Jienan complains that my blogs are way too long and few has the time to read it. So this is a (relatively) short one. (BTW, if you get a chance to read moneyball, it's definitely worth the time. This book is outstanding!)

One interesting headline in msn.com last week was about the ranking of top 50 healthiest large cities in the U.S. (actually, the 50 city/area covers most of the u.s, and the city that ranks the 50th is the least healthy one.) According to the study, San Jose is the healthiest city in the U.S., followed by San Francisco, Washington DC, Seattle, and Salt Lake city. Boston ranks the 11th, New York the 43rd, and New Orleans is the least healthiest.

To measure how healthy a city is, the study ask questions in five areas, including a) mental (sleep problem or not, percentage of depression or so on),  b) life style (how many hours worked per week, vacation days per year and so on), c) physical activities (regular exercise or not and so on), d) health status (cholestrol level, diabetes incidence and so on), and e) diet. After measuring how well a city does on each of the five category, a city's total health points is obtained by adding up separate scores from each of the five.

This is certainly a useful guide for people looking at how healthy a city is and also brings public attention to the health problems. However, I don't quite agree with the way the study calculates the aggregate health level of a city. It seems to me that the two categories above, physical health and mental health, are true indicators of the health levels. The other three, life style, physical activities, and diet, are merely conductive methods to achieve health and should not be counted as part of the health outcome. It is like judgement the performance of a student, we should look at his grades and not how many hours he works.

Eyeballing the health statistics, some interesting patterns are revealed immediately. First, mental health doesn't seem to depend that much by weather. LA has a mental health of 8 (out of 100), San Fran has 30, and chicago has 81.

More interesting, the correlation between mental health and physical health is -0.19. In other words, people with higher physical health level are more likely to have worse mental health. This is certainly a puzzling but absolutely interesting result, since conventional wisdom would suggest physically healthier people are likely to be happier, which leads to good mental health.  Assuming the study is correct, I have two possible explanations for it.

First,  people with low mental health(have trouble to sleep, high stress level) are usually under immense pressure from work. These pressure actually enhances the level of immunity system and thus boosts physical health. One chinese saying for this would be 天行健在自强不息。(Work more, even if under stress, is good for physical health.)

The second explanation also assumes people who have low mental health are under pressure from work. But stressful jobs are typically well paid, and those who get the chance to do such jobs may have better genes. These better genes give them better physical health.

These two explanations are dramatically different and have very different policy implications. One says stressful job is good, even from a health point of view. And the second one makes so much statements and are more likely to be aligned with conventional wisdom.(stress leads to worse physical health.)

It would be interesting to distangle these two theories.  One way is to randomly give people stressful jobs (random promotion) and see how the person's physical health changes. A better physical health suggests explanation 1 is right, and the other way will favor explanation 2. Unfortunately, such practice is quite unethical and is unlikely to be implemented in reality.

Of course, it is actually very likely the study was done improperly and data is unreliable. In this case, the theories above is just total nonsense.

The first time I heard of this book is from Larry Summers, who in a public lecture recommended everyone to read it. By the time I read the book about two weeks ago, it has already been on the NYTimes bestseller for ages. Many people believe that this is the best book of Michael Lewis, and I more or less agree. Although the book is less funny and profane than Liar's Poker, after reading it I feel extremely excited, energetic, and empowered.

The theme of the book is very simple: why market can be inefficient sometimes and how one can exploit it. The scenario to illustrate this theme is not a typical market, but the major league baseball (MLB).  As in any other professional sports, the success of a team these days depends crucially on how much money the team has. A simple regression of number of games won on the amount of money a team has shows that it takes more than $1 mil to win an extra game per season. Some teams, like the Texas Rangers, take about $3 mil to win an extra game. Yet there is one team, the Oakland  A, paid about half a million to win an extra game per season for many years. Furthermore, it is consistently among the teams that win the most games. How does Oakland A do it?

