The IFoA working party on equity release mortgages chaired by our friend Craig Turnbull has just issued an interesting ‘discussion note’ about equity release valuation. You might have thought that the WP might have had something to say about some of the rude things we have said about the subject, or about the IFoA or even about the WP itself, but no.
At one level, the WP’s non response is admirable. From a scientific perspective, however, it isn’t helpful to ignore research that gives conclusions they might not like. Better to confront and rebut, otherwise people might be tempted to draw their own conclusions.
There is also nothing about the ostrich elephant in the room, which is whether the industry are getting it wrong, like getting NNEG valuation an order of magnitude wrong.
Instead, the WP’s note focuses mostly on an issue of no practical relevance to the equity release industry, i.e., the theoretical possibility of a negative deferment rate.
This said, the WP does raise important issues on which further work needs to be done, such as dilapidation and volatility, but does so without benefit of the considerable research that we done on these issues. It’s not as if we went out of our way to hide our lamp under a bushel that they missed.
We shouldn’t moan too much, however. They did get it right (see para 14) on Professor Tunaru’s net rental calculation. Deleting footnotes:
‘Contrary to the tentative suggestion in Tunaru (2019), the loss of potential rental income that arises due to owner-occupation should not be deducted from the estimate of rental income for the purposes of assessment of the deferment rate for use in the NNEG valuation. Owner-occupiers choose to occupy their property because they are willing to bear the opportunity cost of the income foregone by owner-occupation. This does not imply the opportunity cost is zero. It seems self-evidently reasonable to assume the owner-occupier would prefer to own the property today rather than defer ownership of the property to some future date and thus be required to pay rent to the owner in the meantime. This suggests owner-occupation does not imply a deferment rate of zero, and that the deferment rate of owner-occupied properties is a function of the rental income that the property could generate if it were not owner-occupied.’
That’s right. In essence, if the owner rents, he or she gets the benefit of the net rental income. If the owner occupies, he or she gets the benefit of the use value, which he or she deems worth more than the net rental income because otherwise they would have rented out. In either case, there would be some loss to the owner if the contract were for deferred possession instead of current. So deferred possession is worth less than current and the PRA’s Principle III (that the deferment value of an asset is worth less than the spot) holds.
As far as we can tell, this is the first time that the IFoA or anybody writing with an IFoA affiliation has made any criticism of Tunaru’s analysis. It is moreover a fairly fundamental criticism, because it refutes Tunaru’s claim that the net rental rate should be divided by 5 on the grounds that only 20% of UK residential properties are rented.
If one accepts the rest of Tunaru’s analysis, starting with his gross rental rate of 5.6% (p. 31), then subtracting costs to obtain an estimate of the net rental rate, one would end up with a net rental rate of 4% or more. The next logical step would be to ask how correcting for this mistake would affect the NNEG valuation (answer: a lot) and away we go.
But for some reason, the WP chose not to go down this route. Instead, it became fixated with establishing the theoretical possibility that the deferment rate, which is equal to the net rental rate, which as we have just seen would likely be around 4% or more, could be, er, negative.
Clearly, there are people on the WP who have it in for Principle III, the validity of which depends on a positive deferment rate.
And so we arrive at the WP’s main theme:
‘Whilst we would expect the deferment rate of a residential property to usually be positive, we note there is no logical necessity for this to always be the case.’ (p. 4)
As it happens, we addressed this question in our Eumaeus Guide (pp. 101-102). Essentially, there are a number of possible cases. The first
is where the property and the land which it stands are polluted beyond any feasible repair. Chernobyl comes to mind: even if the land could be restored to a usable state, the costs of doing so would be prohibitive. … The property and the land itself would then be abandoned. This type of situation is rare, however.
A comparable case would be where the land on which a house is built erodes off a cliff into the sea. But the counter argument is that we would not expect many lenders to be ERMing properties that could fall into the sea after a couple of storms.’
In this case, Principle III does not hold because the spot and deferment prices would be the same, i.e., zero, but we would not expect equity release firms to be lending to Chernobyl class properties. Chernobyl is not an ERM asset class.
The second and less rare case
‘is where the property is uninhabitable and repair would be uneconomic, but the land itself is valuable. Parts of Detroit come to mind. One might then say that the (current or near current) net rental proceeds were negative, but this situation would not last because the land itself is valuable. The property would be demolished, perhaps after being sold off, and the site redeveloped to restore a positive net rental stream.’
In this second case, some short-term net rentals might be negative, but a positive net rental would eventually be restored. Principle III conceivably may not apply, but this case is still rare and very untypical of an ERM portfolio. As a general rule, ERM lenders avoid ERMing highly dilapidated properties and for obvious reasons. So this case is also of no real relevance to equity release.
A third and more common case is where the property needs repair and repair is economically feasible. The property might not generate any current net rental, but it would be repaired and a positive rental stream restored. This situation is not uncommon, but is still relatively infrequent, in that it does not apply to most properties most of the time.
The general case is that most properties most of the time generate a positive net rental stream. Therefore, when looking for a general rule to assess deferment value, the only sensible rule is to assume a positive net rental stream – and a positive net rental stream implies that the deferment value will be less than the current property value, i.e., Principle III holds.
In short, Principle III makes perfectly good sense for the kinds of properties that equity release lenders lend against. The WP’s emphasis on the theoretical possibility of a negative deferment rate would then appear to be misdirected. There is no point worrying about theoretical possibilities that don’t apply in the real world. But we are talking about actuaries here.
