Labor Market Outcomes of Workforce Development Training Fund Participants

Occasional Paper No. 3

Workforce Development Training Fund Evaluation at the Macro

and Micro Levels

The following represents a summation of tables and figures associated with Workforce Development Training Fund participants and, in some instances, manually matched and statistical control group members (Glover, 2002; Harris, 2002a; Jones, 2004b). The analysis is based on the 3,475 participants appearing at any time during FY1999 to FY2003. Some of the 3,475 individuals appeared in multiple employer contracts. When this occurred, data from the most recent employer contract was utilized. Not all participants are utilized in all parts of the study because not all of them meet specific eligibility criteria for inclusion in various parts of the analysis (e.g., falling outside the top and bottom 2.5% of the wage distribution in the wage change analysis).

Data for this analysis come from: (a) WDTF programmatic data provided by the Department of Workforce Services, (b) historical Unemployment Insurance (UI) Wage Records data maintained by the Department of Employment, Research & Planning (R&P) which include interstate wage records information from Colorado, Idaho, Montana, Nebraska, New Mexico, Oklahoma, South Dakota, Texas, and Utah [referred to as Memorandum of Understanding (MOU) states] in the control group analyses, and (c) driver's license data from the Wyoming Department of Transportation.

Key Findings

    l     Over time, a higher percentage of older and higher wage individuals are being trained with WDTF monies.
    l     The gender distribution of participants has become more balanced.
    l     A broader range of primary industries employ WDTF participants now than in the past.
    l     Trainees do not appear to perform substantial within-industry job changing in Wyoming. They either leave the market or change industries.
    l     The percentage of WDTF participants working for their training provider and in Wyoming drops steadily from one quarter to four years after training.
    l     Slightly more than 55 percent of WDTF participants were working in Wyoming up to four years after training.
    l     Wage progression is statistically significant for participants in the lowest three wage quintiles.
    l     Except for the FY2003 cohort, participants who had wages in the highest two wage quintiles prior to training did not experience significant wage progression after training.
    l     WDTF participants have higher retention in Wyoming than either non-participants (general market comparison) or matched controls.

Age and Gender

Of WDTF participants that had demographic data (6.3% of cases or 219 individuals were lacking demographic data), 1,578 were female and 1,678 were male (see Table 1). Participants were concentrated in age categories ranging from 25 to 54 years of age (70.8% of participants) with the highest percentage in the 25 to 34 age category (26.5%).

The age distribution of participants changed substantially over time with latter cohorts having a higher percentage of older participants. As can be seen in Figure 1, the proportion of participants in the 45 to 54 age category increased substantially over the five fiscal years. More than 40 percent of FY1999 participants were in the 20 to 24 age category, whereas less than 15 percent of FY2003 participants were in this category.

The gender distribution of WDTF participants also changed over the period of FY1999 through FY2003. The first fiscal year (FY1999) was dominated by men and the following year (FY2000) was dominated by women (see Figure 2). Since FY2000, however, the distribution has become more balanced but with more men than women in each of the remaining years.

Industry

The primary industry determination was made the quarter after training ended and is the industry of the firm that paid the participant the most wages during the quarter. Usually this is the same firm that trained the employee, but there may be exceptions. Industry categories come from the North American Industry Classification System (NAICS) and are shown by supersectors ( see Figure 3). WDTF participants are employed across a wide range of supersectors but are most heavily concentrated in Information (16.1%), Retail Trade (14.2%), Health Care & Social Assistance (13.8%), and Leisure & Hospitality (11.3%).

The industry distribution of WDTF participants has diversified over time (see Figure 4). Participants were heavily concentrated in Information and Retail Trade during the first three fiscal years. In the last two fiscal years, Information and Retail Trade employed a smaller percent of participants and the proportion being primarily employed in other industries has increased (e.g., Health Care & Social Assistance). No single industry made up more than 25 percent of total participants in the last two fiscal years, and the vast majority of industries comprised less than 15 percent of the total.

Wyoming Labor Market Activity

Tables 2 through 7 and Figures 5 through 10 examine the Wyoming labor market activity of WDTF participants subsequent to training. Not all cohorts were included in each of the tables and figures because wage records data do not exist for more recent cohorts. All cohorts under study had measures of labor market activity up to one year after training. Measures after one year were limited to available data.

Participants found working with their training employer included those individuals who were reported in UI wage records data with their training employer based upon a match between the UI number reported in the WDTF database and the UI number reported in Wage Records. Those found working in their industry of training included those who remained within the same NAICS supersector as their training employer (e.g., John was trained in the Mining supersector at firm X but took a job with firm Y, also in the Mining supersector, one year later). Because employers can change NAICS codes over time for economic or non-economic reasons, it is logically possible for there to be fewer individuals working in their training industry than with their training employer (e.g., a training employer changes from Retail Trade to Leisure and Hospitality one year after their WDTF contract ends but retains all or most of their employees). Typically, there are more individuals working in their training industry than with their training employer at any given time.

