What is the nature of American poverty?

Robert Samuelson is one of my favorite economics columnists:

Many middle-class families achieved large income gains in the 1990s and — despite the recession and halting recovery — have kept those gains. They’re worse because the increase in poverty in recent decades stems mainly from immigration. Until our leaders acknowledge the connection between immigration and poverty, we’ll be hamstrung in dealing with either.

Let’s examine the Census numbers. They certainly don’t indicate that, over any reasonable period, middle-class living standards have stagnated. Mostly, the middle class is getting richer. Consider: In 2003, 44 percent of U.S. households had before-tax incomes exceeding $50,000; about 15 percent had incomes of more than $100,000 (they’re included in the 44 percent). In 1990, the comparable figures were 40 percent and 10 percent. In 1980, they were 35 percent and 6 percent. All comparisons are adjusted for inflation.

True, the median household income has dropped since 1999 and is up only slightly since 1990. That’s usually taken as an indicator of what’s happened to a typical family. It isn’t. The median income is the midpoint of incomes; half of households are above, half below. The median household was once imagined as a family of Mom, Dad and two kids. But “typical” no longer exists. There are more singles, childless couples and retirees. Smaller households tend to have lower incomes. They drag down the overall median. So do more poor immigrant households.

A slightly better approach is to examine the incomes of households of similar sizes: all with, say, two people. In 2003 those households had a median income of $46,964, off about $900 from the peak year (1999) but up almost 10 percent from 1990. For four-person households, the median income in 2003 was $64,374, off about $2,200 from its peak but still up about 14 percent from 1990. Though unemployment and a decline in overtime have temporarily dented incomes, the basic trend is up.

Now look at poverty. For 2003, the Census Bureau estimated that 35.9 million Americans had incomes below the poverty line; that was about $12,000 for a two-person household and $19,000 for a four-person household. Since 2000 poverty has risen among most racial and ethnic groups. Again, that’s the recession and its aftermath.

But over longer periods, Hispanics account for most of the increase in poverty. Compared with 1990, there were actually 700,000 fewer non-Hispanic whites in poverty last year. Among blacks, the drop since 1990 is between 700,000 and 1 million, and the poverty rate — though still appallingly high — has declined from 32 percent to 24 percent. (The poverty rate measures the percentage of a group that is in poverty.) Meanwhile, the number of poor Hispanics is up by 3 million since 1990. The health insurance story is similar. Last year 13 million Hispanics lacked insurance. They’re 60 percent of the rise since 1990.

To state the obvious: Not all Hispanics are immigrants, and not all immigrants are Hispanic. Still, there’s no mystery here. If more poor and unskilled people enter the country — and have children — there will be more poverty. (The Census figures cover both legal and illegal immigrants; estimates of illegal immigrants range upward from 7 million.) About 33 percent of all immigrants (not just Hispanics) lack a high school education. The rate among native-born Americans is about 13 percent.

The bottom line: The American middle class is doing better than many commentators would have you believe. And while I don’t think we should neglect the welfare of Hispanics, in many cases the correct comparison is with their native countries, not with other U.S. statistical aggregates.

Addendum: Here is a good piece on how the federal government defines poverty.