Friday, October 21, 2011

Review: Presimetrics

I try very hard not to buy paperbooks. They take up a lot of room, aren't as portable, and are difficult to read compared to books on the Kindle. But Presimetrics demands paper. That's because it's full of graphs and charts, which look terrible on the Kindle, and demand high quality presentation. Presimetrics is written by Mike Kimel, who's built statistical software and does a lot of economics analysis. You can in fact, see some of his writing over at Angry Bear. It's intelligent, full of data, and makes good use of data-mining.

This is a great book to read if you've got a statistical bent and are willing to follow the data rather than your pre-conceived notions. For instance, conventional wisdom says that the president can't do very much about the economy. In the first chapter, Kimel debunks that. He shows that control of congress, for instance, doesn't have an impact on the economy, but who the president is matters. Even better, he dis-integrates the effect of the Fed. It turns out that the fed pumps more money into the economy when there's a Republican president than when there's a Democratic president. Who wins? Read the book.

The book is devided into chapters based on issues. If you care about Fiscal responsibility, turn to Chapter 2. If you care about health care coverage, chapter 8 covers it in detail. Crime? Chapter 9. There's also a mish-mash of issues covered under "Democratic issues" and "Republican issues." There's also an entire chapter on taxes and how they affect the economy.

5 appendices round out the text, covering GDP growth in capita and President Obama, who had just had a year into his presidency when the book went to press.

Many people like to say that they're data-driven, but most people actually have prejudices that lead them to believe what they believe, as opposed to actually looking into data and correlations. This book goes a long way towards providing those who want data the actual data with which to base their beliefs on. It won't change the minds of any libertarian fanatics, but those folks are hopeless anyway. The only criticism that could be aimed at the book is that "correlation is not causation", but given the variety of decades covered and the thoroughness in which this book covers the topics (they even do regressions where they assume presidents have a latency in the time they get into power and when they have an impact on the economy), I find this book very convincing.

You might think that this book is boring and tough to read, but it's not. Kimel brought in two other co-authors to help make the book readable (one of them is an Economics writer) and the graphics pretty. This is the kind of book that deserves to sell better than it does. Highly Recommended.
Post a Comment