Sage Journals: Discover world-class research

Abstract

Despite the considerable investments in Chinese public sector innovation in China and its social impact, there is insufficient information on how asymmetries in macroeconomic factors and business cycles affect public sector innovation. Based on provincial data from thirty regions (2000–2016), the study proposes a dynamic model to capture the asymmetrical impact of macroeconomic factors on local government innovation. The innovation-based predictive model indicated that a positive shock to macro-economic factors (i.e., gross domestic product, government administration expenditures, provincial degree of openness in economies, provincial degree of openness in society, higher education research and development expenditures, local revenue, and local expenditure) upsurge government innovation. In contrast, a negative shock to these factors disrupts local government innovation. Third, the model indicated a pro-cyclical relationship between positive and negative shocks to government administration expenditures and provincial government innovation during the boom and recession periods. The study offers an essential policy tool and some important implications for local government officials, future researchers, and policymakers to understand how cyclic and asymmetric behavior in local government innovation shapes the innovation efficiency of the public sector in China.

Subject Classification Codes: O30, O31, O32, 01, O38, O36, O35, O38, P25, Q01, H72.

Plain language summary

The study proposes a dynamic framework to capture the asymmetrical impact of macroeconomic factors on local government innovation. Provincial data (2000–2016) from thirty regions of China were analyzed using a non-linear Auto Regressive Distributed Lag panel-based model. The results indicated that a positive shock to macro-economic factors (i.e., gross domestic product, government administration expenditures, provincial degree of openness in economies, provincial degree of openness in society, higher education research and development expenditures, local revenue, and local expenditure) had upsurged government innovation, even though a negative shock had disrupted local government innovation. A pro-cyclical relationship was observed between positive and negative shocks to government administration expenditures and provincial government innovation during the boom and recession periods respectively. The study offers an essential policy tool and some important implications for local government officials, future researchers, and policymakers to understand how cyclic and asymmetric behavior in local government innovation shapes the innovation efficiency of the public sector in China. Of some limitations that offer future research avenues, this study only explored the connection between selected macroeconomic factors for China. Comparative or single-country studies may be conducted for local governments in other emerging economies, including African nations, Asian countries, European states, OECD nations, G7 countries, and BRICS states.

Keywords

China provincial government research and development expenditures NARDL innovation efficiency

Introduction

Modern public sector organizations constantly pursue innovative strategies to serve the public better and satisfy their stakeholders. Many historical events, including the New Public Management Experiments, the European reconstruction post the second World War, the New Deal (the US), the Welfare Fund development, and the digital government rise, reflect significant transformations in public sector management when public institutions tested with novel processes, ideas, administrative tools, policies, and technologies, for value creation and greater good (Lerman et al. 2021; Pavitt & Walker, 1976). The concept and dynamics of government innovation have attracted considerable attention from policymakers and academicians (Bibri, 2019; Daglio et al., 2014; Jugend et al., 2020). Such trends demonstrate that some conventional methods of dealing with public policy do not offer solutions to contemporary issues the public sector face today, including but not limited to HIV, Climate Change, poverty, and the aging populace), requiring different approaches and perspectives. Thus, the advancement of novel sets of institutional forms, financial support mechanisms, governance structures, accountability and collaboration structures, and public policy methods often distort conventional differentiation between private and public sectors to seek new solutions to deal with some of the pressing global issues (Criado et al., 2013; Schroeder, 2013). The unprecedented rate of advancement nowadays is opening new horizons for public sector organizations to integrate novel tools, methods, and approaches, but it is simultaneously exerting pressure on them to maintain pace with the changing world (Daglio et al., 2014).

Colloquially, government innovation is often associated with searching for novel ways to influence and improve the lives of the public. The same term also reflects finding novel methods to activate citizens (as collaborators) in the shared pursuit of a sustainable future. This strategy implies embracing novel ideas and technologies while simultaneously attempting to abandon and unlearn traditional thinking, philosophies, and structures. Despite the immense challenges of government innovation, its potentials are unlimited. Thus, many governments are altering their traditional work modes to realize the potential of innovation (Walker, 2006). Governments, specifically public sector organizations, are critically revisiting and reshaping the boundaries between their public and themselves (OECD, 2018). Innovations redefining and enhancing the interaction between citizens and governments facilitate the establishment of more transparent, accountable, and inclusive governments that further intensify innovation power. Such innovations encourage public participation and engagement in various phases of public policy development, that is, from idea creation to design, delivery, evaluation, and monitoring (Newman et al., 2001). Such shared endeavors aim to enhance the quality and type of service offered by the governments while changing the culture of public institutions so that the members of the public (citizens) are viewed as collaborators or partners informing and shaping services and policy (Newman et al., 2001).

