|
 Printer Friendly Version • PDF Version
This technical appendix describes the logic and data underlying the
burden model used for estimating the impact of proposed Plan Year 2008
revisions of Form 5500 and associated schedules.(1)
After a general
description of the process, detailed material is presented associated
with the impact model input data, and the basic calculations and
outputs, associated with the version of the model contained in the
spreadsheet “Burden Model 20050901 r1.xls”.
The burden model builds on work done by Mathematica Policy Research (MPR)
in support of ERISA Form 5500 regulatory revisions from 1995 through
2004. During that period, MPR developed detailed estimates of the time
required by sponsors and providers for performing Form 5500 related
tasks based on extensive interviews with filers and their consultants.
The processes and results are described in MPR’s report to the
Department of Labor’s Office of Policy and Research of the Employee
Benefits Security Administration, “Estimates of the Burden for Filing
Form 5500: The Change in Burden from the 1997 to the 1999 Forms,” May
25, 1999.
The first step in modeling the impact of the proposed revisions to
the Form 5500 and associated schedules was to produce a “Current Rule”
summary of hours by type of filer, consistent with the MPR analyses.
Aggregate baseline burden was then estimated, consistent with those
hours and total filings (also consistent with historical MPR estimates).
The second step in the impact modeling was to examine which schedules
would be modified, which sets of respondents would no longer have to
file particular schedules, and which sets of respondents would face new
filing obligations. Where changes were proposed in the information to be
collected, the hours required per respondent were adjusted, consistent
with the underlying assumptions of the previous MPR model.
The final step in the impact modeling was to calculate aggregate
estimates of total burdens under the revised system by multiplying the
new filing counts by the revised burden hours, and converting to dollars
by multiplying by wage rates per burden hour. The calculations were
performed by size and type of filer.
Summary tables from the spreadsheet implementation of the model are
in Table TA_1, Table
TA_2, Table TA_3, Table
TA_4, Table TA_5 and Table
TA_6. TA_1 and TA_2 show
the full current rules and future rules matrices. Total counts of
filings are shown in the top block of each table, broken by schedule by
filer type. The current estimate of hours per schedule for each type of
filer is shown in the second block. The third block gives the total
burden hours, which is the product of the entries in the first and
second blocks. Tables TA_3 and TA_4 give the burdens under the current
rules and the future rules, converted to dollars. They also show the relative burden incurred directly by sponsors and indirectly
through service providers. Approximately 85% of the burden dollars are
associated with costs which pass through service providers. Table TA_5
gives the burden associated with small plans, allocated across the
revisions. Table TA_6 gives the burden for all plans, also allocated
across the revisions.
Burden Hours Per Schedule
The first step in developing the model inputs was to divide the total
Form 5500 burden hours from “[Final Burden Model Dec 2004.xls]74.
Composite Report” by plan volumes from “Final to OPR Burden Hours
Summary-2”, to get hours per 5500 for large and small plans.
Similarly, the aggregate hours per schedule were divided by
corresponding schedule volumes to get hours per schedule for large and
small plans.
The next step was to partition the hours per schedule (including
5500s) by plan type (defined benefit, defined contribution, and
welfare). The starting point was the matrix of aggregate hours by
schedule for detailed plan types from MPR. For large plans, there are 30
plan types, and there are 8 for small plans. Aggregate hours by plan
type were collapsed for each schedule and size to get total hours for
each schedule by the 3 plan types. Plan volumes (5500 filers) were also
collapsed for plan types that may have filled out a particular schedule.
(Model pages have sections shaded in black for schedule/ plan type
combinations that MPR considered not possible.) Dividing aggregate hours
by plan volumes resulted in hours per plan for each schedule by DC, DB,
and welfare. This allowed the calculation of ratios of these hours per
plan by schedule for a particular plan type to overall hours per plan by
schedule. These ratios were multiplied by hours per schedule to get
estimates of average burden hours per schedule for DB, DC, and welfare
plans separately. These average hours per schedule by plan type were
used as inputs to the burden model.
2002 Fractions of Burden Per Agency By Schedule
The 2002 fractions of burden per agency for each schedule do not
appear in the spreadsheets, although they were calculated in supporting
sheets of the full burden model. They were calculated by dividing each
schedule’s burden for a given agency by that schedule’s total burden
hours. Fractions of burden per agency for each schedule were calculated
separately for large and small plans. Data was from “Final to OPR
Burden Hours Summary-2.”
2002 Service Provider/Sponsor Fraction of Burden By Schedule
The 2002 fractions of burden for service providers and sponsors by
schedule were based on inputs to “[Final Burden Model Dec 2004.xls]69.