The following paragraph captures the essence the answer:

"At the bottom of the Oakland experiment was a willingness to rethink baseball: how it is managed, how it is played, who is best suited to play it, and why. Understanding that he would never have a Yankee-sized checkbook, the Oakland A's general manager, Billy Beane, had set about looking for inefficiencies in the game.  Looking for, in essence, new baseball knowledge. "

Yes, the key is to treat baseball as a science and to use scientific method to win as many games as possible with a fixed budget. One feature that facilitates baseball for a scientific study is its abundance of statistics. Furthermore, most of the play in baseball can be considered one to one, and this makes the standard tools of statistics applicable. By attaching all kinds of statistics to a player, this allows the team properly pricing its players, or in the case of Oakland A, also to buy under-valued players and sell over-priced ones.

This sounds all easy enough: step 1, collect the statistics; step 2, perform the statistical analysis and price the players; step 3: buy low and sell high. How hard can this be? Shouldn't all the teams do exactly the same and result in every player being priced properly ( so market becomes efficient)? Part of the answer is that most teams fall under the pitfall of conventional wisdom.

Start from recruiting.  As in basketball, the professional baseball athletes can either be recruited directly from high school or from college. Different from basketball, the recruited players do not play for the team in MLB immediately. Instead, they first need to play in a series of minor leagues and most of them never made the major league.  Conventional wisdom says that high school graduates are more likely to make it into the MLB. As a result, most scouts tour the country mostly visiting high schools. Conventional wisdom also says players with a future are those who have "tools", which include the the ability to run, throw, field, hit, and hit with power.

Statistics, on the other hand, shows that college graduates are more likely to make the MLB than high school graduates. In this case, the conventional wisdom is dead wrong. On the other conventional wisdom, players who can run, throw, and hit are of course better players. But statistics analysis shows that conditional on a player’s standard statistics (batting average, runs batted in, and so on) , being able to run faster or throw faster just doesn’t matter much more. In this sense, the conventional wisdom is misleading.

Because of the conventional wisdom, players, particularly high school graduates who have “tools” are consistently overpriced, and those who have good play statistics but doesn’t have tools (either because of being too fact, too weak, or even because of not good looking enough) are systematically underpriced.

Oakland  A exploited this and was often able to get good players at a bargain. They recruit a lot more college graduates than other teams. They are also willing to recruit players with good statistics but don’t necessarily possess the “tools”. The only extra price they need to pay is their players are often fatter, runs slower, pitch slower, and looks less handsome than those in other teams.

Beyond success in recruiting, the Oakland A also knows how to price existing MLB players better than others. Their advantage comes from being able to better read the baseball statistics, or more correctly, being able to read the right kind of statistics in contrast to just standard ones. Standard statistics in baseball were developed a century ago and have two big drawbacks. First, they are developed quite arbitrarily. Some statistics are chosen simply because they are easier to measure than because they can more accurately capture the players performance.  Second, the rules of game and the equipments the players use today is quite different from those one century ago. As a result, some of the statistics, like errors on the field, were more useful in the past than today. Of course, standard statistics is also awfully incomplete in evaluating a player's performance. 

All these are very obvious, but clubs in MLB somehow never bothered to develop better statistics. Instead, the first step toward better stats comes from a (more or less unemployed) baseball fan named Bill James, who used his own money to publish a book called 1977 baseball abstract. In the book, Bill James argued forcefully why the standard statistics aren’t good indicators of performance and offered to develop his own statistics. Although only 75 people  bought his book, Bill James was actually encouraged and continued to publish his baseball abstract annually. His persistence brought him success: a few years later the book made the NYTimes bestseller.

Ironically, most people bought the book because they find it useful in playing “Rotisserie Baseball”, which is a video game that simulates actual baseball games. Insiders of the MLB, however, mainly treat the book with contempt and apathy. The general managers of most MLB clubs probably trust their own judgments a lot more than baseball statistics, especially statistics developed by amateurs. Billy Beane, the general manager of Oakland  A, is a rare exception. He read Bill James’s book religiously and hired a Harvard graduate who knows statistics. Using the statistics developed by Bill James, Oakland A often offered trade of players that surprise other teams. And very often, many teams soon regret the trade. In fact, the Oakland A has even dubbed many trades as “fucking A trade”, which means that the other team will feel like wanting to fuck Oakland A after the trade.

