Fiscal Policy Uncertainty and Firms’ Production Efficiency: Evidence from China

Abstract

Total factor productivity (TFP) is pivotal to driving sustainable economic growth. This study examines the relationship between fiscal policy uncertainty (FPU) and firms’ TFP with the least squares method. We measure FPU at the provincial level using government work reports from various provinces in China with text analysis and find that a higher degree of FPU is negatively associated with local firms’ TFP. This effect is more significant for firms from regions with lower levels of marketization and government fiscal transparency and those with higher managerial myopia than for other firms. The channel tests show that FPU reduces local firms’ TFP by inhibiting corporate expansionary and research and development investments, and this effect is supported by the intensified financing constraints. Overall, our results suggest that FPU impairs local firms’ production efficiency.

1. Introduction

Fiscal policy is a crucial tool for macroeconomic regulation used by governments to change tax revenue and fiscal spending, and thus affect aggregate demand [1]. A stable and sustained fiscal policy can improve firms’ decision-making in areas such as employment, investment, and financing by enhancing their deterministic expectations of demand [2]. In contrast, a volatile fiscal policy can increase uncertainty about demand expectations and lead firms to misallocate resources [2]. As a result, a higher level of fiscal policy uncertainty (FPU) often leads to greater losses of efficiency; however, limited empirical evidence exists regarding the micro-level effects of FPU. This study examines the externalities of FPU on micro-firm efficiency from the perspective of firms’ production efficiency.

Total factor productivity (TFP) measures the amount of output not directly attributable to tangible inputs, such as materials and capital, and is commonly attributed to improvements in efficiency due to technological progress [3]. It is a crucial factor influencing firms’ long-term sustainability. At the macro level, TFP determines the quality of national economic development, making it a vital focus for economic planners seeking to formulate policies that foster long-term sustainable growth [4]. However, uncertainty in macro policies, including economic [4], trade [5], and climate policies [6], often decreases the TFP of firms. Given the strong relationship between fiscal policy and corporate decision-making [7], a high degree of FPU may also impair TFP at the firm level.

In this study, we hypothesize that a high degree of FPU inhibits the TFP of local firms. For firms, improvements in TFP typically result from the expansion of economies of scale, which relies on investments in the industrial chain, or technical advancements, which are determined by investments in research and development (R&D) [3]. In the context of highly unstable fiscal policy due to, for example, unexpected changes in industrial subsidies or tax breaks, firms’ investment decisions may be negatively impacted. For instance, the “bad news” principle and the difficulty in predicting future cash flows [8,9], (Bernanke’s [8] “bad-news” principle posits that long-term capital investment projects are characterized by some degree of irreversibility; hence, an increase in economic uncertainty can cause a delay in capital investment. According to this view, firms facing uncertainty may prefer to postpone their “irreversible” capital investments such as expansionary and R&D investments), as well as increases in actual financing constraints [10,11], can contribute to firms’ decisions to reduce their investments in industrial expansion and technological innovation at the cost of sacrificing TFP to avoid taking on a high level of liquidity risk [12,13].

We empirically investigate the impact of FPU on firms’ TFP using Chinese data. The Chinese context provides an ideal setting for our study. First, since its decision regarding the Tax-sharing Financial Management System in 1993, China has adopted a decentralized authoritarian system that combines political centralization and economic decentralization, thus granting significant fiscal policy autonomy to local governments (On 15 December 1993, the State Council made the Decision on Implementing the Tax-sharing Financial Management System. Beginning on 1 January 1994, the current local financial contract system was reformed and the tax-sharing financial management system was implemented, covering all provinces, autonomous regions, and municipalities directly under the Central Government, as well as cities with separate plans). As a result, local governments can implement differentiated fiscal policies according to economic development in their own jurisdictions, leading to regional differences in FPU and providing a basis for the empirical analysis in this study. Notably, the report on the work of the government stands out as one of the most important annual official texts issued by local governments, offering a comprehensive reflection of their implementation of differentiated fiscal policies. This report systematically reviews the government’s achievements over the past year and outlines a detailed work plan for the year ahead, including specific measures and arrangements for financial governance. As such, these official texts serve as a critical foundation for analyzing and identifying the distinctive characteristics of local governments’ financial governance. Second, unlike most Western economies, the dominance of manufacturing in China’s industrial structure has created many listed manufacturing firms. TFP has significant strategic implications for these firms, which are closely related to their core competitiveness [14]. Therefore, Chinese data provide an ideal large sample for investigating the determinants of TFP. Third, China’s economy is currently undergoing a shift toward high-quality development, and improving TFP is crucial for the sustainability of high-quality economic growth during this critical stage of transformation. Against the backdrop of the global economic recession, analysis of the relationship between FPU and TFP from the perspective of China can provide valuable insights for informing sustainable growth strategic other economies in both developed and developing economies.

To characterize the intensity of fiscal policies, we use text analysis methods to extract the numbers of terms related to fiscal policy from local government work reports (GWR) in China. We define province-level FPU as the rolling standard deviation of the intensity of fiscal policy in a given province over the past three years. Using Chinese A-share market data from 2003 to 2020, we find that FPU significantly reduces the TFP of local firms. This finding suggests that a higher degree of FPU leads to greater losses in firms’ production efficiency and is consistent with our expectation. To address potential concerns about endogeneity, we perform two-stage least squares (2SLS) regression using two instrumental variables and conduct propensity score matching (PSM) analysis. We also conduct several robustness checks by using alternative measures of the main variables, adding samples from municipalities directly under the control of the central government, and controlling for additional factors at the macro-regional level. Our main conclusion still holds after conducting the above tests.

We conduct additional analyses to gain deeper insights into the relationship between FPU and TFP. Cross-sectional analysis reveals that the negative impact of FPU on local firms’ TFP is more pronounced among firms headquartered in regions with a lower degree of marketization and those with a higher level of government financial opacity than among other firms. Meanwhile, the negative impact of FPU on TFP is more evident for firms with more (vs. less) severe managerial myopia. These findings suggest that central and local governments should promote marketization and enhance fiscal transparency to mitigate the negative impact of FPU on firms’ TFP. Additionally, firm managers should broaden their management horizons to alleviate the negative externalities of FPU. Furthermore, evidence from mechanism tests indicates that FPU inhibits the TFP of local firms by reducing their corporate investments, which include total, expansionary, and R&D investments. Additional tests also indicate that FPU aggravates financing constraints, thus indirectly supporting our views.

Governments frequently use fiscal policies to regulate macroeconomic conditions, making the potential consequences of FPU a popular topic in the academic community. For instance, Fernández-Villaverde et al. [15] investigate the impact of increased uncertainty surrounding the mix and timing of fiscal austerity on current business conditions through its effect on the expectations and behavior of households and firms (The authors estimate tax and spending processes for the U.S. that allow for time-variant volatility. They interpret changes in the volatility of different fiscal instruments as an intuitive representation of the variations in FPU, that is, of the variations in uncertainty about the future path of fiscal policy). Anzuini et al. [16] are concerned with the macroeconomic effects of discretionary fiscal policy and find that an unexpected increase in FPU harms the economy. In a recent study closely related to ours, Wen et al. [7] examine the impact of FPU on investment from the perspective of micro-firms. Their findings suggest that FPU significantly reduces new energy firms’ investment in innovation, and this change is primarily due to the decline in the incentive effect of government support on such investment. In contrast to Wen et al. [7], our study extends the focus from investment to production efficiency and further reveals the inherent connection between fiscal policy stability and high-quality economic development.

