Recently, economic historians have stumbled upon some tendencies or "constants" of social welfare in the United States that span more than 150 years. From the 1820s to the 1990s, the length of time families received private or public assistance was roughly constant; in the 1880s and 1890s and from the 1960s to the 1990s, the percentage of households making the transition from welfare to higher paying jobs was relatively constant; and from 1850 to the 1990s, total spending for private and public welfare as a percentage of average wages was also very stable. The lack of movement in key economic indicators of social welfare suggests that the cost of changing social institutions is rather high.
The Search for Constants
That an economist would search for constants in the economic universe is no less plausible than a physicist searching for constants in the physical universe. It is true that the usefulness and the duration of the constants, economic versus physical, may vary. For example, the manner in which the constant can be defined by the scientist or constituted by the environment is going to have implications for the use value and the duration of the observed constant. The bias points favorably toward the physical. But the economic universe will produce a curiously stubborn constant every now and again. Labor's share of output is probably the leading example.
Obviously, not every finding is going to be a Heisenberg principle, with a limit like Planck's h. Indeed, using synonymously the terms "law" and "constant" and the terms "regular" and "tendency," Lawrence Klein ( 1997: 17) argued in his essay on "Some Laws of Economics" that the term "law" should not be attached to Arthur Okun's finding that a 1 percent increase in cyclical unemployment corresponded tightly with a 3 percent shortfall in actual from potential output during the 1950s and 1960s. Unlike the Heisenberg principle, or the law of gravity, Okun's so-called "law of 3" was of short duration because the empirical regularity it sought to describe was constituted by stable patterns of productivity and of female labor force participation rates that changed in the 1970s. By anyone's definition, Okun's "constant" is no longer perceived to be 3. Nevertheless, his insights did help us understand the economy somewhat better, and that is reason enough to search for constants.
Some "Constants" of Social Welfare
Sometimes the discovery of a constant comes less from conscious searching than from unconscious stumbling, as was the case with Okun's law. I suggest here that economic historians have stumbled upon some constants of social welfare in the history of the United States. The facts are from archival collections and old census manuscripts, which helps to explain why they have not been widely publicized. Enough evidence has been amassed to generalize several key indicators:
* Between the 1820s and the 1990s, the length of time that a household received assistance from public or private sources was essentially the same (Hannon 1984; Kiesling and Margo 1997; Ziliak 1996, 1997; Blank 1989; Borjas and Hilton 1996). The figure lies between 8 and 13 months.
* In the 1880s and 1890s and throughout the period between the 1960s and the 1990s, the percentage of households exiting relief rolls with higher wage jobs is found to be relatively constant (Ziliak 1996, 1997; Blank 1989; Bane and Ellwood 1994; Danziger and Kossoudji 1995). Apparently 36 to 40 percent of all households leave the rolls for higher wage jobs.
* Between 1850 and the 1990s, the generosity of the middle and upper classes to America's poor was hardly variant (Lebergott 1976: 60; Hannon and Ziliak 2001). Over the long sweep of American history, expenditures of public assistance to the poor have been worth about 25 to 30 percent of the average wages of common labor. Moreover, this measure of generosity and its magnitude is not challenged by the years leading up to or directly following the 1996 reform of welfare. …