Frontiers in Business, Economics and Management ISSN: 2766-824X | Vol. 17, No. 1, 2024 324 Heterogeneity Analysis of Economic Policy Uncertainty on Corporate Investment Based on Panel Threshold Regression Chongrong Qiu * China-ASEAN School of Economics, Guangxi University, Nanning, China * Corresponding author: Chongrong Qiu (Email: 121180875@qq.com) Abstract: As the global landscape evolves rapidly, economic policies must constantly adapt to the volatile economic situation, leading to economic policy uncertainty (EPU). Firms, as microeconomic entities, need to continuously adjust their strategies in response to these changes. Given that corporate investment is one of the most crucial factors in business operations, it is of paramount importance to investigate how the characteristics of corporate investment are influenced by EPU. Using the panel threshold regression method, we delve into the nonlinear impact of EPU on corporate investment. By collecting data from 2000 to 2023 for A-share listed companies in Shanghai and Shenzhen, as well as macroeconomic data, we empirically find that EPU exerts significant and differentiated effects on corporate investment across four dimensions: leverage level, company size, profitability, and growth potential. Keywords: Economic Policy Uncertainty, Corporate Investment, Panel Threshold Regression. 1. Introduction Corporate investment, serving as the core driving force of national economic growth, profoundly influences both the nation and the enterprises themselves. By expanding production scale and enhancing production efficiency, corporate investment bolsters market competitiveness and ensures enterprises maintain a competitive edge in the fiercely competitive market. Additionally, corporate investment drives technological innovation and industrial upgrading, facilitating the optimization and transformation of the national economic structure. It also stimulates domestic demand, creates more employment opportunities, and boosts consumer spending power, thereby contributing to the steady growth of the national economy. However, in the face of increasingly complex and volatile economic environments, enterprises face significant challenges in formulating effective investment strategies. Economic policy uncertainty refers to the difficulty economic actors have in accurately predicting whether and when the government will adjust existing economic policies, as well as the specific timing and manner of such adjustments. This uncertainty affects the decision-making processes of market entities, as they must confront the risks associated with potential policy changes. Economic policy uncertainty significantly impacts corporate investment choices. When economic policy uncertainty exists, it increases the potential risks faced by corporate investment, making it more difficult for enterprises to forecast future economic conditions, thereby suppressing their investment intentions. While research on the relationship between economic policy uncertainty and corporate investment has made some progress, there are still some shortcomings. Most current literature focuses only on the average impact of economic policy uncertainty on corporate investment, lacking an examination of the differences in how uncertainty affects enterprises with different characteristics. Additionally, many studies limit their exploration of the heterogeneous effects of economic policy uncertainty on corporate investment to factors such as equity structure and industry affiliation, without considering other company characteristics that may be more important in determining the heterogeneous effects of economic policy uncertainty on investment. Lastly, many studies use direct grouping methods to investigate the heterogeneous effects of economic policy uncertainty on corporate investment, but these grouping criteria lack a solid basis. Therefore, this article employs threshold regression to automatically categorize samples and study the heterogeneous effects of economic policy uncertainty on corporate investment. Employing threshold regression analysis to study the heterogeneous impact of economic policy uncertainty (EPU) on corporate investment holds significant practical implications. Firstly, this method enables the identification of the patterns in corporate investment behavior across different levels of EPU. Specifically, the threshold regression model allows us to examine whether there are significant differences in corporate investment decisions before and after EPU reaches a certain critical value (i.e., the threshold value). This differentiated analysis helps reveal that the impact of uncertainty on corporate investment is not linear but exhibits different response patterns as uncertainty increases. Secondly, through threshold regression analysis, we can more accurately capture the different coping strategies employed by different types or sizes of companies when faced with EPU. Research into these heterogeneous effects provides valuable insights