再論企業的財產保險需求：中國上市公司之實證研究

Revisit the Corporate Demand for Property Insurance: Chinese Evidence

本文檢驗中國大陸上市公司財產保險需求的影響因素。採用Heckman二階段估計法，並且另計算各變數的邊際效果。實證顯示財務槓桿對保險購買決策並沒有顯著影響、經理人持股與購買保險的可能性呈負相關，符合動機一致假說。Zou, Adams, and Buckle (2003)發現公司規模愈大，買保險的機率愈高，本文則發現公司支出的保險費隨規模愈大而下降。相較於Zou, Adams, and Buckle (2003)及Zou and Adams (2006)發現政府持股比率並不影響保險需求，本文的實證顯示政府持股愈高的公司，支出的保險費愈多，實證結果並不支持Charity (moral) hazard的假設。最後，實證結果也顯示由於Heckman第二階段係非線性，必須另外計算邊際效果及其顯著水準才能確認解釋變數對被解釋變數的影響方向及顯著性，換言之，單純觀察迴歸係數可能會導致錯誤的結論。

This study investigates the corporate demand for property-liability insurance among 2,213 publicly listed firms in the People’s Republic of China (PRC) during the 2005-2006 period. Using the Heckman approach, we modeled demand as a binary consumption decision followed by a continuous choice of insurance volume, taken to be premiums scaled by physical assets. We compensated for the effects of sample selection by incorporating the Inverse Mills Ratio as an independent variable during the second stage. To interpret the output, it was necessary to compute the marginal effects of the explanatory variables, which were not equal to their estimated coefficients since both stages of the model were nonlinear. Empirical results show that leverage has no significant effect on the demand for insurance. Unlike previous researchers, we found that managerial ownership was negatively associated with the probability of purchasing insurance, state-owned corporations was positively associated with the demand for insurance, and firm size was negatively correlated with premium expenditures. In addition, while the coefficient of tax rate in the second stage is insignificant, its marginal effect is positively and significantly associated with premium expenditures. This result illustrates the importance of calculating marginal effects of explanatory variables. Simply relying on estimated coefficients may generate false conclusions.