Wednesday, March 17, 2010

Economics 17/03/2010: Nama Estimation Procedures

Yesterday I finally got a few minutes to read through the latest Government documentation on Nama - "Determination of Long-Term Economic Value of Property and Bank Assets" regulations SINo88 of 2010 from March 5 (link here). This delightfully thin document is a treat for anyone who cares to study just how inept our authorities can be when it comes to measuring and assessing/pricing risk.

Timing

Paragraph 2 page 2 defined 'relevant period' - used in assessing long-term economic value - as 'the period that began on 1 January 1985 and ended on 31 December 2005'. Per para 5.ii page 3, this period will be used to estimate reference prices for land based on land hedonic (econometric) links to demographic variables (5.b.i), interest rates (5.b.ii) and GDP (5.b.iii).

The problem here is that there is no clear identification as to which time horizon (the full 21 years worth or some sub-set thereof) the Government will use. And this is crucial, as this period largely covers steadily rising market with almost no corrections. Which means that should Nama use dynamic trend for estimating the land prices, it would be rather accurate within the sample, but will be absolutely ad hoc outside the sample (per Lucas critique).

Of course, Nama won't take dynamic pricing - and as is clear from 5.(1).(a).(i) and (ii) it is going to take 'prices' and 'yields' - i.e. point estimates. Which most likely means some sort of an average. It is important, therefore, to have an exact idea as to within what sub-period of 1985-2005 is this average going to be taken. The note does not identify this exactly, leaving the door open for Nama to deal with the data as it sees appropriate.

Discount factors

There is an added complication to the entire valuations scheme. SSNo88 states in 5.(1).(a) that a discount (adjustment) factor for land located within the State will be based on a difference in price of land on valuation date (nearly absolutely unknown to Nama) and the Long Term (economic) Value (LTEV) price (also unknown to Nama). And this means that Nama will start its valuations process by applying an unknown discount factor to a risky asset it is buying. No data listed within SSNo88 as the basis for valuations allows Nama to escape this grim reality.

Discount Rates

There are other fundamental errors built into valuations process. For example, NAMA discount rates - equivalent to interest rates - on bank assets are (as listed in 2.(2).(a)-(b) on page 2: 4.54% for 3-year rate, 5.57% for 5-year rate, and 6.16% for 8-year rate. This presents a little bit of a conundrum. Firstly, are these rates annualized compounded or simple? Second, and even more important - why are these rates chosen for these maturities? What yield curve has been used to impute these? No clarity on this whatsoever.

ECB gives only two types of average retail rates for non-financial corporations loans: as of January 2010 (the latest), we have existent loans with rate fix up to 1 year are priced at 2.96% which, factoring in Government's 1.7% risk premium margin implies a rate of 4.66%, not 4.54%. Off the starting line, there is a built in subsidy from Nama.

And this subsidy is greater than 12bps spread between the above two rates. How I know this? Well, think - can assets we are buying into Nama attract any loans in the private sector? At any valuations? No. So what justifies that 1.7% risk premium the Government applies (per 2.(2).(a)-(c)) to all loans it will be buying? And why is the risk premium independent of maturity period? Surely it should be rising with maturity?

Let's take a look at what we have from the ECB: non-financial corporate overdraft rate as of January 2010 averaged in Ireland at 5.74%. Suppose this was the basis for our Nama valuations (I still think this is too low of a rate given the assets quality and given the fact that ECB reports 'offer' rates by the banks, not the actual rates on which loans were issued, but what the hell, let's use this as a base assumption).

With Government risk premium, this should imply a 3-year discount rate of 7.44% (still shy of corporate junk bonds, but much better than 4.54% built into Nama). This rate (per 7.(a)) will cover land-backed loans where land value has fallen just 10% below its LTEV - the high quality loans in Nama books. Now, using SSNo88-implied yield curve, 5-year discount rate consistent with our assumed base rate should be 2.08% and for 8-year rate, the risk margin should be 1.9%.

This means that if Nama were to use the riskiest loans rates we have - those for overdrafts by non-financial corporate sector, and use upward sloping risk margins (to reflect the fact that the longer the duration of the rate, the lower is the quality of assets to which it applies - per SSNo88 page 5 own admission), the rates of interest imputed on Nama assets should be:
  • 3-year money - at 7.44%, not Nama-assumed 4.54% (a loss to the taxpayer or a subsidy to developers and banks of 2.9% per annum);
  • 5-year money - at 9.13%, not Nama-assumed 5.57% (a loss to the taxpayer or a subsidy to developers and banks of 3.56% per annum); and
  • 8-year money - at 9.67%, not Nama-assumed 6.16% (a loss to the taxpayer or a subsidy to developers and banks of 3.51% per annum).
Let me give you a look at the Nama interest rate pricing generosity from another angle. Irish Exchequer is currently borrowing at 5% interest rate in the markets. This, presumably does not include Mr Lenihan's 1.7% risk margin. Which means that if Nama was buying loans from borrowers who are as credit-worthy as the Irish Exchequer, it should be applying a rate of 6.7% to these loans. My estimate of 7.44% is much closer to that than SSNo88 statutory fixed rate of 4.54%.

Proposition: under Nama, effective interest rate subsidies for defaulting loans (accruing to banks and developers) will range between 2.9% and 3.56% per annum.

Proof: see two alternative arguments above.

