Now I had something to write about: Salomon Brothers. Wall Street had become so unhinged that it was paying recent Princeton graduates who knew nothing about money small fortunes to pretend to be experts about money. I'd stumbled into my next senior thesis.
The book I wrote was called "Liar's Poker." It sold a million copies. I was 28 years old. I had a career, a little fame, a small fortune and a new life narrative. All of a sudden people were telling me I was born to be a writer. This was absurd. Even I could see there was another, truer narrative, with luck as its theme. What were the odds of being seated at that dinner next to that Salomon Brothers lady? Of landing inside the best Wall Street firm from which to write the story of an age? Of landing in the seat with the best view of the business? Of having parents who didn't disinherit me but instead sighed and said "do it if you must?" Of having had that sense of must kindled inside me by a professor of art history at Princeton? Of having been let into Princeton in the first place?
Lewis is, of course, a talented guy, so it's unlikely he would have led a life of quiet obscurity if he had been seated elsewhere at the dinner table that night. Nevertheless, his story illustrates the role of chance in deciding how we get our start. Many or us will recognize our own version of Lewis' dinner-table moment in this story.
Here are some other connections between the earlier lotteries discussed earlier and the job-selection lottery. If you played varsity lacrosse at the University of Maryland, you will find that the U of M lacrosse alumni on Wall Street are delighted to open doors for you. This is true of most every college sport and most every firm in business or finance. And how do we become varsity athletes? Of course, we work hard, but that doesn't help if we are born with no talent for sports. And as we saw in Figure 9 (Parent Lottery), even with talent, if our parents are poor, we are less like to play sports in high school, regardless of talent.
Is there not also an element of chance in finding that the company or industry you chose for your first job was on its way up or down? Did the Silicon Valley pioneers foresee that the new technology they were fooling around with in college would spawn a group of industries with a combined market capitalization of $6 trillion by the time they reached 45? Could the people who joined Enron in 1990, when it was a second-tier gas pipeline company with a market cap of $2.6 billion, have foreseen that by 2000, it would have a market cap of £60 billion, and would have received Fortune's “Best Company to Work For” award for four straight years? Could the people who joined Enron in 2000 have foreseen that before the end of 2001, Enron would disappear? Could the people who entered law school in 2005 have foreseen that at the time they graduated three years later, firms would be contracting rapidly and getting rid of lawyers as fast as possible? Did Michael Lewis foresee that derivatives were about to revolutionize finance when he took the job at Salomon Brothers?
The accident of being in the right or wrong place at the right or wrong time at the beginning of one's career--or any time along the way--has as much to do with how well we do as hard work or judicious career planning.
Serendipitous forces play out along the road as our careers progress or fail to do so. The quality of the people who manage or mentor us varies enormously. We can be assigned to work in a division or on a deal that turns out to be career elevator up, or a side trip to nowhere. Of course it's true that with persistence, agility and opportunism (not to mention cunning and ruthlessness), we can overcome setbacks from drawing bad tickets in the job lottery. But starting a new job or getting a new boss is not something that happens to people several times every year. If it were, then this would be a lottery in which bad and good luck will tend to even out. But that is not the case.
My final observation on the job lottery is to caution against the fallacy that the best-performing people get the highest-paying jobs. What is generally true is that the highest-paying jobs are those in which people are the most productive, where productivity is measured as the dollar value of output divided by hours of time at work. However, a person's productivity is a function not only of his own effort, talent and intellect, but also of the other “factors of production” that his job puts at his disposal. The best face you can put on the ten-fold increase (1000%) increase in CEO pay relative to the average worker between 1980 and 2005 (Figure 3) is that an hour of CEO time in 2005 added ten times more value than it did in 1980, by virtue of the additional factors of production--mainly in the form of capital investment--that American companies have put at the disposal of their CEO's. It certainly did not do so because the average American CEO was 10 times smarter or worked 10 times harder in 2005 than in 1980.
Nowhere is this clearer than in the finance industry, where the highest-paying jobs are to be found. The reason for this is not that the hardest-working, most talented and most intelligent people end up in finance. The reason is that people who work in finance are, generally speaking, in the business of making bets with other people's money. Financial capital, not human capital, is the primary “factor of production” in this industry, and accounts for most of the value added by an hour of a finance worker's labor. In 1972, the ratio of capital employed to people employed in the top 50 US securities firms was $124,000. In 1987, it was $203,000, and in 2004 it was $1,789,000. (These figures are in constant 2004 dollars. [13]
So, if the average worker in finance in 2004 was paid fourteen times as much in real terms as the average finance worker in 1972, was this because he worked fourteen times harder, or was fourteen times smarter? Of course not. It was because he had fourteen times as much of other people's money to make bets with.
