Lost in Translation: Language Barriers to Global Investment

We use supervised Latent Dirichlet Allocation (sLDA) to measure differences in the content of the Chinese and English translations of the annual reports of Chinese firms. We find systematic differences in content across languages that are not the mechanical effects of translation and that vary predictably with legal requirements for translation, firm size, and incentives to prioritize different stakeholders. We demonstrate that translation differences exacerbate information asymmetries between Chinese-speaking and English-speaking investors on average, as reflected in lower liquidity in the period after reports are released, especially when the information environment is poor. Firms with larger translation differences are more likely to receive government subsidies and raise larger amounts of equity capital from foreign investors, consistent with translation differences constituting part of firms’ strategy to optimize interactions with stakeholders who speak different languages. Our results highlight the impact of language barriers on cross-country business activities, especially in the context of China, which is growing in prominence as a target for foreign capital.