This paper develops a flexible recipe for identifying unobserved input allocations, as well as quantity- and revenue-based total factor productivity (TFP) across product lines, for multiproduct producers, using demand- and supply-side information. Applying variants of this recipe to a panel of plants manufacturing machinery in India from 2000–2007 yields sizable within-plant TFP differences. Removing an average plant’s lowest-performing product increases the unweighted average of plant-level revenue-based productivity by 10%–65%. A 1 standard deviation decline in product-level revenue-based TFP generates around a 6 percentage point increase in the probability of dropping that product.

中文翻译：

---

公司内部生产力分散：估计和影响

本文开发了一个灵活的方法，用于识别未观察到的投入分配，以及跨产品线的基于数量和收入的全要素生产率 (TFP)，用于多产品生产商，使用需求方和供应方信息。从 2000 年到 2007 年，将此配方的变体应用于印度的一组制造机械的工厂会产生相当大的工厂内 TFP 差异。去除平均工厂性能最差的产品会使工厂级基于收入的未加权平均生产率提高 10%–65%。基于产品收入的全要素生产率每下降 1 个标准差，就会使放弃该产品的概率增加约 6 个百分点。