By recruiting better players and making bargain trades, Oakland A won 91 games and their division with $26 mil in 2000, 102 games with $34 mil in 2001, and 103 games with $42 mil in 2002. In the entire MLB, only the Yankees, with a budget of about $150 mil, won as many games as the Oakland A in 2003. Wow!

At the end of the book, Michael Lewis mentioned that the astonishing success of Oakland A was still dismissed as mere aberration by most MLB clubs. Only a few understood why Oakland A has been so efficient; and fortunately for Bostonians, one such person is John Henry, the recent owner of Boston Red Sox. John Henry offer Billy Beane a five year contract of $12.5 mil. Billy Bean backed out in the last minute, but John Henry understood the importance of using good statistics and hired Bill James as a consultant. Boston Red Sox “reversed Baby Ruth’s curse” and won the world series in 2004, the first time since 1918.

Millions of ideas were swirling in my head after reading the book. The most immediate ones are certainly how to price a soccer or basketball players properly. Because of the complementarities (teamworks) between the players, pricing these players are certainly more difficult. But the goalies can probably be priced better, and more challengingly one might think of pricing a package of players together.

Similar ideas can certainly be extended outside the sports field and lead to many interesting profitable opportunities. One fellow Wesleyan and MIT alum, Gary Loveman, now the CEO of the largest casino chain in the world (including Caesars and Harrah's), developed great statistics to price the slot machine and hotels in Las Vegas. By introducing a card that collect people's demographic info, different price will be quoted to different people for a hotel room in Vegas in New Year's eve (when all the hotels are always full). A 60 year old asian lady, typically a big player, will be quoted a very low hotel room price, and a graduate student who is unlikely to gamble, will be charged an exorbitant price for the room. Accroding to Loveman, the average daily revenue of a hotel room in his company has jumped from less than $150 to more than $250 one year after the pricing technique was developed.

As an economist, I am more interested in how firms, governments, and schools can develop better statistics to select people. As an example, is it optimal for Chinese, Math, and English to be weighted equally in the college entrance exam?(Similarly for the SAT exam here). But even within the current exam system, students in some province ( like Shan Dong) need to score a lot higher than Bei Jing students to get into college. Some people think this is grossly unfair. Others believe this is justified because Bei Jing students have higher 综合素质. Which side is correct? From an economics point of view, (if we only care about efficiency), I suggest that we compare students in BeiJing who barely get into colleges from those (who barely got into colleges) in ShanDong. If we can show, say five or ten years after college, the Bei Jng group fare better, then the admission score gap is probably justified. And vice versa. Such test is clearly difficult to carry out, but is certainly worth doing.

For anyone who reads this far, I really appreciate ur patience and tolerance for my babbling. But hopefully, you will also get excited about the book and go ahead read it. :) It would even be greater if some of you can leave a comment and suggest an application of statistics on real life problems. I would love to see those. (If there's one company I want to start, it would be using statistics for human resources or sth like that.)

As the last example of statistics, I will tell a story about Jerry Hausman, a famous economist at MIT. Hausman, in his first class of econometrics, gave a following warning about how to run a regression. He mentioned that Chicago Bull never won an NBA title before he consulted for it. And Chicago Bull (more or less) never won a title after he stopped consulting for it. If one runs a regression of Chicago Bull on whether they have hired Hausman for consulting, then the outcome would be Hausman is single-handed responsible for the Bull's success. Of course, Hausman meant that his consulting probably played little roles. But his contribution to the team, if the story of money ball is correct, might actually be larger than what he thought it is.

A friend of mine (probably Bei ) told me years ago that there were two "holy cross" in the western society/civilization. The first cross is the truly holy one and refers to Christiananity. The second cross consists of two curves.The one sloping downward is called demand curve; the one going upward is called supply curve.

One can hardly exaggerate the importance of these two curves. They are indeed the core of economics. And to many economists, especially those near the Chicago area, these two curves are economics. (I believe Steven Cheung (zhang1 wu3 chang2)used to say that everything in economics (or life?) can be reduced to the "fact" that demand curve is downward sloping.)

