Gain / Loss Analysis Output
Clicking the View button in the Gain/Loss Analysis dialog box generates a list of all the parameters in this Gain/Loss Analysis, a summary of the gain or loss for the period, an analysis of the liability gain / (loss) by source, a breakdown of expected and actual benefit payments with interest to the end of the period and a reconciliation of participant statuses from the beginning to the end of the period.
In the “Liability gain / (loss), by source” exhibit, ProVal analyzes automatically the gain / (loss) for active decrements, inactive mortality and new entrants. Optional sources, such as data corrections, salary growth and regulatory increases, can be skipped if they do not apply to your plan. If you skip the optional sources, any gains or losses due to these optional sources will appear as “unreconciled”, along with any miscellaneous gains and losses (for example, actual marital status versus assumed percent married, actual versus assumed spouse age, etc.).
An explanation of each liability gain / (loss) source, presuming an immediate gain recognition cost method, appears below. For a spread gain recognition cost method, the change in normal cost rate, or change in normal cost as a percent of pay, would be analyzed (instead of the change in liability). For a mathematical derivation of these gain/loss sources, as well as detailed discussion, see the Technical Reference article Gain and loss analysis: a conceptual framework.
Data Corrections
The gain / (loss) for this category is the difference between the liability with uncorrected data and the liability with corrected data.
Typical data corrections include:
Dates of birth and hire changed;
For inactives, form of payment or spouse’s date of birth changed.
For a quick way to create a database with (most types of) data corrections, see Data Corrections.
Active Decrements
This category applies to records with an active status or, in the pension modes (except German and U.K.), a “Vested valued through active” status at the beginning of the period.
The gain / (loss) for this category is the difference between the liability with assumed decrements and the liability with actual decrements. Both of these liabilities use the assumed benefit received. No end-of-period values are reflected here, other than whether the participant actually decremented or not. (This gain or loss is uncontaminated by other sources. It is not the difference between the liability with assumed decrements and the actual end-of-period liability.)
Decrementing Active Benefits Received
This category applies to records with an active status or, in the pension modes (except German and U.K.), a “Vested valued through active” status at the beginning of the period, and an inactive status (including missing participants) at the end of the period.
The gain / (loss) for this category is the difference between the liability with the assumed benefit received and the liability with the actual benefit received. Both of these liabilities use actual decrements. No other end-of-period values are reflected here.
The assumed benefit received includes:
Inactive Mortality
This category applies to records with an inactive status at the beginning of the period.
The gain / (loss) for this category is the difference between the liability with assumed mortality rates and the liability with actual mortality rates (that is, actual deaths). No end-of-period values are reflected here, other than whether the inactive participant actually lived or died. Note that this gain / (loss) is uncontaminated by other sources; it is not the difference between the liability with assumed mortality and the actual end-of-period liability.
The gain / (loss) for “Survivors & beneficiaries” applies to both inactive members with a status of “survivor” and inactive beneficiaries (of records with other inactive statuses) with Payment Forms of the “joint life annuity”, “certain & joint life annuity” and “post-decrement death benefit” types.
Salary Growth & Regulatory Increases
This category applies to continuing actives, i.e., records with an active status at both the beginning and the end of the period.
Certain elements that cause benefit increases from year to year are considered for this “bucket”: salary, statutory and regulatory limits (such as the U.S. Internal Revenue Code Section 415 maximum benefit or the Canadian Income Tax Act maximum pension) and Benefit Formula Components and Accrual Basis Components to which increase rates have been applied.
The gain / (loss) for this category is the difference between the liability reflecting assumed increases over the period and the liability reflecting actual increases over the period.
New Entrants
This category applies to new active participants, i.e., records with an active end-of-period status but no record, or a status other than active, at the beginning of the period.
The gain / (loss) for this category under immediate gain recognition cost methods is the end-of-period liability for these participants; under spread gain recognition cost methods, the change in normal cost rate (or the change in normal cost percent of pay) attributable to these participants is analyzed instead.
Unreconciled Amounts
The sum of items in this category equals the remaining difference between the expected end-of-period liability, after reflecting known experience (i.e., starting with the beginning of period Valuation(s) and reflecting interest, normal cost, data corrections, actual decrements, actual mortality, benefit payments, salary growth, regulatory increases and new entrants), and the actual end-of-period liability. In other words, if there is something the gain/loss analysis can’t account for, it ends up here.
