Employment and Earnings Outcomes

The Proportion of VR
Clients With Any Earnings
From Wages

It is noteworthy that, among those rehabilitated in all three disability
groups, quite a large fraction had earnings in the year of referral (65 to
73 percent) and an even larger percentage in the year of closure (75 to
84 percent). However, the increases were temporary; the proportion with
any earnings dropped to near or below pre-program levels within 2 years
following closure. These patterns are shown in figure 5.2.

Employment and Earnings Outcomes

Consecutive Years of

Post-Program Earnings

aYear of referral varies across clients.

"Clients' cases were closed in fiscal year 1980, which included parts of calendar years 1979 and 1980.

Source: GAO analysis of 1980 combined RSA-SSA data base

In contrast, clients who were not rehabilitated or who dropped out of the program were less likely to have earnings from employment in the years after closure from the program than in the years before referral. The declines occurred within the first 2 years after closure and then leveled off or were slightly reversed. At the end of the period that we studied, from 40 to 50 percent of those clients who were not rehabilitated or who dropped out of the program had some earnings from employment. This compared with the 61 to 66 percent of rehabilitated clients who had earnings.

The post-program patterns of sharp declines, followed by a leveling off, or a slight increase, were similar for clients who were not rehabilitated and clients who dropped out of the program. However, the proportion employed in the former group was slightly lower than in the latter.

We also examined VR clients' continuity of wage-earning after
participation. These results are presented in figure 5.3 and are quite similar
to the results we saw in the previous section. That is, in the years after
closure, a shrinking fraction of clients showed a record of uninterrupted
earnings; at the 8-year point, only about a third had earned wages each
year. Across all three disability types, clients who were rehabilitated were
much more likely to work continuously (according to this measure) than
were those in the other two groups (non-rehabilitants and dropouts)-by
about two to one.

"The measure of continuity is limited by the available data. The SSA wage information consists of a single figure for an entire year. Thus, even those with earnings may not have had steady jobs throughout the year.

Employment and Earnings Outcomes

Notes:

Clients' cases were closed in fiscal year 1980, which included parts of calendar years 1979 and 1980.

Percentages shown for each year represent those with some wages from employment in that year and all preceding years since closure.

Source: GAO analysis of 1980 combined RSA-SSA data base

Annual Earnings

Although the group earning wages shrank noticeably over the years, as
previously reported, average earnings generally rose. We found this to be
true for each of the three major disability groups and for rehabilitants,
non-rehabilitants, and dropouts as well. We first examined earnings for the
entire group of 1980 closures in the RSA-SSA file-that is, for those of
working age each year-and noted increases from 1980 to 1988 of $547 to
$1,430 (depending on the disability type) in the annual average earned by
rehabilitants. (See figure 5.4.) These averages are lowered by including in
the calculation for each year many individuals who earned no wages, so
we recalculated each year's average excluding those with no wages-with
the results shown in figure 5.5. We found increases of $2,052 to $4,592
(again depending on the disability type) for rehabilitants over the 8 years.
Both analyses show that earnings drop-offs before referral to VR were
reversed following closure. And the figures also show that, while both
dropouts and non-rehabilitants had wage gains, rehabilitants did much
better (with gains at least four times as great as gains in other groups),
with the average eventually exceeding that of the before-referral period.
(Some of the other closure groups also recouped earlier wage losses.) We
saw earlier that fewer non-rehabilitants and dropouts worked than
rehabilitants; thus, more had zero earnings. However, even removing these
individuals from the analysis-as we did for figure 5.5-shows that
rehabilitants who were working earned on the average more than those in
the other groups who worked.

"Evaluations of job training programs have documented how participation is not random, and
specifically that individuals whose earnings are unusually low just prior to participation in training are
the ones most likely to enter training. This dip in earnings introduces bias in estimates of program
effects. Not all VR client groups show a pre-program earnings dip in the SSA annual wage data, but
some do. (The week-of-application earnings figure used by RSA may be more affected.) We report in a
later section of this chapter the results of the regression analyses we did to begin to deal with
pre-program factors; further work with RSA-SSA data could explore additional statistical approaches
to these problems common to non-experimental data in order to yield better estimates of the
program's net effect.