Here's a riddle for you.
Andy and Brett both started work at the same age. They both earned the same starting salary. They both got a $1,000 raise every year. Both are still alive and still work for the same company they started with, and have no additional income. But Andy earned twice as much as Brett last year. How is this possible?
If you are confounded by this puzzle, don't despair. You fall into the same trap that intelligent and even Nobel-prize winning economists fall into all the time. Hint: Brett is following in Andy's footsteps. Give up? Andy is 60 years old and Brett is 30.
The obvious fact that annual income statistics will necessarily include people who are young and old, inexperienced and experienced, part-time and full-time, is routinely ignored or dismissed in the one area of economics where differences in income among people are the primary focus: studies of economic (in)equality.
Consider another mental exercise.
Everyone in society E starts work at age twenty with a salary of $20,000. Everyone receives a raise of $1,000 each year (so that, their annual salary is always their age times $1,000). They retire at age 60, earning $60,000 in their final working year. In total, each and every person earns $1,640,000 in their working lives.
Economically, is E an equal society? One could argue that this society is by definition equal. Let us also assume that there are an equal number of people in this society of each working age (one million people of age twenty, one million of age twenty-one, etc.). Now, if in any given year we gather the annual income statistics, we shall find disparities of income: one million people earn $20,000 while another million earn $60,000, with many levels in between. Using annual income statistics without separating for age gives the appearance of an unequal society. The Gini coefficient (a summary statistic used extensively in economic equality studies, where 0 is perfect equality and 1 is perfect inequality) is 0.171. But in our fictitious society of equals this is terribly misleading. Yes, there is income inequality, but it is because people are at different stages of their lives and careers, not because of any permanent or chronic economic disparities among them. If we measured income on a lifetime basis, rather than (the somewhat arbitrary) annual basis, the Gini coefficient would be zero.
Of course, some of the inequality measured by annual income statistics is due to differences among people apart from age. Some people make more money over their lifetime than others. Even though lifetime income differences may be deserved, or may be the result of choices these individuals intentionally make, it is this kind of disparity that concerns most people.
What if a researcher told you that there were vast differences of height among people in America? Would you be thinking, "Oh, that's because babies are short and adults are taller"? Probably not. A normal person would think that fully grown people had height disparities. Would you expect that the researcher had limited the data to adults? You probably would. And the researcher probably would. Yet this is the same sort of problem involved with annual income data. When economists make claims about inequality based on data that is not normalized for age, they are either being incompetent or deceptive. They are making people believe there are differences between the skilled and unskilled, or the lucky and unlucky, or the "rich" and the "poor," when much of these differences are between the old and the young—differences that most everyone will experience.
How much of measured inequality is due to age and how much is due to real differences in lifetime earnings? American economist Morton Paglin tried to figure this out. In a ground-breaking journal article, he took the standard inequality summary statistic, the Gini coefficient, and broke it into two parts: (1) the "Age-Gini," which measured income differences attributable to age, what he called "intrafamily" inequality; and (2) the "Paglin-Gini," which measured real income differences among people, what he called "interfamily" inequality. He suggested that use of cross-section data such as annual income statistics should be broken down in this way to separate innocuous inequality due to annualized statistics from true inequality which reflects differences among people apart from age. Further, he found that "estimates of inequality have been overstated by 50 percent" and concluded that "the overstatement of inequality has lent false urgency to the demand for rectification of our income distribution."
That article was published in 1975. Paglin's method was subjected to criticism, both from a technical standpoint and a conceptual one. There was concern that the method did not make the split cleanly enough, and critics said that in any case inequality in the broader sense (inclusive of differences due to age) was what economists wanted to measure. The dispute between Paglin and his critics was never resolved. As a result, through confusion and willful ignorance, Paglin's method was shelved by most economists who measure inequality, though it does reappear from time to time.
Some might say that the absolute size of inequality, as measured by statistics like the Gini coefficient, is not really at issue. Rather, how the Gini of one country compares to another, or the increase in the Gini in recent years, is the real story. Yet here we have a fallacy of division. Increases in overall annual income inequality do not necessarily imply an increase in true income inequality alone, or an increase in both age-income inequality and true income inequality of equal or similar degrees. It is possible that much of the increase in measured income inequality in recent years is due to age-inequality, i.e. that the ratio of average income of people in middle-age over income of people in their early 20s has been increasing. In fact, it is possible that age-inequality is responsible for all of the increased inequality, and that true inequality has not changed at all. Or that true inequality has actually decreased, and age-inequality increases have been more than total measured inequality increases.
Not only is age-related inequality innocuous, but we might even consider it good. I believe that the differences between these two components of inequality are so distinct that we might call the intra-family (age-related) component "good inequality" and the inter-family component "bad inequality."