So when the WP points out (p. 4) that ‘We would expect these [the PRA’s] four principles [including Principle III] would generally hold in most foreseeable economic circumstances,’ we would respond: under which circumstances relevant to equity release would Principle III not hold, and why are any of these cases, if there are any, of any importance?
One then has to wonder why the WP is spending so much time on irrelevant issues like the theoretical possibility of a negative deferment rate, when the burning issue of the day, market consistent vs ‘real world’ valuation approaches aka the massive NNEG misvaluation scandal, doesn’t get a look in.
There is, however, another way to obtain at least the theoretical possibility of a negative deferment rate: the actuarial profession’s favourite deus ex machina, an illiquidity premium. Unfortunately, this trick doesn’t work either.
Let’s go back to first principles. An asset might be said to be illiquid when it is only possible to achieve a quick sale by accepting a price lower than the going market price. A house is a good example. We could achieve the going market price if we were prepared to wait, but a quick sale requires us to accept an illiquidity discount. Or conversely, we might only be prepared to buy an illiquid asset if the price is marked up relative to an otherwise similar but liquid asset. In this case, we have an illiquidity premium. However, by definition, illiquidity itself is a short term phenomenon. There is no illiquidity in the long run.
So one might say that a contract that gives immediate possession of a house might be subject to an iliquidity discount if we want a quick sale. However, the underlying assets in the individual ‘nneglets’ that make up the NNEG are forward contracts, most of which have long maturities. Take for example the nneglet with a 15 year maturity. The party that is long this nneglet knows that the contract gives them the right but not the obligation to repay their ERM loan by handing over ownership in 15 years’ time. So where is the illiquidity in this contract? The answer is that it doesn’t exist. And if there is no illiquidity, then there is no illiquidity discount or illiquidity premium. These notions simply do not carry over to long maturity forwards.
Or perhaps the intention was to impute not a ‘quick sale’ but a ‘slow sale’ premium, given the 15 year maturity? But that doesn’t work either! The value of a house HP is the sum of the present values of 1 year (e.g.) forward rentals, out to perpetuity:
(1) HP = x1 + x2 + x3+ …
where each of x1, x2, x3, … is positive.
We immediately see that the value of a deferment must be less than HP because we lose one or more of the rentals. E.g. suppose we defer the first rental, and let D be the value of the corresponding deferment, i.e. the sum of the rental values from x2 in perpetuity:
(2) D = x2 + x3 + …
(3) HP = x1 + D
Hence
(4) D < HP.
So why on earth would they think that a lower number is greater than a higher number? Perhaps they may have thought that the HP would be discounted by the illiquidity premium, but that the sum of the rental pvs would not be. Such thinking involves an elementary error, however: you can’t discount the left-hand side of (1) by an illiquidity premium without discounting the right-hand side too.
Now consider the WP’s paragraph 15 (with the footnotes deleted):
There is a technical argument, presented in recent actuarial ERM valuation research, that the presence of an illiquidity premium in the underlying house price should reduce the cost of the NNEG (note that the illiquidity premium of the residential property is distinct from the illiquidity premium of the mortgage). Specifically, the present value of the house price illiquidity premium that will be earned over the life of the option should be added to the house price that is used in the NNEG valuation. This is equivalent to deducting the house price illiquidity premium from the deferment rate used in the NNEG valuation, implying:
Deferment rate = Net rental yield – house price illiquidity premium
In the context of the above discussion of PRA Principle 3, it is again worth noting that there is no theoretical reason why this quantity cannot be negative.
Technical argument or not, we have already explained (a) that there is no good reason to believe that the deferment rate can be negative except under strange and atypical conditions that do not apply to equity release in the first place, and (b) that the notions of illiquidity discount or premium do not apply to NNEGs because the underlyings are forwards rather than spot contracts.
But let’s play along and calibrate this equation. In our Eumaeus Guide we gave (see pp. 39-40) a worked example of how we might obtain a typical net rental rate and we came up with a ballpark estimate of 4.2%. An alternative derivation (p. 40) gives us a net rental rate of 4.3%.
We would normally interpret either of these values as the average (or maybe typical) net rental rate over the borrower’s potential remaining lifetime, which is, typically, a fairly long time.
To obtain a negative deferment rate, we would need a liquidity premium of at least 4.2% or 4.3% per annum and this illiquidity premium would need to be the average illiquidity premium over the same long time period.
Or put it another way, do we really believe that the average long-term illiquidity premium exceeds the average long-term net rental rate?
Consider then the deferment rate of -2.75% recently used by one ERM firm. If the net rental rate is 4.2%, then the implied illiquidity premium is 2.75% + 4.2% = 690 bps, which seems awfully high in comparison to the more sensible estimates of the illiquidity premium (of say, 15 or 20 bps) that one sometimes sees.
To make the example more concrete, suppose we have a 55 year hold female ERM borrower. Assume for simplicity that she is expected to live exactly 30 more years. Assume also that the presumed liquidity premium is 6.9%. Then the illiquid asset trades at exp(-6.9% x 30) = 12.6% of the value of an otherwise similar but liquid asset. If you believe that, we have a bridge to sell you.
The bottom line is that actuaries are wasting their time looking for a long-term liquidity premium to do the heavy lifting in their equity release models. The darn thing is of no significance.
Perhaps our friends on the WP might consider the possibility that the actuarial ‘theory’ on this subject could do with a bit of dephlogistification to clarify matters.