The current study makes several significant contributions. First, we expand the literature on the economic consequences of macro policy uncertainty from the perspective of fiscal policy. Studies primarily concentrate on aspects such as economic policy uncertainty (EPU) [4,17], monetary policy [18], and trade policy uncertainty [5] and emphasize the value of macro policy stability. Although fiscal policy deeply affects the economy [19], the consequences of FPU remain largely unexplored. Accordingly, this study constructs an FPU index by using data from Chinese local GWRs and examines whether and how FPU affects firms’ TFP. Our findings confirm the negative externality of FPU on production efficiency, thereby highlighting the importance of fiscal policy sustainability and enriching the literature on the economic consequences of FPU.

Second, we provide a novel perspective on the macro governance of firms’ TFP. TFP not only serves as a crucial driving force for sustained macroeconomic growth but also underpins the core competitiveness of a firm [14]. The literature examines various determinants of TFP, including macro policies [6,20], technological progress [21], corporate investments [22], and corporate characteristics [23]. Despite the importance of fiscal policy to corporate decision-making and efficiency [24], it remains unclear whether and how FPU impacts TFP. Our findings provide nuanced insights that highlight FPU as a critical determinant of corporate investment decisions and, consequently, a factor that impedes firms’ TFP.

Third, the findings of this study are relevant to the strategies used to implement fiscal policies not only in China, but also in developed economies. Given that China is the second-largest economy worldwide and has a comprehensive industrial system, the quality of its economic development is of utmost importance. This development hinges critically on the governance effect of fiscal policy. Our findings suggest that the government should pursue stable fiscal policy and thus minimize FPU to foster TFP. Moreover, in an environment of high FPU, the government should prioritize fiscal transparency and accelerate marketization. Meanwhile, corporate managers should strengthen governance capabilities, focus on sustainable development, and mitigate the adverse impact of FPU on productivity.

The remainder of this paper is organized as follows. Section 2 reviews the relevant literature and describes the development of our hypothesis. Section 3 describes the sample, data, variable measures, and empirical model. Section 4 presents our descriptive statistics, empirical results, and robustness checks. Section 5 presents several cross-sectional analyses and channel tests, and Section 6 concludes the paper.

2. Literature Review and Hypothesis Development

2.1. Literature Review

TFP refers to the incremental impact of technological advancements on production that cannot be explained by changes in capital and labor factors alone. TFP is a crucial measure of production activity efficiency over a given period and contributes significantly to enhanced regional productivity performance [25]. Studies have highlighted the importance of macro-level uncertainty with regard to TFP. Bloom et al. [26] find that uncertainty shocks reduce gross domestic product and lead to reduced productivity growth. Bonciani and Oh [27] argue that macroeconomic uncertainty shocks disrupt economic activity and reduce both R&D investment and TFP. Li et al. [4] reveal a significant and negative relationship between EPU and growth in the TFP of Chinese firms. Jiang and Zhu [28] suggest that political uncertainty influences both macro-level productivity and micro-level firm performance, including promotion decisions. Research from other perspectives suggests a negative correlation between firms’ TFP and other types of macro uncertainty such as uncertainty in the global market [29], trade policy [5], or climate policy [6]. However, the literature addressing the effects of FPU is scarce.

FPU primarily emerges from changes in fiscal policy tools, which are widely studied. Most relevant studies suggest that tax cuts and fee reductions increase the TFP of firms [30]. In contrast, there is no consensus on the impact of financial subsidies on TFP. Some scholars support the “subsidy support theory”, wherein financial subsidies reduce the marginal cost of production, thus stimulating an increase in TFP [30,31,32]. Fiscal policy also can encourage firms to increase their R&D investment by easing financing constraints [13,31,33,34,35]. Furthermore, political policies can guide corporate decision-making by subsidizing firms and industries that have potential for growth and development [36]. However, the “subsidy inhibition theory” proposes that government subsidies do not affect [37] or may even inhibit firms’ TFP [38,39]. The negative effects of fiscal policy stem from politicians’ distorted incentives, a lack of regulation [40], information asymmetry, and moral hazards associated with firms’ use of government funds [4]. Selective bias in government subsidies [41], rent-seeking, and corruption [42] may compound these negative effects. However, focusing only on fiscal expenditure or income data and ignoring FPU itself would lead to deviation from policy objectives. The fluctuation attribute of policies should be highlighted [43].

Our study focuses on the impact of FPU. Similar to EPU, as defined by Gulen and Ion [44], FPU refers to unexpected adjustments to fiscal policies made by governments that cannot be accurately anticipated or comprehended in advance by economic agents. FPU better reflects economic conditions and contains more information than the absolute values and growth rates of economic indicators [45,46]. FPU represents 40% of EPU in the US according to Baker et al. [47]. It also poses a crucial obstacle to the recovery of the world economy after the global financial crisis [26].

2.2. Hypothesis Development

In China’s decentralized economic system [48], local governments have substantial discretion to implement laws and policies within their jurisdictions [49,50], especially since the implementation of the Tax-sharing Financial Management System in 1993. That is, local governments have considerable autonomy to adjust fiscal policies to promote economic development and boost fiscal revenue in their respective jurisdictions while adhering to the central government’s fiscal policy. Therefore, FPU is an outcome of local governments’ discretion. Global market shocks, financial instability [51], and the mobility of local officials can affect fiscal policies and increase FPU. Increasing uncertainty can distort the beneficial impacts of fiscal policy and exacerbate adverse effects. Given the prevalence of local government-related FPU and its impact on micro behavior, we predict that FPU reduces firms’ TFP.

Studies suggest that the expansion of economies of scale and technological progress are crucial factors that enhance TFP [3]. The former factor involves investment in a firm’s production chain and thus constitutes an expansionary investment, while the latter factor is fueled by research and innovation; notably, R&D investments are key drivers of the enhancement of firms’ TFP [52]. Therefore, both expansionary and R&D investments play a crucial role in boosting firms’ TFP. However, FPU may hinder firms’ investment in several ways. First, both expansionary and R&D investments require firms to make long-term decisions. Bernanke’s [8] “bad news” principle argues that long-term investments are irreversible, and uncertainty may delay investment. Therefore, FPU might encourage firms to postpone their investments until some or all of the policy uncertainty is resolved [53], leading to decreases in expansionary and R&D investments, and ultimately reducing TFP.

Second, when fiscal policy is subjected to significant instability due to factors such as unexpected changes in industrial subsidies or tax breaks, firms find it challenging to accurately evaluate their future cash flows [8,9]. In such circumstances, both expansionary and R&D investments are likely to face increased risks of capital recovery, resulting in reduced industrial expansion and technological innovation and thus potentially sacrificing TFP.

Third, FPU may hinder firms’ willingness to invest by worsening financing constraints [54], a crucial factor that constrains firms’ production efficiency [55,56]. FPU not only exacerbates information asymmetry and adverse selection [57,58], but also increases firms’ external financing costs and intensifies their financial constraints [10,11]; these effects pose challenges to firms seeking external financing. FPU also obscures the signals released by fiscal policy. For instance, local governments that offer subsidies to firms and industries with good market and development potential release positive signals, thereby attracting more external financing. However, vanishing signals increase other types of uncertainty, making investors delay their investments or expect higher investment returns and leading to increased financing costs and reduced investment for firms, which ultimately undermine firms’ TFP [12,13].

Based on the above analysis, we propose that FPU negatively impacts firms’ productivity by reducing expansionary and R&D investments.