Why is "good inequality" good? First, it his how we as individuals get richer, even if society on average does not. Consider again society E above. Would anyone in society E bemoan the development (e.g., through a technological breakthrough, or a better work ethic) of everyone getting a $2,000 raise each year instead of $1,000? And yet, the $2,000 raise is all age-related inequality that increases inequality measured by annual income statistics. Inequality, as measured today, will never be meaningfully reduced or eliminated unless people stop earning more. To achieve equality, as it is commonly defined and measured, there can be no raises, no higher pay for experience or seniority, no personal growth. That entry-level pay you got at your first job at Wendy's is all you can ever aspire to. Second, numerous studies have shown that people derive happiness from personal growth, and specifically income growth. But growth and inequality go hand-in-hand. In a society of differently-aged people, you can not have income growth without inequality. In a growing economy, the more inequality of annual incomes there is, the more growth there is. Ergo, to have happiness there needs to be annual income inequality due to age, i.e. good inequality.
Why is "bad inequality" bad? To the extent that it occurs, some may say that it simply reflects different efforts and abilities, that it is deserved. That may be true, or it may not. But it is hardly arguable that the differences in people's abilities or efforts that cause these inequalities are not lamentable. Some may argue that we should attack the causes of these inequalities, but not the symptom of inequality itself, while others may argue that we should redistribute income to ameliorate the symptom. But this kind of inequality is certainly the inequality that people have in mind when the topic is discussed, and it is this kind of permanent class inequality that is bemoaned, justly or not. Therefore, inequality among lifetime incomes may be considered a "bad inequality."
There is good reason to think that a large portion of measured inequality increases are due to age-inequality. One approach is to use age-cohort income data to see if income differences among age groups have been increasing. And they have. Consider the graph at the top of this column constructed from US Census CPS income data. The figures are in constant 2012 dollars. I included the entire date range (1974-2012) of the Census dataset, but let's look at the generally alleged time period of inequality expansion, 1980-2010. The 15-24 age cohort's average income has stayed essentially flat over this period ($14,265 to $14,305, an increase of 0.3%). But the more aged (and thus more experienced) wage-earners have seen substantial income growth: 25-34 saw 14.2%, 35-44 got 25.6%, 45-54 got 29.1%, and 55-64 got 41.9%. (These were the increases, over 30 years, of the average income for the age cohort, not for incomes of people within the cohort; those were much larger. For instance, people in the 15-24 cohort in 1980 would be in the 45-54 cohort in 2010, so the average income of those actual people increased from $14,265 to $51,712, an increase of 162.5%. That is some perspective for the "income stagnation" thesis. More on that in a later column.) What this shows, then, is that as the U.S. economy grows, people at the beginning of their working lives, with entry level experience, can expect the same pay for their unskilled jobs as people in that same position thirty years ago; and since it is entry level work, this is hardly surprising. But to the extent that people have more job experience, and are closer to the end of their careers, they have seen real rewards to their experience increase by a much greater amount than the premium for experience received thirty years ago.
To some readers, this should cause the metaphorical light-bulb to blink over their heads. The graph also shows that the cohort plots are bunched closer together in 1980 than in 2010. This (roughly) illustrates the inequality due to age in 2010 is greater than in 1980. And this inequality appears not because any one age group is losing out. Rather, it is because as people age they are making more money at a faster rate than they did in the past.
This is pretty solid evidence that at least some (and perhaps most) of the income inequality increase over the past thirty years is age-related and not "timeless" inequality among classes or individuals per se.
The age-income relation is a real problem for using standard annual income data when discussing income inequality or when deriving inequality summary statistics like the Gini. But it is not the only problem. Economists have also pointed to problems with using "market income" rather than "disposable income", i.e. using pre-tax and pre-benefit income data. They have pointed to problems with using "households" or "families" instead of "individuals" as the unit of analysis. Then there is the problem of including part-time workers along with full-time workers. Or the misuse of statistics, such as starting with cyclical economic nadirs and ending with cyclical peaks (among other deliberate tricks). All of these are valid criticisms. But the age-income relation is perhaps the biggest, least reported, and least appreciated problem for the income inequality hysterics. And while there are some economists who may intentionally ignore it, there are others who may be ignorant of it.
There is bad inequality and good inequality. Bad inequality is the differences between people's skills and consequent earning ability that results from differences in their education, effort, and serendipity. Though these differences may be deserved, it is hard to argue that they are not regrettable. Good inequality is the opportunity for growth that results in all people earning more as they grow older with more experience and productivity. It is mathematically impossible to have income growth in a society of differently-aged people and not have greater inequality.
Progressives often use a ladder analogy to describe what they see as growing inequality: "the rungs on the ladder are getting farther apart." The analogy is not inapt, but their interpretation is. Everyone generally starts at the bottom rung of the economic ladder. But increases in annual income inequality, where they do occur, may be the result of people climbing more rungs faster than in the past. Progressives see a rise in annual income inequality and see what they want to see: an ugly story of one class in society dominating another. But what if it simply means that people are getting more rich over their lifetimes than they were before? What if it is the beautiful story of people starting off poor and becoming rich?