Risk and uncertainty: why gilts can be seriously overvalued

Posted on

I wrote what I thought to be an entirely reasonable blog yesterday on why I think the next financial crisis will be more serious than the last, and a host of people who have never appeared in the comments section of this blog before I did so poured in to tell me how wrong I am.

To be precise, they without exception objected to the suggestion within the chart that I reproduced from Prof Daniel Mugge, who is Professor of Political Arithmetic at the University of Amsterdam, that UK gilts are overvalued. If you want to know what political arithmetic is, his explanation is available by following the last link, from which the most telling phrase in the current context may be this:

[W]hen branches of government publish evidence, either to assess past policies or to devise new ones, we demand that they not be colored already. But they inevitably are colored, consciously or otherwise. This applies most immediately to quantitative indicators, and especially to those that are so ubiquitous that we take them for granted: inflation gauges, GDP, unemployment figures, trade statistics, government debt levels. After all, measuring abstract economic quantities is never straightforward, and the choice for one formula over the other carries implications that are rarely understood beyond a narrow circle of experts and communicated to the wider public.

I have not asked Daniel if he approved of what I'd written, but we've mailed since and he raised no issue. He retweeted the blog post. I am guessing he is not taking issue. And I can guess as to why that is. What his work looks at is, amongst other things, the way false perceptions of data can be used to create misunderstanding for political advantage.

I respectfully suggest to those making comment that they are guilty of a number of things. First, they ignored most of my blog. I did not discuss overvaluation of gilts in isolation; I discussed it as part of a phenomenon of over valuation that is likely to correct. I was not, then, discussing a microeconomic phenomenon of whether or not a particular formula applied at a point in time or not  but was instead considering what might happen to gilt valuation in the event of what I called a discontinuity. This idea of a discontinuity seems to be a possibility beyond the comprehension of those commenting. In their world it would seem crashes do not happen, but even if they do, they are always self correcting (hence their obsession with simultaneous liability restatement to match asset price changes). Unlike them I live in a world where crashes do occur, with consequences. It's called the real one. So do they, for the record, but they seem intent on pretending otherwise.

Second, they believe their models. As one commentator said:

Gilts are defined by nominal, coupon and redemption date. Valuation is determined by discounted cash flow against the yield curve alongside the perceived certainty that the payments will be received i.e. they will not default.

Another invited me to use a spreadsheet, which, he claimed would prove that he was right. And I have to tell him, they won't.

So reluctantly I have to explain why that is the case: I say reluctantly only because I would have thought the world would have learned this stuff by now but it seems it has not, meaning that the odds of another crisis both occurring and having worse consequences than last time move from near certainty in my estimation to absolute certainty, precisely because the lessons of 2008 have very obviously not been understood by those who think they know finance.

Those lessons were succinctly summarised by Adair Turner  in his 2009 report on the financial crisis (para 1.4):

At the core of these assumptions has been the theory of efficient and rational markets. Five propositions with implications for regulatory approach have followed:

(i) Market prices are good indicators of rationally evaluated economic value.

(ii) The development of securitised credit, since based on the creation of new and more liquid markets, has improved both allocative efficiency and financial stability.

(iii) The risk characteristics of financial markets can be inferred from mathematical analysis, delivering robust quantitative measures of trading risk.

(iv) Market discipline can be used as an effective tool in constraining harmful risk taking.

(v) Financial innovation can be assumed to be beneficial since market competition would winnow out any innovations which did not deliver value added.

Each of these assumptions is now subject to extensive challenge on both theoretical and empirical grounds, with potential implications for the appropriate design of regulation and for the role of regulatory authorities.

In other words, standard capital asset pricing models do not work.

That said, I have already seen the responses on the blog saying this is not true of gilts, even if it is of other assets. The logic is that because UK gilts cannot fail (a point on which I agree) they are risk free meaning that the factors incorrectly appraised in 2008 cannot recur in the gilt markets now. Ignoring for a minute that this suggests that all commentators appear to agree with my contention on other asset values (and I am taking that as read because no one has said otherwise) let me still address the issue of gilt valuation and why that too may be incorrect.

What the commentators are assuming is that they can correctly appraise risk. That is because they think there is only risk with regard to gilt returns. This is implicit in the comment that we only need to know nominal, coupon, the redemption rate and the yield curve. In isolation, and at a point in time, that model works in isolation.

I stress though it works to produce a current value, not an actual value. Tomorrow's value may be quite different, of course, using the very same model. But  I am sure those commenting use this system day in and day out to trade vast sums and extract a rent from society for doing so, believing that they are incredibly clever for doing so when all they actually do is arbitrage that daily difference in perception which their model cannot value at a point in time because it, quite literally, discounts it, which is the flaw at its very heart.

But the reality is (and why do I still have to point this out?) that we do not live in a risky world, which is what this model assumes; we live in an uncertain one, even when it comes to gilts. And that is why this model is wrong.