A downward sloping demand curve simply says that as the price of a commodity decreases, more people will buy it. And a upward sloping supply curve says that as the price of a commodity increases, more people will sell it. People (like my parents) might ask: what's more common sense than a downward sloping demand curve and an upward sloping supply curve? And why do we need people to study them? 

Simple as it sounds, some very complicated real life problem can be analyzed through a)properly define the commodity to be studied, b)analyzie the factors that influence its demand, c) analyze the factors that influence its supply, and d) determine the resulting price and quantities of the commodity. In applying d), the crucial assumption is that the resulting price should make the demand equal the supply. In other words, the market outcome is where the two curves cross, or in the jargon of economists,  market is in equilibrium. These four steps constitute the paradigm of economic analysis.

The power of the paradigm above lies in the range it encompasses. Commodities that can be analyzed include not only stocks, movies, or pig feets, but also education, marriage, and crime. Economists, pioneered by Gary Becker, have used the paradigm to study problems that typically belong to the field of sociology, anthropolgy, criminology, and etc. This cross made of demand and supply curves is the most powerful weapon economists wave in invading other fields.

However, there is one crucial question within economics that the paradigm stumbles. And the question is: Why wages don't fall during a recession? The paradigm says that a) as the wage increases more people will want to work,(labor supply curve is upward sloping, at least in the relevant ranges) b) as the wage increases, firms want to hire fewer workers (labor demand curve is downward sloping).  The equilibrium employment and wage level is determined by the intersection of the two curves. In a recession, firm's productivity decreases, so at any given wage level they want to hire fewer workers. In other words, the labor demands shifts to the left in a recession, and the wage and employment levels both decrease.  

There are two problems with the theory. First, (nominal) wages seldom fall during a recession. Second, the theory makes the reduction in employment level all sound very voluntary. People don't work because they think wage is too low (they prefer playing xbox at home than toiling in front of a machine at $16 per hour.) If people are willing to work at the prevailing wage, then they will get a job. This sounds better than reality, or even better than communism. :) 

Religious believers of the paradigm cover their eyes and refuse to see involuntary unemployment. One leading mainstream theory in macro economics, called real business cycle, indeed says that people don't work because they value leisure more than the money from working. During a recession, the low productivity drives down the wages, so more people find leisure appealing and less choose to work. For those who suffer a long period of unemployment, I wonder how they feel if they hear the expounders of the real business cycle theory got a Nobel Prize recently for it.

Less religious believers of the paradigm have come up various answers to the puzzle. The two major explanations for the puzzle are based on shirking and goodwill respectively. The first explanation is typically called efficiency wage theory and says that if there's no unemployment, then workers will not be afraid of being fired and will shirk.  Involuntary unemployment exists to scare the workers so that they have to put in effort. The second explanation, called the gift exchange theory, says that workers will be nice to the firm if the firm is nice to them by paying high wages.

Both theories have their pros and cons. But they certainly didn't satisfy the curiosity of one economist. Truman Bewley, a distinguished theorist(micro economist) at Yale, very much want to know in reality why wages don't fall during a recession. As a result, he set out to interview hundreds of human resources managers, top executives, and union leaders during the 89-91 recession to understand the wage policy in real life.

The result of Bewley's project is a book, named properly as "why wages don't fall during a recession", and contains a collection of interview questions and answers. The 500+ pages long book is organized around the determinants of wages (how does a worker's wage depend on the internal and external wage structure and so on), the effects of wages on the company (how does the wage policy and wage changes affect productivity, easiness of hiring, and turnovers) and how the policies are related to existing economic theories (do you think efficiency wage theory is relevant and so on).  

There are several important conclusions from the book:

1: Companies care very much about the morale of the workers. They don't lower wages in fear of lowering the morales, and thus the productivities of workers.

2: Companies are risk averse. Liquidity (cash) matters to companies a lot more when they are tight in money.