for formulating more targeted policy measures. Lastly, the application of this research method can assist governments and policymakers in better understanding the specific mechanisms through which EPU affects corporate investment. It provides a scientific basis for formulating effective economic policies. By gaining a deeper understanding of how the impact of EPU on corporate investment varies with industry characteristics, firm size, and other factors, policymakers can devise more nuanced policy 325 measures aimed at mitigating the negative effects of uncertainty on corporate investment and promoting healthy economic development. Therefore, this study focuses on Chinese listed companies as the research subject, collecting financial data of Shanghai and Shenzhen A-share listed companies from 2000 to 2023, along with selected macroeconomic time series data from China, to deeply explore the heterogeneous impacts of economic policy uncertainty on corporate investment. 2. Literature Review 2.1. Review of Research on Economic Policy Uncertainty The existing research literature on the definition of Economic Policy Uncertainty (EPU) can be broadly summarized from two perspectives: narrow and broad. From a narrow perspective, EPU refers to the difficulty for economic agents in predicting the specific details of monetary policy, fiscal policy, trade policy, and financial regulatory measures that governments may implement in the future. From a broad perspective, EPU encompasses three aspects: firstly, the unpredictability of policy changes themselves; secondly, the underlying factors leading to policy changes, such as political uncertainty or economic fluctuations; and lastly, the various phenomena and economic consequences arising from policy uncertainty. Due to differences in the selection of measurement indicators among different studies, there are also variations in their understanding and interpretation of the concept of EPU. Baker et al. (2016) propose that Economic Policy Uncertainty (EPU) should encompass both policy uncertainty and the economic uncertainty it induces. They point out that policy uncertainty mainly involves three levels: uncertainty among policymakers, uncertainty regarding the timing of policy formulation, and uncertainty about changes in policy content, whether in economic or non-economic areas. Based on this, when constructing a comprehensive EPU index, they selected keywords that reflect uncertainties in all these aspects. Many scholars believe that political factors are a key driver of policy changes and often use indicators measuring political uncertainty as a proxy for policy uncertainty. This practice can sometimes lead to conceptual confusion between the two. Pástor and Veronesi (2013) broadly define political uncertainty as the uncertainty surrounding government actions and the outcomes resulting from those actions. Baker et al. (2016) was the first to employ text analysis methods to construct a comprehensive and scientific Economic Policy Uncertainty (EPU) index. Since then, numerous researchers have cited this index in their empirical analyses or adopted similar methodologies to construct other types of EPU indices, such as indices for trade policy uncertainty, monetary policy uncertainty, and global policy uncertainty. Alexopoulos and Cohen (2015) utilized data from The New York Times to construct a General Economic Uncertainty (GEU) index using three different filtering methods. Furthermore, they developed a policy uncertainty index based on a subset of news articles from the GEU index. Tillman (2016) specifically constructed the Economic Policy Uncertainty (EPU) index for the United States by utilizing discussions about the Federal Reserve's exit from quantitative easing (QE) on the social media platform "Twitter." Alberto et al. (2015) quantifies trade policy uncertainty by utilizing the discrepancy between tariff commitments and actual tariffs, a metric also referred to as "tariff slack." The term "slack" here reflects the likelihood of a country increasing its most- favored-nation (MFN) tariff rates to their binding ceilings without needing approval from the World Trade Organization (WTO). Mumtaz and Surico (2018) constructed a SVAR (Structural Vector Autoregression) model to identify shocks to government spending, changes in tax policy, and the sustainability of public debt. By analyzing the volatility of these shocks, they quantified the uncertainty in fiscal policy. 