Learning from Neighbors and Differentiating Export Quality

• Published date: 01 July 2024
• Author: Qiming Liu,Bin Qiu,Huw Edwards,Bo Gao
• Date: 01 July 2024
• DOI: http://doi.org/10.1111/cwe.12539

©2024 Institute of World Economics and Politics, Chinese Academy of Social Sciences

China & World Economy / 1–32, Vol. 32, No. 4, 2024 1

Learning from Neighbors and Dif‌f erentiating

Export Quality

Qiming Liu, Bin Qiu, Huw Edwards, Bo Gao*

Abstract

This paper explores how learning from neighboring firms affects new exporters’ product

quality. It builds a Bayesian learning model to study how new exporters revise their

prior beliefs about foreign customers’ preferences for product quality from neighboring

pioneering exporters. The model shows that a new exporter improves its product quality

when it receives a positive quality-preference signal from its neighbors. The learning

process of a firm depends on the number of neighbors, the level and heterogeneity of

their export quality, and its own prior knowledge of the market. Highly disaggregated

firm–product–country level transaction data provide robust evidence for this. The results

also suggest that the impact of neighboring signals on a new exporter ’s quality can be

channeled through the importation of high-quality intermediate inputs and more fixed

investment. Learning effects are heterogeneous across firms and learning can influence

other aspects of export performance.

Keywords: Bayesian learning, export quality, information spillovers, uncertainty

JEL codes: D8, F1, F2

I. Introduction

A growing body of literature indicates that quality is essential in determining the pattern

of international trade flows across countries. Quality is also the most significant source

of firm heterogeneity (Hottman et al., 2016) and is increasingly important in firm’s

competitiveness in international market. The existing literature on the determinants of

firm-level export quality focuses mainly on technology and trade costs (Bastos and Silva,

2010; Fan et al., 2015). Little is known about the influence of uncertainty. Compared with

*Qiming Liu, PhD Candidate, School of Economics and Management, Southeast University, China. Email:

lqmhuali@163.com; Bin Qiu (corresponding author), Professor, School of Economics and Management,

Southeast University, China. Email: b_qiu@126.com; Huw Edwards, Senior Lecturer, Business school,

Loughborough University, UK. Email: T.H.Edwards@lboro.ac.uk; Bo Gao, Lecturer, Business school,

Loughborough University, UK. Email: B.Gao@lboro.ac.uk. This research was supported financially by the

National Social Science Foundation of China (No. 20AJY014). The authors thank two anonymous reviewers

for their helpful comments and suggestions for improving this paper.

Qiming Liu et al. / 1–32, Vol. 32, No. 4, 2024

©2024 Institute of World Economics and Politics, Chinese Academy of Social Sciences

2

domestic markets, foreign markets are associated with greater uncertainty because it is

harder for firms to collect and verify information from foreign markets (Appelbaum and

Kohli, 1997). Recent research has highlighted the significant impact of information friction

on inte rnational trade and has documented considerable uncertainty faced by new exporters

as evidenced by turnover rates and firm-level market selection behavior (Eaton et al.,

2008). Firms can adjust their export strategy by self-learning and experimentation but

they also obtain information from neighbors to reduce uncertainty due to the high sunk

costs of self-discovery. There is a small body of literature about the impacts of local

information spillovers on the intensive and extensive margin of trade but little work has

been conducted on its impact on firm-level quality decisions.

This paper establishes a learning model to study how new exporters learn from

their neighbors about quality preferences of foreign markets. The model incorporates

the Bayesian updating process of DeGroot (2005) into a multicountry model with

heterogeneous firms and endogenous product quality, as in Flach and Unger (2022). In

this model, the optimal export quality of a firm is associated with three factors: firm-

specific productivity, product–market specific quality preferences, and firm–product–

market specific appeal. It is assumed that a new exporter knows its own productivity

before entering a foreign market but is uncertain about product–market specific quality

preference and firm–product–market specific appeal. Based on information revealed by

neighbors’ export quality in that foreign market, a firm updates its prior beliefs regarding

the product–market specific quality preference. As the observed neighbors’ export

quality could be influenced by their unobserved firm–product–market specific appeals,

signals about the foreign market’s product quality preferences might be noisy. Based

on the standard learning model, observed signals converge to the actual state of quality

preference when more neighbors provide information and the firm–product–market

specific appeals noise tends to average out to zero. The model predicts that a positive

signal about the quality preference of a foreign market inferred from neighbors selling

there encourages new exporters to increase their export product quality. This learning

effect will be stronger if there are more neighbors to learn from, or if new exporters have

less knowledge about the market, but it will be weaker if the signals are noisier.

We found supporting evidence using transaction-level data covering Chinese

industrial exporters in 2000–2007 from the Chinese Customs Database (CCD) and

the Annual Survey of Industrial Firms (ASIF). We estimated firm-level product

quality following Khandelwal et al. (2013) and measured the average growth rate of

neighboring firms’ export product quality as the signal of product quality preference. We

found that new exporters’ export quality in a foreign market was positively correlated

with the signal and the learning effect strengthened with the number of neighbors that

©2024 Institute of World Economics and Politics, Chinese Academy of Social Sciences

Learning from Neighbors and Dif‌f erentiating Export Quality 3

released the signal. More specifically, a sample mean growth rate of neighbors’ product

quality was associated with 0.216 percent increase in new exporters’ product quality on

average, and at the sample mean of the growth rate of neighbors’ export product quality,

a 1 standard-deviation (SD) increase in the (log) density of neighboring firms exporting

to a country is associated with 0.15 percent higher product quality for new exporters in

the same country. Endogeneity issues were addressed carefully by introducing relevant

controls and fixed effects in the baseline regression, and the estimated impact held in a

series of robustness checks.

To further verify that it was a learning effect, we examined the relationship between

export quality, the precision of signal, and a firm’s own prior knowledge about the market.

Our results showed that dispersed signals reduced the willingness of new exporters to

learn from neighbors and the learning effect also became weaker if the new market had

similar characteristics to markets previously served by the firms. Moreover, the impact of

signals on export product quality depended not only on the neighbors’ geographic distance

from the market but also on the heterogeneity of new exporters. We found that the impact

of signals on product quality of new exporters is channeled through the imports of high-

quality intermediate input and more fixed investment. We also found that a positive signal

encouraged new exporters to increase their initial export price and sales but had a negative

impact on the post-entry growth in their product quality.

The rest of this paper is structured as follows. Section II reviews the literature

closely related to our research. Section III lays out the theoretical model. Section IV

describes the econometric model and the data used in our empirical analysis. Empirical

results are collected in Section V and Section VI concludes the paper.

II. Literature review

This paper is closely related to recent research on information spillovers in international

trade. Existing studies have mainly focused on the impact of firms’ own trade experience

on their export strategy. Arkolakis et al. (2018) provided evidence that learning about

the demands of foreign markets through an accumulation of exporting experience can

be an important driver of firms’ export dynamics. Bai et al. (2021) showed that Chinese

industrial firms can learn from processing trade experience to enhance production

efficiency and to understand how to improve product appeal to a specific foreign market,

which can benefit their ordinary trade performance. Schmesier (2012) found that the

geographic expansion of exporting firms is gradual. They found that older exporters

tended to enter new destinations that were similar to previous export markets to benefit

from the reduction in entry cost and risk. Besides self-learning effects, Fernandes and