Comment by btilly | Hacker News Reader

Comment by btilly | Hacker News Reader

Parent Full thread

The destabilizing threat is real and systemic. Derivatives allow us to hedge local risks at the cost of creating the possibility of the whole system collapsing. It greatly concentrates risk in those who wish to gamble, and if their gambles go south allows them through leverage to threaten the whole financial world.

This is not just a contrarian view. Go to https://www.berkshirehathaway.com/2002ar/2002ar.pdf and start around page 13 where it says "Derivatives" for Warren Buffett's analysis. This is the one where he calls derivatives "weapons of financial mass destruction". (This was several years before the 2008 financial crisis.

He makes a number of good points. One of which is that we value derivatives is based on complex models. Those models inevitably have errors. And the errors are generally in favor of the side that you purchased. So those engaging in the trades always believe that they are making money (and bonuses etc get paid on this basis), no matter how disastrous it may later turn out to be.

In this case it is likely that the pension fund managers never realized what kinds of complex gambles they were making or what the real risks are. They were working according to models in which what they did was wise. And it may well be that they were right in the long run. But as Long-Term Capital famously showed, being right in the long run only matters if you survive to see it. It remains to be seen if these pension funds will survive.

But in this case, there was no error on derivatives. It was unforeseen and very violent market move in the response of Tory’s policy “more tax cuts” and the general loss in the faith that the UK economy will be productive and surplus any time soon.

Those who modelled the risk did not think Tories would do anything that stupid and risk the UK economy.

I would take even a somewhat kinder view of the pension funds involved. With no knowledge of the intimate aspects of the agreements UK pension funds make with employers, those involved knew that you could either pay for the risk margins involved up front, or via emergency capital.

Margin calls are nothing special. Derivatives are used for balancing assets and liabilities, since you cannot match the cash flows on pensions directly given the very long duration (cash flows for pension funds run for nearly a century). Swaptions etc are nothing evil. They are pretty clear cut tools for the job of limiting interest rate exposure for parties with very long liabilities.

This is not a pension fund problem (I agree up front: that's a reductio nearly ad absurdum), it's an employer and pensioner problem, created by a sudden loss in confidence in UK gov in an already very turbulent market. If sterling falls, purchasing power falls, interest rises, there is no way pensions can keep purchasing power or indexation. Margin calls are a tiny symbol that shouldn't worry a risk manager. It's the larger enviroment for the UK - that is very worrying.

It's all hocus pocus but a well run insurer or pension fund should have a pretty solid grasps on interest rates and a pretty low exposure to swings in that rate. Same for FX exposure. End of '21 you should have opened up some upward potential in interest rates and have some room to survive that volatility. The best managed ones keep afloat or even profit, without taking too many excess risks. The worst managed ones are the first to fall. Go and look at the Q3 financial results for the largest _worldwide_ insurers. I predict the largest insurers will show stronger financial capital positions. (Stocks will fall since large insurers are partially valued on general stock market performance since they are in essence also investment companies.)

Again, margin calls are a symptom caused by the underlying agreements, with not a direct relationship to which ones will fail or prosper.

The fact that the model did not account for what actually happened as possible IS a problem in the risk model.

Pension funds are supposed to run for decades. Looking at the last 12 decades, I see many financial crises, a run on the pound, the Great Depression, a couple of world wars and so on. All of which were unthinkable until they happened. Any risk model meant to cover long periods should account for the possibility of similarly extreme events happening in the future, at roughly the historical frequency.

But you're telling me that those who modeled the risk failed to give this ANY consideration at all?? And then you say that you think that this oversight is not an error??

I'm going to have to very emphatically disagree!

1. Just because papers always seem to know why markets move, doesn't make it true. 2. Margin calls happen when you are short, highly leveraged or not hedging properly (derivatives going wrong). Should "pension funds" do any of these things ?

It reminds me of a what a friend of mine said who priced Mortgage Backed Securities before the financial crisis "Oh, we never modeled different mortgage default rates - those were fixed assumptions. The really complicated math was driven by interest rates."

He did leave me with the impression that tnterest rate modeling was technically challenging and had lots of scope for sophisticated work - so that's where the effort went.