2.2. Research on the Impact of Policy Uncertainty on Corporate Investment Arellano et al. (2019) found that increased uncertainty raises external financing costs and increases risk premiums, thereby affecting firms' investment and production activities. Husted et al. (2017) found that an increase in monetary policy uncertainty widens credit spreads and leads to a decline in output, exhibiting similar dynamic effects to those of a tightening monetary policy. Utilizing the Policy Uncertainty Index, Gulen and Ion (2016) conducted research that revealed a significant negative correlation between corporate capital investment and the overall level of future policy and regulatory uncertainty. This relationship is particularly evident for companies with higher investment irreversibility or greater dependence on government spending. The findings support the view that policy uncertainty inhibits corporate investment by inducing precautionary delays in investment due to investment irreversibility. Liu and Zhang (2020) designed a quasi-natural experiment based on China's supply- side structural reform in 2015 to uncover the causal relationship between economic policy uncertainty (EPU) and corporate investment and financing decisions. Using a Difference-in-Differences (DID) strategy, they found that EPU significantly curbed private enterprises' real investment and reduced their net debt issuance, while no similar effects were observed for state-owned enterprises. Z Jing et al. (2023) found that economic policy uncertainty can effectively curb overinvestment by enterprises, thereby reducing the risk of stock price crashes. 3. Empirical Design 3.1. Model Equation When examining factors that influence the dependent variable through affecting a specific variable, traditional methods often employ interaction term tests or grouping tests. However, interaction term tests are limited to scenarios where the impact of the measured indicator is either monotonically increasing or decreasing, which does not align with real economic conditions. Traditional grouping tests, on the other hand, simply divide the sample based on a certain influencing indicator, making it difficult to accurately reflect the impact of various factors on the dependent variable. Given the limitations of these two methods, recent years have seen the development and application of the "threshold regression" method, an improvement over nonlinear econometric models, in many research fields. Therefore, this study employs the following model: 1 2 , , it it it it it it it it Y X Control q Y X Control q      = +   = +    (1) Where itY is the dependent variable, itX is the 326 independent variable, Controlit represents the control variables, and itq is the threshold variable. 3.2. Variable Definitions and Data Sources (1) Dependent variable: Corporate investment: In this study, corporate investment is defined as the ratio of a company’s capital expenditures to its total assets for the current year. Capital expenditures are quantified as the cash paid for the acquisition of fixed assets, intangible assets, and other long- term assets. The variable is abbreviated as INV. The data is sourced from the CSMAR (China Stock Market & Accounting Research) database. (2) Independent variables: Economic policy uncertainty: This study utilizes the monthly China Economic Policy Uncertainty (EPU) Index, jointly released by Stanford University and the University of Chicago, to assess economic policy uncertainty in China. The index is based on data from the South China Morning Post (SCMP), the largest English newspaper in Hong Kong. It identifies articles related to China's economic policy uncertainty published each month in the SCMP and calculates the monthly China EPU Index by dividing the number of such articles by the total number of articles published that month (for detailed construction methods, refer to http://www.policyuncertainty.com/research.html). To calculate the annual economic policy uncertainty, we sum the monthly economic policy uncertainty values for each year and then divide the total by 100 to obtain the annual economic policy uncertainty. (3) Control Variables: This study selects the quick ratio, revenue growth rate, operating cash flow after logarithmic transformation, Tobin's Q, and management expense ratio as firm-level control variables for the regression analysis. Additionally, at the macro level, it includes China's GDP growth rate, the logarithm of China's total retail sales of consumer goods, and the entrepreneur confidence index as external economic control variables. Definitions and variable abbreviations are provided in Table 1. The data is sourced from the CSMAR (China Stock Market & Accounting Research) database. (4) Threshold Variable: This study intends to select the debt-to-asset ratio, total assets, ROE, and net profit growth rate as threshold variables, representing four dimensions of a corporate: leverage risk, size, profitability, and growth speed. The debt-to-asset ratio is a crucial indicator of a corporate's financial leverage and risk. Different levels of debt-to-asset ratio may lead to varying investment behaviors in the face of economic policy uncertainty. Highly leveraged corporates may become more cautious and reduce investment when economic policy uncertainty increases due to higher financial costs and debt repayment pressures, while corporates with lower leverage may have more financial flexibility and continue investing in uncertain environments. Therefore, using the debt-to-asset ratio as a threshold variable allows