The average cost per participant placed in unsubsidized employment describes the average level of resources a WtW program had to invest to achieve a job placement. To develop this measure of average costs, we first had to estimate placements for each WtW program included in the cost analysis.
Placement measures were also constructed using information from the programs' MIS, whenever possible. Since these systems typically recorded more than one placement for a given participant, special attention had to be paid to counting only placements in unsubsidized employment (as opposed to work experience, on-the-job training, or other temporary work assignments) and to counting individual participants who were placed in unsubsidized employment (as opposed to the number of placements). When MIS information was unavailable to conduct such analysis, we also relied on paper-based administrative records, reports to DOL, and other documents to impute these statistics.(16)
|Site/Program Operator||Total Costs||Carry-overs||Enrolled||Ever Active||Total Participant-Months of Service||Average Duration of Participation
|Cost per Participant(a)||Imputed Cost per Participant(b)|
|Note: NA = not available; n.a. = not applicable.
a. Cost per participant = (total costs/total participant months) X average duration.
b. Cost per participant = (total costs/total participants active participants) X (11.1441/average duration).
We computed average costs per participant placed in unsubsidized employment in two ways. First, we divided total costs for the analysis period by the number of individuals ever placed in unsubsidized employment during that year. As an alternative measure, we divided the estimates of average cost per participant by each program's rate of participant placement in unsubsidized employment.(17) In most cases, both calculations produced similar estimates of the cost per placement. Table II.4 summarizes this process and presents the estimates of average cost per placement developed.
While used widely, we must note that measures of cost per placement provide only a partial picture of the success or efficiency that WtW programs achieved in helping participants secure and retain unsubsidized employment. For example, our measures do not take into account the hourly wages that WtW participants received, the total number of hours they were able to work, the amount of time they remained in such jobs, or other dimensions of the quality of such placements. Nor do they fully account for how difficult it was for the WtW program to achieve a placement because of characteristics of its target population, prevailing labor market conditions, or other reasons. Thus, these estimates must be used and interpreted with caution.
|Site/Program Operator||Total Costs||Total Placements||Cost per Placement(a)||Placement Rate (percent)||Cost per Participant||Imputed Cost per Placement(b)|
|Note: Estimates presented in bold are those used for the cross-program analysis in Chapter III.
n.a. = not applicable; NA = not available.
a Cost per placement = Total costs/placements during cost analysis period.
b Cost per placement = Cost per participant/placement rate.
1. Table II.1 reports the analysis periods used. As the table shows, 3 of the 18 WtW programs had cost analysis periods other than January to December 2000 or July 2000 to June 2001. In these cases, the specified time frames corresponded to WtW program operators' contract (and reporting) periods. Using these time frames made data collection easier and should not have affected the basic results of the cost analysis since there is, at most, a two-month difference from the periods specified for other programs.
2. WtW programs went through several phases as they planned, implemented, refined, and phased out their operations. During the cost analysis period, the WtW programs were well established and operations were, for the most part, routine. Cost analysis periods also ended before programs began to close down or alter their structure to move toward more permanent long-term operational arrangements.
3. We sought information on all sources of support for the programs, not just the WtW grant funds. For some programs, this included WtW grant funds, grants from foundations, state matching funds, and funds from other sources.
4. Off-budget costs varied across programs due to differences in program structure and funding strategies. For example, some grantees provided in-kind contributions of office space and facilities, and in some programs participants received transportation passes or other supports from external agencies.
5. The WtW programs participating in the national evaluation were all compensated to offset the burden of introducing customers to the national evaluation, administering baseline information forms, and other activities associated with enrolling WtW participants in a research sample. The compensation these programs received during their cost analysis periods represented, at most, one percent of total estimated costs for these same periods.
6. In such cases, we determined the full overhead costs incurred by the programs on the basis of the contractor's audited overhead rates and accounting procedures. Then, we added in any of these costs not already included in the accounting records.
7. Furthermore, TANF recipients did not have to participate in WtW to receive TANF-funded supportive services.
8. Case management services were provided to WtW participants throughout their enrollment. However, the pre-employment phase of most programs involved primarily case management activities that were closely integrated with overall job readiness preparation. For this reason, we refer to this program component as "job readiness and case management." Case management services provided after job placement are included in "postemployment followup."
9. Cost information was available with such breakdowns from 11 of the 18 WtW programs included in this analysis.
10. From here on, we refer to "average costs per participant placed in unsubsidized employment" as "average costs per placement." The terms should be viewed as equivalent.
11. MIS data were unavailable for WtW programs in Boston. For the WtW programs in Chicago (except Catholic Charities), the available MIS data did not cover the programs' cost analysis periods fully, and there appeared to be significant lags in data entry of participant information for some programs.
12. The estimate of average cost per participant-month masks month-to-month variation in the actual costs of WtW programs. For example, these estimates do not indicate that WtW per-month costs could be much higher while a participant was engaged in temporary work activities that involved wage payments by the program, or that costs were generally lower during the follow-up period after placement in unsubsidized employment.
13. Average duration is not calculated as total participant-months divided by total participants ever active during the cost analysis period. Since the spells of some participants are truncated, but none are overstated, such calculation underestimates average duration. Instead, using MIS data, we estimated the overall duration of participation for individuals active in WtW at any time during the cost analysis period; average duration is the mean value of this variable.
14. As Table II.3 shows, we were unable to estimate total participant-months for nine programs. Also, for the WtW programs in Boston and for Maximus and Easter Seals in Chicago, average duration of participation was estimated based on discussions with program staff. For E&ES and Pyramid (also in Chicago), average duration was estimated using all available MIS participant records (not just those for participants active during the cost analysis year).
15. Using the information for the nine programs for which we had full information, we estimated the average relationship between the total number of individuals ever active in the cost year and the total number of participant-months. We then applied this "adjustment" factor to the crude estimates of average cost per participant for the programs where we were missing information. In algebraic terms, we set (Total Costs/Total Participant Months) * Average Duration = (Total Costs/Total Participants Ever Active) * X, where X is the adjustment factor. Furthermore, we assumed that X was inversely related to average program duration (that is X = Y/Average Duration), which was available for all the programs, so that Average Cost per Participant = Total Costs * Y/(Total Participants Ever Active * Average Duration). Thus, we estimated the average value of Y across those programs with full information (Y = 11.1), and used this adjustment to impute average costs for those programs missing total participant months of service. Alternative models were tested, but the one we used achieved the best fit for the data from programs with full information.
16. Placement rates were imputed for the WtW programs in Boston and Fort Worth, and for the WtW programs operated by Maximus and Easter Seals in Chicago.
17. Importantly, placement rates were computed over each program's full period of operations (not just the cost analysis period), except for Philadelphia-TWC, West Virginia-CEP, and the WtW programs in Fort Worth. For the West Virginia and Fort Worth programs, we could only obtain information on placements occurring during the cost analysis period. For Philadelphia-TWC, we computed an overall placement rate for those participants ever active during the cost analysis period.