3. Research Design

3.1. Sample and Data

We manually collect the text data of 3643 GWRs provided by the website of the government of the People’s Republic of China and the official websites of city governments. Firm-level accounting data are from the China Stock Market and Accounting Research (CSMAR) Database.

Our initial sample contains all firms listed in the Chinese A-share market from 2003 to 2020. The global outbreak of the COVID-19 pandemic in late 2019 significantly disrupted the global economic structure, social progress, and firms’ strategic decision-making. Excluding potential biases and disruptions caused by the major external shock is essential, as the effects of such shocks on firms’ performance often manifest with a time lag. Specifically, the pandemic’s impact on firms’ TFP became particularly evident after 2020. To ensure the accuracy and generalizability of our research findings, we have selected a sample period spanning from 2003 to 2020. This approach allows us to capture long-term trends while effectively minimizing distortions resulting from the pandemic. We exclude four municipalities directly under the central government, firms in the financial industry, and firm-year observations with missing financial data, special treatment (ST) marks, or zero or negative total assets (Such municipalities are directly under the management of the central government, resulting in governments under the direct jurisdiction of the municipality having less discretion in policy-making, which may affect our empirical results. Of course, to prove the robustness of the study results, we include samples from municipalities directly under the central government in our robustness checks). We then winsorize our sample at the 1% level on each continuous variable in each tail to minimize the effect of outliers. We obtain 26,732 firm-year observations of 2805 unique firms in 27 provinces.

3.2. Measure of Fiscal Policy Uncertainty

In recent years, text analysis has become widely utilized in related quantitative research on macroeconomic policies. For example, Baker et al. [59] constructed a quantitative index of uncertainty in US economic policy that has garnered considerable attention. Specifically, the authors count the frequency of texts containing keywords related to policy uncertainty in news reports from 10 major US newspapers, differences among economic forecasters, and the number of expiring US tax provisions, with weights of 1/2, 1/3, and 1/6, respectively. The weighted average of these three components is used to measure uncertainty in US economic policy. Similarly, Huang and Luk [60] analyzed the texts of major news media reports in China and use the frequency of relevant economic policy keywords to reflect the level of social discussion and attention to a particular policy. Following Baker et al. [59] and Davis et al. [61], we use the text analysis constructing variable, FPU, as a measure of FPU in China.

First, we collect a certain proportion of samples in groups by year. We divide 3643 GWRs from 2003 to 2020 into 18 folders according to the year and extract 5 GWRs from each folder according to a specific percentage to yield 90 samples. Due to the varying amount of text in each folder, extracting GWRs according to a certain proportion imparts a certain degree of randomness. Second, we build a dictionary of GWRs based on the extracted samples. The first sample data are segmented using Python 3.10 and “Jieba” machine word segmentation, and the results are corrected to remove the repetitive words and thus form Dictionary 1, which is used as a custom dictionary (“Jieba” is the best Chinese word segmentation component in Python and can be applied to regular Chinese text. However, because the GWRs are regular and special text, to improve the accuracy of word segmentation, in this work we build a custom dictionary of GWRs on the basis of stuttering word segmentation). Dictionary 2 is then obtained using this custom dictionary and “Jieba” word segmentation. Then, Dictionary 2 is matched with Dictionary 1, and new words are added to Dictionary 1. This process is repeated until no new words appear and a custom dictionary is formed. Third, all of the GWR text data are preprocessed. The first step is word segmentation: the text is read in Python and divided into words using “Jieba” and the custom dictionary. The second step involves removing stop words (Stop words refer to words that have no actual meaning and are only used in this paper to connect words and facilitate expression, such as “if”, “and”, “or” and so on). This step filters from the word segmentation, and words that are determined to be included in the stop word list are deleted. Finally, words with local government financial governance characteristics in the GWR are analyzed, and their frequencies are determined. After pooling the word frequency data at the provincial level, we then calculate FPU as the three-year (t − 2, t − 1, t) rolling standard deviation of the frequency of words related to financial governance characteristics in the annual GWRs as a measure of the FPU of each province in China (Specifically, t is the current year, t − 1 means the previous year, and t − 2 indicates the previous two years). FPU is calculated using the actual sample period (Fatás and Mihov [45,46] believe that volatility or standard deviation better reflects economic conditions than the absolute values and growth rates of economic indicators, as volatility contains more information. In academic research, however, volatility is more difficult to address in model setting and construction problems. Even if different control variables are added, volatility or standard deviation is a more robust core explanatory variable than other proxy variables of economic operations).

Figure 1 depicts the spatial–temporal distribution of local FPU across mainland China during 2003–2020. Panel A shows the spatial distribution of the average local FPU. As shown, local FPU is more severe in western China than in eastern China, and has spread over time across mainland China from west to east. Panel B shows the dynamic change in the national average FPU in the time series. The observed regional differences in FPU may contribute to unique regional external corporate environments.

This figure depicts the spatial–temporal distribution of local fiscal policy uncertainty around mainland China between 2003 and 2020. Panel A shows the spatial distribution of the average local FPU, where darker shading indicates a higher degree of FPU in each province. Accordingly, local FPU spread gradually from west to east in mainland China. Panel B shows the dynamic change of the national average FPU in the time series.

3.3. Measure of Firms’ Total Factor Productivity

The core of the Olley–Pakes (OP) method lies in assuming that firms make investment decisions based on their current productivity levels, using current investment as a proxy for productivity (Olley and Pakes [62] established a relationship between capital stock and investment: $K_{it+1}=(1-\delta )K_{it}+I_{it}$, where K represents the firm’s capital stock, and I denotes current investment. This relationship indicates that a firm’s current capital value is orthogonal to its investment). This approach effectively addresses the issue of simultaneity bias. However, in practice, not all firms engage in positive investment every year, and a substantial portion of firm samples (i.e., those with zero investment) may be excluded during the estimation process in the OP method. To address this limitation, Levinsohn and Petrin [63] proposed the Levinsohn–Petrin (LP) method, which uses intermediate input consumption as a proxy for productivity. In short, while the primary distinction between the LP and OP methods lies in their choice of proxy variables, their approaches to estimating TFP are otherwise the same. According to Olley and Pakes [62] and Levinsohn and Petrin [63], we estimate the following model:

$$\begin{array}{ll}\ln Y_{it}& ={\alpha }_0+{\alpha }_1\ln K_{it}+{\alpha }_2\ln L_{it}+{\alpha }_{3}Age+{\alpha }_{4}SOE+\mathrm{Year} \mathrm{Fixed} \mathrm{Effect}\\ & +\mathrm{Industry} \mathrm{Fixed} \mathrm{Effect}+\mathrm{Region} \mathrm{Fixed} \mathrm{Effect}+{\epsilon }_{it}\end{array}$$

(1)

where the subscripts i and t represent the firm and period, respectively, and Y denotes the firm’s operating income. K denotes capital, expressed as the firm’s net fixed assets. L denotes labor investment, defined as the number of employees. Age indicates the age of the firm, and SOE indicates whether the firm is state-owned. We also include year, industry, and region fixed effects.

