Comment by smallnamespace | Hacker News Reader

Comment by smallnamespace | Hacker News Reader

This is old news for large portions of the interest rate derivatives market, which is quickly moving towards OIS[1] rates instead, based on widely traded liquid instruments.

Also, the implication that LIBOR is purposefully a scam is basically untrue.

When LIBOR was first developed, it was an improvement on other interest rate benchmarks, and it also reflected current market conditions at the time, as banks actually did regularly make bilateral interbank (the 'IB' in LIBOR) loans to one another.

There are checks built into LIBOR to discourage fraud: for example, the actual calculation discards the high and low outliers, so individual banks cannot manipulate the benchmark easily[2].

But as the interest rate market continued developing, much of the actual lending transaction volume moved towards other markets; the financial crisis just accelerated that trend.

So now we have a lot of contracts written against a benchmark that slowly stopped reflecting an actual interest rate market. The right path forward would probably be to renegotiate these contracts so that going foward, they're against OIS instead.

[1] https://en.wikipedia.org/wiki/Overnight_indexed_swap [2] https://en.wikipedia.org/wiki/Libor#Calculation

> they're against OIS instead

OIS stands for overnight indexed swap [1]. Like the fed funds rate [2], it's only quoted for one tenor: overnight.

In the United States we're somewhat spoiled with having a deep, reliable, market-based yield curve calculated every business day: the Treasury yield curve [3]. But if you want to approximate the cost of a bank borrowing for a given term on the wholesale unsecured market, Libor is still the default reference.

[1] https://en.wikipedia.org/wiki/Overnight_indexed_swap

[2] https://en.wikipedia.org/wiki/Federal_funds_rate

[3] https://www.treasury.gov/resource-center/data-chart-center/i...

> Like the fed funds rate [2], it's only quoted for one tenor: overnight.

That's not quite true -- for example, 3m OIS swaps have fixings that are essentially the 3m average of FF over the period in question.

In fact, this is mathematically a bit cleaner. If you construct an interest yield curve off of compounded 1m LIBOR vs. 3M LIBOR, you get rather different answers, whereas OIS yield curves constructed from different tenors are much closer.

The argument for using OIS instead of Treasury yields is pretty simple: Treasuries reflect the cost of borrowing for the government and are implicitly affected by the government's creditworthiness (e.g. not raising the debt ceiling), while banks can actually borrow from the Fed at FF. While the US gov't funds at pretty close to 'risk free', in other markets the credit component would be significant.

OIS is generally closer to LIBOR than a Treasury yield curve is.

> this is mathematically a bit cleaner

But also misleading. Bootstrapping requires making assumptions about term structure [1].

> OIS is generally closer to LIBOR than a Treasury yield curve is

After the crisis the Libor-OIS spread was observed as an indicator of bank instability. They're close, but not the same. When they diverge, it's for reasons incredibly important to certain users of Libor. You're correct in the OIS rate being better than the Treasuries for estimating banks' borrowing costs.

[1] http://www.investopedia.com/terms/t/termstructure.asp

I would suggest [1] instead of Investopedia for a modern treatment of yield curve bootstrapping. The only difficulty with OIS is in my opinion the low liquidity for higher maturities where basis swaps have to be used to estimate a spread to LIBOR IRS.

[1] https://papers.ssrn.com/sol3/Papers.cfm?abstract_id=2219548

As an aside, as someone that works in this space (yield curve construction, pricing, risking etc), does anyone know of a forum where news items have this sort of discussion?

Nuclearphynance and quant stack overflow are basically it. Otherwise best discussed on the trading floor!

Wilmott may be your best bet

https://forum.wilmott.com

I think you are mixing overnight index and overnight index swap. Derivative discounting is based on the index referenced in the CSA, which is typically an overnight index (Fed Fund in USD). To build a forecast curve for this index (and therefore a discount curve) you need to use swaps paying that index, ie OIS, hence the term OIS discounting. You can have any tenor you want for an OIS. When people refer to an OIS and don't specify the tenor, they usually mean a 3m swap, but you can 10y OIS.

You could document a loan paying quarterly with an interest calculated as the average o/n index for these 3 months. In fact that's how weekly CSA work. But that's not very practical for smaller non financial clients. It would be better if there was an index they can observe directly.