Unreconciled amounts can represent:
Miscellaneous gains and losses,
Flaws in setting up the Valuation (e.g., an inadvertent plan change or assumption change) and/or
Sources of gain / (loss), e.g., data corrections, that have not been reflected.
To make sure your Valuation is not flawed and you haven’t missed a source, it is important to be able to explain the unreconciled amounts. To make this task easier, ProVal reports unreconciled amounts by status transition. To find out which records are generating unreconciled amounts, you can also save Individual Results as part of the gain/loss analysis.
You’ll find some typical causes of unreconciled amounts illustrated below.
Unreconciled Amounts: Continuing Actives
The gain / (loss) for this category is the difference between the expected end-of-period liability, reflecting actual salary growth and regulatory increases and the fact that the participant remained in active status, and the actual end-of-period liability.
The typical causes of unreconciled amounts for continuing actives are:
Actual increase or crediting rates different from what was assumed. Look at your Valuation Assumptions to see if they contain increase or crediting rates for Benefit Formula Components or Accrual Basis Components. Also, look at your Plan Definition to see if it contains a "cash balance" Accrual Definition. In either case, an unreconciled amount may be due to actual increase rates or crediting rates different from those assumed.
Actual service accrual different from the assumed service accrual. If any continuing actives had service accruals other than the assumed accrual for the period (e.g., other than 1 per year), this will produce an unreconciled gain or loss amount.
Skipping the Continuing Actives topic (i.e., not reflecting salary growth and regulatory increases in the analysis).
Skipping the Data Corrections analysis.
Unreconciled Amounts: Decrementing Actives
The gain / (loss) for this category is based on the expected end-of-period liability, reflecting data corrections and actual decrements, and the actual end-of-period liability. Generally, a small unreconciled amount will be generated by most (if not all) decrementing actives, because of the number of variables that might not have been anticipated perfectly before decrement.
The typical causes of unreconciled amounts for decrementing actives are:
Actual benefit amount different from the expected benefit amount.
Actual percent married different from that assumed. For example, 80% married is assumed, but when a participant actually decrements, he/she is either 100% married or 0% married.
Actual difference in age between participant and spouse not the same as the assumed age difference. For example, you assumed a three year spouse age difference, but (the end-of-period data indicates that) the spouse actually is 7 years older than the participant.
Actual Payment Form different from that assumed. For example, you assumed a life annuity would be paid, but (the end-of-period data indicates that) the actual payment form is a joint & survivor annuity, or even a lump sum. Note that participants who have received lump sum cash outs during the gain/loss period may be indicated as such under the Non-Participating Statuses topic. In this case, the amount of liability that would have otherwise been considered unreconciled will be added to the expected benefit payments as the “additional liability cashed out” and the unreconciled amounts will be zero.
Inactive mortality in the year of decrement. When an active participant retires with an immediate single life annuity, there’s a small assumed probability that the participant will die within a year after he/she retired – this is due to the inactive retirement mortality assumption. Because an active who retires doesn’t die (otherwise he/she would be reported as a death at the end of the period, instead of as a retirement), there’s a small unreconciled amount of gain or loss.
Actual election probabilities different from that assumed. For example, 80% of retiring participants are expected to take a lump sum and 20% to take the annuity, but all new retirees took their benefits on a single payment form, either the annuity or the lump sum. Thus, for each record, there was 100% election of the actual payment form – whether a life annuity or a lump sum. Whichever form was taken, both of these outcomes are markedly different from the assumed 80% or 20%.
Skipping the Data Corrections analysis.
If you were to create three one-life Gain/Loss Analysis runs, all processing an active participant who retires with a lump sum, where:
the first run assumed the participant would take a life annuity,
the second run assumed the participant would take a lump sum, and
the third run assumed the participant would take 20% of the benefit on a life annuity and 80% as a lump sum, then
the first Gain / (loss) Analysis run (life annuity) can generate an unreconciled amount for any of the “typical” reasons listed above.
the second Gain/Loss Analysis run (lump sum) doesn’t generate any unreconciled amounts because both the expected end-of-period liability and the actual end-of-period liability are 0. The difference between the expected lump sum amount (beginning-of-period data) with interest and the actual lump sum amount with interest ends up in the “Benefit payments” gain / (loss) (discussed below).
the unreconciled amount for the third Gain/Loss run (20% life annuity + 80% lump sum) will be mostly due to the difference between the assumed election probabilities and the actual election probabilities (20% life annuity + 80% lump sum vs. 100% life annuity or 100% lump sum). If, instead of a lump sum, the retiree actually took a life annuity, then unreconciled amounts could be generated for any of the “typical” reasons listed above.