Tables 2 through 7 summarize the data displayed graphically in Figures 5 through 10 . Since the pattern of results are more easily seen in the figures, the discussion of results in this section will focus on Figures 5 through 10. Readers may refer to the tables for detailed data.

Regardless of the time frame, there was little difference between the number of individuals who remain working with their training employer and the number of individuals who remained in their training industry (see Figures 5 through 10). Job changing appears to be associated with either leaving the market altogether (i.e., not being found in Wyoming Wage Records) or changing to another industry.

When examining all cohorts together, over 80 percent of WDTF participants were working with their training employer three months after training and over 90 percent were working in Wyoming (see Figure 5). At six months, just over 70 percent of WDTF participants were working with their training employer and just under 90 percent were found in Wyoming (see Figure 6). One year after training 60 percent of participants were with their training employer and just over 80 percent were working in Wyoming (see Figure 7). Some variation exists among the individual cohorts over these three time periods. Both FY1999 and FY2001 participants were less likely to be found working with their training employer or working in Wyoming at six months and one year after training.

Figures 8, 9, and 10 will be discussed separately as they involve different cohorts. Figure 8 shows labor market outcomes two years after training for FY1999 through FY2002 cohorts. Just under 40 percent of FY1999 through FY2002 WDTF participants remained with their training employer two years after training. Approximately 70 percent were found working in Wyoming two years after WDTF training.

About 30 percent of FY1999 through FY2001 participants were found working with their training employer three years after training (see Figure 9). Just over 60 percent of FY1999 through FY2001 participants were working in Wyoming three years after training.

Four years of wage records data exist for the first two WDTF cohorts (see Figure 10). Twenty-three percent of FY1999 through FY2000 WDTF participants were found working with their employer four years after training. Fifty- seven percent of FY1999 through FY2000 WDTF participants were found working in Wyoming four years after training.

The percentage of WDTF participants working with their training provider or in Wyoming dropped steadily from one quarter to four years after training. Data beyond one year are limited to available wage records. Slightly more than 55 percent of WDTF participants were found working in Wyoming up to four years after training.

Wage Progression (one-group pretest-posttest)

One-group pretest-posttest wage progression analysis indicates that all but the highest wage quintile (based upon wages before training) show wage progression from one quarter before to one quarter after training (see Table 8). The magnitude of the wage progression is largest for the lowest wage quintile ($2.96 and $2.59 assuming either a 35- or 40-hour work week, respectively) and is smaller in each of the three succeeding quintiles (to $.54 and $.48 assuming either a 35- or 40-hour week, respectively). There is a slight decline in hourly wages for the fifth quintile.

Wage Progression (manually matched and statistical control groups)

Results of both the manually matched and statistical control group evaluations are sub-divided into the three lowest and two highest wage quintiles (based upon wages prior to training). The purpose of control group evaluations is to show whether wage progression experienced by participants (as shown in Table 8) is also occurring for individuals who did not receive training. If non-trainees also experience similar wage progression, then training does not cause the subsequent increase in wages but some other factor does (e.g., wage growth in the economy due to a boom in oil and gas). Such large scale economic factors affect both the participant and non-participant groups. Control group analyses better capture these factors and help to determine the impact of training rather than that of general economic conditions within the state.

For the manually matched control group analysis, participants were matched with controls based upon wage and demographic data. Specifically, both participants and controls must have been found in Wage Records for at least any two quarters in the year prior to training and at least any one quarter in the year training ended. Participants and controls were also matched on age and gender to obtain similar demographic distributions. Participant and control cohorts should be most similar in average quarterly wages the year prior to training (this can be seen graphically on Figures 11 through 15 and Figures 17 through 21).

The discussion will first focus on graphs showing participant and control cohorts in the lowest three wage quintiles (see Figures 11 through 16). The last series of figures will show participants and controls in the highest two wage quintiles (see Figures 17 through 22). For FY2003, or Figures 15 and 21, one year of wage data exists after the training year.

For those in the lowest three wage quintiles, all years but FY1999 show greater wage progression (see Figures 11 through 15) for participants (see Figure 16 for statistical tests). The statistical tests shown in Figure 16 utilize Ordinary Least Squares regression techniques. Equations are solved separately for males and females and indicate the average quarterly wage of workers in a services-providing industry one year after training who earned the average quarterly wage prior to training.

Based upon analysis shown in Figures 11 through 16 it appears that WDTF participants in the lowest three wage quintiles experienced wage progression greater than that experienced by control group members (all but one cohort had a statistically significant relationship).

The analyses for the highest two wage quintiles were less distinct (see Figures 17 through 22). Results indicate no statistically significant difference between participants and controls for the FY1999 through FY2001 cohorts (see Figure 22). Further, FY2002 produced a statistically significant difference but results indicated that control group members, on average, were earning more than WDTF participants one year after training. Only FY2003 produced a statistically significant effect in the expected direction. It appears that WDTF training is less effective in producing the expected wage progression for participants in the two highest wage quintiles prior to training.

