When we talk about the economy, we hear it said over and over: things have radically changed since the days of the mass-production economy. Upward mobility is more difficult, wages of people with ordinary middle-class qualifications are in decline.

But when we talk about immigration, we are supposed to forget all those things, and to believe that patterns that held true between 1913 and 1963 will hold equally true over the next half century.

David Leonhardt in the New York Times on Sunday:

“These fears about immigrants have been voiced many times in American history, and they’ve never proven true,” Alan M. Kraut, a history professor at American University, in Washington, told me. “It doesn’t happen immediately, but everything with Latinos points to a very typical pattern of integration in American life in a generation or two.”

But there is no "pattern"! There is only the experience of a prior historical period - and the mystery of the extent to which that experience will be applicable in a very different period.

FOR decades, the average Latino immigrant has had slightly more than a junior-high school education. An average child of a Latino immigrant today completes high school and attends almost one year of college. A typical grandchild attends more college, Mr. Smith found. In the last decade alone, according to the Pew study, the number of Latinos graduating from college has roughly doubled, to more than 250,000.

To put that latter number in context: about 1.8 million Americans complete a bachelor's degree every year. Another 937,000 complete a 2-year associate's degree. Leonhardt's 250,000 graduates stat refers to both groups. Do the arithmetic, and you see that the 23% of young Americans who are Latino earn only about 9% of the country's associate diplomas and bachelor degrees. That does not portend "catch up" anytime soon.

Indeed, there is good reason to fear that we will see retrogression in third-generation Hispanic immigrants. In their study of Mexican-American immigrants, Generations of Exclusion, Edward Telles and Vilma Ortiz found that educational progress slowed to a halt after the second generation.

One reason for that slow-down is another pattern that Leonhardt implausibly cites as good news: the disintegration of Latino family structures on contact with American realities.

Even one alarming trend among the children of Latino immigrants highlights their increased American-ness: younger Latinos are having more children outside marriage than their parents did, just as whites and African-Americans are.

Right. And among whites and African Americans, non-marital childbearing is associated with wage stagnation and downward economic mobility. Latinos are unlikely to be spared those same dire outcomes.

The United States was once a country of dramatic class mobility. It was once a country where ordinary educational attainments plus a strong work ethic could achieve a solid middle-class lifestyle. It was once a country where high-school graduates were just as likely to sustain stable families as college graduates.

It is not that country any more.

Yet the immigration debate proceeds in blithe disregard as if nothing has changed since the days of our grandfathers and grandmothers - and ironically, it is those writers who (in other contexts) are most aware of the way the US economy has changed who most insist on forgetting everything they know when it comes time to write about immigration.