strategies / dynamics
strategy 06 · dynamics
Microcosm Dynamics ages people forward through earnings, family structure, disability and mortality. Analysts of retirement and social-insurance policy need that aging. Established U.S. models that provide it, among them the Urban Institute's DYNASIM, the Social Security Administration's MINT and the Congressional Budget Office's CBOLT, run inside their institutions. CBO does not publish CBOLT's code, and we found no public code with which to rerun DYNASIM or MINT. Microcosm Dynamics is open. It adds a longitudinal layer to Microcosm's certified cross-sectional data, a single snapshot that we calibrate to one year of administrative targets.
Microcosm Dynamics fits most transitions of individual lives on the Panel Study of Income Dynamics, draws claim ages from the Social Security Administration's published awards and runs these transitions forward year by year. Its earnings, family, work-limiting disability and population-transport components each pass a gate: a test on held-out or external data that we register publicly before we run it, with a pass rule fixed in advance and changed only by public amendment. The term comes from the quality gates that code must clear before it ships and from the go/no-go decisions of clinical trials. Four gates pass only under amendments we adopted after an earlier attempt failed: those for generated earnings histories, family formation, the transport onto the CPS frame and the whole engine's 2015–2019 holdout, which certifies nothing about mortality drift. No projection starts from Microcosm's CPS frame yet: every projection so far starts from PSID members. Four blind tests compare the model with DYNASIM3's published results, and none of them uses a gated component. The two tests that project to 2030 draw no earnings after 2010, apply no retirement earnings test and model no claiming response to reform. The model produces no cost estimates.
Generated earnings histories, the forward earnings model (the paper calls it the forward earnings law), family formation, household composition, work-limiting disability and the transport of PSID-estimated processes onto Microcosm's CPS frame each pass a gate. Mortality, claiming, disability-insurance entitlement and the immigrants who enter the population on a schedule run without one.
A one-point cut to cost-of-living adjustments, which acts mechanically on benefits already being paid, lands within 0.74 percentage points of DYNASIM3 in every age group. Each of the other three tests finds a gap between the two models. We post each test's plan in public, make one scored run and save its output to the public repository before comparing it with DYNASIM3.
Gate 3 is meant to score near-term outputs, such as beneficiary counts and average benefits, against administrative publications. We have named it but have not yet set the specific statistics it will score or how close counts as a pass, and it has never run. For the long horizon, we plan to publish sensitivity surfaces over documented assumptions and to present none of them as the model's forecast.
We plan to connect the forward earnings model to the projections, apply the retirement earnings test and add behavioral claiming. None of these pieces joins the model until it passes its own gate. The first test of carrying earnings forward from 2010, B2, failed in October 2026, so a successor model has to pass first.
The scorecard lists the model's components with their sources under where the model stands, and every gap in its exercise sections. The working paper reports every gate verdict.
We validate the model on one program first: U.S. Social Security. We can score near-term outputs against its administrative statistics, and we compare the model with DYNASIM3's published results for it.