交通事故外部成本估計：倖存者的長期影響

Abstract

交通事故對社會造成龐大的經濟與健康負擔，然而現行政策評估與安全投資分析多半聚焦於死亡事故，對於倖存者所承受的長期成本關注有限。本文利用臺灣完整的行政資料，系統性估計非致命交通事故傷害對勞動市場與醫療使用的長期影響。

本文建構一個涵蓋近二十四年的縱貫資料，將警政署交通事故紀錄與全民健康保險資料庫加以串聯，涵蓋幾乎所有臺灣人口。我們採用具異質處理時間的事件研究法，並以「未來將受傷者」作為對照組，以避免使用從未受傷者所導致的比較偏誤，進而辨識交通事故傷害的因果效果。

實證結果顯示，非致命交通事故對勞動市場造成顯著且持久的衝擊。個人年薪與全職就業機率於事故後第五年下降約9%，且事故當年就業中斷風險大幅上升，進而對後續就業與收入軌跡產生長期影響。在健康面向上，事故導致住院天數、門診次數及全民健保給付之醫療支出顯著增加，且部分效果具有持續性。

異質性分析顯示，傷害嚴重程度是影響事故後結果差異的主要來源；所得差異主要反映在醫療使用上，而非勞動市場結果；性別差異則主要體現在勞動市場衝擊上，男性的收入與就業損失明顯較大。不同用路人之間的差異，在經濟結果上較健康結果更為明顯。

進一步將觀察到的醫療支出與所得損失加總進行概略估算，結果顯示非致命交通事故所造成的累積社會成本相當可觀，且隨時間快速累積。當與常用的生命統計價值作為比較基準時，倖存者所承受的經濟損失在量級上已不可忽視。本文指出，僅以死亡事故作為評估基礎，將嚴重低估交通事故的整體社會成本，未來交通安全政策與保險制度設計應納入對倖存者長期影響的考量。

Road traffic injuries impose substantial social costs, yet policy evaluation and safety valuation have traditionally focused on fatalities while largely overlooking the long-run consequences borne by survivors. This paper quantifies the economic and health impacts of nonfatal traffic injuries using comprehensive administrative data from Taiwan. We construct a population level longitudinal dataset linking police-reported traffic accidents to national health insurance and employment records, covering nearly the entire working-age population over a 24-year period.

We estimate causal effects using a staggered-adoption event-study design with later-treated individuals as controls, which allows for heterogeneous treatment timing and avoids biases arising from comparisons with never-injured individuals. We find that nonfatal traffic injuries lead to large and persistent losses in labor-market outcomes: annual earnings and full-time employment probabilities decline by about 9 percent five years after the accident, with job separations spiking immediately after the accident and generating long-lasting scars on employment trajectories. Health care utilization also rises sharply, including substantial increases in inpatient stays, outpatient visits, and publicly financed medical expenditures under Taiwan's National Health Insurance system.

Heterogeneity analyses show that injury severity is the dominant source of variation in post-accident impacts. Income differences primarily shape health care utilization, while labor-market losses are systematically larger for men than for women. Differences across road users are more pronounced in economic outcomes than in health outcomes.

A back-of-the-envelope aggregation of observed medical spending and earnings losses implies that the cumulative social costs of nonfatal injuries are large and accumulate rapidly over time. When benchmarked against commonly used values of a statistical life, the implied losses from surviving injuries are comparable in magnitude, underscoring that evaluations based solely on fatalities substantially understate the true social burden of traffic accidents. The findings highlight the importance of incorporating long-run survivor costs into transportation policy, safety investment appraisal, and social insurance design.