Estimating Equation (1) gives us a pair of parameters, ${\alpha }_1$ and ${\alpha }_2$, which represent the capital and labor output elasticities, respectively. A firm’s TFP in each year can thus be calculated using the following formula:

$$TF{P}_{it}=\ln Y_{it}-{\alpha }_1\ln K_{it}-{\alpha }_2\ln L_{it}$$

(2)

3.4. Regression Model

To analyze the impact of local FPU on firms’ production efficiency, we construct the following benchmark regression model:

$$\begin{array}{cc}\hfill TF{P}_{it}(TFP\_LP/TFP\_OP)& ={\beta }_0+{\beta }_{1}FP{U}_{it}+\beta Control{s}_{it}+\mathrm{Year} \mathrm{Fixed} \mathrm{Effect}\hfill \\ & +\mathrm{Industry} \mathrm{Fixed} \mathrm{Effect}+\mathrm{Region} \mathrm{Fixed} \mathrm{Effect}+{\epsilon }_{it}\hfill \end{array}$$

(3)

where the dependent variable, TFP_LP [TFP_OP], indicates firms’ production efficiency estimated according to Equation (1) [Equation (2)], which follows Levinsohn and Petrin [63] [Olley and Pakes [62]]. The variable of interest is local FPU. We include year, industry, and region dummies in the regression model to control for year, industry, and region fixed effects, respectively. We cluster the standard errors at the firm level.

Controls in Equation (3) is a matrix comprising additional control variables included in the above benchmark model. Referring to Li et al. [4] and Zhang et al. [20], they pointed out that higher asset returns indicate better financial performance and contribute to increased productivity. At the same time, the size of a company has a significant impact on its productivity, and excessive leverage can increase financial risks, potentially inhibiting productivity growth. Additionally, in previous studies, factors such as tangible assets, company age, and governance structure characteristics were considered important control variables. Therefore, we introduce control variables at the firm level, including the return on assets (ROA), sales growth rate (Growth), natural logarithm of firm size (Size), natural logarithm of the firm’s listing year (Age), asset–liability ratio (Lev), proportion of tangible assets (Tangible), state-owned firm dummy (SOE), natural logarithm of board size (Boardsize), proportion of independent directors (Indep), and shareholding ratio of the top 10 shareholders (Tophold). We further control EPU (EPU) to distinguish it from FPU. Appendix A includes the definitions of all of the variables.

4. Empirical Results

4.1. Descriptive Statistics

Table 1. Descriptive statistics.

Panel A: Summary Statistics
Variables	Obs.	Mean	SD	P25	Median	P75
TFP_LP	26,732	8.881	1.093	8.144	8.797	9.517
TFP_OP	26,732	7.948	0.985	7.269	7.859	8.535
FPU	26,732	0.159	0.298	0.066	0.110	0.180
ROA	26,732	0.035	0.066	0.013	0.034	0.064
Growth	26,732	0.181	0.427	−0.018	0.116	0.280
Size	26,732	21.980	1.210	21.102	21.833	22.687
Age	26,732	2.132	0.722	1.609	2.197	2.708
Lev	26,732	0.444	0.203	0.285	0.441	0.596
Tangible	26,732	0.931	0.083	0.919	0.958	0.980
SOE	26,732	0.419	0.493	0.000	0.000	1.000
Boardsize	26,732	2.262	0.179	2.197	2.303	2.303
Indep	26,732	0.370	0.053	0.333	0.333	0.400
Tophold	26,732	0.349	0.149	0.233	0.328	0.450
EPU	26,732	4.977	0.636	4.559	4.823	5.627
Panel B: Univariate Tests
Variables	High_FPU = 1		High_FPU = 0		Difference Tests
	Obs.	Mean	Obs.	Mean	Difference	t-Value
TFP_LP	10,803	8.866	15,929	8.891	0.025	1.860 *
TFP_OP	10,803	7.931	15,929	7.960	0.027	2.223 **
ROA	10,803	0.035	15,929	0.034	−0.001	−1.652 *
Growth	10,803	0.197	15,929	0.170	−0.026	−4.929 ***
Size	10,803	21.970	15,929	21.986	0.016	1.092
Age	10,803	2.119	15,929	2.141	0.023	2.514 **
Lev	10,803	0.436	15,929	0.449	0.014	5.336 ***
Tangible	10,803	0.928	15,929	0.933	0.006	5.552 ***
SOE	10,803	0.408	15,929	0.426	0.018	2.859 ***
Boardsize	10,803	2.261	15,929	2.263	0.002	1.057
Indep	10,803	0.369	15,929	0.370	0.001	1.298
Tophold	10,803	0.350	15,929	0.349	−0.001	−0.331
EPU	10,803	4.980	15,929	4.976	−0.004	−0.506

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

presents the descriptive statistics for our main variables. Panel A reports the summary statistics. The LP (OP) method yields a mean firm TFP of 8.881 (7.948) with a standard deviation of 1.093 (0.95). TFP shows considerable cross-sectional heterogeneity. The mean and median FPU are 0.159 and 0.110, respectively, indicating that FPU is a significant problem, with substantial differences across regions.

Panel B of Table 1 presents the results of univariate tests in which we compare firms headquartered in high- and low-FPU regions. Firms with headquarters in high-FPU areas have a mean TFP_LP of 8.866, whereas those in low-FPU areas have a mean TFP_LP of 8.891, a difference of 0.025. Similarly, firms’ headquarters in high-FPU regions have a mean TFP_OP of 7.931, which is lower than the mean TFP_OP of 7.960 among firms with headquarters in low-FPU regions, a difference of 0.027. These differences are statistically significant at the 10% and 5% levels, respectively. This finding preliminarily supports our hypothesis that local FPU damages firms’ production efficiency. Moreover, compared with firms with headquarters in low-FPU areas, those in high-FPU areas have higher returns on assets and sales growth rates, a lower asset–liability ratio and balance of tangible assets, and a smaller percentage of SOEs. In general, these findings support our hypothesis that local FPU inhibits firms’ production efficiency.

4.2. Main Regression Results

We use the benchmark model represented by Equation (3) to examine the effect of FPU on firms’ production efficiency and report the baseline results in

Table 2. Impact of fiscal policy uncertainty on firms’ TFP.

Variables	TFP_LP	TFP_OP
	(1)	(2)
FPU	−0.031 **	−0.053 ***
	(−2.14)	(−3.68)
ROA	2.162 ***	1.741 ***
	(17.95)	(15.44)
Growth	0.172 ***	0.186 ***
	(15.53)	(16.75)
Size	0.666 ***	0.557 ***
	(66.40)	(58.73)
Age	0.039 ***	0.044 ***
	(2.62)	(3.07)
Lev	0.687 ***	0.603 ***
	(12.09)	(10.88)
Tangible	0.588 ***	0.717 ***
	(6.18)	(7.86)
SOE	0.040	0.029
	(1.59)	(1.17)
Boardsize	0.012	−0.053
	(0.21)	(−0.89)
Indep	−0.116	−0.091
	(−0.69)	(−0.56)
Tophold	0.281 ***	0.185 ***
	(4.13)	(2.82)
EPU	0.122 ***	0.176 ***
	(7.89)	(11.14)
Cons.	−8.092 ***	−6.779 ***
	(−26.72)	(−23.29)
Fixed Effects	Yes	Yes
Obs.	26,732	26,732
Adj. R2	0.751	0.702

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. *** and ** denote statistical significance at the 1% and5% levels, respectively.

. In Columns 1 and 2, the coefficients of FPU are −0.031 and −0.053, and are significant at the 5% and 1% levels, respectively, suggesting that an increase in local FPU reduces firms’ TFP_OP. In terms of economic magnitude, Column 1 (Column 2) suggests that a one-unit increase in local government FPU explains approximately 2.84% (5.58%) of the standard deviation of the dependent variable (As shown in Table 2, the standard deviation of TFP_LP (TFP_OP) is 1.093 (0.95). As the independent variable (FPU) has a significant coefficient of −0.031 (−0.053) in Column 2 (Column 4) of Table 2, a one-unit increase in the independent variable can explain 0.031/1.093 × 100% = 2.84% (0.053/0.95 × 100 = 5.58%) of the standard deviation of the dependent variable (TFP_LP or TFP_OP)). These results indicate that local FPU is significantly negatively correlated with firms’ TFP.

The results obtained for the control variables are generally consistent with our expectations. Specifically, firms’ TFP is positively associated with returns on assets, sales growth, size, age, the asset–liability ratio, the proportion of tangible assets, and the shareholding ratio of the top 10 shareholders. In addition, EPU can promote firms’ production efficiency to a certain extent. This result distinguishes FPU from EPU. Specifically, EPU shapes the overall economic environment in mainland China, and an appropriate EPU is conducive to benign competition among firms. In contrast, FPU is related to local governments, and thus firms in different regions face different levels of FPU.

4.3. Endogeneity

Although we control for several variables in the regression equations, the potential for endogeneity between financial uncertainty and firms’ production efficiency remains.

4.3.1. Instrumental Variables/2SLS Approach

Typically, the central government, which is not influenced by local micro-firms, is the direct source of fluctuations in local fiscal policy. However, FPU at the provincial level might still be driven by the performance of locally listed firms. To address this potential endogeneity, we take an instrumental variable approach. Our instrumental variables are local government political turnover (PT) and the fluctuation in fiscal policy due to the Five-Year Plan of the Chinese Communist Party (VOL).

First, PT is equal to 1 if the secretary or mayor is replaced in a city where the firm’s headquarters are located in a given year, and 0 otherwise. China has a decentralized economic system in which local governments have a great deal of autonomy in financial matters [48]. As the legal representatives of the government, government officials have the legal capacity to change local development policies [64]. Thus, replacing government officials has become an essential channel of policy uncertainty [53,65]. In addition, the personal experiences and characteristics of government officials and their financial policy decisions are generally heterogeneous during their tenures [66]. In summary, a change in local government officials should increase local FPU. This uncertainty may arise from uncertainty regarding the sustainability of existing fiscal policies, or from uncertainty regarding future fiscal policies. Local government politicians in China are not elected by local voters, but are appointed by higher-level Chinese Communist Party officials [67]. Therefore, the replacement of local officials in a city where a firm’s headquarters is located is exogenous. Thus, our empirical results suggesting that a change in local officials increases local FPU, as shown in

Table 3. Fiscal policy uncertainty and TFP: IV approach.

	1st Stage	2nd Stage		1st Stage	2nd Stage
Variables	FPU	TFP_LP	TFP_OP	FPU1	TFP_LP	TFP_OP
	(1)	(2)	(3)	(4)	(5)	(6)
PT	0.019 ***			0.011 ***
	(8.91)			(4.46)
VOL	0.520 ***			0.702 ***
	(25.20)			(43.96)
FPU		−0.032	−0.087 ***
		(−1.00)	(−2.67)
FPU1					−0.039 *	−0.071 ***
					(1.83)	(−2.62)
Cons.	2.401 ***	−2.246 ***	−1.550 ***	−0.405 **	−2.657 *	−1.771 ***
	(17.40)	(−3.59)	(−2.71)	(−2.23)	(−1.69)	(−3.15)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	23,790	23,790	23,790	23,475	23,475	23,475
R2	0.424	0.751	0.70	0.423	0.74	0.70
F-statistics	280.60	215.70	183.36	217.44	168.57	182.27
Sargan test (p-value)		0.729	0.445		0.868	0.473

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the city level and firm level in the 1st stage and the 2nd stage, respectively. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

, are consistent with the literature.

Second, according to the temporal planning included in the Five-Year Plan, we calculate VOL, another instrumental variable used to measure FPU in Chinese local governments, as the standard deviation of the frequency of annual financial governance feature terms in the GWRs during each Five-Year Plan period (China’s Five-Year Plan outlines the five-year plan for the national economic and social development of the People’s Republic of China. It is an essential component of China’s national financial strategy. China’s Five-Year Plan indicates the future path of economic policy and the potential growth opportunities that may arise. China formulated and implemented its first Five-Year Plan in 1953 and is currently at the end of its fourteenth Plan. The focus of the development goals varies across plan periods). The Five-Year Plan is a top-down development plan formulated by the central government of China. Changes in national development goals and the placement of focus on other planning areas naturally leads to increased policy uncertainty [68], including FPU. The effect of the China Five-Year Plan on FPU is exogenous.

We present the results of the instrumental variable approach in Table 3. The first-stage regression estimates reveal that two instrumental variables, PT and VOL, are positively correlated with FPU (FPU or FPU1) at the 1% significance level, as shown in Columns 1 and 4 (FPU1 is the proxy variable of FPU. The specific calculation method is shown in Section 4.4 (Robustness Tests). It is used here as an additional test of the instrumental variable approach). Moreover, the F-statistics value is greater than 10 in Columns 1 and 4, and thus the null hypothesis that these two variables are weak instrumental variables is rejected. For our instrument to be valid, it must be correlated with the variables in question and exogenously given. The p-values obtained with the Sargan test show that PT and VOL are exogenously given (Table 3). These test statistics suggest that the instrumental variables are appropriate. The second-stage results reported in Columns 2, 3, 5, and 6 indicate that the negative relationship between FPU and firms’ TFP generally remains true after correcting for potential endogeneity bias; in other words, our results are not driven by potential endogeneity bias.

4.3.2. Propensity Score Matching Approach (PSM)

The traditional multilinear approach used in our main regression tests requires a linear relationship between the outcome and explanatory variables. If the linear form assumption is violated, the model suffers from functional form misspecification and can produce biased coefficient estimates. To address this econometric concern, we use the PSM approach, wherein pairs of firm-years are matched according to similarity in characteristics that are likely to be related to FPU, but dissimilarity in the actual occurrence of FPU. After matching, any difference in the firms’ TFP can be more appropriately attributed to the existence of FPU than to differences in other characteristics, regardless of the underlying functional form.

Table 4. Identification: propensity score matching approach.

Panel A: Covariates Balance Before and After PSM
	The Mean Value Before PSM			The Mean Value After PSM
Variables	Treat	Control	Difference (Treat-Control)	Treat	Control	Difference (Treat-Control)
ROA	0.035	0.035	−0.160	0.034	0.035	0.312
Growth	0.178	0.184	1.073	0.178	0.183	0.646
Size	22.001	21.960	−2.781 ***	21.949	21.935	−0.726
Age	2.168	2.100	−7.738 ***	2.114	2.119	0.420
Lev	0.444	0.443	−0.247	0.447	0.449	0.719
Tangible	0.932	0.930	−1.130	0.934	0.935	0.844
SOE	0.426	0.412	−2.246 **	0.423	0.435	1.590
Boardsize	2.266	2.260	−2.679 ***	2.263	2.269	1.799 *
Indep	0.369	0.370	2.572 **	0.369	0.367	−1.326
Tophold	0.349	0.350	1.074	0.353	0.353	0.226
EPU	4.980	4.975	−0.550	4.944	4.925	−1.876 *
Obs.	12,696	14,036		7751	7751
Panel B: Regression Results
Variables	TFP_LP			TFP_OP
	(1)			(2)
FPU	−0.036 **			−0.058 ***
	(−2.090)			(−3.398)
Cons.	−5.332 ***			−4.116 ***
	(−8.377)			(−7.621)
Controls	Yes			Yes
Fixed Effects	Yes			Yes
Obs.	15,502			15,502
Adj. R2	0.929			0.912

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

presents the results obtained using the PSM approach. In this setting, we define our treatment group (DUM_FPU = 1) as firms whose headquarters are in high-FPU areas and the control group as those whose headquarters are in low-FPU areas (DUM_FPU = 0). We use logit regression to predict a firm’s propensity for being subjected to FPU and variables that are likely to affect the existence of FPU, including the control variables included in Equation (3) and year, industry, and province dummies. We then match each high-FPU firm with a low-FPU firm according to the closest propensity score without replacement in the same year. As Panel A of

Table 5. Robustness tests: alternative measures of main variables.

Panel A: Alternative Measures of Policy Uncertainty
Variables	TFP_LP	TFP_OP	TFP_LP	TFP_OP	TFP_LP	TFP_OP
	(1)	(2)	(3)	(4)	(5)	(6)
FPU1	−0.030 **	−0.050 ***
	(−2.01)	(−3.35)
FPU2			−0.039 **	−0.061 ***
			(−2.20)	(−3.33)
FPU3					−0.040 **	−0.062 ***
					(−2.11)	(−3.32)
Cons.	−7.813 ***	−6.328 ***	−8.057 ***	−6.726 ***	−8.128 ***	−6.705 ***
	(−26.59)	(−21.20)	(−26.58)	(−23.05)	(−25.17)	(−21.47)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	26,101	26,101	26,732	26,732	23,450	23,450
Adj. R2	0.749	0.700	0.751	0.702	0.748	0.695
Panel B: Alternative Measures of TFP
Variables	TFP_OLS	TFP_FE	TFP_GMM	TFP_OLS	TFP_FE	TFP_GMM
	(1)	(2)	(3)	(4)	(5)	(6)
FPU	−0.027 *	−0.025 *	−0.055 ***
	(−1.93)	(−1.75)	(−3.13)
FPU1				−0.026 *	−0.023	−0.055 ***
				(−1.72)	(−1.54)	(−2.95)
Cons.	−10.132 ***	−10.658 ***	−2.270 ***	−9.822 ***	−10.353 ***	−1.938 ***
	(−34.85)	(−36.29)	(−6.20)	(−35.14)	(−36.82)	(−5.09)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	26,732	26,732	26,732	26,101	26,101	26,101
Adj. R2	0.826	0.836	0.388	0.826	0.836	0.383

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

shows, the treatment and control groups differ significantly in terms of these firm characteristics in the full (i.e., before PSM) sample. After PSM, only Boardsize and EPU remain significant. Thus, the PSM sample achieves covariate balance in the first moment (i.e., the mean) for these variables. In Panel B, Equation (3) is re-estimated using the PSM sample and the entropy balancing approach. The resulting coefficients on FPU are negative and statistically significant, consistent with the results obtained in our main tests (Table 2). This consistency between the results of our main hypothesis tests based on the larger cross-sectional sample and those based on the PSM sample further indicates that our findings are not driven by bias resulting from the potential functional form misspecification of our linear model.

4.4. Robustness Tests

In this section, we perform several tests to confirm that our results are robust. First, we consider how alternative measures of local FPU might affect firms’ TFP. We calculate the proxy variable FPU1 in the same way as FPU. However, unlike FPU, which uses an actual period, FPU1 excludes samples with less than three years of data on word frequency. FPU2 and FPU3 are five-year rolling standard deviations of the annual frequency of financial governance characteristic words in the GWRs. FPU2 is calculated using the actual sample period, while the calculation of FPU3 excludes samples with less than five years of data on word frequency. They also differ in whether they are calculated using the actual period. The results reported in Panel A of Table 5 confirm our main finding that FPU decreases firms’ production efficiency.

Second, we change the method used to measure the dependent variable to address possible measurement bias. To examine the impact of FPU on firms’ production efficiency, we analyze the dependent proxy variables TFP_OLS, TFP_FE, and TFP_GMM using the ordinary least squares, fixed effects, and generalized moment estimation methods, respectively. The results in Panel B of Table 5 show that our finding that local FPU reduces firms’ TFP remains true.

Third, our main sample excludes the municipalities of Beijing, Tianjin, Shanghai, and Chongqing, which are directly managed by the central government. In addition, the government’s official administrative level in these four municipalities differs from the executive levels of other cities. To prove that the impact of FPU on firms’ productivity is not affected by these factors, we add the sample of municipalities directly under the control of the central government back to the main samples. The results reported in Panel A of

Table 6. Robustness tests: the change of sample and control variables.

Panel A: Including Firms with Headquarters in Municipalities
Variables	TFP_LP		TFP_OP		TFP_LP		TFP_OP
	(1)		(2)		(3)		(4)
FPU	−0.020		−0.041 ***
	(−1.43)		(−2.89)
FPU1					−0.025 *		−0.047 ***
					(−1.74)		(−3.26)
Cons.	−7.393 ***		−6.190 ***		−7.370 ***		−6.093 ***
	(−30.87)		(−26.40)		(−30.18)		(−25.25)
Controls	Yes		Yes		Yes		Yes
Fixed Effects	Yes		Yes		Yes		Yes
Obs.	33,500		33,500		32,612		32,612
Adj. R2	0.755		0.708		0.755		0.709
Panel B: Introducing Macro-Level Control Variables
Variables	TFP_LP				TFP_OP
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
FPU	−0.033 ***	−0.036 ***	−0.028 **	−0.031 **	−0.054 ***	−0.056 ***	−0.038 ***	−0.038 ***
	(−3.01)	(−3.14)	(−2.71)	(−2.57)	(−3.54)	(−3.67)	(−3.47)	(−3.14)
AGDP	0.227 ***	0.070 *	0.074	0.050	0.266 **	0.091 *	0.149 **	0.141 *
	(3.48)	(1.82)	(1.20)	(0.76)	(2.56)	(1.78)	(2.20)	(1.91)
CSHL		−0.603 *	−0.576 *	−0.623 **		0.074	0.146	0.231
		(−1.71)	(−1.82)	(−2.44)		(0.19)	(0.52)	(0.66)
BE			−0.014	−0.001			−0.055 *	−0.045
			(−0.60)	(−0.04)			(−1.84)	(−1.52)
IEU				0.036				0.590
				(0.06)				(1.15)
Cons.	−8.010 ***	−7.229 ***	−7.480 ***	−7.487 ***	−6.860 ***	−5.405 ***	−6.311 ***	−6.211 ***
	(−16.02)	(−10.04)	(−11.74)	(−11.02)	(−13.23)	(−7.20)	(−9.76)	(−8.79)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	26,732	26,126	24,598	19,408	26,732	26,126	24,598	19,408
Adj. R2	0.751	0.752	0.749	0.740	0.702	0.704	0.702	0.691

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

continue to show that FPU decreases firms’ TFP, and indicate that our results are not driven by the administrative level of a municipality.

Finally, we include macro factors that may affect firms’ productivity as control variables to test the robustness of our results. Using Equation (3), we control the GDP per capita of the province where the firm is headquartered, the level of urbanization development, the business environment score, and the uncertainty of the industrial environment in turn, and the results are shown in Panel B of Table 6. The results continue to suggest that FPU reduces firms’ TFP, and the results are unaffected by the inclusion of macro factors.

5. Further Research

5.1. Cross-Sectional Analysis

In this section, we further investigate the heterogeneity in the impact of local FPU on firms’ TFP from three perspectives, namely, differences in marketization, fiscal transparency, and managerial myopia.