This difference between risk and uncertainty is fundamental, but seemingly unknown to my commentators, just as it was to Adair Turner, as he has readily admitted it was to him and most of the City pre-2008. Risk means we can attach a probability to an event. In fact, not only can we attach a probability to an event, but we can even attach a probability to when an event might happen. So, in gilt terms, the only variable that the commentators thinks of consequence is if there is an interest rate change and when, and both these can be predicted and so priced, hence their apparent certainty.

Uncertainty on the other hand means we simply do not know what will happen. A probability cannot be attached to an  uncertain event. We do not know what will happen, or when, and what the consequence might be. We can guess, but we will be wrong. No amount of modelling can deal with the uncertainty: the future of uncertain events is not known.

My first contention is that a discontinuity - a disruption resulting from a financial crisis - creates uncertain consequences, even for gilts. My second - which I will come to - is that the consequences of uncertainty need time to play out. As a result, I argue gilt revaluation can be created by the type of crisis three of us predicted on Thursday.

Let me deal with two issues in turn. First, I am presuming for reasons already noted that those commenting accept that it is possible that stock market, property, personal and mortgage debt valuations on bank balance sheets, corporate debt, and derivatives based on all these things, appear over-valued at present. We can broadly argue that this is the result of misguided use of QE: I argued that this would happen in 2010  meaning I have some form in this area. In that case let's assume a crash can happen.

This is my 'discontinuity'. We do not know what will trigger this. We do not know where it will start. We do not know its scale, although I think it will be big. We cannot predict the durability of institutions in the face of it. We cannot predict their liquidity at a point of time during it. We are unable to say how long the consequences will last. And quite critically we do not know how governments will react. If those things are not a measure of uncertainty I am not sure what is and of course each has spillovers for the gilt market.

What are those spillovers? First, there could be market breakdown: liquidity failure might ensure that. We cannot know that market players will not fail, especially when their current solvency is based solely on inflated asset values that will disappear overnight.

Second, there could be forced selling of gilts: the supposedly reliable asset may be sold in the face of liquidity failures to realise cash wherever it is available.

Third, there could alternatively be a flight to safety and a massive demand for gilts, assuming that liquidity survives. So prices might irrationally rise: the excess value now is in no small part a measure of risk aversion and not just yield. This excess may increase.

But let's also be clear that, fourthly, the government may respond to this by flooding the market with gilts to maintain positive interest rates by reversing QE. It would be an act of folly, but governments are capable of those.

Fifth, things might go the other way. The government might have to (I think it will) inject cash into the economy. But it could not do QE again because there may not be enough gilts to do it with. So the gilt market may be abandoned and direct cash issuance could happen. It happened in WW1. It could happen again. And that changes the whole long term nature of the market.

Sixth there are unpredictable inflation consequences impacting yield expectations.

Seventh, interest rates could move in a number of ways.

But, eighth, just to add some fun serious tax changes could have serious impacts on asset values. Throw in a wealth tax and see what happens.

My point is, we just don't know. But the commentators ignore all this. They say we are dealing with risk when glaringly obviously in a discontinuity that need not, and almost certainly will not be the case. There model refuses to recognise that this could seriously disrupt the gilt market. Mine accepts it could. They're different models. And I say I am right because my model is evidence based: I am discussing the world that is. Their model is based on a world that clearly does not exist. That's why they're wrong.

And then, I come to my second theme, which is that contrary to the suggestion from my commentators, markets will not react immediately or rationally to any of these scenarios and the others that are possible (I am not suggesting the above list is complete). The idea that markets clear immediately is what is called 'freshwater' thinking because it is associated with Chicago. The assumption is that rational people rationally change their minds immediately when facts change because a) they know precisely what is happening and b) can process it immediately. This, of course is absurd. That cannot happen. Instead, there is a lag in processing, if it ever happens at all (the fact that I am writing this blog shows that for some this clearly has not occurred since 2008). This is 'saltwater' thinking because it is the basis for the neo-Keynesian of east coast USA universities.

Saltwater thinking is, of course, correct. But we also do not know how long correction takes, or if at all (and by raising that point I make my own position clear) But what we do know is that whilst it occurs models break down, sometimes completely. So the discontinuity has an impact, even on gilt markets, whilst reappraisal to take account of the new reality occurs.

And the only thing we know about that new reality is that gilts will be redeemed at par, which means that the only rational valuation in the face of discontinuity is something not too far from that value. But right now we are far from it. And I'd suggest that makes the current premium looks decidedly uncomfortable.

You can disagree, of course. But please don't say there is no logic to my disagreement, because there most certainly is, and unless anyone criticising it is willing to claim their clairvoyance their counter-argument based on a spreadsheet is actually utterly irrational, because making such a claim just shows  what they do not admit that you do not know and what assumptions you will not embrace.

At least I admit the scale of my uncertainty. It really is time my commentators did theirs.

PS This has been written in haste in little more than 90 minutes - which is pretty good going for 2,200 words. It could be improved  As a result and I may return to it to do so at some time.