Monday, August 15, 2005
The rich get richer...
This post is a mea culpa. A year ago, I knew that the U.S. poverty rate -- the portion of people in households with incomes low enough to qualify as officially poor -- has been stuck at around 10% for the last 30 years. I also knew that real GDP per capita -- the amount of stuff that the average American consumes each year -- had just about doubled over that same time. (From $18k in 1970 to $35k in 2000; those are inflation-adjusted figures from the Economic History Services.) I assumed that the explanation for the apparent discrepancy between these two facts was simple: the U.S. government had revised the poverty threshold upward over time, as we as a society had come to expect that everyone eat lunch out, that siblings should not have to share bedrooms, and that every family requires two cars.
I was wrong. I first discovered I was wrong when I finally bothered to look up historical poverty line figures and determined that the 2003 value ($9393 for a single person) is exactly the same as the 1967 value when adjusted for inflation. The government hasn't been changing the goal-posts after all.
So how is it possible that our society has gotten twice as wealthy, but we still have just as many poor people? The answer is that, while average incomes have doubled, the lowest incomes haven't budged. The situation is illustrated by the chart below, which shows the evolution of inflation-adjusted household income for each quintile of the population separately. (This cart is generated from U.S. census bureau data.)
The lowest-earning 20% are represented by the red line, the next-lowest-earning 20% by the orange, and so on, up to the highest-earning 20% represented by the purple line. Notice that the red line is essentially flat, indicating that, as mentioned above, the incomes of the poor remain essentially unchanged. And it's not just the poor who have fared worse than the average. The incomes of the next 20% have gone up a bit, but still have nowhere near doubled. As we move up through the higher income quintiles, each one exhibits larger gains than the last. The incomes of the top 20% really did double, and if we were to plot the incomes of the top 5% or top 1%, we would see that they did even better.
It's important to understand that no mathematical necessity arising from our segregation of the population into quintiles dictates that gains must be distributed in such an unequal way. We could have doubled average income, for example, by doubling the income of each quintile; such a distribution of gains would have cut the poverty rate approximately in half. Nor does any economic necessity require that the rich get richer faster than anyone else. Although these government time-series don't go back far enough to show it, the incomes of the lowest 20% must have increased significantly before 1970. (In 1940, average income was below the current poverty line, so the lowest 20% of incomes must have been far lower, and increased dramatically in the 30 years that followed.)
A number of explanations for this sad phenomenon have been proposed. Certainly membership in trade unions, which aim to increase the bargaining power of the lower classes, has declined precipitously over the last 30 years. (Of course, this explanation begs the question of why, if unions provide such significant benefits, union membership has declined so precipitously.) Also during this time, changes in technology have increased the productivity of capital and of skilled workers far more than they have increased the productivity of unskilled workers. So the difference in the distribution of gains between the period 1940-1970 and the period 1970-2000 may reflect the different sorts of technological process that were made in the two eras.
It is worth pointing out that the U.S. "imports" a large number of poor people to replace those that climb the income ladder -- the number of generations a person's family has been in the country correlates strongly with income. Seen in this light, our constant poverty rate doesn't look so bad, since it means that we are giving new generations of immigrants the chance to better themselves as previous generations lift themselves out of poverty. Conservatives -- at least those that can stomach immigration -- tend to emphasize this effect. I have not yet attempted to calculate what proportion of poverty it could explain. While I do not doubt that this effect does account for a part of the poverty rate, it does not seem to offer any explanation of the difference in the evolution of poverty between the 1940-1970 and 1970-2000 eras.
Progressives, on the other hand, tend to view the contrast between our growing wealth and our unchanging poverty rate as an indictment of the system. But the historical record should give them pause. The late 1960s, which separate our two eras, mark the birth of the war on poverty and the ensuing great society programs intended to reduce the poverty rate. While one could argue that the poverty rate today would have been even worse in the absence of these programs, the prima facie interpretation of the facts before us would be that these programs have failed. In fact, the controversial sociologist Charles Murray, in his book Loosing Ground, makes an occasionally persuasive argument that at least some aspects of these programs incentivized the poor to take decisions that tended to perpetuate their poverty.