Is the "tax cut" justification the real cause though? The Euro and Yen plunged in lockstep with GBP, which suggests a more fundamental issue beyond just the UK market. I have to wonder if the mini-budget was just used as justification.

There’s no complex modeling required here on an interest rate swap. The payout on swaps is easily calculable and the range of outcomes knowable. There’s not much overlap in the context of what Buffett talked about and what happened here besides the word “derivatives” and the idea of exogenous shocks tending to lead to outcomes outside your range of modeled assumptions.

I’d be glad to engage more but it seems like you are fixated on a slant and cannot be convinced otherwise.

I think the equivalent would be if, a very modestly able coder, were to insist on this forum that nobody should ever have loops in their code because sometimes they do weird things.

> The payout on swaps is easily calculable and the range of outcomes knowable.

While this is technically correct, it's the solvency of the counterparties that is not easily calculable or knowable. Derivatives must be viewed in the totality of the system in which they are used. Focusing on the fact that payouts are calculable and therefore perfectly safe in isolation is extremely myopic.

> I think the equivalent would be if, a very modestly able coder, were to insist on this forum that nobody should ever have loops in their code because sometimes they do weird things.

I think the equivalent would be if, a very modestly able coder, were to insist on this forum that because the database is up 99.9% of the time, there's nothing wrong with not handling errors, because the DBA team is running with an average IQ of 143, they all went to MIT, they know what they're doing, black swans do not exist, and that 99.9% might as well be 100.

Derivatives are inherently dangerous because they are a knowable mathematical system coupled quite tightly to an unknowable system of risks. You can't mentally decouple them and claim that welp, the math works, it's fine, ship it.

This expresses it perfectly.

And given the web of interdependencies among the counterparties, every counterparty should be assumed to be overly exposed to a systemic risk that everybody is currently discounting. If that risk happens, it will be difficult to predict who will be affected, and it will be impossible to unwind the deals in a hurry.

We have had a number of financial crises that have required external intervention to keep the whole system from melting down. Most famously, 2008. And they show how quickly a solid counterparty that everyone trusts becomes a source of contagion that poses a risk to everyone touching them.

If we have a long planning horizon, what kinds of shocks should we be prepared for? Well here are a few sample scenarios.

1. Russia collapses, leading to nukes floating around to bad hands.

2. China finally invades Taiwan, threatening everything that needs computer chips.

3. A US government shutdown finally results in non-payment of Treasuries, causing a "risk-free" investment to have to be priced for risk.

4. A repeat of the 1859 Carrington Event happens, with damages in the trillions of dollars. No seriously, https://www.space.com/the-carrington-event shows how plausible this is.

The financial system treats all of these as unthinkable and therefore impossible. But between all of them, over a period of decades, the risk of SOMETHING on this order of magnitude happening is significant. It is a problem that the entire financial system fails to appreciate that there IS a risk, let alone fails to do anything to mitigate it. Indeed, quite the opposite, the 2008 bailout has let the financial system believe that the government will always bail them out. And so has become complacent about their role in creating various minor financial crises.

But there is no guarantee that the government will always be in a position to do that. The result has been a normalization of deviance that has to end badly at some point. See https://www.ostusa.com/blog/normalization-of-deviance-defini... if you're not familiar with the phrase "normalization of deviance".

Thank you. Although I think the risk of #4 is greatly overblown. We watch the sun constantly now, and detection of an X20+ class flare aimed at us gives us about 10 hours or so to disconnect as much as we can, limiting damage to maybe only a trillion or so.

My personal favorite pet risk is a combination of #1 and #4, though. A few HAEMPs de-orbited and detonated over North America by a collapsing or collapsed Russia gives us no time to disconnect anything. And the fantasy that neither defense.gov, nor mil.ru, have many of these in orbit right now because "OMG it would be a Space Weapons Treaty violation and nobody would ever do that", is as naive as the pre-Snowden "OMG the NSA would never spy on Americans because it's illegal" normie zeitgeist.

> the equivalent would be if, a very modestly able coder, were to insist on this forum that because the database is up 99.9% of the time, there's nothing wrong with not handling errors,

That’s what the margin calls are, no?

Reply on news.ycombinator.com