Fostering theoretical innovation is conducive to building an independent knowledge system. Photo: TUCHONG

Throughout the history of economic thought, the evolution of economic theory has been closely tied to the development of primary productive forces and changes in social relations of production. Given that the growth model and underlying logic of the digital economy are fundamentally different from those of the traditional economy, established economic theories are unable to adequately explain the new phenomena and characteristics brought about by the digital economy. Therefore, it is necessary to accelerate the innovation of economic theories that recognize data as a key productive factor and digital technology as a productive force in order to provide robust theoretical support and practical guidance for the sustained flourishing of the digital economy.

Unique theoretical properties 

Data is the core driver of digital economic development and a defining characteristic that distinguishes it from traditional economic forms. The key to theoretical innovation in the digital economy lies in understanding and grasping the economic attributes of data. Unlike traditional factors of production such as capital and labor, data exhibits two distinct characteristics: endogenous generation and non-rivalry in its use. These traits make data more akin to the technology factor representing total factor productivity. Specifically, data is generated endogenously. The more dynamic the economic activity becomes, the more data is produced, which in turn further stimulates economic growth. This mechanism mirrors Paul M. Romer’s Endogenous Growth Model, where technological innovation drives growth from within. Additionally, with a given set of inputs, enterprises can achieve more valuable outputs by utilizing big data, which is consistent with the principle of enhancing output efficiency by using technology or total factor productivity in standard economic models. Finally, data and technology are non-rivalrous. A single technology can be used by multiple entities simultaneously without interference, just as a data set can be owned and utilized by multiple entities or used repeatedly by the same entity. This stands in stark contrast to the characteristics of capital and labor factors.

However, data, as a factor, exhibits significant economic differences from the technology factor in terms of its economic attributes. In the production process, technological innovation and progress usually require large resource inputs, while the generation of big data is not dependent on the incentive mechanisms of property rights protection or profit sharing.  In the allocation phase, the ownership of technology is usually clear and easy to define, but establishing the ownership of data is more complex. Since data assets are usually composed of numerous individual contributions, it is difficult to assign corresponding ownership to each part of the data asset. In the circulation phase, large platform enterprises can more effectively control the directed circulation of data. The circulation of technology is often dependent on human capital, and the free flow of employees between enterprises makes it impossible to control and track the transfer of technology. In contrast, data exists in the form of complex data sets and cannot be easily transferred through human capital. In the consumption phase, data has higher tradability than technology due to its divisibility and describability. In addition, sellers of data products can clearly describe the content of the data set without revealing specific information, an advantage not shared by technology products.

Economic measurement scale

Economic measurement is the prerequisite and foundation for empirical research. It aims to provide a comprehensive description, portrayal, and summary of economic operations and development based on empirical data, and its multi-disciplinary integration feature is becoming increasingly prominent in the digital economy era. Key elements of digital economic measurement include the valuation of data assets and the assessment of the digital economy’s scale. 

Data assets are the key production factor for the deepening development of the digital economy, encompassing various types of information and digital assets that both possess economic value and are measurable. Due to data’s non-physical nature, its diverse forms, volatile value, and the potential for processing, the academic and business communities have not yet reached consensus on how to define data assets. However, there is broad agreement that two core conditions must be met for data to be considered an asset: clear economic ownership and profitability. The valuation of data assets can draw upon methods used for intangible assets, primarily the revenue approach, market approach, and cost approach, along with their derivative methods. Among them, the cost approach can estimate the benchmark value of data assets, but may lead to conservative valuations.  The calculation of the digital economy’s scale should dynamically adapt to the evolving definition of the digital economy, encompassing all economic activities driven by the latest digital technologies and their integration with traditional industries. Nevertheless, several theoretical challenges persist in this area. 

The first challenge is the issue of measurement caliber. Narrow-caliber digital economy scale measurement mainly focuses on the digital industrialization field. It is generally aligned with traditional national economic accounting frameworks. In contrast, wide-caliber measurement also covers the industrialization of digital technologies, reflecting the penetration of digital technology into socioeconomic development. These differing approaches offer various perspectives on economic development and form a crucial foundation for empirical research on the digital economy. 

