Borrowed Youth: Demographic Change and Multinational Location Advantage

Why can two multinationals prefer different hosts for the same plant? A manufacturer needs access both to workers at opening and to the cohorts that will replace them. Home-country recruitment and a team shared with another foreign affiliate change the value of that succession. Jaguar Land Rover chose Nitra over Puebla on a twenty-year appraisal. Slovakia had the larger initial adult share, while Mexico had stronger cohort replacement. If every 2015 resident stays and survives, 76 people reach twenty per hundred reaching sixty-five in Slovakia, against 203 in Mexico. Treating aid as cash caps Nitra's funded lead at EUR91 million. Holding that yardstick fixed, losing 223 net placements per year after substitutes costs EUR91.05 million over twenty years and EUR65.54 million over the fourteen-year vehicle machinery life. It takes 310 placements per year to exceed EUR91 million over fourteen years. The calculation assumes immediate operation, exponential productive tenure averaging 4.57 years, EUR10,000 net per staffed year and five-percent continuous discounting. Actual site placements and the fourteen-year margin are unobserved. Online Appendix K links the UK operating home and Indian parent to both candidate hosts. All four five-year national flow reconstructions favour Slovakia for UK-origin arrivals and Mexico for Indian-origin arrivals. Mexico's estimated Indian advantage is 179–207 people annually. With the same placement share at both sites, the largest gap is worth about EUR85 million over 20 years or EUR61 million over 14 at full conversion and the stated staffing benchmark. Per ten percentage points of additional placements at both hosts the short-horizon range is EUR5.3–6.1 million and the long-horizon range EUR7.3–8.5 million. An illustrative EUR7 million Nitra lead changes hands only over the longer horizon in all four reconstructions, holding other recruitment fixed. Increasing the UK placement share by 0.2 percentage points at both hosts keeps Nitra ahead at twenty years in one reconstruction. With both origins delivering one extra hire per ten arrivals, Nitra starts EUR7 million ahead and remains the preferred site at both horizons in all four reconstructions. Page 16 displays both strategies side by side. Plant recruitment shares were unobserved. Run site_succession/compare_origins.py for Table K3. Online Appendix K reports two comparisons. The country regression uses identical outcomes, fixed effects and 1,637 observations. Revaluing the three demographic services over twenty years in 86 motor-vehicle cells across 15 hosts changes the full-sample arrival association from 6.5% to 6.1% (95% interval 1.2–11.2%, p=0.018). For the primary firm sample, the twenty-to-fourteen-year ratio of conditional asset multipliers for 10,000 additional network-weighted annual arrivals is 1.19 (95% interval 0.66–2.15, p=0.506). Neither panel supports Hypothesis 1 on relative duration weights. The case supplies a decision threshold and the panels measure allocations. No new observations have been added. The later estimates preserve the directions of both national corridors. Over the later block, Nitra matches Puebla's intake across all four reconstructions when a UK arrival is one-quarter as likely as an Indian arrival to become a hire, holding each origin's share equal across hosts. Replace the first five appraisal years and restore earlier rates thereafter. At the illustrative EUR7 million lead the Indian-only agreement now selects Puebla at twenty years in two reconstructions. With equal additional hiring shares on both routes Nitra wins throughout. For the Indian agreement alone the common-lead interval for a unanimous horizon reversal is empty. The largest updated twenty-year gain at full Indian-route conversion is EUR87.66 million at the maintained staffing benchmark, below the EUR91 million maximum funded lead. Table K4 and its source cells and replication code report these retrospective checks. The anonymous manuscript, online appendix, LaTeX, figure files, editable tables and replication are available here. README.txt gives restoration instructions. Run site_succession/compare_horizons.py for Table K1 and site_succession/refit_vehicle_horizon.py for Table K2. There are 2,044 replication files with 2,043 checksums. Licensed WRDS rows require subscriber access. Queries and aggregate estimates are supplied. The main PDF contains 9,997 words including exhibits. There are 20 full pages of substantive text before the references. ChatGPT and Codex (OpenAI) were used in October 2026 for coding help, literature search, and red teaming and review. All figures were drawn with AI-assisted code from the model, demographic data and statistical estimates.