Estimation process

Another issue, related to interest rates arises when the SSNo88 outlines 3 sets of variables Nama will use to impute LTEV based on 'correlation' (whatever this means in econometric terms, I have no idea): 'between land prices and interest rates in the State' (5.(1).(b).(ii)). Can someone explain to me which interest rates Nama has in mind? Central Bank rates? Retail rates? Retail rates for non-financial corporations loans?

The same applies to paragraph (5.(1).(c)).

Valuations timing

The document SSNo88 was published in March 2010. Nama will not begin valuations before April and will not finish these before the end of 2010 (optimistic projections). So why are all valuations being made by Nama will take into account only market values of land prior to January 10, 2010? Surely, falling markets mean - and consensus forecast expects - at least 5%+ decline in house prices (what can one say about land!) over 2010. Is Nama going to ignore this reality and price the assets in buys in, say, November 2010 at prices valuations for January 2010?

Let me explain. In a year when I bought my house, within 11 months of my purchase, home prices fell roughly speaking 3%. Do I get to go to my bank and tell them - 'Folks, shave off 3% from my total original mortgage as Nama is doing for you?'

5% is a non-trivial number, adding up, over the loan book Nama id about to take up roughly €3.9 billion overvaluation, spread over 15 years with interest and market discounts accruing to it.

Use of GDP

SSNo88 applies to valuations of land and real estate assets. These are non-exportable. Why is then the Government planning to use the rates of growth in GDP, not GNP? Does our booming (1985-2005) pharma sector has anything to do with the fundamentals on which land prices are set? Using GDP for the estimation process instead of GNP introduces a distortion of between 12% and 18% depending on the range of years used. This distortion risks further overvaluing the LTEV for land.

Building in dreamy planning

Paragraph 5.1.d.ii states that Nama will use "existing and future transport planning and the associated supply and demand projections for land use" in its valuations of LTEV. What does this mean? Will Nama use full extent of NDP plans? Including the frozen and indefinitely postponed plans? There is significant divergence between what is planned and what is delivered. And in the next 10 or so years, this divergence will be in the direction of planned transport investment being well in excess of real investment. If Nama were to use the former as a basis for estimates, then there will be land with LTEV in excess of realizable value because its estimated LTEV will be based on excessively optimistic plans for transport investment.

At any rate - the current phrasing of the SSNo88 does not provide for a rigorous definition of what planned infrastructure will Nama actually factor in. This makes the LTEV estimation process outlined in LTEV completely undefinable under the current legislation.

Land based outside the State

The entire section 6 outlining the LTEV estimation process for land outside the State is simply a carbon copy of the valuation processes for LTEV within the State. This suggests that - given that assets being valued are different in nature (legal and economic), risk and geographies - the Minister has no idea how these different risks should be reflected in the estimation process.

The section does not even mention exchange rate risks or derivatives on Forex exposure attached to loans. There is no clear understanding as to what interest rates should apply and how to deal with the carry trades.

Operational costs

Section 8 of the note states that the assumed 'due diligence costs, incurred or likely to be incurred by Nama over its lifetime' is 0.25%. This is simply unrealistic. Industry average operating cost on performing loans - across all subheads - is equal to about 0.75-1% in times when operating conditions are deemed to be normal. As of the end of February 2010, Nama, reportedly already has blown through its entire 2010 legal budget. How can 25bps cover its due diligence cost over life time?


In short, SSNo88 is yet another document that shows just how exposed Nama remains to shoddy planning and poor estimation procedures, courtesy of the DofF that simply cannot deliver realistic and transparent operational guidelines.

3 comments:

Anonymous said...

What is your conclusion regarding the reasoning for designating the relevant period 85 - 2005 (which omits the 2007-onwards crisis and also the severe Irish recession of the early 1980s (not to mention the anemic economic performance in the 1970s)?

Why do you think data and analyses produced after 10 Jan 2010 may not be used in calculating LEV?

Why do you think GDP was chosen over GNP with forecasting using empiricals given that GDP has been described as vanity and GNP is sanity?

I'm not expecting a political answer but if you have an economics insight which might support the above rules, that would be very helpful.

Terrific article by the way!

patrick1978 said...

Constantin,
Can you see a surge in bank shares this week ahead of the announcement of "big bang" recapitalisation of AIB and BOI.

TrueEconomics said...

Patrick, I really can not understand what valuations can drive people to go long on AIB&BofI at these prices. The two will need anywhere between 7bn and 10bn in capital post Nama. The two will not have any viable business model left after they are recapitalized. The total state capital holdings in both banks will be well over 50%. And their market cap will have to be x7 current levels to support this equity issuance. Can anyone imagine the two banks returning to profitability levels of 2005 that this type of equity issuance would require? Remember, back in 2005 there was a prospect of Irish economy growing at 5-6% pa into the medium term future. Anyone expects this to happen now?

So let's track back. If absent toxic loans BofI fundamental value of around 4 per share. This means that any investor buying today assumes that at the very least BofI will only have to raise equity of ca 2.5bn (capital)/1.5 (dilution factor) in the next 2-3 years. But this is simply mad! BofI will need way more than that by a factor of at least 2 times! So current price is at least 50 percent overvalued already. Unless the markets expect the Exchequer to provide BofI capital free of charge - with no equity issuance. That would not have been possible even were BofI capable of paying preference shares dividend, since in that case they would end up with huge overhang of preference shares over equity, breaching capital requirements standards.

I just cannot get my head around the latest rally...