Note that if the participant is indicated under the Non-Participating Statuses topic as cashed out at the end of period, none of the runs above will generate an unreconciled gain/loss. Instead, the liability that would have otherwise been considered unreconciled will be added to the expected benefit payments as the “additional liability cashed out”.
Unreconciled Amounts: Continuing Inactives
The gain / (loss) for this category is based on the expected end-of-period liability, reflecting data corrections and actual mortality, and the actual end-of-period liability.
The typical causes of unreconciled amounts for continuing inactives are:
Actual COLA different from the assumed COLA. Look at your Valuation Assumptions to see if they contain a COLA assumption. If so, the unreconciled amount may be due to a COLA different from the assumed COLA.
In OPEB mode, actual lapse probabilities different from the assumed. For example, 10% of inactive participants are expected to lapse coverage, but based on the new census data, 50% actually lapsed coverage.
Lump sums. An inactive participant starting the period with a different Payment Form from the end-of-period Payment Form; for example, starting with a life annuity and ending the period with a lump sum will produce an unreconciled amount.
Skipping the Data Corrections analysis.
Unreconciled Amounts: Inactive Member Deaths
The gain/loss for this category is based on the expected end-of-period liability, reflecting data corrections and actual mortality, and the actual end-of-period liability.
The typical causes of unreconciled amounts for inactive participant deaths are:
A spouse’s benefit different from that assumed.
Skipping the Data Corrections analysis.
Unreconciled Amounts: Rehires from Inactive Status
Unexpected status changes, such as rehires, can cause gains and losses. In these cases, the entire gain / (loss) may be allocated to the unreconciled category.
Unreconciled Amounts: Other Changes in Inactive Status
For example, vested terminated participants who retire usually will generate a small unreconciled amount, for the same reasons that decrementing actives do. These include deviations in the benefit amount, percent married, spouse age difference and payment form.
Benefit Payments
This category is based on a comparison of expected benefit payments, which are based on beginning-of-period data, and the actual benefit payments that are reported under the Assets and Expenses topic. The expected benefit payments for a Gain/Loss Analysis are not the same as the expected benefit payments in a Valuation. For gain/loss analysis, expected benefit payments are computed using actual experience for active decrements and inactive mortality, but Valuation expected benefit payments are based on expected experience for active decrements and inactive mortality. In a gain/loss analysis, expected benefit payments also include interest to the end of the period, but Valuation expected benefit payments won’t include interest.
For decrementing actives expected to be in pay status and getting an annuity, there is a small gain / (loss) reported under the “Benefit payments” category due to inactive mortality in the year of decrement.
For the three Gain/Loss runs outlined above for “Unreconciled Amounts: Decrementing actives”, the benefit payment gain / (loss) would be:
Run 1 (assumed life annuity): the difference between the expected annuity payment, based on beginning-of-period data, and the actual lump sum amount with interest reported under the Assets and Expenses topic, plus a small inactive mortality gain / (loss).
Run 2 (assumed lump sum): the difference between the expected lump sum amount, based on beginning-of-period data, and the actual lump sum amount with interest reported under the Assets and Expenses topic.
Run 3 (assumed 20% life annuity + 80% lump sum): the difference between the expected benefit payments (80% of the expected lump sum + 20% of the expected life annuity benefit payment), based on beginning-of-period data, and the actual lump sum amount with interest reported under the Assets and Expenses topic. There is a small expected benefit payment gain / (loss) due to inactive mortality in the year of decrement (20% of what you would see in the first run’s gain/loss analysis).
Note that if the participant is indicated under the Non-Participating Statuses topic as cashed out at the end of period, the liability that would have otherwise been considered unreconciled will be added to the expected benefit payments as the “additional liability cashed out”. The benefit payment gain/loss would be the difference between the actual lump sum amount with interest reported under the Assets and Expenses topic and the sum of the expected benefit payments (determined in each example above) plus the additional liability cashed out.