5.1.1. Role of Marketization

Marketization is an essential factor in an investigation of firms’ production efficiency. Research shows that for economies with perfect market mechanisms, resource allocation has been optimized, and improvements in TFP are mainly attributable to firms’ micro-technical progress [69]. However, in China and other countries with transitional economies, market-oriented reform can promote factor allocation efficiency and thus improve TFP. During the 1997–2007 period, marketization contributed 1.45 percentage points to China’s annual economic growth rate, accounting for 39.2% of TFP [69]. We predict that the location of a firm’s headquarters in an area with a low degree of marketization enhances the negative impact of FPU on the firm’s TFP. We divide the sample into two groups, high and low, according to the province-level median degree of annual marketization in the location where the firm is headquartered.

As shown in

Table 7. Moderating effect of marketization.

	TFP_LP		TFP_OP
Variables	(1)	(2)	(3)	(4)
	High	Low	High	Low
FPU	−0.209	−0.025 *	−0.0867	−0.038 ***
	(0.51)	(−1.97)	(−0.60)	(−3.12)
Cons.	−7.722 ***	−8.094 ***	−5.646 ***	−6.734 ***
	(−54.20)	(−30.20)	(−22.62)	(−47.79)
Controls	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes
Obs.	8655	18,077	8655	18,077
Adj. R2	0.764	0.729	0.668	0.722
Empirical p-value	0.000 ***		0.050 **

Notes: We calculate the empirical p-values, which test the differences in the coefficients of FPU, based on Fisher’s permutation test by bootstrapping 1000 times. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

, we obtain evidence consistent with our prediction. The coefficients of FPU are significant and negative when analyzing the subsamples of firms with headquarters located in an area with a low degree of marketization (i.e., Columns 2 and 4), and these coefficients are statistically more significant than those obtained when analyzing firms with headquarters located in an area with a high degree of marketization (i.e., Columns 1 and 3). These findings indicate that firms that are headquartered in an area with a low degree of marketization are more sensitive to local FPU than are firms located elsewhere. Accordingly, it is essential to accelerate the marketization process, deepen market-oriented reforms, give full play to the market’s role, and optimize the allocation of resources.

5.1.2. Role of Local Governmental Fiscal Transparency

We first examine whether the negative impact of local FPU on firms’ production efficiency is exacerbated by inefficient fiscal transparency in local governments. Increasing local FPU can hinder firms’ ability to forecast their future financial policies, disrupt their business plans, delay significant business decisions, and even cause them to make wrong judgments that can lead to irreversible losses. The negative effect of FPU on firms’ TFP is more severe among firms whose headquarters are located in cities with inefficient fiscal transparency than among firms located elsewhere, and this result is due to an increase in information asymmetry.

To proxy for fiscal transparency in cities, we use the fiscal transparency scores of Chinese municipal governments, which are issued by the Institute of Public Management of Tsinghua University. Taking the annual city mean of the proxy values as the threshold, we divide the sample into two subsamples, high and low, according to the degree of fiscal transparency in the city where the firm’s headquarters are located.

Table 8. Moderating effect of fiscal transparency.

Variables	TFP_LP		TFP_OP
	(1)	(2)	(3)	(4)
	High	Low	High	Low
FPU	0.127 **	−0.328 **	0.112 **	−0.412 ***
	(2.53)	(−2.31)	(2.38)	(−3.05)
Cons.	−7.202 ***	−7.115 ***	−5.831 ***	−5.384 ***
	(−19.56)	(−16.35)	(−16.52)	(−13.56)
Controls	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes
Obs.	10,897	4827	10,897	4827
Adj. R2	0.756	0.746	0.717	0.714
Empirical p-value	0.000 ***		0.000 ***

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. We calculate the empirical p-values, which test the differences in the coefficients of FPU, based on Fisher’s permutation test by bootstrapping 1000 times. *** and ** denote statistical significance at the 1% and 5% levels, respectively.

reports the regression results obtained using these subsamples. At the bottom of the table, we report the empirical p-values calculated using Fisher’s permutation test using 1000 bootstraps. Columns 1 and 2 compare the effect of FPU on TFP_LP between firms in cities with low and high degrees of fiscal transparency. Columns 3 and 4 compare the effect of FPU on TFP_OP between firms in cities with low and high degrees of fiscal transparency. In Column 1, the coefficient of FPU is 0.127, and this is significant at the 5% level. In Column 2, however, FPU has a coefficient of −0.328, which is also significant at the 5% level. Furthermore, the difference in the coefficients of FPU between the two subsamples is significant at the 1% level, with an empirical p-value of 0.000. Similar findings are shown in Columns 3 and 4. In summary, the results in Table 8 suggest that the effect of local FPU on firms’ TFP is more pronounced in firms whose headquarters are located in cities with a low (vs. high) degree of fiscal transparency.

5.1.3. Role of Managerial Myopia

We next investigate the moderating effect of managerial myopia in listed firms on the relationship between local FPU and firms’ TFP. Given the importance of managers in firms’ decision-making, we predict that the effects of FPU on firms’ production efficiency are stronger in firms subject to managerial myopia than in other firms.

The term “managerial myopia” refers to a situation wherein managers have a relatively short decision-making horizon. Managers tend to pay more attention to interests of the firm that can be immediately met than to the future development of the firm. Thus, managerial myopia reflects a manager’s time cognition. According to upper echelons and time orientation theories, a manager’s time cognition determines their behavior and thus affects organizational outcomes [70]. Managerial myopia inevitably leads to managers who, restricted by a short decision-making horizon, sacrifice the long-term interests of firms when making decisions, and thus enhance the negative effect of local FPU on firms’ TFP. Following Hu et al. [71], we calculate the index of managerial myopia as the ratio of the total word frequency of “short-term vision” words to the total word frequency of Management Discussion and Analysis Published in the Annual Report of Listed Companies. We then use the median annual index of managerial myopia to divide the sample into subsamples of firms with high and low levels of managerial myopia.

The empirical results are shown in

Table 9. Moderating effect of managerial myopia.

Variables	TFP_LP		TFP_OP
	(1)	(2)	(3)	(4)
	High	Low	High	Low
FPU	−0.054 ***	−0.002	−0.068 ***	−0.034 *
	(−3.24)	(−0.10)	(−3.72)	(−1.71)
Cons.	−8.396 ***	−7.742 ***	−7.279 ***	−6.062 ***
	(−25.37)	(−23.30)	(−22.45)	(−18.00)
Controls	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes
Obs.	13,859	12,873	13,859	12,873
Adj. R2	0.747	0.758	0.702	0.703
Empirical p-value	0.016 **		0.089 *

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. We calculate the empirical p-values, which test the differences in the coefficients of FPU, based on Fisher’s permutation test by bootstrapping 1000 times. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

. Columns 1 and 3 indicate that the effect of FPU on firms’ TFP is enhanced in the subsample of firms with a high level of managerial myopia, and this result is significant at the 1% level. However, the effect of FPU on firms’ TFP is nonsignificant in the subsample of firms with a low level of managerial myopia. The empirical p-values of these comparisons are significant at the 5% and 10% levels, respectively. In summary, the findings in Table 9 indicate that local FPU significantly reduces the TFP of firms with a high level of managerial myopia.

5.2. Mechanism Tests

Thus far, our results show that local FPU reduces firms’ TFP. Compared with other firms, this effect is more prominent among firms headquartered in regions with a lower degree of marketization, those with a higher level of government financial opacity, and those with a high level of managerial myopia. Recent studies suggest that FPU is likely to have a negative impact on both investment and R&D [12,13], ultimately leading to a decrease in firms’ TFP. Therefore, we conduct further analyses to reveal the mechanisms underlying these effects of FPU and confirm the negative externalities of local FPU on corporate investment and R&D.

