Borrowed Youth: Demographic Change and Multinational Location Advantage

Borrowed Youth: Demographic Change and Multinational Location Advantage explains how national cohort succession and international recruitment become firm-specific location advantages over a production commitment. Home-linked relationships, recruitment shared across foreign affiliates and admission institutions determine attainable replacement labour.

Version 3.38.0 adds joint economic bounds computed from the held coefficient estimates and complete covariances. The bound concerns the weakest allocation response across stated demographic increments. For the country panel it is 5.26 percent per million discounted service-years. For firms it is 3.42 percent across 100,000 network-weighted opening residents and 10,000 annual arrivals throughout twelve-to-twenty-seven-year machinery lives. Within the joint cluster-reference region, both increments cannot raise conditional capital allocation by more than 3.42 percent at every life. On identical 299-observation support, the engineering-life bound is 2.75 percent and the accounting-life bound is 12.89 percent. These are conditional allocation-response bounds, with approximate cluster inference; Appendix O reports reference-distribution sensitivity and exact optimisation certificates. The replication adds no observations or regression re-estimation. The article makes the theoretical contributions to resource bundling, human-capital mobility and foreign-affiliate portfolios explicit. The introduction states the research question and three contributions; the discussion explains how the results change the comparison within each debate. The empirical section links demographic exposure, service-delivery requirements and observed allocation in the sequence of the location decision. Unused notation and repeated explanations are removed. The article retains direct, decision-centred exposition. The demographic argument connects opening workforces to the organisation supplying their successors. Five main propositions state the conditions for host rankings, shared recruitment, admission constraints and service-delivery deadlines. A governance remark gives the threshold at which internal recruitment beats contractual access. The retained main text states a closed-form deadline under independent permanent route closure; the appendix derives the separate survival integral for preserved incumbent rights and retains all sixteen deadline scenarios.

The retained Appendix L proves a local observational-equivalence result: unknown access conversions and operating life can generate identical population coefficient products. The result identifies the restrictions or independent measurements needed to separate those components. Stable positive relative conversion preserves the ordering of arrival versus opening-service weights. The held expenditure design has 82 matched records and 36 after fixed-effect support. Its five focal regressors have rank three, leaving two coefficient directions unidentified. Public aggregate rank reconstruction and licensed rerun code are supplied.

The engineering-life asset comparison contains 317 allocations from 32 manufacturers in nine hosts. Its rotation contrast is -110.25 with a 95 percent interval of -520.28 to 299.78 and positive-direction p=0.724. Joint support and familywise contradiction probabilities are 0.763 and 0.712. The corresponding country probabilities are 0.974 and 0.077. All exceed 0.05. Holm-adjusted block probabilities are 0.113, 0.085 and 0.656. The observed quantities are conditional demographic-exposure associations; structural recovery uses the stated choice, duration, access-conversion, aggregation and disclosure conditions.

The held Nitra-Puebla appraisal supplies a EUR106-226 million residual strategic hurdle and 2,834 planned direct jobs. At a five-percent reference rate this is EUR2,959-6,308 per planned job-year. The frontier prices the additional productive services required to cover an assigned share. The risk and governance calculations state their assumed inputs explicitly. All prior empirical estimates, covariance matrices, formal statements and six figures are preserved.

The manuscript has 9,989 counted words, a 174-word abstract and 23 full substantive pages before references. The main PDF has 43 pages and the appendix 146. Regression tables report three-decimal probabilities without significance stars; the text reports economic magnitudes. Counts include abstract, text, references and transparency; separately placed exhibits and the online appendix are excluded.

Start with README.txt. The article, appendix and complete LaTeX source are directly downloadable. Three verified ZIP blocks and nine frozen complementary archives reconstruct the exact 1,952-file replication tree with 1,951 checksum entries. Subscriber queries reconstruct licensed WRDS rows; public aggregate outputs are included. Third-party rights remain. Literature PDFs are linked.

ChatGPT and Codex (OpenAI) were used in October 2026 for coding help, literature search, and red teaming and review. All figures were created using AI. Reviewer-upload files omit author identifiers. Public repository metadata identifies the author. Journal status is unchanged by this public release.