for examining the differences in investment strategies among corporates with different financial leverage levels under economic policy uncertainty. The total asset size reflects a corporate's overall strength and resource base. Large corporates typically possess more resources and stronger risk- bearing capacity, potentially maintaining stable investment levels even under economic policy uncertainty; in contrast, small corporates, due to limited resources, may quickly reduce investment to mitigate risk when uncertainty increases. By using total assets as a threshold variable, we can explore the investment behavior differences among corporates of different sizes in an environment of economic policy uncertainty. ROE is an important indicator of a corporate's profitability, reflecting the return on equity. Corporates with high ROE generally have strong profitability and market competitiveness, potentially maintaining high investment intentions even when economic policy uncertainty increases; conversely, corporates with low ROE may become more conservative and reduce investment in uncertain environments due to weaker profitability. Selecting ROE as a threshold variable allows for analyzing the investment decision differences among corporates with varying profitability levels under economic policy uncertainty. The net profit growth rate reflects the speed and trend of a corporate's profit growth. Using the net profit growth rate as a threshold variable enables the examination of investment behavior changes among corporates with different growth speeds under economic policy uncertainty. Definitions and variable abbreviations are provided in Table 1. The data is sourced from the CSMAR (China Stock Market & Accounting Research) database. (4) Overall Variable Data Handling: After collecting the data, we performed the following overall data handling: (1) Excluded samples from listed companies in the financial industry. (2) Excluded samples from ST-classified listed companies. (3) Applied a 5% winsorization to firm-level variables. Table 1. Variable Descriptions Variable Abbrevi ations Variable Name Variable Descriptions INV Corporate investment Capital Expenditures / Total Assets EPU Economic policy uncertainty China Economic Policy Uncertainty Index calculated by Baker et al. (2016) based on the South China Morning Post QA Quick ratio (Current Assets - Inventories) / Current Liabilities RGR Revenue growth rate Revenue Growth Rate within the Current Year CASH Net Operating Cash Flow Logarithm of the Net Cash Flow from Main Business Activities in the Current Year TQ Tobin's Q Market Value / Total Assets MER Management Expense Ratio Management Expenses / Revenue GDP GDP Growth Rate (Current Year GDP / Previous Year GDP) * 100 CONSU ME Logarithm of Total Retail Sales of Consumer Goods Logarithm of Total Retail Sales of Consumer Goods for Each Year CONFI D Entrepreneur Confidence Index Average Entrepreneur Confidence Index for the Current Year LEV Debt-to-Asset Ratio Total Liabilities / Total Assets SIZE Total Assets Total Assets ROE Return on Equity Net Profit / Shareholder's Equity NPGR Net Profit Growth Rate (Current Year Net Profit - Previous Year Net Profit) / Previous Year Net Profit 327 3.3. Descriptive Statistics As shown in Table 2, the standard deviation of corporate investment is 0.05, indicating relatively low overall volatility, with a maximum value of 0.2 and a minimum value of 0. This suggests that some companies may not have undertaken new investments, while others have engaged in investments at a higher proportion. The standard deviation of the economic policy uncertainty index is 30.40, with a minimum value of 6.68 and a maximum value of 95.02, indicating significant variation in economic policy uncertainty across different times and regions, suggesting substantial volatility in economic policy uncertainty. Table 2. Descriptive Statistics Variable count mean sd min max INV 59347 0.05 0.05 0.00 0.20 EPU 59466 44.97 30.40 6.68 95.02 QA 59465 1.87 1.99 0.23 9.71 RGR 59166 0.15 0.34 -0.47 1.32 CASH 46497 19.04 1.60 15.60 22.61 TQ 59466 1.92 1.09 0.92 5.90 MER 59407 0.09 0.08 0.01 0.38 GDP 59466 107.27 2.70 102.35 114.23 CONSUM E 54258 12.32 0.74 10.57 13.00 CONFID 41429 123.66 7.95 106.93 141.93 LEV 59466 0.44 0.21 0.08 0.89 SIZE 59466 8.46e+0 9 1.59e +10 3.96e+0 8 8.23e+ 10 ROE 58818 0.05 0.13 -0.48 0.25 NPGR 59241 -0.21 2.40 -11.11 4.63 Observati ons 59466 4. Empirical Results Below, we will conduct threshold regression based on the threshold variables mentioned above one by one. We will present the results of the threshold effect test and the coefficients from the threshold regression. For the sake of brevity and clarity in presenting the results, we will only showcase the regression coefficients for EPU, and omit the regression coefficients for other control variables. The results from Table 3 reveal that economic policy uncertainty exerts varying effects on corporate investment at different levels of corporate leverage. The threshold effect