systemgmmkit 0.5.2

Documentation-only release refreshing the PyPI long description after the 0.5.1 native System GMM xtabond2 parity certification. No estimator behavior changes.

systemgmmkit 0.5.1

Release 0.5.1 updates the public package after certified native System GMM xtabond2 parity.

Highlights:

• Certifies native System GMM baseline as PASS_STRICT_XTABOND2_SYSTEM_GMM_BASELINE.
• Verifies parity against Stata xtabond2 for sample size, instrument count, coefficients, Windmeijer-corrected two-step standard errors, Hansen p-values, and signed AR(1)/AR(2) diagnostics with p-values.
• Adds multi-spec AR diagnostic parity validation for baseline, no-controls, three-way interaction, and decomposition System GMM specifications.
• Updates README and reviewer-facing parity artifacts.

Scope note:
This is benchmark-specific parity certification, not a universal claim of Stata identity across every possible xtabond2 configuration.

systemgmmkit v0.5.0

systemgmmkit v0.5.0

This release adds native Windmeijer-corrected two-step covariance support for dynamic-panel GMM and documents benchmark-specific Stata xtabond2 parity for native System GMM.

Highlights

• Native System GMM now supports Windmeijer-corrected two-step standard errors.
• The certified benchmark matches Stata xtabond2 e(V) standard errors within a maximum relative difference of approximately 0.000579.
• Native System GMM benchmark documentation now covers coefficients, raw residual moments (Z'u), group-scaled two-step weighting matrix (A2 / n_groups), Hansen J, and Windmeijer-corrected standard errors.
• The old uncorrected two-step clustered covariance benchmark path is preserved through SYSTEMGMMKIT_NATIVE_WINDMEIJER=0.
• Repository hygiene was improved by removing tracked parity/debug Python snapshots and restoring a clean Ruff/pytest/CI gate.

Validation status

The Windmeijer parity result is benchmark-specific. It should not be presented as universal Stata identity across all possible datasets, lag windows, missing-data patterns, instrument classifications, covariance assumptions, or finite-sample settings.

Installation

python -m pip install systemgmmkit==0.5.0

Recommended extras

python -m pip install "systemgmmkit[all]==0.5.0"

systemgmmkit v0.4.2

systemgmmkit v0.4.2

Patch release preparing systemgmmkit for public PyPI installation and improving README presentation.

Changes

• Add public PyPI installation instructions.
• Add README badges for PyPI version, supported Python versions, license, CI, publish workflow, and downloads.
• Refresh package metadata for PyPI release.
• Keep native Difference GMM and native System GMM validation status clearly documented.

Installation

python -m pip install systemgmmkit

Recommended extras

python -m pip install "systemgmmkit[all]"

systemgmmkit v0.4.1

Public release hardening: citation metadata, CI branch trigger update, and runnable examples.