本研究探討COVID-19 疫情期間泰國一般保險公司風險承擔行為的決定因素， 並關注資本充足性、 流動性及疫情衝擊的影響。 研究以道德風險理論為基礎， 並採用2019 年第一季至2022 年第四季間35 家泰國保險公司的縱橫資料，運用固定效果迴歸模型分析資本充足率（ CAR） 、 獲利能力、 槓桿比率與流動性等財務指標對風險承擔之兩項反向衡量指標——RTA 和RTE 的影響。

研究結果顯示， CAR 對於企業風險或總體風險承擔均無顯著影響， 質疑資本監管作為單一風險控制工具的有效性。 COVID-19 虛擬變數與RTA 呈正相關， 表示疫情期間企業風險承擔行為有下降趨勢。 而CAR 與流動性之交互作用並未顯現顯著的調節效果。 在控制變數方面， 公司規模與獲利能力皆顯著， 顯示規模較大與獲利能力較佳的公司傾向承擔較低的風險。

此外, 研究結果發現， 保險公司風險行為的影響因素中， 內部財務健康狀況比起監管機構設定的資本門檻更為關鍵。 此發現對監理機關具有重要啟示， 尤其在危機情境下， 除了資本要求之外， 亦需更全面的監理工具以減輕道德風險。 本研究的限制包括未納入部分未上市公司以及使用二元虛擬變數衡量疫情影響。 後續研究可考慮採用更細緻的危機指標及延長觀察期間以提升分析深度。

This study investigates the determinants of risk-taking behavior among general insurance firms in Thailand during the COVID-19 pandemic, focusing on the roles of capital adequacy, liquidity, and the influence of pandemic-related shocks. Based on the moral hazard theory and panel data from 35 Thai insurers from the first quarter of 2019 to the last quarter of 2022, this research applies fixed effects regression models to analyze how financial indicators such as Capital Adequacy Ratio (CAR), profitability, leverage, and liquidity affect two inverse measures of risk-taking: RTA and RTE.

The findings indicate that CAR does not significantly influence either business or total risk-taking, calling into question the effectiveness of capital regulation as a standalone risk control tool. The COVID-19 dummy variable is positively associated with RTA, suggesting a reduction in risk-taking during the pandemic. Mean- while, the interaction between CAR and liquidity shows no significant moderating effect. Among the control variables, firm size and profitability are consistently significant, with larger and more profitable firms engaging in less risky behavior.

The results imply that internal financial health, rather than regulatory capital thresholds, is more crucial in the impact of insurers’ risk behavior. These findings have important implications for regulators, suggesting a need for broader oversight tools beyond CAR requirements to mitigate moral hazard, especially in crisis conditions. Limitations include the exclusion of non-listed firms with unavailable data and the use of a binary pandemic proxy. Future studies may benefit from more detailed crisis indicators and extended time frames.