test confirms the presence of a double threshold effect. Furthermore, Table 4 reveals that when a company's debt-to- asset ratio is below 0.3189, the coefficient is -0.0001171; when the ratio is above 0.3189 but below 0.6558, the coefficient is -0.0000445; and when the ratio exceeds 0.6558, the coefficient is -0.0001364. This may be attributed to the fact that companies with lower debt-to-asset ratios generally enjoy more stable financial conditions and possess more internal funding sources. However, in the face of EPU, these companies may adopt a more cautious stance, reducing their reliance on external financing and consequently decreasing investment activities. Conversely, companies with higher debt-to-asset ratios may already be exposed to significant financial risks. When confronted with uncertainty, they prioritize short-term debt repayment capabilities and decrease long-term investments to avoid further financial burdens. For companies with intermediate debt-to-asset ratios, they may possess a relatively balanced financial structure, allowing them to leverage external financing channels without facing substantial financial pressure due to excessive debt. Therefore, under EPU, their investment decisions are relatively less affected. Table 3. Threshold Effect Test with LEV as the Threshold Threshold Number Threshold Value Fstat Prob 95% Confidence Interval Double 0.6558*** 0.3189 34.29 0.000 [0.6502,0.6585] [0.3020,0.3208] Table 4. Regression Results with LEV as the Threshold Variable EPU0 -0.0001171*** (-4.55) EPU1 -0.000445* (-1.86) EPU2 -0.0001364*** (-4.67) CONTROL YES Observations 31680 R-squared 0.0336 Note: The numbers in parentheses are t-values or z-values, with *, **, *** denoting significance levels of 10%, 5%, and 1%, respectively. The same applies below. The results from Table 5 indicate that the double threshold effect test is not significant at the 95% confidence level, but the single threshold effect test is passed. This significantly demonstrates that the impact of economic policy uncertainty on corporate investment differs among companies of varying sizes. The results from Table 6 indicate that small-scale corporate investment is more significantly affected by economic policy uncertainty compared to larger corporations. Under the influence of economic policy uncertainty, smaller companies tend to invest less, possibly due to their typically limited financial reserves and narrow financing channels, which make them more vulnerable to external shocks. In the face of economic policy uncertainty, small-scale companies may adopt a more cautious approach to cost and expenditure control, reducing investment activities to mitigate risks. In contrast, larger corporations, with their stronger financial capabilities and diversified financing options, are better equipped to handle the challenges posed by uncertainty. Table 5. Threshold Effect Test with SIZE as the Threshold Threshold Number Threshold Value Fstat Prob 95% Confidence Interval Single 5.6596e+09*** 24.08 0.000 [5.4284e+09, 5.7279e+09] Double 6.1208e+09*** 6.1471e+08 8.27 0.088 [5.6962e+09,6.1660e+09] [6.0830e+08,6.2291e+08] 328 Table 6. Regression Results with SIZE as the Threshold Variable EPU0 -0.0000981*** (-4.68) EPU1 -0.0000547** (-1.92) CONTROL YES Observations 31680 R-squared 0.0312 The results from Table 7 indicate that the double threshold effect test is significant at the 95% confidence level, demonstrating that the impact of economic policy uncertainty on corporate investment differs among companies with varying profitability. Furthermore, the results from Table 8 reveal that companies with poorer profitability are relatively more affected by economic policy uncertainty, resulting in lesser investment under such uncertainty. Conversely, companies with better profitability, especially those exceeding the second ROE threshold, the effect is not significant. Table 7. Threshold Effect Test with ROE as the Threshold Threshold Number Threshold Value Fstat Prob 95% Confidence Interval Double 0.0606*** -0.0615 26.36 0.000 [0.0591,0.0610] [-0.0736, -0.0507] Table 8. Regression Results with ROE as the Threshold Variable EPU0 -0.00021*** (-6.87) EPU1 -0.0001105*** (-4.50) EPU2 -5.53e-06 (-0.23) CONTROL YES Observations 31401 R-squared 0.0372 The results presented in Tables 9 and 10 below indicate that companies with the lowest and highest net profit growth rates are more affected by economic policy uncertainty, leading them to invest less under such uncertainty. Conversely, companies with moderate net profit growth rates are the least affected by economic policy uncertainty and tend to invest more under such conditions. These findings may be attributed to the fact that companies with the lowest net profit growth rates, due to their poor financial condition and potentially unstable cash flow, adopt a more conservative approach when faced with economic policy uncertainty. They reduce investments to ensure survival and avoid financial crises. In contrast, although companies