Similarly, many experts consider innovation in the public sector a key indicator of socioeconomic progress. In 1978, the opening up and reform period in China was initiated after the 11th Central Committee third Plenary Session, which paved the way for historic changes toward socioeconomic progress. The inadequate building, infrastructure, and governance experience in the reform phase contributed to the development of many novel practices and innovations in local governments that took advantage of low hierarchical control from the central government in managing local affairs. Extant Chinese publications on government innovation reflect that most economists predominantly focused on innovation and organizational changes resulting from the opening up and reforms. The notion of government innovation in China has gained significant momentum, especially considering the following: eight government sessions on innovation, evaluation of 2,000 innovation programs that included more than 150 finalists and eighty winners, covering areas such as social service, poverty relief, government service, culture, eco-protection, education, and healthcare (Yu & Guo, 2019).

Past reports confirm that the financial allocation proportion of science and technology in the total allocations upsurged from almost 34% in 1998 to around 50% in 2007. As a result, local governments have significantly contributed to national innovation and technology processes (Bai & Li, 2011). Besides macroeconomic variables, including government expenditure, gross domestic product, and local government innovation nexus, showing mixed results (e.g., Bai & Li, 2011), more research is needed on other macroeconomic factors enabling local government innovation in China. Further, no study explains if asymmetries and cyclicality in macroeconomic factors symmetrically or asymmetrically affect local government innovation. Ignoring the aspects of asymmetries and cyclicality leads to misleading predictions and policy recommendations.

Therefore, the present article aims to assess the potential asymmetrical effect of macroeconomic variables on local government innovation at the regional level. This study contributes to the existing knowledge in the following ways. Most previous studies (e.g., Bai & Li, 2011; Zhao, 2012) have examined the linear relationship between macroeconomic factors (e.g., government administration expenditures, local expenditures, and gross domestic product) and local government innovation in China. Past studies on China have underestimated the effects of multiple booms and recession periods and the role of business cycles in the dynamics of government innovation in China. In other words, these studies have ignored the cyclical influence of selective macroeconomic indicators, including government administration expenditures on local government innovation efficiency during the contractions and expansion periods. To the best of our knowledge, no prior study has investigated the impact of macroeconomic variables on local government innovation in the presence of economic recessions and boom periods. Examining asymmetries is pivotal because of their dynamic role in the economy. Generally, asymmetries emerge due to positive and negative shocks or structural oscillations in macroeconomic factors, for example, total innovation or aggregate supply, demand, productivity, and real output. Changes in these factors may profoundly impact economic growth and innovation activities. Therefore, underestimating the cyclic aspect of innovation could generate inaccurate, spurious, and misleading empirical outcomes. This study adds to the empirical literature by examining selected macroeconomic factors’ dual (cyclical) influence on local government innovation in China.

The rest of the article is organized as follows: Section 2 comprises a review of literature; Section 3 constitutes methodology, including data, variables, sources, and analysis techniques; Section 4 entails results and discussion; Section 5 includes conclusions, limitations, and policy implications, and future direction.

Literature Review

In the energy economics literature, the asymmetric nature of innovation and its environmental impact, especially in the public sector, has gained considerable academic interest. The institutional theory and the innovation diffusion theory provide some insight into the asymmetries in various factors that affect the public sector or local government in China. Proponents of the former theoretical paradigm advocate that certain macroeconomic forces affect public sector innovation asymmetrically (Halvorsen et al., 2005; Kline & Rosenberg, 2010). These theorists argue that economic development or growth exhibits a non-linear interaction (inverted U-shaped curve) with innovation when new ideas are promoted or developed through enhanced demand and resources, indicating that moderate economic development is linked to high innovation levels (Kline & Rosenberg, 2010). Jordan et al. (2020) explain that a decline in innovation after a certain threshold is associated with institutional factors and resource constraints. In other words, the effects of economic growth on innovation in the public sector are asymmetric and dependent on indigenous-specific situations (Asunka et al., 2020; Zagler & Dürnecker, 2003). Additionally, public sector innovations are also affected by technological factors, yet their impact is asymmetrical in nature (Borins, 2001). Advancements in technology, for example, digitalization and technological infrastructure developments, promote public sector innovation. Still, asymmetries exist due to significant differences in technological progress across various Chinese regions. Youtie and Shapira (2008) clarify that highly developed areas, regions, or municipalities with advanced technological competencies in China may exhibit superior innovation output than regions lacking technological resources and infrastructure. This notion suggests that the effects of technological factors on innovation in the public sector are dependent on context and may generate varying results (Vassallo et al., 2023).

From the innovation diffusion theory standpoint, experts argue that certain institutional factors may affect public-sector innovation asymmetrically among local governments in China (Kankanhalli et al., 2017). Researchers explain that governance arrangements and administrative structure among different municipalities profoundly influence innovation outcomes, that is, varying levels of decentralization, red tape, and decision-making hierarchies could prevent or promote innovation. Generally, a high degree of decentralization in administrative structure, flexibility, and autonomy generates superior innovation outcomes by allowing local governments to experiment and learn in the innovation process. Therefore, different regions may exhibit different levels of innovation due to the asymmetrical influence of governance arrangements and administrative structure (Bhatta, 2003). Beyond the above factors, some economists agree that the effect of a regulatory framework and policy environment on public sector innovation may be non-linear and asymmetrical (Bekkers et al., 2011). Academicians believe supportive policies can increase innovation outputs, including regulatory flexibility, innovation-led procurement practices, and public-private sector partnerships. Still, the impact of such policies could vary among different regions due to contextual factors, causing asymmetric innovation outcomes in the public sector. Buys (2007) clarifies that the complex interaction between local conditions, regulatory frameworks, and policy settings establishes a dynamic environment shaping the asymmetrical impact of these factors on innovation. Nonetheless, the two theoretical paradigms discussed above offer invaluable information on the nuanced interactions between institutional arrangements, macroeconomic factors, and innovation in the public sector of China. Therefore, Zhu et al. (2022) asserted the need to recognize the heterogenous nature and impact of the pre-stated factors to assist practitioners and policymakers in developing targeted and robust policies to enhance innovation in different regions with different macroeconomic conditions.

Beyond the two theoretical perspectives noted above, academic literature explaining the nexus between government innovation and macroeconomic factors in China is scant and limited. Of a few published articles, Bai and Li (2011) explored how macroeconomic factors, including government support, R&D, bank loans, total infrastructure, and labor force, affect government innovation. The study revealed that bank loans adversely affected local government innovation. The scholars added that, unlike government support, lending by the private sector or financial companies was only manifested to attain profits by investing in good outlook, low-risk, short-term, high-success probability projects. Such innovation projects would have attracted funding, with and without government support. These experts observed no rise in the total R&D capital because the loans (offered by the lenders) shrank the investment potential of enterprises. Another important observation was that, instead of small and medium enterprises, the lenders preferred funding large enterprises that did not require infusing the R&D capital. This situation promoted a monopoly of R&D, where small and medium enterprises struggled. In new project evaluation, the lenders or investors often marginalize the patent number and focus on the commercial value, possibly explaining the positive and significant coefficient variable of finance.

Furthermore, Bai and Li (2011 reported a positive interaction between government innovation and support. The researchers argued that government support aims to promote enterprise investment, mitigate the public and private sector imbalance because of technology outflow), and reduce firm R&D costs. However, in effect, it is a double-edged sword. The authors observed that the total R&D supply did not rise because the public sector involvement in the project occurred at the expense of limited firm investment. The same support contributed to the high demand for material and non-material resources, for example, R&D professionals. The low flexibility in procuring such resources led to high pricing (high salary), resulting in high R&D costs. This situation encouraged enterprises to channel their resources toward other profitable ventures, thereby shrinking R&D funding and investments. Goolsbee (1998) supports that the salaries of R&D experts and talents occupy a significant portion of government support, contributing less to upgrading the R&D quality. Beyond the above, Bai and Li (2011) noted the favorable effects of infrastructure and labor force on local government innovation. As per the authors, improvement in labor quality and infrastructure generated high innovation, even though there were differences among the three studied regions, as evidenced by the negative value of the dummy. As per data estimates, the western and central regions demonstrated lower innovation efficiency than the eastern regions due to moderate or low economic growth, infrastructure, and less capability to nurture R&D talent.

Later, Zhao (2012) focused on the impact of different macroeconomic factors—such as the openness degree (society and economy), government administration spending, regional openness, local revenue, local expenditure, and gross domestic product per capita—on local government innovation. Although a positive interaction existed among all the selected variables, local government expenditures and access to revenues were the primary catalysts driving provincial innovation efficiency, as evidenced by the level of awards won by provinces (90% CI). Zhao (2012) argued that the importance of different macroeconomic factors, including administrative spending, GDP, openness, local revenues, and spending, suggests that the fiscal and economic capacity across regional and provincial governments in China is wide-ranging. At the same time, there is a significant disparity (high, moderate, and low) among regions vis-a viz the implementation of government innovation. Thus, the patterns and behaviors of local government innovation are highly diverse and complex, differing across space and time. At the local level, the internal resources of organizations and external environmental factors affect innovation behaviors and patterns. Zhao (2012) believed past studies had identified a few local government innovation socioeconomic antecedents for China from a single policy perspective. Nevertheless, the effects of factors promoting [or limiting] the aggregate local government innovation behavior at the Chinese provincial level, covering multiple policy areas combined, have yet to attract much attention. Instead, this paper combines empirical and theoretical aspects of local government innovation antecedents.

Following Zhao (2012), after a considerable gap, Deng et al. (2019) explored the local government innovation dynamics using different variables and provincial data of China from 2001 to 2006. The super-efficiency data envelopment analysis (SE-DEA) was used to analyze the macroeconomic factors and local government innovation nexus. The estimates revealed the following: i) technology spillover can be instrumental in improving local government innovation; ii) provincial engagement in regional innovation projects and initiatives occurs due to the competition of local governments to win foreign direct investment (FDI); iii) enhancing the intensity of eco-related regulations improve local government innovation by innovation compensation effect; iv) the probability of local government innovation diminishes when local governments undermine or reduce the intensity of eco-related regulations to attract more FDI; v) there is considerable constancy among regions in the effects of the eco-related regulations intensity and local government competition on local government innovation in China, linked to regional disparities in economic reforms, factor endowments, and economic progress.

More recently, some authors conducted studies on local government innovation in China using data from thirty provinces. For instance, Nie et al. (2021) analyzed city-wise data from 2010 to 2018 to estimate the implications of local government competition and support contribute to green economic progress using a spatial distance weight indicator. The author concluded that consistent local government competition for eco-innovation through setting up of economic zones enhanced the level and efficiency of green development. Another study by Khattak, Khan, Khan, et al. (2022) measured the promoting role of regional local government innovation in uplifting industrial structure. The threshold effect model examined China’s provincial-level data (2005-2019). Unlike the previous study outcomes, the estimates revealed that local governments’ inclination toward green government innovation had disrupted industrial structure growth. More so, there was a significant “U-shaped nexus between industrial structure upgrading and environmental regulation intensity across different regions, similar to Deng et al. (2019) conclusion.

Despite Zhao’s (2012) assertions, no previous studies have accounted for the complex role of asymmetries and business cycle dynamics when considering the link between local government innovation and macroeconomic factors. This study is an initial step in explaining an empirical and theoretical framework.

Methodology

Variables and Data Sources

Data for various study variables for thirty provinces (2001–2016) were compiled from the Statistical Yearbook of China. These variables included the data for winners and nominees of the Excellence in Chinese local governance, gross domestic product, government administration expenditures, provincial degree of openness in economies, provincial degree of openness in society, higher education research and development expenditures, local revenue, and local expenditure. Following Bai and Li (2011) and Zhao (2012), this study uses excellence in Chinese local governance as a proxy for government innovation. Table 1 depicts a description of all variables.

Table 1.

Summary of the Variables.

Variables	Notation	Description	Source
Innovation and excellence inChinese local governance	IECLG	Provincial numbers of winners of the IECLGAwards Program	Statistical Yearbook of China (2000–2016)
Gross domestic product	GDP	100 million yuan	Statistical Yearbook of China (2000–2016)
Government administration expenditures	GDE	1,000 yuan	Statistical Yearbook of China (2000–2016)
Provincial degree of openness in economies	PGOE	Percentage	Statistical Yearbook of China (2000–2016)
Provincial degree of openness in society	PDOS	Percentage	Statistical Yearbook of China (2000–2016)
Higher education research and development expenditures	HER&DE	1,000 yuan	Statistical Yearbook of China (2000–2016)
Local revenue	LR	1,000 yuan	Statistical Yearbook of China (2000–2016)
Local expenditure	LE	1,000 yuan	Statistical Yearbook of China (2000–2016)

Theoretical Setting and Model Specifications

Following Salisu and Isah (2017, a non-linear ARDL panel-based model was constructed to estimate different relationships between the study variables. This model is a non-linear illustration of the dynamic heterogeneous panel data model appropriate for large T panels. The non-linear model was adopted for the following three main reasons: i) the model enables capturing of asymmetries and nonlinearities in the panel data: ii) the model efficiently addresses heterogeneity in the panel data (as commonly observed in local government innovation); iii) the model is more suitable if there is mixed integration order or unit-root presence. Blackburne and Frank (2007) argue that large T and large N dynamic panels differ from small T and large N panels asymptotic. Small T panel computation often depends on random-and fixed-effect methods or a combination of instrumental variable and fixed-effect methods (Arellano & Bond, 1991), for example, the generalized method-of-moments technique (Blackburne & Frank, 2007). However, a key finding of previous literature on large T and large N is that the prediction of slope parameter homogeneity is usually inaccurate (Blackburne & Frank, 2007). Therefore, this study has adopted the dynamic heterogeneous panel data model, given the nature of panels used, that is, large T panels (Westerlund & Kaddoura, 2022). For stock price indexes, the Pesaran (2007) CD method also indicates the existence of heterogeneity. Generally, the mean group (MG) estimator and the pooled mean group (PMG) estimator are the prominent estimators for dynamic heterogeneous panel data models. The former depends on computing N time series regressions and coefficient averaging, while the latter comprises pooling combination and coefficient averaging (Blackburne & Frank, 2007). The Hausman method is often applied to assess disparities between the two estimation techniques. The two estimators also compute individual unit results apart from estimating panel regression outcomes. Hence, evaluating the response of individual government administration expenditures, local expenditures, and gross domestic product to shocks in local government innovation (for the asymmetric and symmetric situation) (when needed) is more computationally rigorous than the time series method for estimating the same. In addition, the response of both short-and long-run of two groups to shocks in local government innovation can be estimated.

In this study, the assumption that local government innovation symmetrically responds to government administration expenditures, local expenditures, and gross domestic product variations was relaxed to account for negative and positive government administration expenditures, local expenditures, and gross domestic product variations. Hence, the following panel ARDL symmetric version was obtained.

$\begin{matrix} Δ IECL G_{it} = α_{0 i} + β_{1 i} IECL G_{i, t - 1} + β_{2 i} GD P_{t - 1} \\ + β_{3 i} GA E_{t - 1} + β_{4 i} PGO E_{t - 1} + β_{5 i} PDO S_{t - 1} \\ + β_{6 i} HER & D E_{t - 1} + β_{7 i} L R_{t - 1} + β_{8 i} L E_{t - 1} \\ + \sum_{j = 1}^{N 1} ω_{ij} Δ IECL G_{i, t - j} + \sum_{j = 0}^{N 2} ψ_{ij} Δ GD P_{t - j} \\ + \sum_{j = 0}^{N 3} χ_{ij} Δ GA E_{t - j} + \sum_{j = 0}^{N 4} ϕ_{ij} Δ PGO E_{t - j} \\ + \sum_{j = 0}^{N 5} υ_{ij} Δ PDO S_{t - j} + \sum_{j = 0}^{N 6} τ_{ij} Δ HER & D E_{t - j} \\ + \sum_{j = 0}^{N 7} ς_{ij} Δ L R_{t - j} + \sum_{j = 0}^{N 8} σ_{ij} Δ L E_{t - j} + μ_{i} + ε_{it} \end{matrix}$ (1)

where $i = 1, 2, \dots \dots, N; t = 1, 2, \dots \dots, T$ ; $IECL G_{it}$ represent the log of excellence in Chinese local governance; GDP indicates gross domestic product; GAE reflects government administration expenditures; PGOE shows the provincial degree of openness in economies; PDOS signifies the provincial degree of openness in society; HER&DE indicates higher education research and development expenditures; LR shows local revenue; LE is local expenditure; $t$ represents the number of periods; $i$ is the sampled units; and $μ_{i}$ signifies the group-specific effect.

The above equation can be rearranged to get the error correction model:

$\begin{matrix} Δ IECL G_{it} = ζ_{i} υ_{i, t - 1} + \sum_{j = 1}^{N 1} ω_{ij} Δ IECL G_{i, t - j} \\ + \sum_{j = 0}^{N 2} ψ_{ij} Δ GD P_{t - j} + \sum_{j = 0}^{N 3} χ_{ij} Δ GA E_{t - j} \\ + \sum_{j = 0}^{N 4} ϕ_{ij} Δ PGO E_{t - j} + \sum_{j = 0}^{N 5} υ_{ij} Δ PDO S_{t - j} \\ + \sum_{j = 0}^{N 6} τ_{ij} Δ HER & D E_{t - j} + \sum_{j = 0}^{N 7} ς_{ij} Δ L R_{t - j} \\ + \sum_{j = 0}^{N 8} σ_{ij} Δ L E_{t - j} + μ_{i} + ε_{it} \end{matrix}$ (2)

Where $ζ_{i}$ is the error-correcting speed of adjustment term for each unit; and $υ_{i, t - 1}$ is the linear error-correction term.

In contrast to the symmetric scenario, this panel ARDL version (a non-linear panel ARDL) accounts for asymmetries in the IECLG in relation to GDP, GAE, PGOE, PDOS, HER&DE, LR, and LE variations. In this case, alternate shocks are predicted to have a dissimilar effect on IECLG. Hence, equation (1) was modified into the following equation (2) to obtain the asymmetric version.

$\begin{matrix} Δ IECL G_{it} = α_{0 i} + β_{1 i} IECL G_{i, t - 1} + β_{2 i}^{+} {GDP}_{t - 1}^{+} \\ + β_{3 i}^{-} {GDP}_{t - 1}^{-} + β_{4 i}^{+} {GAE}_{t - 1}^{+} + β_{5 i}^{-} {GAE}_{t - 1}^{-} \\ + β_{6 i}^{+} {PGOE}_{t - 1}^{+} + β_{7 i}^{-} {PGOE}_{t - 1}^{-} + β_{8 i}^{+} {PDOS}_{t - 1}^{+} \\ + β_{9 i}^{-} {PDOS}_{t - 1}^{-} + β_{10 i}^{+} HER & {DE}_{t - 1}^{+} \\ + β_{11 i}^{-} HER & {DE}_{t - 1}^{-} + β_{12 i}^{+} {LR}_{t - 1}^{+} + β_{13 i}^{-} {LR}_{t - 1}^{-} \\ + β_{14 i}^{+} {LE}_{t - 1}^{+} + β_{15 i}^{-} {LE}_{t - 1}^{-} \\ + \sum_{j = 1}^{N 1} ω_{ij} Δ IECL G_{i, t - j} + \sum_{j = 0}^{N 2} \\ (ω_{ij}^{+} Δ {GDP}_{t - 1}^{+} + ω_{ij}^{-} Δ {GDP}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 3} (ψ_{ij}^{+} Δ {GAE}_{t - 1}^{+} + ψ_{ij}^{-} Δ {GAE}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 4} (χ_{ij}^{+} Δ {PGOE}_{t - 1}^{+} + χ_{ij}^{-} Δ {PGOE}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 5} (υ_{ij}^{+} Δ {PDOS}_{t - 1}^{+} + υ_{ij}^{-} Δ {PDOS}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 6} (τ_{ij}^{+} Δ {PDOS}_{t - 1}^{+} + τ_{ij}^{-} Δ {PDOS}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 7} (ς_{ij}^{+} Δ {LR}_{t - 1}^{+} + ς_{ij}^{-} Δ {LR}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 8} (σ_{ij}^{+} Δ {LR}_{t - 1}^{+} + σ_{ij}^{-} Δ {LR}_{t - 1}^{-}) + μ_{i} + ε_{it} \end{matrix}$ (3)

where $GD P^{+}$ , $GA E^{+}$ , $PGO E^{+}$ , $PDO S^{+}$ , $HER & D E^{+}$ , $L R^{+}$ , and $L E^{+}$ denote positive shocks to gross domestic product, government administration expenditures, provincial degree of openness in economies, provincial degree of openness in society, higher education research and development expenditures, local revenue, and local expenditure. Besides, $GD P^{-}$ , $GA E^{-}$ , $PGO E^{-}$ , $PDO S^{-}$ , $HER & D E^{-}$ , $L R^{-}$ , and $L E^{-}$ shows a negative shock to gross domestic product, government administration expenditures, provincial degree of openness in economies, provincial degree of openness in society, higher education research and development expenditures, local revenue, and local expenditure. These shocks are respectively computed as positive and negative partial sum decompositions of each variable changes as defined below:

${GDP}_{t}^{+} = \sum_{k = 1}^{t} Δ {GDP}_{ik}^{+} = \sum_{k = 1}^{t} \max (Δ GD P_{ik}, 0)$ (4)

${GDP}_{t}^{-} = \sum_{k = 1}^{t} Δ {GDP}_{ik}^{-} = \sum_{k = 1}^{t} \min (Δ GD P_{ik}, 0)$ (5)

${GAE}_{t}^{+} = \sum_{k = 1}^{t} Δ {GAE}_{ik}^{+} = \sum_{k = 1}^{t} \max (Δ GA E_{ik}, 0)$ (6)

${GAE}_{t}^{-} = \sum_{k = 1}^{t} Δ {GAE}_{ik}^{-} = \sum_{k = 1}^{t} \min (Δ GA E_{ik}, 0)$ (7)

${PGOE}_{t}^{+} = \sum_{k = 1}^{t} {Δ PGOE}_{ik}^{+} = \sum_{k = 1}^{t} \max (Δ PGO E_{ik}, 0)$ (8)

${PGOE}_{t}^{-} = \sum_{k = 1}^{t} {Δ PGOE}_{ik}^{-} = \sum_{k = 1}^{t} \min (Δ PGO E_{ik}, 0)$ (9)

${PDOS}_{t}^{+} = \sum_{k = 1}^{t} Δ {PDOS}_{ik}^{+} = \sum_{k = 1}^{t} \max (Δ PDO S_{ik}, 0)$ (10)

${PDOS}_{t}^{-} = \sum_{k = 1}^{t} Δ {PDOS}_{ik}^{-} = \sum_{k = 1}^{t} \min (Δ PDO S_{ik}, 0)$ (11)

$\begin{matrix} HER & {DE}_{t}^{+} = \sum_{k = 1}^{t} Δ HER & {DE}_{ik}^{+} \\ = \sum_{k = 1}^{t} \max (Δ HER & D E_{ik}, 0) \end{matrix}$ (12)

$\begin{matrix} HER & {DE}_{t}^{-} = \sum_{k = 1}^{t} Δ HER & {DE}_{ik}^{-} \\ = \sum_{k = 1}^{t} \min (Δ HER & D E_{ik}, 0) \end{matrix}$ (13)

${LR}_{t}^{+} = \sum_{k = 1}^{t} Δ {LR}_{ik}^{+} = \sum_{k = 1}^{t} \max (Δ L R_{ik}, 0)$ (14)

${LR}_{t}^{-} = \sum_{k = 1}^{t} Δ {LR}_{ik}^{-} = \sum_{k = 1}^{t} \min (Δ L R_{ik}, 0)$ (15)

${LR}_{t}^{+} = \sum_{k = 1}^{t} Δ {LR}_{ik}^{+} = \sum_{k = 1}^{t} \max (Δ L R_{ik}, 0)$ (16)

${LR}_{t}^{-} = \sum_{k = 1}^{t} Δ {LR}_{ik}^{-} = \sum_{k = 1}^{t} \min (Δ L R_{ik}, 0)$ (17)

Equation (3) error-correction model generates the following equation:

$\begin{matrix} Δ IECL G_{it} = ϱ_{I} ϖ_{I, t - 1} + \sum_{j = 1}^{N 1} ω_{ij} Δ IECL G_{i, t - j} \\ + \sum_{j = 0}^{N 2} (ω_{ij}^{+} Δ {GDP}_{t - 1}^{+} + ω_{ij}^{-} Δ {GDP}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 3} (ψ_{ij}^{+} Δ {GAE}_{t - 1}^{+} + ψ_{ij}^{-} Δ {GAE}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 4} (χ_{ij}^{+} Δ {PGOE}_{t - 1}^{+} + χ_{ij}^{-} Δ {PGOE}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 5} (υ_{ij}^{+} Δ {PDOS}_{t - 1}^{+} + υ_{ij}^{-} Δ {PDOS}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 6} (τ_{ij}^{+} Δ {PDOS}_{t - 1}^{+} + τ_{ij}^{-} Δ {PDOS}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 7} (ς_{ij}^{+} Δ {LR}_{t - 1}^{+} + ς_{ij}^{-} Δ {LR}_{t - 1}^{-}) \\ + \sum_{j = 0}^{N 8} (σ_{ij}^{+} Δ {LR}_{t - 1}^{+} + σ_{ij}^{-} Δ {LR}_{t - 1}^{-}) + μ_{i} + ε_{it} \end{matrix}$ (18)

