Failed performativity in Iceland

September 25, 2010

(Many thanks to a friend of the blog, who shell remain anonymous, for pointing out the initial story)

A new documentary film, Inside Job, describes the financial crisis and discusses, among other things, the role that academics played in setting the scene for the crisis. Fredric Mishkin, a finance professor from Columbia business school is interviewed in the film, where he is asked about a May 2006 report he wrote (with Tryggvi Thor Herbertsson), titled “Financial Stability in Iceland”. The report was funded by the Icelandic Chamber of Commerce, who also paid Mishkin $124,000, a fact that Mishkin did not disclose in the report itself. The film focuses on the possible biases in the report and makes a general argument that financial economists were part of a broad network that facilitated the motivations and the market devices that brought about the 2008 crisis.

An SSF look at the report, however, produces a more nuanced insight, I believe. Mishkin and Herbertsson try to measure the likelihood of a banking crisis in Iceland by using the spread on Credit Default Swaps (CDS). The CDS spread of Icelandic banks is used to arrive at the probability of a financial crisis. The methodology in the report refers to a Barclays Bank report from April 2006 where the analysts present possible future scenarios of Icelandic banks and where they pose a CDS spread of 500 basis points as a ball park figure that would present a financial catastrophe: “In the worst case, a financial crisis could potentially lead to banks widening to an average of 500 bp, say.” Based on the hypothetical figure of 500 where there is a 100% probability of a crisis (because an actual crisis would cause such widening of spreads), the analysts look back at the current spread of 67 basis points and calculate a probability of 7% for a crisis.

Mishkin and Herbertsson use the figure of 500 basis points as upper limit for a linear relation they assume exists between CDS spreads and the probability of a crisis (see figure 20 in their report). The validity of this exercise relies on how well the CDS market translates default risks to prices. The report, however, does not only present these model-based findings, but also tries to affect the reality in the Icelandic financial markets. After all, the Icelandic Chamber of Commerce funded the research to show that the markets there were stable and sooth investors. In other words, Mishkin and Herbertsson were involved in a performativity attempt: presenting a model-based projection in the hope that market participants would react to the projection in way that will materialise it in the market.

This attempt, as we all know, failed. The question why it failed is intriguing. Finding potential explanations for the failure can shed new light on the conditions in financial markets before the crisis and on the behaviour of financial markets in general. For example, it could be that the CDS-implied probability performativity attempt failed because it ‘came from above’? That is, there were not enough users and market devices that depended on the model when acting in the market to support increasing similarity between the model and the market. Such a process, of co-institution of markets and models (where both gain validity and viability) did not evolve sufficiently in the case of CDS, again, for obvious reasons.

So, the story of the Icelandic report, a candidate for a failed performativity attempt (note: this is different from counter-performativity), is an intriguing way to look at the crisis and its potential causes.

3 Responses to “Failed performativity in Iceland”

  1. danielbeunza Says:

    Great post! Leaving the sensationalist story aside, what strikes me as fascinating is that the technique used by Mishkin is exactly what I studied in my paper with David — reflexive modeling. They find out what “the market” thinks from applying math to the prices of assets. And, as with the case hapless merger arbitrageurs that we followed, the technique blows up on their face.

    Watch this space for a longer post in which I compare and contrast the two.

  2. danielbeunza Says:

    By the way, I forgot to add that I disagree with you. You claim that the authors were trying to engage in performativity. I don’t think so. Performativity, according to *your own* paper with Donald, is when the accuracy of a model is improved. But the author’s claim not about any model — i.e., the use of the no-arbitrage condition to back out crisis probabilities on the basis of interest rate spreads. Instead, they insist on a specific prediction of the model, namely, that the crisis was unlikely. It is more of a typical process of enactment: trying to influence a collective by making a bold claim about the future. So you’re reading more nuance than there is.

  3. yuvalmillo Says:

    So, you say that because someone actually wanted to bring about a certain result and acted toward that goal this is not a case of performativity, but of ‘enactment’. Two things:
    First, that might have been the case if that someone was, say, Ben Bernanke. But, if that someone is some professor of finance (even a famous professor of finance), then the strongest tool they can use is their theoretical models and the predictions these models produce.

    Second, the ‘accuracy of the model’ is improved, according to the performativity argument, when there is verisimilitude between the model’s predictions and the market. However, how this situation is brought about is a different matter. The options traders (and clearninghouses, etc) unintentionally brought about performativity by doing model-based arbitrage. Mishkin et al intentionally tried to bring it about by publishing a report showing that the Icelandic market is unlikely to fail. Mishkin’s intended outcome was to bring about a situation whereby the Icelandic banking market won’t fail and, thus, increase the similarity between the model’s prediction and reality.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: