A young Pakistani livestreamer is promoting products to international consumers from a Hainan-based tech company. In recent years, Hainan Province has made cross-border e-commerce a key focus for advancing its digital economy. Photo: IC PHOTO

As the dominant economic form following the agricultural and industrial economies, the digital economy has become an important engine driving global economic growth and a key instrument in transforming modes of development. The dynamism of the digital economy is vividly reflected not only in real-world economic activities but also in its deep penetration into academic research in economics. At present, studies, conferences, and institutions related to the digital economy have proliferated across virtually every corner of the academic community.

However, current research on the digital economy faces several prominent issues in areas such as conceptual definition, indicator selection, and discussion of underlying mechanisms. A particularly common problem is that many studies fail to illuminate the core characteristics of the digital economy. Instead, they merely examine how one indicator associated with the digital economy affects another economic or social variable—without clarifying which specific feature of the digital economy such an effect reflects, or how it differs from patterns observed in the pre-digital era. The abundance of such research has, to a certain extent, resulted in a situation where academic output on the digital economy in China has expanded rapidly in volume, while its overall quality remains in need of improvement.

Given this reality, the academic community urgently needs to reflect on what constitutes a sound research paradigm for the study of the digital economy, and how future scholarship in this field can advance in terms of research objects, analytical perspectives, data, and methodological approaches.

Current research: overly generalized, lagging behind innovation

A review of the literature on the digital economy reveals two notable phenomena: first, most studies are empirical in nature; and second, compared with foreign-language publications, research on the digital economy receives much more attention within Chinese-language scholarship. From the perspectives of conceptual definition and quantitative measurement, research objects, socio-economic impacts, as well as data and identification methods, a brief review of the current state of empirical studies shows that research objects tend to be broad and generalized, with measurement often conducted at an excessively microscopic level. In addition, research perspectives fail to fully highlight the core characteristics of the digital economy. Third, research data remains relatively traditional. Fourth, research techniques have not kept pace with the frontier of technological innovation.

Although China’s digital economy ranks among the most advanced in the world, the quality of domestic academic research on the topic still has considerable room for improvement. Progress should be pursued by reforming research objects, perspectives, data, and technological methods, thereby driving a paradigm shift in the study of the digital economy.

Research objects: from ‘broad and general’ to ‘focused and refined’

While the digital economy encompasses a wide range of domains—including digital industrialization and industrial digitalization—empirical studies that attempt to examine its impacts in a sweeping manner often lack analytical depth. As several years have passed since the concept of the digital economy was first introduced in China, and as its development in real-world economic activities has been remarkably robust, academic inquiry should now strive to engage with digital economy issues at a deeper level and abandon traditional, coarse measures.

To enhance the depth and credibility of future research, scholars should shift their focus from “broad and general” analyses to “focused and refined” investigations. Only by addressing major questions through narrowly defined and well-specified lenses can domestic research engage in substantive dialogue with authoritative international scholarship and further enrich the discipline of economics.

First, researchers could conduct detailed analyses within specific subfields of the digital economy—such as digital infrastructure development, digital technology innovation, digital firms’ location decisions, or international cooperation in the digital economy—rather than attempting to capture the entire concept in a single study. Second, research could focus on representative economic forms within the digital economy. Third, in the process of refining research objects, scholars should remain attentive to their diversity, conducting systematic comparative studies across different yet related subjects.

Research perspectives: revealing new socioeconomic dynamics through core features of digital economy

The most crucial task of academic research on the digital economy is to uncover how its deep penetration transforms, amplifies, weakens, or reverses certain traditional behaviors, thereby reshaping specific economic and social phenomena or relationships in the digital age. The logical mechanisms behind these transformations should reflect the core features of the digital economy. In essence, the digital economy is characterized by its ability to reduce search, replication, transportation, tracking, and verification costs. Although a single paper cannot cover all these characteristics, it should at least clearly engage with one of them.