Labor Market Retention in Wyoming and Interstate Outcomes

Figures 23 through 32 show the average number of quarters worked in Wyoming for WDTF participant and manually matched control group cohorts. Across both the lowest three (see Figures 23 through 27) and highest two (see Figures 28 through 32) wage quintiles, WDTF participants work more quarters, on average, subsequent to training than control group members.

Figures 33 and 34 show interstate labor market outcomes (see Tables 9 and 10). Figure 33 indicates that, in comparison to non-participants (e.g., a general labor market comparison), a much higher percentage of WDTF participants (across the four cohorts shown) were found primarily working in Wyoming during the second year after training. Typically, less than 20 percent of WDTF participants and non-participants were found working in an MOU state during the second year after training. The graph also indicates that during the second year after training a much higher proportion of non-participants were not working in either Wyoming or an MOU state. Additionally, higher proportions were found not working than found working in an MOU state.

Similar patterns are also found when comparing WDTF participants with matched controls, although the differences are less pronounced (see Figure 34, page 49). Higher proportions of WDTF participants were found primarily working in Wyoming than matched control group members.

Wage outcomes for those primarily working in Wyoming during the second year after training and those working primarily in an MOU state show no clear pattern (see Tables 9 and 10). Some cohorts indicate that those primarily working in an MOU state have higher average quarterly wages whereas other cohorts show those primarily working in Wyoming have higher average quarterly wages.

Summary and Conclusions

Over time, a higher percentage of older and higher wage individuals are being trained with WDTF monies. Potentially this trend mirrors the general aging of Wyoming's population (Liu, 2003). The gender distribution has become more balanced since inception. Additionally, a broader range of primary industries employ WDTF participants now than in the past. The diversification of industires is likely responsible for the changes in gender concentration. Some industries in Wyoming are known to have large gender imbalances (Jones, 2004a).

Trainees do not appear to perform substantial within-industry job changing in Wyoming. They either leave the market or change industries. This may be due to the generally under-diversified nature of Wyoming's economy (Harris, 2002b). Industry changing may indicate the development of transferable skills through WDTF training.

The percentage of WDTF participants working for their training provider and in Wyoming drops steadily from one quarter to four years after training. Slightly more than 55 percent of WDTF participants were working in Wyoming up to four years after training. However, statistical evidence indicates that WDTF participants have higher rates of retention in Wyoming than either non-participants (general market comparison) or matched controls. Wyoming's labor market is extremely dynamic (Harris, 2003). WDTF sponsored training may buffer economic ups and downs that cause fluidity in the labor market.

With the exception of the first cohort (FY1999), wage progression is statistically significant for participants in the lowest three wage quintiles prior to training. Although wage progression is evident for both participant and control cohorts (as can be seen in the upward sloping lines shown in Figures 11 through 15), separation between WDTF and control groups visually seen in Figures 12 through 15 and results of OLS statistical tests (see Figure 16) indicate that, on average, WDTF participants in the lowest three wage quintiles experience wage progression greater than that experienced by control group members.

Except for the FY2003 cohort, participants who had wages in the highest two wage quintiles prior to training did not experience significant wage progression after training. It may be unrealistic to expect wage progression among individuals already at the higher end of the wage distribution.

References

Glover, T. (2002, June). Compared to what? The purpose and method of control group selection. Wyoming Labor Force Trends. Retrieved February 11, 2005, from http://doe.state.wy.us/LMI/0602/a2.htm

Harris, M. (2002a, June). Measuring the impact of Wyoming's workforce development training fund: Part two. Wyoming Labor Force Trends. Retrieved February 11, 2005, from http://doe.state.wy.us/LMI/0602/a1.htm

Harris, M. (2002b, September). Is Wyoming's economy diversifying and is economic diversity in Wyoming desirable? Wyoming Labor Force Trends. Retrieved February 11, 2005, from http://doe.state.wy.us/LMI/0902/a1.htm

Harris, M. (2003). An examination of the fluid nature of Wyoming's labor supply. In S. Murray and K. Shinkle (Eds.) Employment Outlook: 2010 (pp. 47-50). Casper WY: Wyoming Department of Employment, Research & Planning.

Jones, S. (2004a). Earnings by Age, Gender, and Industry. Retrieved February 11, 2005, from http://doe.state.wy.us/LMI/wfdemog/toc3.htm

Jones, S. (2004b, April). Examining Workforce Information Act programmatic outcomes using the Wyoming wage records universe as a statistical comparison group. Wyoming Labor Force Trends. Retrieved March 3, 2005, from: http://doe.state.wy.us/LMI/0404/a1.htm

Liu, W. (2003). Wyoming's resident population: Historical and projected data. In S. Murray and K. Shinkle (Eds.) Employment Outlook: 2010 (pp. 41-46). Casper WY: Wyoming Department of Employment, Research & Planning.