Return to PSG Models documentation page
GEMINI output file .SCN
--- .SCN file contains a variety of aggregate cohort sample statistics
for each scenario.
The GEMINI runNNNNN.scn file has no heading lines and no summary lines.
There is one line for each scenario in the run.
The statistics on a line are separated by the tab character.
Cohort sample individuals who are foreign born or emigrate or have
no retirement years (as defined below) are excluded from the
group of individuals used to calculate .scn statistics.
Retirement begins at disability or retirement, whichever is first, and
these events occur at the start of the year.
Retirement ends with death, which occurs at the end of the year.
So, for example, disability at age 57 and death at age 58 implies this
individual has two retirement years.
Real pretax retirement income (RRI) in a retirement year is the total of
individual's pretax OASDI benefit (adq:2 or share of adq:4) and
ind's pretax employer-sponsored pension benefit (adq:7 or share of adq:8),
where the total is converted from nominal to real terms by expressing it
in thousands of SSASIM:COHORT.cpi_year dollars.
The value of STATS.scnsid_ind determines whether pretax retirement income
is computed using the individual's benefits or the individual's adult-
equivalent share of the couple benefits. The nature of the sharing
is controlled by the three STATS table parameters that are used in
the National Academy of Sciences formula (see documentation of the
GEMINI STATS table).
So, for example, if a scenario sample has 10,000 individuals who each have
15 retirement years on average, then there will be 150,000 retirement
years in the whole scenario sample, and there will be 150,000 RRI values
for the whole scenario sample.
AWI denotes "average wage index", the mean annual earnings used in the wage
indexing of various OASDI amounts, where AWI is converted from
nominal to real terms by expressing it in thousands of
SSASIM:COHORT.cpi_year dollars. Note that the AWI in the retirement years
of the cohort individuals refers to the average earnings of mostly younger,
working-age people (not the cohort individuals whose retirement income is
The pretax scenario statistics on each line are as follows:
( 1) scenario number
( 2) aggregate number of retirement years in scenario sample
( 3) number of individuals in scenario sample with retirement years
( 4) mean real AWI (averaged over all retirement years for all scenario indivs)
( 5) mean RRI (averaged over all retirement years for all scenario individuals)
( 6) mean OASDI benefit (a component of mean RRI defined above)
( 7) mean pension benefit (sum of pension components of mean RRI defined above)
( 8) mean steady AWI-indexed earnings (sum:40 or sum:41 depending on scnsid_ind)
( 9) mean first retirement age (averaged over all scenario individuals)
(10) certainty-equivalent RRI over all scenario individuals (see method below)
(11) certainty-equivalent sben over all scenario individuals (see method below)
(12) certainty-equivalent pben over all scenario individuals (see method below)
So, (5)=(6)+(7) for each scenario, apart from rounding error.
But (10) is not necessarily equal to (11)+(12) because of method (see below).
Method for Calculating a Certainty-Equivalent Amount
As described in the documentation of the .sid output results file, a
certainty-equivalent amount is computed for each individual even if the
.sid output file is not being written. Using the certainty-equivalent
amounts for each individual in a scenario, the following methods are used
to compute a certainty-equivalent amount for the scenario sample.
Standard expected-utility-theory methods are used with the addition of
a special rule to handle the individuals with a zero certainty-equivalent
amount. A constant relative risk aversion (CRRA) utility function is
assumed with the crra parameter value being set by the scnsid_ra field in
the STATS table of the GEMINI input database. That utility function is used
to compute the certainty-equivalent amount for individuals with positive
certainty-equivalent amounts, and that scenario certainty-equivalent amount
is multiplied by the proportion of scenario individuals who have positive