Washington University Professor James Morley on typical recession shapes and why they suggest we might see a strong recovery.
Harvard Professor Lucian Bebchuk on how to buy troubled assets while avoiding some of the problems pointed out by many analysts.
Oil 101 looks like a useful new book by commodity trader Morgan Downey.
And the Shadow Open Market Committee is back in business.
Waded through Prof. Morley’s paper and found it terribly incomplete. By virtue of looking at only post-WWII US recessions, he fails to capture any of the unique characteristics of a financial crisis driven recession (per Reinhart-Rogoff) or the consequences of simultaneous recessions in our globally-connected world. Sorry, but it’s not really that useful–and may well be misleading.
I agree the Oil101 book looks interesting. In an April 13 post on his own blog Mr Downey posted a poorly specified regression model. Makes me wonder if he is fit to be writing such a book. If you don’t understand regression don’t use it. It is better to not use a tool than to misuse it. And it makes me wonder what other tools he has misused along the way which may have contributed to this book.
http://Find Bad Model Here/
Oh please. The graph is fine. There is no 100% “correct” model for plotting oil demand. All statistics in economics are rough at best and every statistic can be nit picked to death. It is very clear and obvious to me from the chart what Downey is saying. What specifically do you find is “bad” about it?
I have read a good chunk of this book – As a macro trader on a prop desk, I am familiar with oil fundamentals in general, and this type of reference book is exactly the type I was looking for. I was referred to it by some of my oil trading colleagues.
It’s mainly about fundamentals, rather the author’s opinion or predictions. It takes a big picture view of the oil industry – it explains the facts, but it also explains why those facts are important.
That chart looks ok to me – As one who has read the book, it’s clear the point of graph is intended to plot a bunch of data in a succinct way than to do a predictive regression analysis.
I think JS may be taking it out of context in that way. The part of the text the chart accompanies is a section that puts into perspective oil consumption from different countries and consumption over time, so that was my takeaway from the chart. I guess it could have served just as well as a simple scatter, but the line doesn’t really bother me either.
I guess its just a different environment – I was a math major, but in real life on the trading desk we use charts and draw lines to visualize information succinctly and quickly. We do not really use them to predict x given y given a normal distribution for data like this. If that’s what you’re trying to get out of it, it may not be for you.
Terry: Please read the section entitled, “Is This Time Different?”, particularly footnotes 4 and 6 which directly address the Reinhart and Rogoff paper on financial-crisis-driven recessions and the IMF’s view that globally-synchronized recessions (e.g., 1973-75 and 1981-82) produce weak recoveries.
Perhaps my criticism is a little nitpicky but that does not change the fact the model is mis-specified.
Economic statistics arent 100% accurate but that does not mean we should not do our best to use, treat and model them correctly. And that isnt the case here.
Some of the discussion about the book has focused on it being about fundamentals. That is the danger here concluding a fundamental relationship, linear demand, from a mis-spec-ed model. Maybe something as simple as running logs fixes the problem and solidifies the conclusion but maybe not..
I would think rigor would be just as important in evaluating a graph and regression like this on a trading desk as anywhere. To me demand looks as if demand maybe quadratic and quasi-asymptotic. What if you are trading on the linear growth assumption but the real relationship is approaching the flatter region. Finally, if an econometric model cant be critiqued on a UCSD related blog then where can it? Everyone I know from UCSD has attempted to convince me that it is the alpha and omega of Econometrics
Thats exactly my point – I agree with you in that it’s not robust, but the key is, nothing’s being ‘modeled’ in this particular instance. Would you have a problem with it as a simple scatterplot? It’s a simple depiction of information, that’s all. The regression itself is not the point of this chart. I wouldn’t care if he used a cubic spline.
It is important that pricing and trading models must be robust, but comparing that type of information with this type of information is like discarding a good research report on copper because there was a misspelling in it.
Traders get a lot of different type of information, and we apply different interpretations depending on what it is.
I don’t want the argument to come to semantics. When I say model in this case I am not referring to a pricing or other model that, we, traders would use. But rather a model that depicts a relationship such as the one presented by Mr. Downey, GDP v. Oil Consumption.
I would be more comfortable if it were just a scatter plot but I dont think that would tell the story. I reread his post and still contend he did a poor job of graphically substantiating his thesis. His narrative seems spot on with what the data points tell me. Unfortunately the regression he has attempted to present tells a different story and is incorrect in doing so. When I look at the graph that is what I see, its stark. And if the regression is not important to the picture or logic dont include it. The fact that he included it is what gaves me pause.
Perhaps this was oversight on Mr. Downeys part like a grammatical error in a copper report. But I see this as a little bit more serious. Maybe it is maybe it isnt, that is why I raised the question. Trust me I dont want to miss the forest for the trees here. I contend that if this narrative and graph were sent to a journal the referee would reject it. This was a blog post, I understand that. If the book has figurative spelling errors Im fine with that, logic errors not so much.