(FORTUNE Magazine) - I have good news and bad news. The good news is that income inequality in the U.S. -- after 30-plus years of steadily increasing -- may be decreasing. The bad news is why that trend is reversing. It looks like another lesson in how profoundly a globalizing economy is upending what we thought we knew.

Rising income inequality has settled comfortably into America's big economic picture as a reliable--and much lamented--megatrend. Starting around the late 1960s, U.S. incomes started to become more disparate. The trend was remarkably steady. Recessions might slow it down or briefly reverse it, but mostly it just marched on.

While such a large tendency has many causes, the chief explanation centered on education and skills. The late 1960s were arguably high summer of the era in which a man with 12 years of schooling could work in a unionized factory or trade and earn a solid middle-class or even upper-middle-class income. Then began the age of the info-based economy in which higher education really started to pay. The "skill premium" began growing dramatically. The college graduate's income started beating the high school graduate's income by a wider margin every year--and income inequality began to swell.

That explanation makes sense, and the data support it. But now it appears just possible--based on the latest research available--that the whole chain of causation is falling apart. Wait before you cheer.

The evidence is in a new Fed study of family finances, the latest in a triennial series. It shows modest but clear signs of incomes converging rather than diverging. Between 2001 and 2004 (the most recent year for which data are available), incomes of the poorest 20 percent of families increased while incomes of the richest 20 percent fell. Basically, the poorest families' share of total incomes grew, and the richest families' share shrank. Incomes became just a little less unequal.

What could that trend reversal mean? The most obvious explanation seems highly counterintuitive: The skill premium, the extra value of higher education, must have declined after three decades of growing. The Fed researchers didn't pursue that line of thought, but economists Lawrence Mishel and Jared Bernstein at the Economic Policy Institute did, and they found supporting evidence in the new Economic Report of the President, issued within days of the new Fed survey. It cited Census Bureau data showing that the premium had indeed fallen sharply between 2000 and 2004. The real annual earnings of college graduates actually declined 5.2 percent, while those of high school graduates, strangely enough, rose 1.6 percent.

That is so contrary to the conventional view of this major economic trend that it demands explanation. One possibility is that it's just a blip. Could be, but remember that 2004, when the readings started going haywire, was a year of strong economic growth, low unemployment, and rising productivity, offering no obvious reason to expect weird results.

The other main possibility is that something unexpected and fundamental is changing in the way the U.S. economy rewards education. We don't yet have complete data, but anyone with his eyes open can see obvious possibilities. Just maybe the jobs most threatened by outsourcing are no longer those of factory workers with a high school education, as they have been for decades, but those of college-educated desk workers.

Perhaps so many lower-skilled jobs have now left the U.S.--or have been created elsewhere to begin with--that today's high school grads are left doing jobs that cannot be easily outsourced--driving trucks, stocking shelves, building houses, and the like. So their pay is holding up.

College graduates, by contrast, look more outsourceable by the day. New studies from the Kauffman Foundation and Duke University show companies massively shifting high-skilled work--research, development, engineering, even corporate finance--from the U.S. to low-cost countries like India and China. That trend sits like an anvil on the pay of many U.S. college grads.

We need more evidence before concluding that we're at a major turning point in the value of education to American workers. But it certainly feels like one, based on what we can observe. Higher education still confers an enormous economic advantage. Just not as enormous as it used to be.

As for income inequality, pretty much everyone has always hated it, and its growth was a certain cue for handwringing and brow furrowing. Well, it's not growing anymore. Because our best-educated workers are earning less, and the incentives for higher education may thus be declining, the result could be a more uniform--and lower--standard of living. Be careful what you wish for.

FEEDBACK gcolvin@fortunemail.com