Sure, high earners work more hours than low earners, but not 1000, 100, or even 10 times more hours. Yet their salaries are 1000 times higher. There is a limited number of these highly productive jobs in which you get the chance to work long hours for high earnings. Few of the people who get these jobs create the high levels of productivity their jobs exhibit. In most cases, that was done progressively over time by their predecessors. For this reason, these prize jobs are the object of an intense competition, in which the contestants fight to win using their smarts, education, social skills, and job experience. This is, however, a lottery within a lottery, in which a large number of equally-qualified people struggle over a much smaller number of winning positions.
If I were to include a seventh lottery here, I would call it “Personal Life.” This is the lottery in which we connect with other people as friends or as partners. Clearly chance plays a large role here. Consider how serendipitous were the circumstances under which you met your best friend or partner. Consider also that, although most of us marry someone we are convinced is Mr. or Ms Right, in hindsight we usually realize that we had no clue whether this would turn out to be true. Finally, consider how much difference the right or wrong choice of partner makes to how well we function in the world. This lottery may therefore be as important as any of the others in determining how well we do in life.
Why We Have What We Have
In discussing the lotteries, I have stated repeatedly that these accidents “contribute significantly,” “strongly predict,” “are very important in explaining,” and so forth. Such phrases beg this question: what is the relative strength of the accidental and the intentional in determining what we have? Is it 90/10, 50/50, or 10/90? If we can't quantify the answer, then the bottom line of the six lotteries would be “so what?” and I wouldn't have bothered to write this. But if we can answer “Why do we have what we have” quantitatively, this will provide firm ground on which to discuss how government should take what it needs for charity, which is the goal of the exercise.
The lotteries discussion has suggested a number of ways in which chance affects lifetime earnings prospects: gender, IQ, looks, smarts, parents, health, and unpredictable, unintentional, but life-changing things that may happen at school, on the job, or in our personal lives. I will argue that two of these variables--gender and parents--explain (in the statistical sense) at least ha
lf of the observed variation in the population's lifetime earnings.
We begin with gender. In our discussion of the Education Lottery, we saw how, in order to understand the importance of educational attainment in explaining earnings, it is not enough to look at average values (as in Figure 4.1 and Figure 4.2). We also had to look at the variation of each individual's earnings around the median for his education class (as in Figure 10). The same is true for gender.
Figure 4.2 tells us that for people who work full-time for forty years, within any education class, the average woman will earn only 75% as much as the average man. But there is no such thing as an average woman or man. Figure 12 fills in this picture by showing the variation of individual men and women's earnings around the gender medians. [14] The median value of lifetime earnings for men is $1.4 million, while for women the median is $750,000.
Figure 12
Now that we know where these distributions lie, we can predict the likelihood that a woman selected at random will earn more than the “average person” (the population median), or more than a man selected at random. It turns out that the random woman has 40% chance of earning more than the population median, and a 30% chance of earning more than the random man.
The gender variable clearly has some "explanatory power" in predicting what you or I can expect to earn during our lifetime. Statisticians measure explanatory power with something they call the "coefficient of determination," usually referred to as “R2”. For the relationship between gender and earnings, the correlation is 0.22. [15] Since, in statistics, “explanation” is the square of correlation, this gives a value for R2 of 0.10, meaning that gender “explains” around 5% of the observed variation in the lifetime earnings of individuals above or below the population median of $1.1 million.
So what explains the other 95% of the variation in the earnings of individuals? Naturally, it is a combination of things. There are intentional factors, which include, in addition to hard work, such things as ambition, persistence, self-discipline, risk-taking, aggression, cunning, deception, and ruthlessness. In the academic literature that I have looked at, I found hardly any studies that included these "intentional" variables as predictors of economic outcomes. This is not because intention is not important--it obviously is--but because it is hard to measure in a way that can be coded into a data set and included as an independent variable in multiple regression analysis.
There are, however, other variables that have an effect on earnings and can be measured, allowing calculation of their R2. One of these is parents' “socio-economic status” (education and income), which turns out to be a powerful “predictor” of children's income, much more powerful than gender.
Quite a lot of research has been done on the “inter-generational transmission of socio-economic status.” Sorting out the relative importance of the genetic and the environmental factors that transmit economic status from one generation to the next is important to the public policy debate because, if it's mostly all genetic, as Charles Murray argues, there's not much that public policy can do to reduce economic inequality. For our purposes, however, it doesn't matter whether genes or environment predominate, because both forms of transmission are entirely accident from the child's point of view. The child chooses neither her genes nor her parents.