Report Work Grid.” Service provider hours were those from service
providers when service providers file and from accountants and actuaries when service providers or sponsors file.
Sponsor hours were sponsor hours when service providers or sponsors
file. Aggregate service provider hours for each schedule were divided by
aggregate hours for that schedule to get the fraction of hours for a
schedule that was incurred by service providers. The 2002 fractions of
burden for service providers and sponsors by schedule were calculated
separately for large and small plans.
Controlling Schedule Counts to MPR
For each schedule (including 5500s) current estimates of 2002 large
plan schedule counts were multiplied by the ratio of MPR’s 2001 large
plan schedule count to current estimates of the 2001 large plan schedule
count. This was done to maintain consistency with the underlying MPR
data. (Newer counts of filings suggest the original model may somewhat
overstate total volumes of schedules, in which case the resulting
estimates are conservative). The same procedure was used to estimate
schedule counts for small plans.
Burden Hours per Schedule After Revisions
The change in burden associated with the revised filing requirements
was estimated based on the number of items added to or deleted from a
schedule, and the underlying estimated burden associated with the
original form. This was done either as a factor applied to the current
burden for the schedule, or as an add-on number of hours. For revisions
where the change was adding items to a schedule, the impact on hours per
schedule was calculated by multiplying the current estimated hours for
the schedule by the ratio of the number of items on the schedule after
the change to the number of items on the schedule currently. For
revisions where parts of another schedule were moved to a different
schedule, or items were added to a schedule which were similar to items
on another schedule, the burden associated with the revision was
estimated by adding the associated hours for the new parts to the
current hours for the schedule.
The burden hour estimate for the new short form required both
transfers of items from existing schedules and development of estimates
for new items. The final estimate of the proposed Short Form 5500’s
average burden per filer is 2 hours, 5 minutes. This is an average of
burden hours per small filer for the Form 5500, Schedule I, Schedule R,
and Schedule A, weighted by the proportion of elements from each of
these sources that were “donated” to the Short Form. It also
includes an estimate of burden attributable to several new data elements
on the Short Form. Average burden hours per small filer for each source
were taken from Mathematica’s Burden Model Composite Burden Hours (‘Final
Burden Model Dec 2004.xls’, tab ’74. Composite Report’). Using a
mock-up of the proposed Short Form, data elements were identified and
located on the current Form5500 and Schedules. To obtain the weights,
the number of Short Form elements from each source was divided by the
total number of elements from that source. Three items on the proposed
Short Form were not found on current schedules. These items are included
in the “Compliance Questions” portion of the Short Form, a section
that is otherwise taken from Schedule I. Therefore, it was assumed that
these three new elements would have the same burden as an average
Schedule I element. The Schedule I burden of 2.84 hours was divided by the number of Schedule I elements
(58), and multiplied by 3 to obtain a burden of 0.147 hours for the
three new elements. Adding this burden to the 1.925 figure results in a
total burden estimate for the Short Form of 2 hours and 5 minutes
(rounded up to the next whole minute), or 2.08 hours.
A macro was written to sum, for each agency, entity (sponsor, service
provider), and proposed revision, the number of affected filers, the
number of hours spent by those filers pre-revision, and the number of
hours post-revision. For each revision, the change in hours was then
calculated by subtracting the pre-revision hours from the post-revision
hours. In order to calculate the change in cost in dollar terms, a labor
rate of $84/hour was applied to the service provider change in hours,
and a rate of $59/hour was applied to the sponsor hours. The service
provider labor cost is consistent with the rate used in the underlying
MPR model ($75 in June, 2001 updates for 1999 levels), projected to 2002
at 4% per year (consistent with MPR update estimates from October 12,
2004). The sponsor hours costs are derived from wage and
compensation
data from the BLS, projected to 2002
and loaded for overhead. The resulting changes in cost for both service
providers and sponsors were summed to obtain the total change in cost.
The output applicable to the Executive Order and Regulatory Flexibility
is displayed on tabs ‘ExecOrder’ and ‘RegFlex’, respectively,
from ‘Burden Model 20050901 r1.xls’. Summaries are shown in
Tables TA_5 and TA_6.
The tabulations for the proposed revisions were carried out
sequentially. The following list explains the details and underlying
assumptions for each tabulation.
Removal of IRS Schedules
The change in burden due to the removal of IRS schedules was
tabulated in isolation, as if none of the other proposed revisions had
taken place.
Number of Affected Filers - The number of filers affected by the removal of schedules E, P, and
SSA is estimated using the sum of 1) the number of pension plans
(including both Short Form-eligible and ineligible) and 2) the number of
Schedules P filed by large welfare plans and small welfare plans.
