Panel Paper: Making Information Work for Low-Income Students in College Choice: Experimental Evidence Under Centralized Admissions

Saturday, November 4, 2017
Water Tower (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Xiaoyang Ye1, Yanqing Ding2, Liping Ma2, Yinduo Wu2 and Jin Yang2, (1)University of Michigan, (2)Peking University


In this paper, we provide novel evidence on the role of information in low-income students’ college choices under a centralized admissions system. We characterize the differences between information and knowledge that students need to navigate the complex college application process under centralized admissions. Using a stratified school-cluster RCT among all public high school graduates (about 60,000) in one of the poorest provinces in China, we find substantial heterogeneities in treatment effects across intervention designs and student preferences.

The central policy problem is that, even when low-income students are qualified for college admission and could have attended high-quality colleges with low costs, they still make undermatched college choices (Hoxby & Avery, 2013; Dillon & Smith, 2017). Over the past decade, a growing body of literature and policy discussion has focused on the role of behavioral interventions in improving college-going decision making, given the fact that low-income students lack information and guidance about effective college choices (e.g., Deming & Dynarski, 2010; Bettinger et al., 2012; Hoxby & Turner, 2013; Dinkelman & Martínez, 2014; White House, 2014; Castleman et al., 2015; Hastings et al., 2015; Page & Scott-Clayton, 2016).

Existing evidence concentrates on the decentralized admissions systems (particularly in the United States). We still know little about how information works under centralized systems, which exist in many countries (e.g., Australia, Brazil, Chile, China, India, UK) and have very different institutional features. Decentralized systems may close the income gap in informational barriers by streamlining and simplifying the process by which students make college choices; however, there are also features that bring informational barriers to low-income students relative to their high-income peers, for example, the risky “gaming” application like in the K-12 centralized school choice (Chen & Kesten, 2017).

To make progress on this important question, we design and implement a school-cluster RCT in China, where there is the largest student-college matching market in the world with a well-established centralized college admissions system. We have employed four treatment arms based on the levels of instruction, including (1) general information about returns to colleges and majors (“simple information”), (2) a printed guidebook about how to make informed college application (“comprehensive information”), (3) school-level public lectures on analyzing all relevant information (“semi-knowledge”), and (4) individualized application assistance through online platforms (“knowledge”). Combining complete administrative data on student demographic, test, application, and admission information with survey data on student background, preferences, and career expectations, we estimate both the average and heterogeneous treatment effects.

Our paper contributes to the literature and relevant policies in several ways. We provide some of the first evidence on the cost-effectiveness of different informational interventions on low-income students’ college choices under centralized admissions systems. We have tested not only the popular and promising programs under decentralized systems, but also programs with new designs for centralized admissions. We investigate a wide range of heterogeneities in the treatment effects to explore potential underlying mechanisms. Our results will inform policymaking in improving low-income students’ college choices in both developing countries and developed countries.