The second is the measurement of zero-price goods such as digital services. The social, information acquisition, and entertainment functions provided by digital platforms significantly enhance the utility value for individuals and families. Based on traditional utility theory, concepts such as equivalent variation and compensation variation, in combination with the latest findings from experimental economics, can be used to accurately capture the increase in utility value these zero-price goods offer. 

The third is the production measurement from the perspective of saving output. The widespread application of information technology has significantly reduced the demand for traditional consumer goods and services (such as transportation) without diminishing their utility. For new measurement, due consideration should be given to the consumer surplus value of such saving-type behaviors. 

The fourth is the measurement of subjective factors such as emotions and sentiment. Participants in the digital economy possess consciousness and cognitive abilities, meaning that traditional economic assumptions of rational actors are insufficient. To gain a more comprehensive understanding of economic behavior, subjective factors must be incorporated into the measurement framework of the digital economy.

Casual inference 

The digital economy provides a “treasure trove” of massive data that reveals the development patterns of this emerging field. To effectively mine this treasure, appropriate methods and tools are essential. Therefore, digital economy research should fully integrate the findings from various disciplines, particularly econometrics and statistics, alongside innovative methodologies that have emerged with the advancement of digital technologies, such as data science and artificial intelligence. Applying these methods to the empirical study of the digital economy can effectively manage large-scale data in complex scenarios, helping to uncover the underlying patterns of  digital economic development. Although existing big data analysis methods and tools have become increasingly sophisticated, the exploration of the intrinsic logic in the data—especially causal relationships—still relies heavily on the guidance of economic theory. 

How to measure the multiplier effect of data is an important theoretical and practical challenge. Here, the causal inference methods of econometrics can play a crucial role. Causal inference, which won the Nobel Prize in Economics in 2021, focuses on the economic causal relationships between variables rather than superficial statistical correlations, allowing for the optimizing of decision-making strategies. By applying causal inference, key, causal, and decision-making information extracted from one dataset can be generalized and applied to multiple scenarios, thus realizing the multiplier effect of data. For example, in the field of public health, causal inference can identify the most effective health interventions from health data sets and promote these measures to improve overall health outcomes. However, if a data set does not contain the causal relationship among key variables and merely records individual descriptive statistical quantities, the information contained in the data set is difficult to generalize across different scenarios, and the multiplier effect of the data remains unrealized.

To advance theoretical research on the multiplier effect of data factors from the perspective of causal inference, two key areas should be explored. First, it is suggested to further develop causal identification methods suitable for new scenarios. In the digital economy scenario, datasets primarily originate from non-experimental data generated by economic activities, making direct causal relationships difficult to identify. This requires relying on existing econometric theories while innovating causal identification methods for observational data in the digital economy, using the common structural characteristics found in homogeneous datasets. Second, it is suggested to construct quantitative indicators of causal relationships within data sets, considering multiple factors such as data quality, the strength of causal relationships, and data applicability. Such indicators can also provide auxiliary support for data product pricing.

It is evident that advancing theoretical innovation, measurement innovation, and research method innovation is a necessary condition for building the independent knowledge system of the digital economy. A comprehensive understanding of the natural attributes and economic characteristics of data, along with an accurate grasp of the distinct roles that data factors play in social reproduction compared to traditional factors, provides a solid foundation for theoretical innovation in the digital economy. Accurate economic measurement is the prerequisite for empirical research in the digital economy, and addressing the challenges of defining and valuing data assets can help provide a comprehensive and precise depiction of the development of the digital economy. Research methodology innovation is an important means of advancing theoretical innovation in the digital economy. This innovation should integrate emerging technological and methodological approaches from econometrics, statistics, data science, and artificial intelligence, and explore the economic causal relationships contained in the data under the guidance of economic theory.

Hong Yongmiao is a professor from the School of Economics and Management at the University of Chinese Academy of Sciences. Xie Haitian is an assistant professor from the Guanghua School of Management at Peking University. Xi Jin is an assistant research fellow from the Academy of Mathematics and Systems Science at the Chinese Academy of Sciences.

Edited by ZHAO YUAN