Michael Burry

Factual · Economics and markets

“And to be clear, there was only one risk model that they all used.”

Michael Burry, lecture speaker · 5 April 2011 · Transcript of Michael Burry's talk, Missteps to Mayhem (Vanderbilt News)

Wall Street firms all used one single risk model in the mid-2000s.

Mostly false False in substance, right in a detail.

The second judge would not resolve this claim; the verdict shown is the first judge's.

What would count as true or false

Judged as of 2011-04-05, about the mid-2000s period. True: evidence (regulatory reports, the FCIC report, academic or industry studies) shows that major Wall Street firms all used one single, common risk model. Mostly true: they used the same core model or methodology, such as Value-at-Risk under Basel rules or the Gaussian copula for CDO pricing, with minor firm-specific variation, so that 'one model' is a fair description. Mixed: one methodology dominated an important area (for example the Gaussian copula for CDOs, or the rating agencies' models) but firms used meaningfully different models elsewhere. Mostly false: there was real homogeneity (a widely shared VaR approach or copula framework), but the literal claim that all firms used one single model overstates it, because firms built and ran their own distinct proprietary models. False: the evidence shows widely varied, independent risk models with no dominant common model.

Evidence

1. [PDF] SEC Risk Management Review of Consolidated Supervised Entities U.S. Securities and Exchange Commission, Office of Prudential Supervision and Risk Analysis (internal memorandum, FCIC document archive) · 6 February 2008

   “Value-at-Risk ("VaR") continued to be relatively flat across CSEs, but this overall”

2. [PDF] SEC Risk Management Review of Consolidated Supervised Entities U.S. Securities and Exchange Commission, Office of Prudential Supervision and Risk Analysis (internal memorandum, FCIC document archive) · 6 February 2008

   “The firm also increased its firmwide VaR limit to $150 million, although the divisional limits were not increased in conjunction with this decision.”

3. The Gaussian Copula and Modelling Practices in Investment Banking ("The Formula That Killed Wall Street") Donald MacKenzie and Taylor Spears, University of Edinburgh (published in Social Studies of Science) · 2013

   “was, as suggested above, characterized increasingly by a dominant approach.”

4. The Gaussian Copula and Modelling Practices in Investment Banking ("The Formula That Killed Wall Street") Donald MacKenzie and Taylor Spears, University of Edinburgh (published in Social Studies of Science) · 2013

   “It has been developed mathematically in different ways by different people in different contexts.”

Reasoning

The SEC's 2008 review of consolidated supervised entities shows that Value-at-Risk was a common metric across the major firms (E1). The same review records one firm raising its firmwide VaR limit to $150 million (E2). MacKenzie and Spears describe derivatives modelling as increasingly dominated by one approach (E3), but they also say the Gaussian copula was developed in different ways by different people (E4). Together this points to real homogeneity in methods, but not to the literal single shared model the claim describes.

Cite this claim

Michael Burry, “Wall Street firms all used one single risk model in the mid-2000s.” (said 5 April 2011). Verdict: mostly false. Reliability API, claim c_ea2255e8cfce, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_ea2255e8cfce