Drought events have critical impacts on agricultural production yet there is little consensus on how these should be measured and defined, with implications for drought research and policy. We develop a flexible rainfall-temperature drought index that captures all dry events and we classify these as Type 1 (above-average cooling degree days) and Type 2 droughts (below-average cooling degree days). Applied to a panel dataset of Indian districts over 1966–2009, Type 2 droughts are found to have negative marginal impacts comparable to those of Type 1 droughts. Irrigation more effectively reduces Type 2 drought-induced yield losses than Type 1 yield losses. Over time, Type 1 drought losses have declined while Type 2 losses have risen. Estimates of average yield losses due to Type 1 droughts are reduced by up to 27 per cent when Type 2 droughts are omitted. The associated ex-post economic costs in terms of rice production are underestimated by up to 124 per cent.

中文翻译：

---

干旱指数的选择是否会影响对印度干旱引起的水稻损失的估计？

干旱事件对农业生产具有重要影响，但对于如何衡量和定义这些影响，以及对干旱研究和政策的影响，几乎没有达成共识。我们开发了一个灵活的降雨-温度干旱指数来捕捉所有干旱事件，我们将这些干旱事件分类为 1 类（高于平均降温日数）和 2 类干旱（低于平均降温日数）。应用于 1966-2009 年印度地区的面板数据集，发现 2 型干旱具有与 1 型干旱相当的负面边际影响。与类型 1 相比，灌溉更有效地减少了 2 型干旱引起的产量损失。随着时间的推移，第 1 类干旱损失减少，而第 2 类损失增加。如果忽略第 2 类干旱，对第 1 类干旱造成的平均产量损失估计值最多可减少 27%。相关的事后稻米生产的经济成本被低估了124%。