Clearly this is an important riddle, and I hope to come back to it as I uncover and analyze more data.
I was wrong. I first discovered I was wrong when I finally bothered to look up historical poverty line figures and determined that the 2003 value ($9393 for a single person) is exactly the same as the 1967 value when adjusted for inflation. The government hasn't been changing the goal-posts after all.
So how is it possible that our society has gotten twice as wealthy, but we still have just as many poor people? The answer is that, while average incomes have doubled, the lowest incomes haven't budged. The situation is illustrated by the chart below, which shows the evolution of inflation-adjusted household income for each quintile of the population separately. (This cart is generated from U.S. census bureau data.)
The lowest-earning 20% are represented by the red line, the next-lowest-earning 20% by the orange, and so on, up to the highest-earning 20% represented by the purple line. Notice that the red line is essentially flat, indicating that, as mentioned above, the incomes of the poor remain essentially unchanged. And it's not just the poor who have fared worse than the average. The incomes of the next 20% have gone up a bit, but still have nowhere near doubled. As we move up through the higher income quintiles, each one exhibits larger gains than the last. The incomes of the top 20% really did double, and if we were to plot the incomes of the top 5% or top 1%, we would see that they did even better.
It's important to understand that no mathematical necessity arising from our segregation of the population into quintiles dictates that gains must be distributed in such an unequal way. We could have doubled average income, for example, by doubling the income of each quintile; such a distribution of gains would have cut the poverty rate approximately in half. Nor does any economic necessity require that the rich get richer faster than anyone else. Although these government time-series don't go back far enough to show it, the incomes of the lowest 20% must have increased significantly before 1970. (In 1940, average income was below the current poverty line, so the lowest 20% of incomes must have been far lower, and increased dramatically in the 30 years that followed.)
A number of explanations for this sad phenomenon have been proposed. Certainly membership in trade unions, which aim to increase the bargaining power of the lower classes, has declined precipitously over the last 30 years. (Of course, this explanation begs the question of why, if unions provide such significant benefits, union membership has declined so precipitously.) Also during this time, changes in technology have increased the productivity of capital and of skilled workers far more than they have increased the productivity of unskilled workers. So the difference in the distribution of gains between the period 1940-1970 and the period 1970-2000 may reflect the different sorts of technological process that were made in the two eras.
It is worth pointing out that the U.S. "imports" a large number of poor people to replace those that climb the income ladder -- the number of generations a person's family has been in the country correlates strongly with income. Seen in this light, our constant poverty rate doesn't look so bad, since it means that we are giving new generations of immigrants the chance to better themselves as previous generations lift themselves out of poverty. Conservatives -- at least those that can stomach immigration -- tend to emphasize this effect. I have not yet attempted to calculate what proportion of poverty it could explain. While I do not doubt that this effect does account for a part of the poverty rate, it does not seem to offer any explanation of the difference in the evolution of poverty between the 1940-1970 and 1970-2000 eras.
Progressives, on the other hand, tend to view the contrast between our growing wealth and our unchanging poverty rate as an indictment of the system. But the historical record should give them pause. The late 1960s, which separate our two eras, mark the birth of the war on poverty and the ensuing great society programs intended to reduce the poverty rate. While one could argue that the poverty rate today would have been even worse in the absence of these programs, the prima facie interpretation of the facts before us would be that these programs have failed. In fact, the controversial sociologist Charles Murray, in his book Loosing Ground, makes an occasionally persuasive argument that at least some aspects of these programs incentivized the poor to take decisions that tended to perpetuate their poverty.
Clearly this is an important riddle, and I hope to come back to it as I uncover and analyze more data.