3: A worker's salary is not determined only by his productivity, but depends crucially on what others are paid at the firm. Within any firm, there is typically a stable  internal pay structure that are relatively protected from outside labor market.

4: Workers don't seem to know much about the pay level in other firms. Therefore, firms are relatively unaffected by the outside labor market and have very large power in setting pay.

5: Firms care very much about how the workers thinking about them. They want the workers the feel that the firm is one big happy caring family.

6: Fiirms in secondary sectors, which typically hire temporary workers, behave a lot like what the economic theory prescribes.

There conclusions reflect mostly the subjective beliefs of human resource managers and sometimes beliefs from firm owners or labor union leaders. Most of the conclusions are reasonable and intuitive. Everyone knows morale is important and people care about fairness in pay within an organization. Bewley concluded  from his interview experience in the end that: workers are sensible and reasonable, and the wage practice in the firms is in general optimal.

It would be great for economists to write down simple models capturing the interdependence of people. Traditional economics is mostly about individuals maximizing their own utilities without caring about others. And it is exactly this feature that makes economics simple and powerful. But concerns about fairness and morality are important aspects of human behaviors and deserve to be studied and modelled more seriously. In this sense, the book serves as an useful encouragement of economic research in this direction. In fact, I become more inclined to write models on these aspects after reading the book.

As an economist, however, I tend to skeptical about the conclusions, especially because most of conclusions come from subjective beliefs of human resource officers(HRO). My source of skeptism comes from 1) HROs do not always tell the truth, especially in areas where their interests are not perfectly aligned with the firm's; 2) the HRO's might not be perfectly rational, and their subjective beliefs are likely to be wrong. 

For the first objection, it is sensible that human resource managers design policies that maximize their own well-being, not the firm's. for example, in choosing from laying off workers versus a pay cut, even if it is better to have a pay cut, the human resource manager might choose laying off instead, especially if they hate to see employees whining in the companies all day long.(and of course, they don't need to cut their own pay this way.)

The point above is certainly more general and is a special case of the so-called principal/agent problem in economics. The agent (the HROs) maximize their own utility, not that of the principal's (the firms.) As a result, the overall performance is not very efficient. One important example in stock market is that fund managers might be overconcerned with their own job security and avoid taking proper risks. (This, in my opinion, might be the biggest reason why market isn't efficient). 

My bigger objection to Bewley's conclusion is that managers aren't really rational, and there are two separate reasons for it. First, some managers are just incompetent. More than once in the interviews some managers mention that companies aren't very careful or disciplined in booms. I take this as managers become lazy or too optimistic in good times, which is probably true to most people of varying degrees. In this sense, the wage policy can hardly be optimal.

Second, most managers design their policies according to some kind of conventional wisdom. For example, to encourage the loyalties of the workers, they might want to treat the workers as family members, and this might be done at huge cost. In doing so, instead of maximizing the net value of the company, the managers are maximizing the happiness of their employees, which is an adorable act, but hardly efficient.

In fact, I tend to think it is exactly this kind of obediance to conventional wisdom that makes most firms mediocre. Great firms are those that understand what the right objectives are and are willing to discard the conventional wisdom. To this, I will have two reviews coming up in the next week one on a baseball team and the other on GE.

This book has received ravish review both from the nytimes and the economist magazine. The only problem, both review mentioned, is that the name of the book is not serious enough. I agree with the comments. I think the name sounds freakish (but jienan believes that economists are freaks anyway. :) ). But the rest of the books are truly treasures deserve careful hunting.

The book is composed of six separate independent chapters. Each chapter more or less tells an interesting story based more on some paper that Steve Levitt, the main author of the book wrote.

Steve Levitt, a fellow MIT phd and now a professor at University of Chicago, made his name in 2001 when he coauthored a paper with Dohohue at stanford.(This is also one chapter in the book.) Their paper propose an interesting, reasonable, but highly controversial theory to explain why the crime rate ( especially of violent crimes) has dropped in the U.S. in the 1990s.

