Former welfare recipients have fared differentially in the 1990s. Some are now earning high incomes while others are still below poverty, despite work efforts. Still others have no recorded work involvement. We have attempted to learn why. The vast majority of research studies to date on the employment and earnings outcomes of former welfare recipients have focused exclusively on the characteristics of the women recipients. Much of the analysis of earnings among all workers has also focused on characteristics of the workers. This study is an exception: it focuses on the characteristics of the employers that chose to hire these former recipients. It asks two questions. One is that of the overall importance of employer characteristics in determining employment and earnings outcomes. The second is that of the relative roles of various employer characteristics in determining earnings outcomes. It is the first comprehensive, longitudinal examination of the topic.

The study utilizes a special database created by the Wisconsin Department of Workforce Development. To make the study manageable, we analyze the work and earnings experience of the 96,000 women who were on Aid to Families with Dependent Children (AFDC) in Wisconsin sometime in 1990 and who worked in Wisconsin in employment covered by Unemployment Insurance sometime between January 1, 1990 and December 31, 1998. We choose to examine this one cohort because of its size and the nine years of experience it presents. It traverses a period from traditional welfare through the start of W-2.

The answer to the first question on the relative role of the characteristics of employers on employment and earnings of these former recipients is that employer characteristics by themselves are important, but not nearly as important as the workforce commitment of the women and the characteristics of the women.

The answer to the second question is more complicated. Earnings are influenced by the number of quarters a woman works and the characteristics of the employer(s) for whom she works. The most critical characteristics of the employers appear to be the average earnings per worker with that employer and the industry in which the employer is located. Some industries, such as manufacturing and construction, should be sought; others, such as retail, should be avoided. But what are also influential are the proportion of former welfare recipients on the payroll and the turnover rate of employees. The higher these ratios, the lower the average earnings of the former recipients on their payrolls. The list below contains the highlights of a number of characteristics that influence the earnings and employment outcomes of these former recipients. Appearing first, however, are some basic numbers describing the welfare population:

. • Over 96,000 women in Wisconsin were on AFDC in 1990. By 2000, that number had dropped to fewer than 8,000

. • Most of the 96,000 women recipients worked sometime during the 1990s. In fact, 88% of them were employed in at least one quarter.

• In 1998, the latest year for which we have data, over 62,000 of these women (65%) worked in "covered" employment in Wisconsin. Where the remaining 34,000 were in 1998 remains to be explored.

• Of all those who did work in any given year, an increasing percentage worked in all four quarters. That percentage grew from 37% in 1990 to 71% in 1998.

• Average earnings per woman did increase (in constant dollars) over the nine years: from $4,865 in 1990 to $12,787 in 1998. That average earnings figure is just below the poverty line for a family of three ($13,133 in 1998). Contributing to the increase were twice the number of women working in all four quarters and additional work experience.

• Those women who worked all four quarters in 1998 earned, on average, almost enough ($16,389) to put a family of four over the poverty level ($16,588 in 1998). With the use of Earned Income Tax Credits (EITCs), their incomes were over the poverty level.

• Those women who committed to only three quarters of work in 1998 earned $10,000 less ($6,295), on average, illustrating the importance of the commitment to work to earnings success.

• Similar to the overall workforce, these women were concentrated in three industries: Services (49%), Manufacturing (17%), and Retail (19%). But former recipients are over-concentrated in Services and underrepresented in Manufacturing. The distribution of all workers in Wisconsin in March of 1999 was: Services (25%), Manufacturing (23%), and Retail (18%).

• Despite the overrepresentation in Services, these women in 1998 were not highly concentrated in particular industries. Five percent or fewer worked in Eating and Drinking establishments, Nursing Homes, or Temporary Help Services, the three most common, specific industries.

• The average earnings per worker varied with the size of their employer on their "main" job, the job on which they earned the most income. In 1998, those who worked for employers with fewer than 20 employees earned an average of $8,668. Those who worked for employers with at least 500 employees earned an average of $12,525 on their main job.

• Those women who worked four quarters in 1998 in Public Administration ($21,042), Construction ($18,468), or Manufacturing ($18,422) earned far more on their main job than those women who worked in Retail Trade ($10,338), Agriculture and Mining ($11,248), or Services ($13,436)

• Of the 62,605 women who worked in 1998, some 37% were able to earn more than $15,000. Those who fell short fell far short, with average earnings of $5,643. Those who exceeded the $15,000 threshold did so convincingly, with average earnings of $22,868

• Those who earned over $15,000 were much more likely to have worked for employers that, on average, paid their employees much more ($26,836, on average) than those that employed women who earned less than $15,000 (that paid $12,512, on average).

• Another possible explanation for low earnings is that the women started in the "wrong" industry, that is they started their careers in "dead end" jobs. These jobs are generally thought to be in such entities as Retail Sales, Temporary Help, and Health Services, such as in a nursing home or the bottom of the ladder in hospitals. Some 23,136 (56% of those working in 1998) of these women did start here. But only 17% of those who worked in both 1990 and 1998 were in one of these industries in 1998. These women did move on to other industries; the jobs were not "dead end."

• On the other hand, those who started in these "wrong" industries and who worked four quarters in 1998 were at a disadvantage: their average earnings were just $14,795 compared to the $16,389 earned by all women who worked four quarters in 1998. Those who started in some industries (hospitals, for example) were at no disadvantage, but others in certain industries, especially those starting in Temporary Help and Eating and Drinking places, earned less in 1998

• The differential in earnings among women who were on welfare is, in part, attributable to the characteristics of the employers for whom they choose to work. But that choice of employer is not a one-way street: it is a joint decision by both employer and employee. Thus, earnings are partly a worker’s decision in terms of one’s commitment to work, and partly an employer’s decision to hire a given employee.