Language barriers in multinational companies: Evidence from Myanmar


Published 17.06.22

English language courses for domestic managers improve their interactions with foreign managers, and potential employers value these interactions

Countries routinely use incentives to attract foreign investment, but cultural distance also shapes economic flows across borders (Rauch and Trindade 2002, Burchardi et al 2019). Cultural barriers also create challenges within multinational corporations (MNCs). Subsidiaries bridge this gap by having experienced foreign managers (FMs) supervise less-experienced domestic middle managers (DMs). But this simply pushes the inherent cultural challenges down to the level of the subsidiary. As a result, cultural differences within MNC subsidiaries could reduce the interaction of managers across layers, impact operational performance, and potentially impede beneficial transfers from the MNC to the host country.

In Guillouet, Khandelwal, Macchiavello and Teachout (2021), we identify one potential friction to knowledge transfers among MNCs operating in Myanmar: language barriers. The subsidiaries in our sample – MNCs operating in Myanmar’s Thilawa Special Economic Zones – are structured as a three-tier hierarchy, with DMs sitting between the FMs and production workers. DMs perform a range of standard middle-management tasks, spending on average an hour a day interacting with FMs. Communication between domestic and foreign managers occurs in English, a non-native language for both layers. However, English tests (and anecdotes) reveal that the average DM’s English proficiency is low and communication is frequently ‘lost in translation’. For example, one FM at a Korean firm told us: “One problem is that it is difficult to teach Myanmar workers the details of their job due to language barriers.” A DM at a Japanese firm reported: “Although the boss can speak English, if the issue is important, we use translator. The [Myanmar] factory manager cannot speak English at all. So, when the translator is not there, we have to talk to him with body language or by drawing pictures. It takes more time.” Indeed, DMs with better English proficiency report interacting more frequently with FMs and are more involved in the company’s management, a finding consistent with communication being important to learn a tacit and hard-to-codify skill. 

Anecdotes and correlations, however, are only suggestive. Companies manage these barriers, for instance, through translators, bilingual DMs, and technology (e.g. Google Translate). Reducing language barriers may allow FMs to communicate the same information in less time, thereby reducing DMs’ interactions with FMs. So, it is not clear that language barriers are binding, despite the anecdotes. Moreover, FMs may communicate specific knowledge about MNC operations, while DMs want to learn general management skills. It is not obvious that reducing language barriers would increase (1) communication and (2) the transmission of general knowledge.

We test this first idea through a randomised trial that offers free English language courses to DMs. At endline, treatment DMs have improved their English proficiency relative to control DMs. Moreover, have more frequent interactions with FMs (about 34 minutes more per week), higher (hypothetical) willingness to pay for additional meetings with FMs, and are more involved in the management of personnel. We complement these self-reported measures through controlled simulation that mimics the organizational structure of MNCs. An English-speaking enumerator acting as an ‘FM’ provides instructions to a DM to manage a packaging task. The DM then manages two enumerators (who have no English knowledge) acting as production workers to complete the task. Treatment DMs spend more time communicating with the ‘FM’, confirming that reducing language barriers increases communication, but at a time cost for FMs. Treatment DMs also supervise the completion of the packaging tasks faster than control DMs (with no differences in mistakes).

To explore the second idea – that communication with FMs imparts general skills – we recruit human-resource managers from domestic firms to assess the market’s demand for DMs who have communicated with FMs. In the first block, HR managers rate hypothetical resumes of job applicants that vary in their characteristics. HR managers place a 16% premium on English ability, a 10% premium on MNC work experience, and an additional 3.6% premium on candidates with both. This is consistent with lower communication barriers being valued in the market. Since these attributes could proxy for other unobservables, we further implement a second block to test the communication mechanism further. We ask HR managers to assess (hypothetical) candidates in their answers to three ‘interview’ questions: How frequently did they use a MS office (a proxy for a hard skill)? How involved were they in management? And how frequently did they communicate with FMs? The valuation placed on frequently communicating with FMs is similar to that of the hard skill and management experience, confirming that communication with FMs, per se, is valued by the domestic market. 

We end with a discussion of why the observed communications between DMs and FMs are inefficient, where these frictions likely arise, and potential policy remedies. Through a simple model of organisations, we demonstrate that the amount of communication within the MNC will be suboptimal if two conditions arise: (1) DMs cannot contract with FMs with the amount/content of communication; and (2) communication with FMs imparts general skills to the DM. Condition 1 is a standard assumption in workhouse models of organisations (e.g. Dewatripont and Tirole 2005), and survey evidence reveals that it applies in our setting. Condition 2 is established through the resume rating protocol described above. As a result, there is scope for policy intervention, and we show that language subsidies could raise communication between DMs and FMs because of the results from the language experiment: lowering communication barriers improves interactions.

Our project explores the conditions for policy intervention in Myanmar, but we believe that such conditions arise in other contexts. The adoption of English as the corporate lingua franca is increasingly common (The Economist 2014), particularly outside of Anglophone countries. Additionally, recent decades have seen a sharp rise in South–South foreign direct investment flows, such as the surge of Chinese investments in Africa. There are, of course, many justifications for such language policies, and our study highlights one specific benefit: potential knowledge transfers from MNCs. 

Editors' note: An earlier version of this column appeared on VoxEU.


Burchardi, K B, T Chaney, and T A Hassan (2019), “Migrants, Ancestors, and Foreign Investments,” The Review of Economic Studies 86: 1448–86. 

Dewatripont, M and J Tirole (2005), “Modes of communication”, Journal of Political Economy 113: 1217–1238. 

Guillouet, L, A Khandelwal, R Macchiavello and M Teachout (2021), “Language barriers in multinationals and knowledge transfers”, NBER Working Paper 28807.

Rauch, J. E. and V. Trindade (2002): “Ethnic Chinese Networks in International Trade,” The Review of Economics and Statistics, 84, 116–130.

The Economist (2014), “The English empire”.