“In recent years, academia and industry have paid close attention to economic policy uncertainty. Indeed, the greatest certainty of this era is that it is full of uncertainty. The main indicator currently used to measure economic policy uncertainty is EPU index developed by BBD (2016). However, there is only one EPU index for each region or country at each time point, so it cannot be distinguished from time fixed effects in measurement, and it cannot distinguish the policy uncertainty of different enterprises. Feel the difference.”

——Public account: Professor Nie Huihua

 In 2020, Teachers Nie Huihua, Ruan Rui and Shen Ji published the article "Enterprise Uncertainty Perception, Investment Decisions and Financial Asset Allocation" in the "World Economy" magazine, providing a way to calculate the economic calculation at the enterprise level. policy uncertainty index method, and use this data to analyze the impact of uncertainty feelings on corporate investment and financing. After the paper was published, it received widespread attention from the academic community.

 In order to promote research on economic policy uncertainty and reduce unnecessary time costs, Professor Nie Huihua and his research team made the Enterprise Economic Policy Uncertainty Perception Index available to the academic community. With authorization, CnOpenData has included this data in the public data section of the CnOpenData official website for the convenience of scholars.

  To download this data and for more information, please jump to White Shark Online.

Database Application Guide

Data construction method

 This article's indicators for measuring economic policy uncertainty are extracted from the text of annual reports of listed companies using text mining methods.

 With the development of computer technology, it is increasingly common to introduce unstructured data such as text into corporate finance research (Tetlock, 2007; Li, 2008; Tetlock et al., 2008; Loughran and McDonald, 2014, 2016). This article refers to the practices of Baker et al. (2016) and Hassan et al. (2019) and uses the "vocabulary method" to filter specific content texts. If specific words appear in a piece of text, the text is identified as expressing certain specific meanings. text. This article believes that if "policy words" and "uncertainty words" appear in a sentence at the same time, it is considered that this sentence is the content of the annual report writer's statement that the company faces economic policy uncertainty.

 The specific methods are as follows:

  • First, use a format conversion tool to convert the PDF file of each listed company's annual report into a text file, and use regular expressions to extract the content of the "Management Discussion and Analysis" (MD&A for short, in some annual reports it is the "Board of Directors Report"). Remove all numbers, English letters, and all punctuation and special symbols except periods.
  • Then, the MD&A text is divided into sentences using Chinese periods as delimiters. Considering Chinese language habits, this article uses sentences as the basic unit of analysis. Assume that the number of MD&A sentences in the annual report of listed company i in year t is s. Use the programming language Python to call the jieba word segmentation module to segment each sentence, and remove stopwords while segmenting. In order to minimize the ambiguity caused by word segmentation, this article defines a user vocabulary list during word segmentation. The vocabulary list includes the full names and abbreviations of all A-share listed companies, accounting account names, words indicating uncertainty used in subsequent text processing, and words related to Words related to government (policy) meaning. After word segmentation, each sentence becomes a combination of a series of words, and then the following operations are performed on each sentence (s) one by one: search for the words that appear in each sentence. If a word indicating uncertainty appears, it is considered to indicate uncertainty. Deterministic sentences; if words related to government, policy, etc. and words expressing uncertainty appear simultaneously in a sentence, it is considered to be a sentence expressing policy uncertainty (P). The economic policy uncertainty (FEPU) faced by enterprises is measured by the ratio of the number of uncertainty words (n) in the economic policy uncertainty sentence to the total number of words in MD&A (N).


Note: Any use of enterprise-based economic policy uncertainty perception index must indicate the source of citation:

  • Nie Huihua, Ruan Rui, and Shen Ji, 2020, "Corporate Uncertainty Perception, Investment Decisions and Financial Asset Allocation", "World Economy", Issue 6, pp. 77-98.

Data update frequency

Updated from time to time