Customer Reviews

Predicting Presidential Elections and Other Things (Stanford Business Books)

5 star:		(4)
4 star:		(0)
3 star:		(0)
2 star:		(2)
1 star:		(0)

 

 

This is a short book (155 pages, 10 chapters). And, the dominant theme is how to predict Presidential election (occupies 4 chapters). Ray Fair writes more like a bestseller writer than an economics professor at an Ivy League that he is (Yale). His writing style is alive and journalistic. As a result, his book is a lot more fun to read than your usual dry economics textbook. You don't need to be a quantitative type to enjoy this book.

I came across this book by doing an Internet search on Presidential election forecasting models. By far the best was Ray Fair's model, as it accurately predicted the outcome in 18 of the past 21 elections. Really intrigued, I read his technical notes extracted from his working papers published within econometric journals. But, I could not fully understand some of the key details underlying the foundation of his model. Fair indicated he had written a book explaining his model for the layperson. I jumped at the occasion, and bought the book.

The book met and exceeded my expectation. Within the second chapter, the author gives an excellent lesson in linear regression and multivariate regression analysis. The entire chapter is written in plain and understandable English without recourse to a single Greek letter. Armed with this renewed and deepened understanding of linear regression, I could easily follow the most relevant details underlying the foundation of his Presidential election forecasting model.

This model is really pretty amazing as it forecasted several quarters before the polls did, and before we even knew who the Republican nominee would be that the Gore - Bush election was going to be just about too close to call. The model uses several macroeconomic factors (inflation, overall GDP per capita growth, number of quarters with very rapid GDP per capita growth) and several political factors (who is the party in power, how long has it been in power, what party is in power) to come up with the incumbent party's voting share of the two party total votes.

Right now in March of 2004, this model indicates that Bush will get about 55% of the two party total votes. This is pretty interesting, given that currently the polls often have Kerry ahead by several percentage points. Additionally, the model suggests that its estimates of Bush's vote share is unlikely to change radically between now and November. In other words, it does not foresee a Kerry win under most likely economic scenarios between now and then. I know this may not be what readers would like to hear, but that is a social science observation, it is not a political opinion. Mr. Fair's objective is to develop the most accurate model he can. Granted, the model can always be wrong one more time. But, given its track record I would not bet big money against it.

The book is not just about Presidential elections; it also addresses the following issues:

1)Who is more likely to engage in extramarital affairs;
2)How to forecast the quality of a new wine;
3)How to forecast grades in a college level economics class;
4)How to predict by how much people slow down with age
5)How to forecast short term interest rates; and
6)How to forecast inflation.

Issues 5 and 6 are classic macroeconomic concerns. The author addresses them in a predictable way (using other relevant macroeconomic variables to predict the dependent variable).

But within these six issues, two stood out for me. These include wine quality and how much we slow down with age. In both cases, the model's output was amazingly interesting.

Regarding, wine quality, it suggests you can forecast the quality of a young wine years down the road just by looking at several climate variables around the time the wine was produced. The back testing of the model was really successful. In other words, just by observing data and using his multivariate regression model, you can come up with an assessment of a wine quality that would probably be more accurate than the subjective opinion of most wine experts.

Considering aging, the author looked at age based world records in middle to long distance running disciplines. From the data, he derived what is the pace of slowing down from one year to the next after age 35. From looking at his data, you will be amazed of how little the body slows down between 35 and 59 years old. Sixty appears to be a transition, when the slowing down becomes a bit more rapid. But, still people who keep healthy and fit can remain pretty competitive well in their early seventies. This section gives you a totally different perspective about aging; and, frankly a much more optimistic perspective at that.

In the acknowledgement section of this book, Prof. Ray Fair notes that his kids talked him out of titling this book "What Econometricians Can Do" or "Economterics Made Easy." Either of those titles would have actually been more descriptive than the admitedly accurate title of "Predicting Presidential Elections and Other Things."

This book isn't really about predicting presidential elections -- even though it goes to great detail on how to do just that. This book is really about giving an accessible account of how to use statistics/econometrics to study questions we face in society or daily life. Presidential elections is a clever device/examplar to illustrate his ideas, but other issues are covered as well (e.g., value of wine, forecasts of extra-marital affairs, and more standard econ stuff like interest rates and inflation).

This book would be a great supplement to someone who is taking an econometrics course or a social science statistics course. It would also be of value to researchers, students, policymakers, et al., that are looking to have an accessible account of how to structure statistical research. E.g., Ray Fair gives a checklist/breakdown of how social science statistical studies should be carried out that should be useful in almost every relevant circumstance.

I also recommend "A Concise Introduction to Econometrics: An Intuitive Guide" by Franses.

Being a "stat freak" from a long-time interest in various sports, especially baseball and track & field, I very much enjoyed this book. Of special interest to me was the chapter on determining slowdown rates in the marathon as well as in shorter running distances. Professor Fair takes marathon age records beginning at age 35 -- the age when we are supposedly "over the hill" -- on up to age 84 and comes up with some interesting statistics. The slowdown rates at shorter distances vary a little, indicating speed deteriorates a little faster than endurance, at least in the middle years.

The findings relative to running times may very well apply to other activities. Author Fair concludes that societies may have been too pessimistic about losses from aging for individuals who stay healthy and fit.

Other chapters deal with presidential elections, extramarital affairs, wine quality, college grades, interest rates, and inflation. What do they have in common? As Fair explains, they can all be explained and analyzed using the tools of social science and statistics.

Fair does a great job treating a diverse set of topics (presidential elections, marathon performance at different ages, wine quality, probability of having a marital affair, effect of class attendance on students' grades) in a very technical but understandable way. He does have a chapter on the basic statistics involved in econometrics which is fairly easy to understand but if you want to avoid any math you can simply read about the problems and his solutions and still get a ton out of this book. The true insight in the book is Fair's commentary on how he approaches each problem and refines his model and that's where understanding the math becomes important.

A MUST READ for anyone interested in getting a more technical background in the quantitative approach to problem solving made popular by books like Freakonomics and used everyday by quant funds on Wall St, artificial intelligence researchers, and business intelligence / data mining tools.

I thought this book was interesting from standpoint of the topics that were discussed but it was poorly written. I have read a number of other economic books written for the lay public and they were much better written. The author in a couple of chapters starts out with a good theory but then never proves it. After reading those chapters I wondered what point he was trying to make. I don't think he knew either. I cannot recommend this book.

A friend suggested this book, given my interest in predictive analytics. I realize this text predates the likes of Freakonomics and Super-Crunchers, but I find the examples to be somewhat random and the text lacking readabilty. Ehh...