THE EFFECT OF OLD AGE ASSISTANCE ON RETIREMENT: EMPIRICAL RESULTS 5

Additional Results

This section exploits the individual level data available in the Censuses to show that some groups of the population were more sensitive to OAA than others. It may not be obvious, though, why the impact of OAA should differ across demographic groups. We do expect OAA to have a bigger impact on individuals with a lower wage. Also, because it is fully means tested, those with greater non-labor income or wealth should be less sensitive to OAA. However, with earnings and wealth data lacking for much of the sample, a feasible alternative is to allow OAA to vary with other observable characteristics which are likely to be correlated with earnings and other family resources. The resulting coefficients indicate which groups of the population responded most to rising OAA benefits. Table 8 reports the coefficients from column (4) on the demographic characteristics interacted with OAA benefits. Although many of the estimated coefficients are not individually significant because of relatively small cell sizes, some interesting patterns emerge.

The education terms are of particular interest because of the strong link between education and earnings. The expected relationship is reflected for those with twelve years of school, who were much less sensitive to OAA than those with only 0-8 years of school and, in all likelihood, poorer earnings opportunities. The coefficient on the interaction term of 0.131 (0.077), compared to the coefficient on the OAA main effect of -0.188 (0.085), implies that those with a high school education were about one-third as sensitive to OAA as those who did not attend high school. The other education coefficients are far from significant, however.

The connection to household composition is also interesting. The interaction terms indicate that someone who was not the household head, but related to the head, was insensitive to OAA. Meanwhile, those who were neither household heads nor related to the head were about as sensitive to OAA as those who were themselves household heads. This is likely to reflect the determination of benefit eligibility and levels, which was meant to take into account the support available from kin. It suggests that household status was the principal means for detecting or enforcing kin support. Furthermore, the greater sensitivity of household heads to OAA compounds the impact on participation of changes in household composition, since the participation of household heads fell from 54.0% in 1940 to 51.8% in 1950, while household headship rose from 83.7% to 85.0%. The rise in household headship further implies that the link between household structure and OAA recipiency may have raised the propensity of the elderly to maintain their own households.

TABLE 8

OAA main effect -0.188 (0.085)
9-11 years of school -0.074 Lives in central city of SMA 0.046
(0.074) (0.053)
12 years of school 0.131(0.077) Lives in SMA, not in central city 0.028(0.049)
13-15 years of school -0.006(.105)
16+ years of school -0.103(0.113) Lives on a farm 0.070(0.041)
Widowed or divorced 0.016(0.048)
Never married -0.090(0.085) Foreign born 0.051(0.068)
Father or father-in-law of household head 0.301(0.063)
Other relative of household head 0.399(0.094) Nonwhite -0.402(0.062)
Not related to household head -0.001(0.085)

/tr