Implicit Assumption Changes
This category applies to valuations utilizing dynamic mortality tables or interest rates that vary by duration from the valuation date, or certain other assumptions where a roll-forward of beginning of period liability does not perfectly match end of period liability (e.g. different pre-decrement and post-decrement interest rates).
Gain/loss analysis should be run with the same assumptions for the beginning of period and end of period Valuations (except for the valuation interest rate, as discussed in the Interest Rate Change section of this article, below); otherwise an unreconciled gain or loss will be generated. Here, “same” means cosmetically identical. For example, if your beginning-of-period valuation assumptions use dynamic mortality table X, so should your end-of-period valuation assumptions. In some cases, though, even cosmetically identical assumptions imply an assumption change.
The most common implied assumption changes are due to dynamic mortality tables and interest rates that vary by duration from the valuation date. In these cases, merely changing the valuation date changes the assumption. These assumption changes, due solely to the change in valuation date from the beginning of the period to the end of the period, are labeled “implicit assumption changes”.
For dynamic mortality tables, the implicit assumption change is the difference in liability between freezing the mortality rates at the end of the period versus freezing them at the beginning of the period (note that dynamic mortality can be assumed for many purposes in ProVal, including active decrement, inactive mortality, lump sum factors, 415 maximum benefit limits, etc.).
With respect to interest rates, this is an implicit assumption change because the rate that applies in a particular calendar year changes from one valuation date to the next (as the calendar year’s duration from the valuation date shortens). For spot interest rates, the implicit assumption change is the difference in liability between using the beginning of period rates "rolled forward" to the end of the period and using the beginning of period assumptions at the end of the period, i.e., entering the same spot rates in the Valuation Assumptions for both the beginning of year and the end of year Valuations. For further discussion of ProVal’s roll forward methodology for interest rates when a PPA liability is the Gain/Loss Analysis basis, see Gain and loss analysis: U.S. PPA Target Liabilities.
Certain other assumptions may cause implied gains and losses and will be included as “implicit assumption changes”. These include any situation where a cosmetically identical end of period assumption set will not produce the same results as rolling forward the beginning of period liability. Examples of this include:
differing pre-decrement, in-deferment and post-commencement interest rates;
using a U.S. social security PIA custom operator that has a salary override based on the valuation date;
using any German State Pension operator (either standard or custom);
using the PPA option to reflect mid-year changes in accrual rates as if they had been effective on the valuation date (found under the Liability Methodology topic of Valuation Assumptions);
using retirement rates that vary by age/service on valuation date.
Interest Rate Change
In the event that a Gain / Loss Analysis is run with a valuation interest rate that differs cosmetically between the beginning and the end of the period, ProVal will display separately the gain or loss associated with the change in interest rate. Note that in the event spot or segment rates are used, and also differ between the beginning and the end of the period, ProVal will first calculate the impact due to the implied assumption change, as discussed above. The interest rate change will be the difference between the values in the actual end of period run and those in an end of period run using interest rates that are cosmetically identical to those used in the beginning of period run.
Report Writing
When you are satisfied with the gain/loss results, you can Print them or save them to a File by clicking the appropriate button.
In addition to saving to Excel and text files (as you can with all ProVal output), you can also save gain/loss results to an Access database for use with a report writer. After you save the results, the Access database will contain the following tables:
ProVal_GainLoss_Params: Basic information about the gain/loss analysis in the database, including the name of the run and the ProVal mode of operation.
ProVal_GainLoss_Summary: Summary of gain/loss.
ProVal_GainLoss_BySource: Liability gain / (loss), by source.
ProVal_GainLoss_StatusRec: Status Reconciliation.
To facilitate use by other programs, such as report writers, WinTech keeps many elements in the database stable. That is, these elements won’t change from one ProVal version to the next unless necessary, such as to incorporate a law change. The stable elements are:
Table names (e.g., ProVal_GainLoss_Params)
Field (i.e., column) names. These are meant to be used for column lookups.
The Name field in each table. This is meant to be used for row lookups.
Not the Description field in each table. This is meant to improve the readability of the database when viewed with human eyes. The descriptions might change from one ProVal version to the next and therefore should not be used by other software applications.