> It would be better if there was an index they can observe directly

Agree. Even for sophisticated clients, throwing a direct reference into a contract is cleaner than translating a mathematical construction into legally-binding text.

Also most banks got badly burned with fines after the Libor scandal and introduced a pretty strict process to ensure that doesn't happen again.

The "fantasy" the article is referring to is the fact that large banks have to submit a number for every currency and every tenor every day, even if they didn't fund in that currency and that tenor that day. They will usually interpolate based on other tenors.

However this is only a problem for the lesser used tenors (like 2m, 8m), the most referenced tenors in private contracts (1m, 3m, 6m) tend to trade very frequently.

The most substantial change was reducing the number of currencies and tenors Libor is quoted in. Gone are the Australian dollar, Canadian dollar, New Zealand dollar, Danish krone, Swedish krona, 2 week, 4 month, 5 month, 7 month, 8 month, 9 month, 10 month and 11 month Libors [1]. That's an 80% reduction in the number of currency-tenors Libor is quoted in.

[1] https://en.wikipedia.org/wiki/Libor#Currency

I wasn't aware they had been dropped. Thanks!

That's what the article stated - that at one point it made sense but banks found cheaper ways to get capital and slowly stopped lending to each other as much.

> There are checks built into LIBOR to discourage fraud

https://en.wikipedia.org/wiki/Libor_scandal

'Discourage', obviously, is distinct from 'prevent'. But those checks were enhanced in the wake of the scandal, and it is based in large part on the strength of those enhanced checks that this conclusion about the underpinning validity of LIBOR has been drawn.

We can all agree, though, that the next iteration could do more to stop manipulation and fraud.

Just by your terminology, I trust that you know (at least somewhat) what you're talking about, so... is there an ELI5 for this stuff? I'm completely lost in these types of discussions. They seem absurdly complicated for (good|bad) reasons? Are we looking at another complexity bubble that's going to burst in 5-10 years... to the detriment of everyone but the "top execs" who'll be bailed out?

My impression after reading "Too Big To Fail" -- which, BTW is an amazingly well-written book[1] -- is that "they" will probably get away with all this confusion and obfuscation... right?

I really do want to learn at least a little bit, but it seems like such an impenetrable world of concepts, and I really do want to avoid the "news-based" or even "journalist-based" approach to learning about this subculture. (When a Dartboard fares better that your average journalist/investor-proxy, you know you have a gambling problem.)

[1] Not sure how accurate it is, but it's really well-written and exciting. Unfortunately, it doesn't delve that much into the technical concepts in Finance :(, though it does mention LIBOR at least once.

Regularly read Matt Levine's Bloomberg column and the articles he links to and you'll pick up more than you ever thought you would know about finance. It is also fun, how he explains things.

Note: not a substitute for formal study of these things.

If you read the market section of the FT every day for a year, you will probably read enough articles with an introduction to all of these concepts to have a good enough understanding. But like any other profession, if you dive into the details, it can get quite complicated very quickly. There are lots of introductary books to each area of finance. For interest rates, the best book I have read is Interest Rate Swaps and Their Derivatives: A Practitioner's Guide by Amir Sadr. It's been written a few years ago and some things have moved on since but that's still an excellent, very practical, not too mathematical introduction to interest rate products.

Good reasons, imho. The complexity is not (purely) a smoke screen thrown by conspiratorial insiders.

Most of the harder tech HN articles would seem just as impenetrable to regular Bloomberg Terminal users.

The difference being that in tech the complexity just happens as a byproduct.

With the "finance industry" the complexity is not a bug, it's a feature, to keep ahead of regulators. Every time some immensely profitable "market" gets regulated the "finance industry" does it best to create a new "financial vehicle" to feed that unhinged greed for the fictional perpetual economic and financial growth.

It's an endless cat&mouse game that has been going on way too long and got us into this current messy situation where the "finance industry" might just as well be considered a fantasy football league, completely decoupled from the actual day to day realities of the vast majority of human beings, yet still holding massive influence over their futures.

I disagree; I think that's just a lazy way to look at an unfamiliar field.

Likewise, plenty of clueless non-techies suspect techies are intentionally making tech seem complicated. Why can't Apple make a back door for the FBI and nobody else?

Reply on news.ycombinator.com