How did I know that Oil prices would drop in 2008
One of the questions that smart people ask me when they learn that I won a 2008 Forbes investing contest with only a few trades is: “How did you know that oil prices would decline in July 2008?” The answer is that I did not know, but the fundamental data seemed to indicate an improbable growth in productivity, and I bet against the veracity of the data.
In plain terms, my bet wasn’t that I could predict the price of oil day-to-day. Instead, I bet that the entire industry was deceiving itself about the quality of its data and a reckoning was near. I did not know back then that the questioning of data and analysis validity would become the central point of my PhD dissertation, but it shouldn’t be a surprise to anyone. David King, my academic adviser at Harvard, calls me a contrarian. But, given my work and academic history, I like to think of my approach to life as skeptical. After all, up until 2008, my adult lifespan had included nothing but economic bubbles. I had worked through the 1990s/2000s tech bubbles as an insider, watching how the earnings numbers were manipulated and learning about the myths that analysts accepted as truisms but were clearly myths. And I had experienced the housing market bubble, where real estate brokers, consumers, and other professionals believed in the myth that housing prices would always go up. It seems like great hindsight to say that I was extremely skeptical of the unprecedented bubble in the markets in 2008, except I have a huge body of supporting documentation to validate that I really was extremely skeptical.
So, when I read recent analyses of history by the Federal reserve and other economic experts that says that prices were based on economic fundamentals, I must ask the seemingly obvious question: “What are you smoking?”
Economic fundamentals are only as good as the data measurement methods and practice for analyzing them. During the Ford administration, economists realized that they had crude metrics for many aspects of measuring economic conditions and one of the great triumphs of this era was the increase in the quality of metrics. But, the metrics are still very imperfect. If you meet an analyst that regularly works with BLS data and doesn’t have some question about the data, I would like to meet them. So, if there is a weakness in the reporting of fundamentals or in the traditional economic analysis’ treatment of the data, how would current economists find it using their traditional analysis methods?
To jog your memory about what was happening in 2008, the spot price of a barrel of oil on the New York Mercantile Exchange had gone from $116.32 on May 2, 2008 to $145.29 on July 11, 2008. Economists have suggested that this price was rational at the time because all oil was being consumed. This may be true, but for how long can this pricing level be sustained? Some economists were arguing that fundamentals and inelasticity of demand by consumers would continue to support this pricing level as global consumers clamored for the last container of oil each month. To believe the Federal Reserve models for economic fundamentals at the time, one had to assume that worker productivity in the United States had doubled in a three year period. (Note: I must thank Dr. Akash Deep from Harvard’s Kennedy School for working with me at Harvard to develop this component of my system while I was a student.) But, more importantly, in June 2008, the rate of growth of estimated U.S. productivity might increase by another 2x+ in the July to August period alone.
I have daily interactions with insurance companies, banks, mortgage processors, lawyers, technology professionals, food service workers, etc. My bank has trouble keeping my name, address, and account balance correct. As a skeptic, do you think I can believe that the productivity of the workers in these institutions is doubling?
So, the question is, how do you operationalize the skepticism? In the case of oil, I began searching news reports for stories that didn’t add up, given the assumptions in the data. Economists told me floating oil inventory was declining, but, in China, local reports were indicating that there was a huge gap between the reported size of the floating inventory and the actual size. At the time, China was reporting near zero floating inventory and the day before I made the first oil trade, China’s local news reporters had taken a picture of miles of tankers parked off the coast, filled with Iranian crude, that clearly no one was counting in the official numbers. A hyper-local Chinese story (by a blogger) was about a dispute that was erupting between China and Iran because there was simply no place to park the tankers any longer and no ability to store the oil.
This was my first introduction to Chinese accounting, which, like U.S. corporate accounting, has been documented by scholars to contain a certain element of institutionalized fiction. (See my other blog posts for more data about this. I may eventually return and update this story with the links.)
To summarize ….
I am a data scientist. My PhD dissertation is about subjectivity in machine learning and data analysis. If you believe your data analysis methods are objective, I want to short your trades. All data analysis is subjective because the data production and analysis methods are subjective. When our society fools itself into believing improbable theories, usually through the rigorous application of the best scientific methods of the time, I want to be there to question it.
