Identity-Based Discrimination: Evidence from Turkish and Syrian Firms

As part of a larger randomized controlled trial—soon to be pre-registered—that offers small and medium-sized enterprises (SMEs) in Gaziantep, Turkey, access to a business matchmaking initiative designed to reduce local networking frictions, we will collect survey data during the recruitment phase. Embedded within this baseline data collection, we will implement an additional intervention to examine how different informational treatments affect firms’ willingness to participate in the matchmaking initiative.
As part of the baseline, firms are randomly assigned to view one of three video versions describing the programs. The core content of the videos is identical, except for two key experimental variations: (1) whether the video explicitly states that both Turkish and Syrian firms will participate in the matchmaking program (the identity treatment), and (2) whether it includes disaggregated statistics comparing the export performance of Syrian and Turkish firms (the disaggregated statistics treatment).
Our goal with this design is to test two key questions. First, to what extent does the salience of ethnic identity influence firms’ willingness to participate in the matchmaking program and shape their attitudes toward inter-ethnic business collaboration? Second, can information about the economic performance of firms from different ethnic backgrounds affect firms’ decisions—particularly those from a different ethnicity—to engage in partnerships with more diverse business counterparts?

2025-04-22

2025-06-30

Our primary outcomes are: (1) Firm’s willingness to participate in the matchmaking program; and (2) Belief updates, measured immediately after the video, capturing shifts in perceptions of Syrian firms’ export performance and product quality.

These include but are not limited to: (1) Attitudes toward outgroup firms, including perceived trust, competence, and product quality, and (2) Business aspirations among Syrian respondents.

We are implementing a randomized video experiment embedded within the baseline survey. This is different from our main intervention, which will be pre-registered later once the baseline data collection is completed. Firms recruited for the baseline survey were sampled from business registries provided by the Gaziantep Chamber of Commerce and Building Markets.

All participating firms are shown a generic video introducing the study, its partners, and the services to be offered in the main intervention (including a business matchmaking service). In addition to this generic video, firms will be randomly assigned to also watch one of three videos (and we have Turkish and Arabic versions of all videos). These videos are identical except for the following two characteristics.

1. Group 1: No Information
a. The video does not explicitly mention that the matchmaking service can result in interactions with Turkish or Syrian firms.
b. The video presents aggregate statistics (not discriminating by Turkish or Syrian firms) on the export activity of firms in Gaziantep. Specifically, the video states that 38% of all SMEs outside the organized industrial zone (OSB) export internationally, and that 20% of them export to Germany, a market known for its high-quality standards.
2. Group 2: Identity
a. The video explicitly mentions that the matchmaking service can result in interactions with Turkish or Syrian firms.
b. The video presents aggregate statistics (not discriminating by Turkish or Syrian firms) on the export activity of firms in Gaziantep (the same sentence as the No Information Group).
3. Group 3: Disaggregated Statistics
a. The video also explicitly mentions that the matchmaking service can result in interactions with Turkish or Syrian firms.
b. The video presents disaggregated statistics on the export activity of Turkish and Syrian firms in Gaziantep. In particular, it reports that 65% of Syrian SMEs export, compared to 30% of Turkish SMEs, and that 20% of exporters in both groups trade with Germany.

Firms assigned to Groups 1 and 2 receive the same aggregate statistics on the export behavior of SMEs in Gaziantep. Only firms in Group 3 are shown disaggregated, group-specific data comparing the export performance of Turkish and Syrian firms. The comparison between Groups 1 and 2 identifies the net effect of explicitly learning about group identity, while the comparison between Groups 2 and 3 captures the effect of receiving disaggregated performance information (conditional on identity being made salient).

Not available

Randomization done in office by a computer

Randomization occurs at the individual firm level and is stratified by ethnicity (Turkish/Syrian).

No

We expect between 2,200 and 2,300 Turkish firms and between 700 and 800 Syrian firms to complete the baseline survey. The exact numbers will depend on take-up rates and the capacity of the survey firm.

We expect between 2,200 and 2,300 Turkish firms and between 700 and 800 Syrian firms to complete the baseline survey. The exact numbers will depend on take-up rates and the capacity of the survey firm.

No Information: 50%, Identity: 25%, Disaggregated Statistics: 25%.