation process isn't (apparently) as simply as "we pay more for performing loans". It is supposed to be done loan by loan property by property, valuation by valuation. Using the high level figures we have and applying them here if the loan was 100M in 2006 on a 75% LTV does that not imply a 133.33mn asset originally? So (to follow through on the NAMA numbers) if that asset has fallen in value by 47% it implies an underlying asset value of c 72mn. So it all depends on what they pay for the (now) 118mn loan. As it is completely non-performing to date with no interest or capital payments I assume they would hit it with a fairly hefty haircut. Presumably 35% ish? Too much? Maybe. Anyway something along those lines - that means they'd pay the bank something around 77mn for the loan. That 77mn would by them an "obligation" on 118mn plus an asset worth 72mn. I'd agree with you that the "obligation" is probably worth feck all.
ReplyDeleteAm I missing something? Can the banks really get paid more simply by adding rollup covenants to existing non performing loans? Seems unusual?
Adrem - you are right on numbers - see revised.
ReplyDeleteI am of course assuming that the work-out process for Nama loans pricing will be not exact. This is an assumption. Now, if it is and Nama disregards rolled up interest, guess what - somehow this has to be covered, for the entire amount of the current loan (rolled up interest inclusive and charges and costs rolled up inclusive, plus whatever the bits that are cross-collateralized - in short all there is on the books today, not on the date of loan origination) require cover by capital base. If Nama were to repair capital bases of the banks, it will have to take these amounts on-board and give banks something in return.
It is, really, that simple.
Cross-collateralization: banks gave multiple loans on same properties in several forms -
1) most commonly, a property was valued several times consecutively and whatever capital gains acrued on the property, these gains were re-mortgaged under new loans;
2) also commonly, capital gains were priced out of new building permits being extended to the properties. I am aware of several cases of mega deals (hundereds of millions borrowed) where a developer/investor bought a site with the site itself being collateralized for this first round of borrowing at the market value, then rezoned it, taking out a new loan against the site value after rezoning in excess of original loan, then obtained a planning application and re-collateralized the site again;
3) less commonly, the banks simply did not check if a colleteral property has already been pledged elsewhere.
What happens here then?
Suppose a site was bought for 100mln at 75LTV, so that the developer borrowed 75mln for it. New zoning applied lifting the site value to 200mln, providing another 75mln loan facility at 75%LTV on 200-100mln capital gains. The building permission was then granted that, say lifted the site value to 300mln, and a new loan was taken out at 75LTV. Total value of the site was 300mln. Suppose each step in borrowing and capital gains took 1 year (a very short period of time), suppose interest rate was 5%. This means that
Loan 1 now totaled Euro83mln
Loan 2 now totaled Euro78.8mln
Loan 3 now totals Euro75mln.
At loan 3 origination, LTV ratio on the entire site was 79%.
I assumed in my calculations on the blog that 20% of loans are written against sites that are cross-collateralized - so that other banks hold claims against the same site.
This assumption is based on a guess. It can be challenged if someone has any evidence on better numbers.
Now, that means in example above that some 20% of the site value was cross-collateralized with another bank. If it was the first loan that was cross-collateralized, LTV rises to 257.2/300 or 85.7%. If all three loans were cross-collateralized at 20%, the resulting LTV is 284.2/300 or 94.7%.
So here you have the maths on Nama - 75LTVs on each loan in reality mask 95% LTV of total loan package.
I wonder could you rig it so LTV is greater than 1 simply because the 1st tranche of the loan was much larger. Now there'd be a money maker - for someone.
ReplyDelete