Replication archive: Displacement, Sequencing, and the Second Best in Systems Markets

Complete replication archive for the paper "Displacement, Sequencing, and the Second Best in Systems Markets". The paper studies a systems market in which one firm sets a capability that determines how much of the market a complementor can toll, and buyers sink investment only after both decisions are fixed. It characterises the second-best frontier — the capability that maximises welfare taking the complementor's best response as a constraint — and shows that four standard conclusions reverse against it. The displacement mechanism itself is the investment squeeze of Farrell and Katz (2000); the contribution is the benchmark. The model is applied to the licensing of standard-essential patents. CONTENTSpaper/ manuscript and online appendix (.tex and .pdf)theory/ model, proofs, exact Sturm certificates, parameter sweeps, figures 1-4empirics/ data acquisition scripts (build_01 to build_20), cleaned datasets, and all raw source filesfigures/ every figure in PDF and EPSMANIFEST.json every file with byte size and SHA-256README.md reproduction instructions RAW SOURCES INCLUDED IN FULLSEC EDGAR: 26 Qualcomm 10-K filings, FY2000-FY2025, plus XBRL company facts for Qualcomm, InterDigital, Nokia, Ericsson and Alcatel-Lucent.ETSI IPR bulk export from docbox.etsi.org/IPR/Open: ISLD-export.zip (4,972,603 rows) and GD-export.zip.IEEE PatCom Letter of Assurance spreadsheets: 15 XLSX files, 1,751 rows.Internet Archive captures of patent-pool rate cards (SIPRO, Via Licensing, Sisvel).Rendered page captures and screenshots from the browser-driven collection. The archive is deliberately complete. Raw sources are bundled rather than left to be re-downloaded, because a script pointing at a live URL is not replication: hosts change, bulk exports are replaced, and archived captures move. Total 222.0 MB across 169 files. REPRODUCTIONpython theory/run_all.py — analytical resultspython empirics/build_06_qcom_yield.py — Qualcomm royalty yieldpython empirics/build_18_etsi_reclass.py — ETSI declaration panelpython empirics/build_20_did_inference.py — difference-in-differences and pre-trend tests Requires Python 3.9 or later with numpy, scipy, sympy, pandas and matplotlib. The acquisition scripts additionally need requests, bs4, lxml and playwright, but nothing needs re-fetching: the raw files are in empirics/data/raw/. RESULTS THAT FAILEDThree identification designs were attempted. Two failed and one is only suggestive. All three, and the tests that killed them, are reported in the online appendix and reproduced by the scripts. build_14 fails its own event-study pre-trend test and is retained deliberately, not by oversight.