We see we have carried out an "experiment" in reducing welfare benefits: since 1970 AFDC payments, in constant dollars, sharply declined by 33 percent. Surely this decline was moderated, but by no means offset, by in- kind payments. And what were the effects?

The number of AFDC families continued to rise, but the number of AFDC children leveled off in the mid- 1970s -- a demographic effect. So this is my difficulty with (Charles) Murray's argument in "Losing Ground": the increase in dependency in the 1970s that so concerns him continued taking place even as the value of welfare benefits declined.

Startling evidence of this comes from Mary Jo Bane and David Ellwood up at Harvard. They have now completed an exhaustive series of cross- correlations examining various family patterns and the impact of the AFDC program across the 50 states and over time. Since each state has a somewhat different program, welfare benefits vary greatly. So it is easy to relate differences in the welfare-benefit levels to variations in family structure.

They show conclusively that differences in welfare-benefits are not the primary cause of variations in family structure. Unmeasurable differences in culture, in attitudes and expectations seem to account for most of the differences in birth rates of unmarried women and in divorce and separation patterns among families with children. Welfare benefits are largely impotent as explanatory variables. . . .