Codes & clues, credit & crisis

Still conference blogging. Now an early morning session with Michael Lissak who has changed his original talk to a complexity approach to the current financial crisis. Michael is never knowingly un-opinionated, one of the major whistle blowers on Wall Street in the past and from anyone’s perspective a controversialist (he makes me seem mild mannered and reasonable) so this should be interesting. He’s a former merchant banker turned academic (ethics, complexity) and real estate agent so he is in a good position to talk about this.

Starts with a quote from Rorty, Knowledge is not a matter of getting reality right but rather a matter of acquiring habits of action for coping with reality and moves from that to a summary of complexity issues:

• Complexity worries about compartmentalisation
• identity of actions situations and contexts is seldom stable and often time proceeds in multiple directions
• Emergence and weak signals/ raises questions about models metrics and goals. Multiple perspectives are key.

If we see the world as stable we name things (ascribed labels) and respond accordingly. A complex perspective says that things change and there is an emergent narrative. Mistakenly confuses this as realism v constructivism but sees me wince! Lovely picture of a sign (Japanese or Korean) which says: slip and fall down carefully. (see above). This is used as an ironic example of management instructions. The main theme of Michael’s current thinking is to distinguish codes and clues.

• Codes – the names we give things have assigned meaning and we can look up those meanings in a code book. There are one to one mappings and the label carries it meaning with it. So when a manager talks they are speaking from the code book. Quotes Derrida’s view that perception is dependent on coding. Codes are part of the modern. Codes allow detection of ambiguity. Science places emphasis on codes.
• Clues – complexity understands context, perspective etc so cannot over-rely on codes. Interesting quote from Roland Barhes: We know now that a text is not a line of words releasing a single ‘theological’ meaning (the “message” of the Author-God) but a multidimensional space in which a variety of writings, none of them original, blend and clash. The text is a tissue of quotations drawn from the innumerable centres…. Moving on to a series of quotes the best of which is: Texts are code-productive and code-transgressive as well as code-confirming from Terry Eagleton. Clues give us an idea, they can support models but we can not be sure what is going on.

Clues are the big idea here, so he is now expanding it. Taking a post-modernist view on interpretation and different perspectives which is fine, has not yet fallen over the edge into relativism and denial of reality as happened yesterday. Pointing to clear clues in the build up to the financial crisis. Cheap credit. adverts offering click for cash adverts offering money in seconds to people with bad credit ratings. Really scary chart of the distribution and number of sub-prime loans in the US, and when I say really scary I mean really really scary.

Good point about coding fallacy – if you have a triple-A rating it must be OK, if the credit score says fine you are creditworthy. However, the clues were all there (well they always are lets be honest). Says that the issue is what do you choose to pay attention to. Implication that those in the know who realised the implications would use shorting to protect their personal position. I am reading that into what he says as its not been explicitly stated, but it makes sense.

Not sure about where he is going now. Saying that the world divides into two types: those who believe in code and those who believe in clues. Code-people have a hard time when clue-people start to create alerts and tend to ignore them. A bit too much white-hat/black=hat for me, but it may just be a rhetorical point so lets wait and see.

Issue at hand

Moving on to what we do and how (or least says he will, a lot of people do that but they carry on saying what is wrong with existing practice. Mike has the capacity to break this so lets see) He starts with three key issues:

• Complexity reduction in the interest of regulation and transparency
• the degree of complexity present is the degree to which our chosen method of reduction has failed
• as the reduction method for “gamed” complexity increased – a lot of people on Wall Street are paid a lot of money to game the system

Now asking Why model? Regulators have to act within the scope of a model, but it doesn’t mean they have to act within a coded-model such as bell curves where the implicit assumption is independence of agent and random behaviour. OK, we know this, still diagnostic but worth doing. Law of large numbers suggests that weak signals are noise – complexity says not it’s important. This is a critical point and well made. Bullet point now: Ascribed labels have no room for emergence, which is largely true. The indexical quality may have reached its limit and there is something we are missing. Still talking about negative aspects of codes.

Uses the famous Einstein quote: Everything should be made as simple as possible … but not simpler. Saying that market economics as gone for simpler by assuming there will always be a market. Another scary picture of negative equity over the US. Valid point that there is no market in these circumstances. Logical inconsistencies of banks having to value houses at market price, even though most people will not sell. So book value is being badly stated and making the problem worse. Some great cartoons now showing up the foolishness of the current situation.

First clues were out over a year ago (is this retrospective coherence?). Saying that no banks have a cash problem, its the liquidly ratios in a falling market which are killing them. i.e the code (leverage is good) is killing us. Because people think is good they go for more of it and then we get the problem. Still hammering away at market-to-market values but as yet (and looking less likely) an alternative. Don’t get me wrong, I think market economics on their own are evil (and I use the word deliberately) but I’m interested to see what a former US merchant banker comes up with, especially as he a friend I respect!

New SEC rules allows management estimates that incorporate future cash flow numbers to replace market-to-market which will, when things catch up in balance sheets etc. allow money to flow and the problem will ease up. No one has really noticed the rule change yet due to panic, but they will. Once panic sets it you have to wait for people to feel tired of panicking before they pay attention to changes you have made. This is a very important point with wider applicability but its still descriptive/diagnostic rather than intervention/alternative model and its over. Pity, it was a great session and I enjoyed it, but a little flat at the end as no alternative, no way forward other than wait it out.

Nice reference to our methods for weak signal detection during the questions. Question session overall is getting interesting. Peter Allen is talking about models. I reference our work on language, and the need for uncertain environments to be established before that language can be used together with the need to ritualised social constructs for sense-making rather than focusing on individual decision making. Today is a real contrast with yesterday which is good news. Serious shit as they say.