26 September 2017
Gert Raeves

Data management 101

Gert Raeves - TowerGroup

10Posts 46,279Views 1Comments
Finextra community

Data Management 101

A community blog about data and how to manage it

Get with the Toxic Asset De-Tox Programme

13 July 2010  |  5177 views  |  1

Hello, my name is the international investment community and I have been brought low by my addiction to toxic assets. It took away my bonus, my job, my lifestyle and my reputation. As I am trying to make sense of what happened and how I can get my life back - I wanted to share with you a few things I learned about toxic assets the hard way:

1. Toxicity is inversely proportional to transparency - or welcome to the world of known unknowns.

These assets are not toxic just because they are bad investments, i.e. assets that are worth less than what you paid for them . They are toxic first and foremost because of their size. The exposure to toxic assets threatens the survival of the company that holds them - ie if all payment obligations need to be met at market condition n - there will not be enough cash in the pot (illiquidity) or assets on the balance sheet (insolvency) .
So the markets worry about the Rumsfield Ratio - how much do you know you don't know, or don't know that you don't know? The result is systemic opacity, distrust, and unwillingness to transact - as witnessed by the shutting down of the credit markets after the Lehman bros collapse.

2. "Does my risk look big in this?" It is all about perception, which feeds trust, which feeds reality.

The toxicity of an asset is not just about the empirical size of the bad stuff on the balance sheet- a toxic asset infects the value of all other investments. Toxic assets replace the systemic trust relationship that underlies all banking (if I give an amount of money to a bank, I will get back at least the same amount of money if I need it), with systemic distrust. Toxic assets destroy that trust - and replace it with the suspicion that no counterparty can be trusted.
There are two main root causes of this distrust - the envelope problem - and the nodal network problem. The value of a toxic asset is determined by lots and lots of data - and because we assume there is too much data to compute - its value is based on assumptions, scenarios, conventions, and ultimately human psychology.

3. The truth about the Envelope Issue - What is inside this thing?

There is a popular misconception that no-one knows what is inside these toxic assets. Of course we know - it is just that there is too much low-level data that needs to be tracked to be practical. The best examples of toxic assets are envelope type products - which take a large number of real-world inbound cash flows (mortgage payments, car loan payments), and package them into a single debt instrument. The bank puts all the real-word stuff in one big envelope and sells it to wholesale investors, and the envelope is associated with the creditworthiness and risk rating of the bank, as well as the characteristics of the real-world stuff.
Because human behaviour is more predictable at the macro level than at the individual level, the price, return and terms and conditions of the envelope are based on assumptions. So the value of a mortgage-backed derivative is not based on actual tracking and analysis of individual repossessions, mortgage arrears, payment defaults and other relevant real world events. Instead it is mostly based on socio-demographic averages and macro-economic trends.
If more-people-than-usual cannot make their mortgage payments, that does not impact the value of the envelope immediately. Instead that fact (along with millions of other facts) patiently sits there until a news organisation or government agency, picks it up - aggregates, normalises and computes it - and releases it as a new set of yesterday's facts to turn into today's assumptions. These in turn impact the price that people are willing to pay - and in normal market conditions this is the normal interplay or supply and demand, of risk and return. But when the difference between yesterdays and today's assumptions is bigger than expected, the house of cards falls.

4. It is very hard to solve a "nodal" problem - how can you know your counterparty's counterparty?

I have bought some kind of envelope product - which is a contract between me and the bank that put the envelope together. The contract says that this bank will pay me an agreed amount of money if certain things happen. So as long as I believe my counterparty is able to make those payments, I can sleep at night - right?
Wrong - because I know that there are envelopes within the envelope - contracts between my bank and some other bank . How can I possibly know what will happen if that third bank becomes insolvent? If they don't pay, what does that mean for my bank - will they struggle to release enough cash from their balance sheet, and ultimately will there be enough assets to prevent insolvency? This is why simple calls to "know your counterparty" are not enough - to have a complete view of all potential risks to my investment, I need to know my counterparty's counterparty's counterparty's counterparty... Some of this can only be solved by the use of centralised trading and clearing models - which allows all counterparty nodes, all risk and all risk mitigation to be shared across an entire market - resulting in systemic transparency.


My Three Step program - Back to reality, trust, and trading

But I have not lost faith - it is possible to do more than simply wait for new regulation, new infrastructure, or a new academic consensus on valuation methods and models.

1. Remind yourself that the value of toxic assets has been de-coupled from reality - it is based on assumptions and models, supposedly because there are too many facts to be practical/necessary to allow the markets to accurately value the asset.

2. If the markets cannot provide usable prices because there is not enough activity to level out individual supply/demand or risk/reward ("mark-to-market" is not available) - there are two options: using a set of models and assumptions ("mark-to-model", but there is no academic consensus about the right approach), or a more fundamental data-driven analysis of true value. I guess you could call it "mark to data".

3. Do not accept that there is too much data to be useful. This is 2010. There is a tremendous amount of intellectual and technological firepower available to compute fact-based assessments of value and risk. The problem is not one of too much data - it is the opposite. The times demand radical data availability and transparency. We all need more data, and better data to drill down on, examine, normalise, aggregate, refine, crystallise and use to make sense of and start trading again.


TagsRisk & regulationWholesale banking

Comments: (1)

John Fitzgerald
John Fitzgerald - AIB - Dublin | 14 July, 2010, 08:54

Wasn't step 3 supposed to be done by the rating agencies? They saw all the data (all of it) and summarised it using their massively complex and extremely reliable models, ensuring consistency of assessment and saving the rest of us the trouble.

It wasn't lack of data that punched holes in that process.

Be the first to give this comment the thumbs up 0 thumb ups! (Log in to thumb up)
Comment on this story (membership required)

Latest posts from Gert

Get with the Toxic Asset De-Tox Programme

13 July 2010  |  5177 views  |  1 comments | recomends Recommends 0 TagsRisk & regulationWholesale bankingGroupData Management 101

If you are reading this, it means...

17 May 2010  |  4838 views  |  0 comments | recomends Recommends 0 TagsRisk & regulationPost-trade & opsGroupData Management 101

Price and Prejudice and Zombies

03 December 2009  |  6635 views  |  0 comments | recomends Recommends 0 TagsPost-trade & opsWholesale bankingGroupData Management 101

Pricing - Mind the Gap

09 September 2009  |  4188 views  |  0 comments | recomends Recommends 0 TagsPost-trade & opsWholesale bankingGroupData Management 101

It always is, was and will be about the Data, Stupid

07 May 2009  |  5135 views  |  0 comments | recomends Recommends 0 TagsPost-trade & opsWholesale bankingGroupData Management 101

Gert's profile

job title Research Director Capital Markets
location London
member since 2008
Summary profile See full profile »

Gert's expertise

Member since 2005
10 posts1 comments
What Gert reads
Gert's blog archive
2010 (2)2009 (6)2008 (2)

Who's commenting on Gert's posts