For example, regarding transportation costs, advances in information technology enable information to circulate across vast spaces at nearly zero marginal cost. The near-elimination of transportation costs has led many scholars to reconsider whether geographical distance remains an important factor in the digital era—an issue that has sparked substantial debate in academia. Another example concerns online consumption, which differs from offline consumption in its dependence on geographic space. Analyzing how variables such as housing prices (or other spatial factors) affect online consumption patterns through the lens of spatial distribution constitutes an excellent research perspective on the digital economy.

Moreover, while most existing studies emphasize the positive outcomes of digital economic development, current literature tends to lack “cool-headed reflection” on the problems it exposes. Future research should thus examine the potential challenges and negative externalities of the digital economy.

In short, discussions of the mechanisms through which the digital economy exerts its influence should increasingly originate from its core characteristics, rather than from conventional analytical frameworks or “formulaic” mechanisms.

Research data: employing multi-source big data under deep digital penetration

From the perspective of research data, much of the existing literature still relies on traditional datasets that lack distinctive features. In fact, as the digital economy has become deeply embedded in all sectors, an extensive variety of big data has been accumulated, offering rich possibilities for exploration and utilization by researchers. Three typical categories of digital economy–era big data are illustrative:

The first is platform enterprise data. Platform enterprises are central to the digital economy. Beyond facilitating the digital transformation of everyday life (e.g., online shopping, ride-hailing), they accumulate massive amounts of user information. This data is invaluable for monitoring macroeconomic dynamics and analyzing economic activities and can be effectively leveraged in empirical studies.

The second category is social media data. With the advancement of information and communication technologies, social media platforms have become crucial channels through which individuals generate, share, and exchange information and opinions. Owing to their decentralized nature and massive user bases, these platforms produce vast quantities of analyzable data that can significantly enrich empirical research in economics.

The third is economic geography data. Economic geography big data refers to datasets that include spatial location information attached to various economic indicators. Compared with traditional economic data, it possesses explicit spatial attributes and high-resolution granularity. Such data has been widely applied in urban and regional economics and development economics, and is equally valuable for research in the digital economy.

Research techniques: harnessing frontier AI for empirical analysis

In the era of the digital economy, the way data is generated has undergone profound change. Digitalization now permeates all aspects of daily life—consumption, transportation, housing, and beyond—producing massive amounts of traceable big data. Yet such raw data is often “noisy, unstructured, and messy,” comprising text, images, and audiovisual materials.

The need to process these complex datasets has greatly accelerated the adoption of artificial intelligence (AI) methods, particularly machine learning, in digital economy research. Although such algorithms have long been used by internet companies and financial institutions, their use in academic studies still has considerable room for expansion. Incorporating frontier AI technologies—represented by machine learning—into academic research will be a crucial methodological trend in future studies of the digital economy. Specifically, these techniques can be applied in three core areas: prediction, variable generation, and causal inference.

In prediction, unlike traditional econometric research, which emphasizes unbiased and efficient parameter estimation, machine learning identifies complex patterns and trends from vast historical datasets in a data-driven manner, yielding highly accurate predictive results. For variable generation, machine learning methods can extract and construct novel indicators from massive, unstructured data sources and introduce them into traditional social science research. In causal inference, identifying causal relationships among variables in complex economic systems is a central task in the social sciences, especially in economics. Machine learning offers valuable tools for improving causal identification by better detecting and controlling confounding factors, constructing comparison groups under frameworks such as difference-in-differences or matching methods, and identifying heterogeneous causal effects and external validity.

Furthermore, the development and application of AI technologies are likely to shift the research paradigm from theory-driven to data-driven inquiry. Such a transformation can reduce biases introduced by theoretical preconceptions, enable researchers to explore data more freely, and uncover new causal relationships and economic mechanisms—thereby fostering a new, data-centered scientific mindset.

Guo Feng (professor) and Cao Youbin are from the School of Public Economics and Administration at Shanghai University of Finance and Economics.