While the number of schedules filed by plan type constitutes a major
component of the Burden Model, counts of filers that file a particular
schedule were not explicitly retained in the Model. In general, the
counts of filers and counts of schedules should be the same for all
schedules except schedules A, P, and R. In the case of pension plans,
the implicit assumption is that all pension plans file at least one of
the Schedules E, P, or SSA. Welfare plans can file Schedule P but
neither E nor SSA, so the number of Schedules P filed by welfare plans was used as an approximation for the number of
affected welfare filers.
Change in Burden Hours - The decrease in burden hours resulting from the removal of schedules
E, P, and SSA is calculated by summing the hours required to complete
each schedule for all pension and welfare plans.
Short Form 5500/Small 403(b) Plans
The change in burden resulting from the introduction of the Short
Form 5500 for eligible small plans was calculated with the proposed
elimination of the IRS schedules in place. This avoids double-counting
of the burden hours associated with the IRS schedules. Additionally, the
Short Form revision includes a provision from the 403(b) plan revision,
in that all small 403(b) plans were counted as Short Form-eligible.
Number of Affected Filers - The number of filers affected by the proposed introduction of a Short
Form 5500 is estimated at 90% of all small 5500 filers (both pension and
welfare plans).
Change in Burden Hours - In order to estimate the change in burden hours, the total number of
hours required by eligible plans to complete the Form 5500 and all
schedules was subtracted from the total number of hours required to
complete the Short Form 5500.
Schedule B: Asset Allocation Questions
Number of Affected Filers - The number of filers affected by the proposed new asset allocation
questions on Schedule B is the number of defined benefit plans with more
than 1,000 participants.
Change in Burden Hours - The change in burden hours is calculated by subtracting the total
number of hours spent on Schedule B for the affected plans in 2002 (“pre-revision”)
from the total number of Schedule B hours for affected plans in 2008 (“post-revision”).
Schedule C: Additional Questions for Fiduciaries and certain other
Service Providers
Number of Affected Filers - The number of filers affected by the proposed additional questions on
Schedule C is estimated by taking the number of Schedules C filed by
large pension and welfare plans. The number of schedules filed is
considered a reasonable proxy for the number of filers because only one
Schedule C is to be filed per plan.
Change in Burden Hours - The change in burden hours is calculated by subtracting the total
number of hours spent on Schedule C for the affected plans in 2002 (“pre-revision”)
from the total number of Schedule C hours for affected plans in 2008 (“post-revision”).
Questions for Multiemployer Plans on Form 5500 and Schedule R
Number of Affected Filers - The number of filers affected by the introduction of new questions
for multiemployer plans is estimated using the number of large
multiemployer pension and welfare plans. The Burden Model does not track
counts of small multiemployer plans.
Change in Burden Hours - The change in burden hours is calculated by subtracting the total
number of hours spent by large multiemployer plans on the Form 5500,
pre-revision, from the total number of hours post-revision. To this, the
difference in hours needed to complete the Schedule R for large defined
benefit multiemployer plans is then added.
Miscellaneous Revisions
Several questions are proposed to be added to the 2008 Form 5500,
Schedule A, Schedule H, Schedule I, and Schedule R.
Number of Affected Filers - The number of filers affected by the miscellaneous revisions is all
large pension and welfare plans and all small, Short Form-ineligible
pension and welfare plans.
Change in Burden Hours - The change in burden hours is calculated as the difference between
hours required, pre-revision and post-revision, to complete the Form
5500 and Schedules A, H, I, and R.
Elimination of Special Treatment for Large 403(b) Plans
Estimates of the impact of elimination of special treatment for
403(b) plans were calculated with all previously discussed revisions in
place. This means that small 403(b)s were modeled as if they all filed
the Short Form 5500; only large 403(b) plans figure into the 403(b)
revision burden estimate.
Number of Affected Filers - The number of affected filers is the number of large 403(b) plans.
Due to limited reporting requirements, 403(b) plans do not have to
report participant data; therefore, the large/small split of 403(b)
plans was estimated using outside data.
Change in Burden Hours - The change in burden hours is calculated as the number of hours
required for large 403(b) plans to complete the Form 5500 only,
subtracted from the number of hours required for large 403(b) plans to fully complete the Form 5500 and all relevant
Schedules (including the applicable revisions discussed above).
Technical Documentation of Form 5500 Revision
BurdenModel.mod30May1400.v3
30 May 2006 1400
jwm 5/o/t
-
When the project was started, the Form 5500
Revisions described here were proposed for 2007.
|