The typical theories of why the crime rate has dropped in the 1990s include 1: a strong economy, 2: increased police force, 3: tougher gun control , 4: innovative policing strategies (espcially in NYC under Mayor Guilliani), 5: increased reliance on prisons, 6: decrease in the crack and drug markets, and etc.

I have to say that I find most of the explanations above quite reasonable (except possibly 3 and 4), yet it turns out that other than 5 and 6, the rest of factors have virtually no effect on the crime rates. One common theme of all these explainations is that they use sth happened recently(strong economy, more policeman and so on) to explain sth that happens recently(drop of crime rate). Levitt's theory goes back 20 years back, and his answer is: Roe vs. Wade.

Roe vs Wade might be the single most well-known case in the U.S. history: it legalizes abortion. Without legal abortion, many women who don't want to give birth will have to deliver their children reluctantly. These children have a high risk of becoming a future criminal, particularly if the mother is black, a teenager, or have low income. With abortion, women have a choice and can abort the unwanted children. These abortions in turn cut the supplies of potential future criminals.

Levitt and Donohue did some further checks for the validity of their story. For example, some states legalized abortion earlier than the rest of the country(Roe and Wade is a decision by the supreme court, which made abortion legal in the entire country. Before that, several states like new york and hawaii have already legalized it. Just like Massachusetts legalizes gay marriage earlier than the rest of the country(if they ever do)) It turns out the earlier was the abortion legal in a state, the earlier did the crime rate drop in that state. And the time difference in crime drops roughly equals the time difference in legalling abortion. Furthermore, the higher was the abortion rate immediately after abortion became legal in a state, the larger the drop in crime rate in that state in the 90s.

Simple and reasonable as it is, this story is the source of huge controversies. In fact, nearly everyone hates it. The conservatives hate it because the paper seems to suggest abortion is good. The liberals hate it because the story suggests that poor and black women can be the source of future crimes. The only group I know of that likes it is economists. To us, the story has a bigger theme: sad as it is, reality often does not conform to ideologies. It is the task of economists, or social scientists in general, to discover how the society functions, not how it ought to function.

This is but one of the six stories in the book, and in fact the rest of the book have many fascinating stories. I list below a few of them (questions and facts, no answers) and recommend everyone to read the book.

1: how do we catch school cheating and athlete cheating?

For promotion and pecuniary reasons, many teachers actually change the answers of their students in standarized exams. How might we catch them? In sports, can we test whether the sumo wrestlers (xiang4 pu1) are corrupted?

In fact, i have long wanted to test whether the referees in chinese soccer leagues are corrupt! Probably I will write a paper sometime.

2: Why do drug dealers live with their moms?

Guess what's the salary of a typical drug dealer? Hint: it's not very high. How are the drug dealers organized? Hint: look at MacDonalds and Burger King. Why do people want to become drug dealers? And finally, how do economists get to know so much about drug dealing???

3: Do names matter?

Some guy in new jersey named one of his sons "winner", and another, yes!, "loser". How did these two do? In california during the 90s, eight babies are named as "harvard", 15 as yale, 18 as princeton, 3 as lawyer, 9 as judge, and two as president. It will be fun to find out how they turn out.

Last but not least, to Pin-Xiang: Happy Birthday(one day late :)) ! BTW, Benjamin is one the most popular names from a high-end (rich well-educated) family recently and is likely to be more popular in the future.

Many people have New Year Resolutions, but rumor says that few accomplish theirs. I am going to make an end of semester resolution now, and it is probably even less likely to be realized. But one resolution is better than none, and by writing it down here may actually help discipline me.

My resolution is to read at least one book a week and document my thoughts about the book on my blog.

In the past, I have started many books, but interests often waned after 100-200 pages or so. Since my new year resolution this year is to be more persistent, I actually have finished more than usually number of books this year. But the frequency that this happened is still way too low, so hopefully this resolution help to speed things up. To be scientific (and to satisfy my own little vanity), I list below the books I have read so far this year, and we will see how many I will have read by the end of the year.

1: Biography of Bob Rubin. 2: Peddling Prosperity. 3: Liar's Poker. 4: One L. 5: The twenty years' crisis.