with the highest net profit growth rates have strong profitability, the market has high expectations for their sustained growth. The uncertainty in economic policy increases the unpredictability of future earnings, prompting these companies to adopt a more cautious investment strategy and reduce capital expenditures to avoid stock price fluctuations and maintain investor confidence. In comparison, companies with moderate net profit growth rates possess both financial stability and profitability without the pressure of excessively high growth expectations. In an environment of economic policy uncertainty, these companies can more flexibly assess the risks and returns of investment projects and increase investments in a timely manner to seize market opportunities. Therefore, they are relatively less affected by economic policy uncertainty. Table 9. Threshold Effect Test with NPGR as the Threshold Threshold Number Threshold Value Fstat Prob 95% Confidence Interval Double -2.1116 *** 1.0021 30.09 0.000 [-2.3998, -1.8583] [0.8065,1.0131] Table 10. Regression Results with NPGR as the Threshold Variable EPU0 -0.0001838 *** (-5.81) EPU1 -0.0000476 ** (-2.04) EPU2 -0.0001547 *** (-5.85) CONTROL YES Observations 31680 R-squared 0.0345 5. Conclusion and Policy Recommendations This study utilizes financial data from China's Shanghai and Shenzhen A-share listed companies, along with China's macroeconomic data, and incorporates the China Economic Policy Uncertainty Index calculated by Baker et al. (2016). Using the panel threshold regression method, it investigates the impact of economic policy uncertainty on corporate investment under different corporate characteristics. The empirical results indicate that: (1) companies with high and low leverage conduct the most investment searches when economic policy is uncertain; (2) small-scale companies reduce their investments more significantly than large-scale companies during times of economic policy uncertainty; (3) economic policy uncertainty has a greater impact on the investments of companies with weaker profitability; and (4) companies with moderate growth capabilities experience the smallest impact on their investments from economic policy uncertainty, while companies with higher and lower growth capabilities are most affected by economic policy uncertainty in terms of their investments. It is evident that economic policy uncertainty has significant heterogeneous effects on corporate investment. The aforementioned empirical conclusions are of great significance in promoting healthy and prudent investment among enterprises. Based on the above conclusions, we propose the following policy recommendations: The 329 government should provide differentiated financial support to high-leverage and low-leverage companies to alleviate their financial burdens; offer financing facilitation to small-scale enterprises to enhance their ability to resist uncertainty risks; grant financial subsidies and technical support to companies with weaker profitability to improve their competitiveness; and establish a market information platform to provide policy guidance for companies with high and low growth potential, helping them effectively manage uncertainty risks. Through these measures, the negative impact of economic policy uncertainty on corporate investment can be effectively mitigated, promoting the stable development of enterprises. References [1] Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring economic policy uncertainty. The quarterly journal of economics, 131(4), 1593-1636. [2] Pástor, Ľ., & Veronesi, P. (2013). Political uncertainty and risk premia. Journal of financial Economics, 110(3), 520-545. [3] Alexopoulos, M., & Cohen, J. (2015). The power of print: Uncertainty shocks, markets, and the economy. International Review of Economics & Finance, 40, 8-28. [4] Tillman, P. (2016). Uncertainty about Federal Reserve policy and its transmission to emerging economies: Evidence from Twitter. [5] Osnago, A., Piermartini, R., & Rocha, N. (2015). Trade policy uncertainty as barrier to trade (No. ERSD-2015-05). WTO Staff Working Paper. [6] Alberto, O., Roberta, P., & Nadia, R. (2015). Trade policy uncertainty as barrier to trade. World Trade Organization Working Paper, No. 5. [7] Mumtaz, H., & Surico, P. (2018). Policy uncertainty and aggregate fluctuations. Journal of Applied Econometrics, 33(3), 319-331. [8] Arellano, C., Bai, Y., & Kehoe, P. J. (2019). Financial frictions and fluctuations in volatility. Journal of Political Economy, 127(5), 2049-2103. [9] Gulen, H., & Ion, M. (2016). Policy uncertainty and corporate investment. The Review of financial studies, 29(3), 523-564. [10] Liu, G., & Zhang, C. (2020). Economic policy uncertainty and firms' investment and financing decisions in China. China Economic Review, 63, 101279. [11] Jing, Z., Lu, S., Zhao, Y., & Zhou, J. (2023). Economic policy uncertainty, corporate investment decisions and stock price crash risk: Evidence from China. Accounting & Finance, 63, 1477-1502.