In the first step, we identify corporate investment as a potential mechanism through which FPU affects firms’ TFP. We use firms’ total investment (TI) and expansionary investment (EI) as proxy variables for corporate investment. For a firm, TI during an accounting period is calculated by adding the cash paid to purchase and construct fixed assets, intangible assets, and other long-term assets, as well as the net cash paid to acquire subsidiaries and other operating units. This is then reduced by the net cash received from the disposal of fixed assets, intangible assets, and other long-term assets, as well as the net cash acquired from the removal of subsidiaries and other operating units. EI is calculated as TI minus the investment needed to maintain normal business operations. Panel A in

Table 10. Possible channel.

Panel A: Channel Analysis of Total Investment and Expansion Investment
Variables	TI	TFP_LP	TFP_OP	EI	TFP_LP	TFP_OP
	(1)	(2)	(3)	(4)	(5)	(6)
FPU	−0.121 **			−0.048 *
	(−2.58)			(−1.86)
TI		0.199 ***	0.161 ***
		(16.87)	(15.56)
EI					0.032 **	0.033 ***
					(1.96)	(2.84)
Cons.	13.131 ***	−1.377 ***	−1.048 **	−0.254	1.232 ***	1.071 ***
	(20.07)	(−3.04)	(−2.53)	(−0.46)	(2.88)	(2.78)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	22,667	22,667	22,667	24,145	24,145	24,145
Adj. R2	0.322	0.591	0.555	0.141	0.515	0.494
Panel B: Channel Analysis of R&D Investment
Variables	RDI	TFP_LP	TFP_OP	RDP	TFP_LP	TFP_OP
	(1)	(2)	(3)	(4)	(5)	(6)
FPU	−0.369 ***			−0.719 ***
	(−3.66)			(−5.86)
RDI		0.087 ***	0.043 ***
		(9.84)	(5.00)
RDP					0.056 ***	0.061 ***
					(4.82)	(5.50)
Cons.	−4.915 ***	−5.925 ***	−5.093 ***	−9.595 ***	−6.583 ***	−8.767 ***
	(−6.52)	(−19.09)	(−16.38)	(−12.37)	(−14.54)	(−20.18)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	17,834	17,834	17,834	12,013	12,013	12,013
Adj. R2	0.466	0.772	0.724	0.436	0.756	0.837

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

presents the regression results pertaining to the channel role of expansionary investment. Columns 1 and 4 demonstrate significant and negative correlations between FPU and both TI and EI, respectively. Additionally, Columns 2–3 and 5–6 provide evidence that expansionary investment improves firms’ TFP, which further supports the channel role of investment.

Next, we consider firms’ R&D investment as a potential mechanism, given its crucial role in driving firms’ TFP growth [72]. We use R&D investment (RDI) and the number of R&D personnel (RDP) as proxy variables. Panel B in Table 10 presents the regression results. Columns 1 and 4 show that FPU is significantly correlated with both RDI and RDP, indicating that FPU reduces firms’ R&D investment. The results in Columns 2–3 and 5–6 provide evidence that increasing R&D investment increases firms’ TFP, supporting the prediction that FPU decreases firms’ TFP by reducing R&D investment.

Finally, firms subjected to intensifying financing constraints are less willing to invest [54]. Therefore, this study tests the role of financing constraints and provides additional evidence for the channels of investment. Following Hadlock and Pierce [73], we use firm size and age to construct the SA index; the larger the absolute value of SA, the more severe the financing constraint on the firm.

Table 11. The role of financing constraint.

Variables	TFP_LP	TFP_OP
	(1)	(2)
SA*FPU	0.026 *	0.033 **
	(1.79)	(2.18)
SA	0.015	0.009
	(0.31)	(0.18)
FPU	−0.113 **	−0.156 ***
	(−2.08)	(−2.81)
Cons.	−7.672 ***	−6.466 ***
	(−7.37)	(−6.34)
Controls	Yes	Yes
Fixed Effects	Yes	Yes
Obs.	25,129	25,129
Adj. R2	0.753	0.704

Notes: We report the t-statistics presented in parentheses below each estimated coefficient based on robust standard errors adjusted for clustering at the firm level. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

presents the regression results obtained when interacting financing constraint (SA) with FPU. The interaction term coefficients between SA and FPU are 0.026 and 0.033, which are significant at the 10% and 5% levels, respectively. Based on the above analysis, our estimation supports the prediction that FPU exacerbates firms’ financing constraints and weakens their expansionary and R&D investments, thereby decreasing their TFP.

6. Conclusions

6.1. Conclusions and Policy Implications

In contrast to most studies, which concentrate on political and economic policy uncertainties and their effects, we take a more nuanced approach by distinguishing between FPU and macroeconomic EPU. We examine the relationship between local government FPU and firms’ TFP, using a sample of Chinese listed firms from 2003 to 2020. Our results demonstrate that FPU has a significant and negative effect on firms’ TFP. We confirm our finding by correcting for potential endogeneity bias using instrumental variables and PSM, applying a variety of alternative measures of the primary variables, using alternative samples that include firms in four municipalities directly under the central government, and adding control variables to the model. The cross-sectional analyses reveal that, compared with other firms, the negative externality of local FPU on TFP is more significant in firms whose headquarters are located in areas with low levels of marketization and fiscal transparency and that are subject to a high level of managerial myopia. We also conduct channel tests and find that FPU can exacerbate firms’ financing constraints and decrease their investments, including expansionary and R&D investments, thus reducing their TFP.

The following policy suggestions are based on our findings. First, considering the negative impact of FPU on market entities, local governments should prioritize the sustainability of policies when designing or adjusting policies, to promote the long-term sustainable development of the economy and society. Second, to mitigate the adverse externalities of political or policy uncertainty on firms, the central government should push for more balanced regional development, accelerate market-oriented reforms, and optimize resource allocation through the invisible hand, creating a favorable external environment for firms’ sustainable development. Third, the fiscal transparency of local governments plays a crucial role in mitigating the negative impact of FPU on firms’ productivity. Therefore, governments should strive to promote fiscal transparency, which can reduce information asymmetry, minimize firms’ decision-making errors, and prevent the loss of good investment opportunities. Finally, firms should prioritize improving their governance practices, particularly those related to their long-term interests and sustainable development. Such improvements will enable firms to better navigate the impacts of external policy uncertainties and ensure high-quality, sustainable growth.

6.2. Limitations and Future Potentials

This paper still harbors some shortcomings. While the utilization of text analysis to quantify FPU marks an innovation, there remains ample room for enhancing accuracy. Second, the research provides a limited analysis of the channel mechanisms through which FPU affects firms’ TFP, as well as its heterogeneous effects across different contexts. Investigating the significant relationship between FPU and corporate sustainability is an important topic in economics, warranting further scholarly exploration. Future research endeavors should prioritize the following key areas: First, refining measurement techniques for FPU to achieve greater accuracy and timeliness. Second, undertaking a more granular examination of the distinct impacts of uncertainty within various fiscal policies (e.g., tax vs. expenditure policies; expansionary vs. contractionary fiscal measures) on firms’ sustainable development, thereby furnishing policymakers with more tailored guidance. Third, delving into the multiple pathways through which FPU influences firms’ TFP, to fully elucidate the underlying mechanisms and provide both theoretical and practical insights to foster sustainable economic development.