The best research summary I could find is a collection of papers published under the title “Unequal Chances: Family Background and Economic Success”. The key paper is “The Apple Falls Even Closer to the Tree Than We Thought,” in which Bhashkar Mazumder reports that when incomes are measured over sixteen years, the “intergenerational income elasticity” reaches 0.6. Under certain assumptions, this would mean that the correlation is also 0.62, in which case parents' incomes would “explain” 36% of children's incomes (explanation being the square of correlation). [16]
Another paper in Unequal Chances, authored by Thomas Hertz, is titled “Rags, Riches and Race: The Intergenerational Economic Mobility of Black and White Families in the United States.” Among other things, Hertz focuses on the problem of data quality. The measured correlation between parents' and children's income gets stronger as the time period over which incomes are measured gets longer. This is not surprising, since the effect of transitory factors-which create “noise” in the data-diminishes as longer periods are considered. Hertz reports that when incomes are measured are over seven years (as opposed to sixteen), the correlation drops to 0.45. We can only speculate on how high the correlation would be if incomes were measured over thirty or forty years. Unfortunately, the challenge of obtaining data of sufficient quantity and quality over such a long period for both generations is too great.
What does this correlation tell us about causality? If there is a correlation between A and B, three things are possible. A could cause B, B could cause A, or C could cause both A and B. It's obvious that children's incomes don't cause their parents' incomes. But the ways in which parents' incomes might “cause” their children's incomes are not entirely clear.
In Parents Lottery we delineated the primary “vectors of transmission” by which this might come about. For example, race and parents' education “explain” a significant part of the relationship between parents' earnings and children's earnings. In the language of multiple regression analysis, when race and parents' education are introduced directly into the equation as independent (explanatory) variables, along with parents' income, the coefficient for parents' income drops significantly.
The same might also be true if we introduced parents' IQ, parents' health, or the traits in parents' personalities that make them good or bad performers at work. The direct transmission of these underlying parental variables to their children, rather than the things that parents' income itself can buy, may explain most of the “inter-generational transmission of economic status.” Since these parent variables also explain why parents' incomes are what they are, we will observe a strong correlation between parents' income and children's income.
For our purposes, however, it doesn't matter whether the relationship between parents' income and children's income is direct, or derived from a direct relationship between parents' race, IQ, education, health, or personality and their children's income. All these underlying explanatory parental variables are accidents that happen to children in the birth lotteries.
The question of inter-generational transmission of economic status is closely related to the question of intergenerational mobility. Table 5.10 in the Thomas Hertz paper presents a “transition matrix” showing the probability that children born to parents in a given income bracket will end up in the same or a different bracket. The matrix is thus a convenient way to summarize the degree of “intergenerational income mobility” in our society. Table 1 is a simplified version of Hertz's transition matrix.
Table 1
Table 1 tells us that if your parents' earnings were in the bottom 20% of the population, there is 48% chance that you will end up there, too, and only a 4% chance you will end up in the top 20%. If your parents were born in the top 20%, there is a 39% chance that you will end up there, and a 6% chance you will end up at the bottom. The more detailed matrix in the Hertz study shows that the child of parents earning less than the median has a 34% chance of earning more than the median.
In a perfectly mobile society, the transition matrix would look like Table 2. There is no correlation at all between your income and your parents’ income, and your chance of ending up in a given bracket bears no relation to your parents’ bracket
Table 2
In a perfectly immobile society, the matrix would look like Table 3. Your income and your parents' income are perfectly correlated, and no-one escapes his parents' bracket.
Table 3
Hertz's transition matrix, we can construct Figure 13, showing the distribution of lifetime earnings for people whose parents' earnings were below the median (“poor children”), and the distribution for people whose parents' earnings were above
the median (“rich children”). [17] However, since Hertz's matrix is based on data measuring parents' and children's incomes over only 11 years on average, and since, as mentioned, the measured correlation gets stronger as incomes are measured over longer periods, Hertz's matrix understates the true degree of inter-generational immobility, and the “rich children” and “poor children” distributions in Figure 13 would not overlap as much as they do if Hertz's data had measured incomes over longer periods.
Figure 13
Now, if gender ”explains” around 10% of our lifetime earnings and parents' income explains at least 38%, does that means that these two variables together explain around 43%? The answer is yes, but only if there is no meaningful correlation between gender and parents' income. If there were a correlation, it would not be appropriate to add the two R2 because this would be double-counting. If we were talking about India or China, where “gender-planning” is widely practiced, it would not be unreasonable to expect a degree of correlation between parents' income and children's gender. However, since gender-planning is not prevalent in the USA, we may safely add the R2 of the two variables without fear of double-counting. We can therefore conclude that at least 43% of our earnings are determined by our gender and our parents' income.
Remember also that gender and parents' income are only two of the accidents we identified in the Six Lotteries that influence our lifetime earnings. Only a portion of our IQ is inherited from our parents (estimates indicate it is around 50%), and the rest is an accident having nothing to do with parents. The same is true for health, looks, athletic or artistic talent-much of what we have is genetic, and much is not. The part that is not is an accident that has nothing to do with parents. Then there are the accidents that happen in the school and college admissions lotteries, the jobs lottery, and the partner selection lotteries. The outcomes are all to some degree independent of the parents accident.
The Grasshopper and the Ant, or the Beautiful and the Damned? Why We Have What We Have, and How Government Should Take What it Needs for Charity Page 4