ECON 2342: China Economy Seminar

PAPER: "Does Dialect Serve as a Signal of Social Networks in Job Search? Evidence from China" (paper joint with Xianxiang Xu and Yang Jiao) 

ABSTRACT: 
Previous studies have documented that language as manifested by communication skills (Chiswick & Miller 1995) and social or cultural identity (Lazear 1999; Alesina et al. 2003; Desmet et al. 2017) affect economic outcomes. In this paper, we provide a third channel through which language can influence economic decisions - by serving as a signal of social capital. We study the companies’ hiring decisions among candidates from diverse linguistic backgrounds. Growing companies benefit from wider cross-regional social networks if they plan to expand their market nationwide. This is especially true when the costs of overcoming regional barriers to exchanging information are high and when cultural and institutional differences are large. As a result, in order to broaden their social connections, firms may favor job applicants with social networks beyond those already represented in the  company. As dialect is shaped by past interactions, religious divisions, transportation and social networks, one’s social network can thus be measured by his or her dialectal background (Falck et al. 2012, 2016). High dialect similarity to what is already present in an organization means less access to diverse social networks. The greater the distance between an applicant’s and the firm’s dialect the more likely he or she is to be hired. To verify this hypothesis, we utilize a novel database drawing from the 2010 – 2016 episodes of the Chinese television show Only You. The results show that the probability of being hired is 10% higher when the applicant’s dialect differs from that of the owner of the company.

Data from this live TV show has several advantages that fit the purpose of our study. First, the format of the show’s interview process allows us to clearly identify the social network signaling effect that language can serve. The interview consists of two stages. In the first stage, applicants briefly introduce themselves. They identify the area of the country they are from during this phase. The dialect spoken in the applicant’s region will be known to the business owner making the hiring decision. Applicants who successfully proceed to the second stage present more information about themselves (such as personal experiences and skills). According to Spence the signaling effect which dialect serves would be weaker when more information is observed by the firm’s manager. The coefficients of dialectal background on the hiring decision is smaller from the second than from the first stage of the interview. Second, during the whole interview process, people communicate using standard Chinese (Mandarin) rather than dialects. Because owners are unaware of the applicant’s dialect skills, we can eliminate the possibility that hiring outcomes are determined by individual language ability. Our results that firm owners select candidates from a different dialect region than their own. This suggests that firm’s demand workers with different dialect backgrounds because of their possession of social capital rather than their language skills. An applicant with a dialect distant from that of the owner provides the firm with an economically beneficial entrée into social networks particular to that dialect which the firm may not heretofore been able to access. Third, some may argue that what we’ve found could be the results of cultural preference (bias) or social identity. For instance, people feel more connected and culturally affiliated with someone from the same town or same region. If that is the case, we would find that it is more likely the jobs would be offered to the applicant who came from the same town (region) as the firm managers. In fact, our data does not support this. Because we measure social capital by dialect similarity, and Chinese dialectal division is not consistent with administrative border, we could further separate hometown fellow effect from dialectal effect. Fourth, each applicant is interviewing by ten firms at the same time. Their personal attributes do not change by who interviews them, or by whether they have receive the offer. It is possible for us to capture both observable and unobservable personal abilities by controlling for fixed effect. Fifth, with the information of both applicant educational background and the industry of the firm, we are able to test whether and how well applicant’s occupational ability matches the demand of firms.

Our preliminary results support language as signal of social network hypothesis, and even more, we find that dialect’s role as social capital is even stronger when we focus on growing firms, on sales positions where the job performance is highly depended on the working place. Our study provides a third channel through which dialect could influence economic decisions by serving as a signal of social capital by playing important role in firm’s hiring decisions. The novelty of the database used in this study provides information from both the demand and supply sides of the job search process. It helps us to understand the black box of hiring decisions. Our study may also improve the current understanding on the positive relation between labor diversity (including education, ethnic, language and social capital) and the firm’s productivity.