As shown in equation (18), the error-correction term $(ϖ_{I, t - 1}$ ) detects the long-run equilibrium in the asymmetric panel ARDL. In contrast, its linked parameter $ϱ_{I}$ explains the adjustment speed term by computing the time the system needs to converge to long-run equilibrium (in the existence of a shock).

Results and Discussion

Stationarity and Unit Root Testing

The third-generation unit root test (TGUNT) is depicted in Table 2. This test showed that all the variables were stationary at the first difference in the presence of structural breaks, implying that all the series were integrated of order one $I (1)$ . These results allowed us to conduct the co-integration analysis and employ the NARDL approach.

Table 2.

Third Generation Unit Root Test (TGUNT).

Stationary testing
At Level
Variables	$Z$	$P_{m}$	$P$	Break Years
IECLG	1.16	1.21	29.23	2001–2003–2005–2009–2012
GDP	1.27	−2.17	22.89	2001–2003–2009–2013–2015
GAE	1.23	−1.18	33.73	2002–2003–2005–2007–2010
PGOE	1.29	1.06	20.24	2001–2002–2004–2010–2015
PDOS	1.08	1.03	19.93	2002–2003–2005–2009–2012
HER&DE	1.230	−1.07	18.75	2001–2003–2006–2008–2011
LR	1.05	1.51	18.45	2002–2003–2007–2011–2014
LE	1.11	−1.04	17.33	2003–2005–2009–2010–2013
At First Difference
IECLG	−1.83**	12.74***	60.54***	---------------------------------
GDP	−4.43***	16.96***	270.68***	---------------------------------
GAE	−3.87***	11.29***	203.63***	---------------------------------
PGOE	−5.49***	39.25***	134.52***	---------------------------------
PDOS	−1.29***	11.82***	209.93***	---------------------------------
HER&DE	−3.76***	18.58***	289.93***	---------------------------------
LR	−1.21***	10.44***	201.34***	---------------------------------
LE	−4.66***	37.38***	130.48***	---------------------------------

Note. IECLG = innovation and excellence in Chinese local governance; GDP = gross domestic product; GDE = government administration expenditures; PGOE = provincial degree of openness in economies; PDOS = provincial degree of openness in society; HER&DE = higher education research and development expenditures, LR = local revenue, LE = local expenditure.

, ** and *** represent 1%, 5%, and 10% level of significance, respectively.

Co-integration Testing

Table 3 shows the findings of the Westerlund (2007) panel co-integration test for the four models. The results showed that all four test statistics, Gt, Ga, Pt, and Pa, were statistically significant, accepting the null hypothesis of co-integration. This result confirmed all variables’ long-run parallel and co-movements in the presence of cross-sectional dependency.

Table 3.

Westerlund (2007) Co-integration Test.

Model	Gt	Ga	Pt	Pa
Model 1	−7.550***(−30.131){0.000}	−17.686***(−2.753){0.003}	−115.847***(−86.380){0.000}	−33.918***(−16.384){0.000}
Model 2	−7.105***(−27.517){0.000}	−19.578***(−4.114){0.000}	−144.387***(−110.29){0.000}	−33.184***(−15.88){0.000}
Model 3	−7.391***(−29.199){0.000}	−18.246***(−3.156){0.001}	−172.573***(−133.91){0.000}	−32.585***(−15.470){0.000}
Model 4	−4.808***(−14.008){0.000}	−21.274***(−5.333){0.000}	−118.167***(−88.324){0.000}	−33.811***(−16.310){0.000}

Note. () = z-Value; {} = respective probability value.

***

Represents 1% level of significance.

The Westerlund and Edgerton (2008) co-integration test results are presented in Table 4. Both the test statistics $Z_{τ} (N)$ and $Z_{ϕ} (N)$ confirmed the existence of a long-run association among GDP, GAE, PGOE, PDOS, HER&DE, LR, LE, and IECLG when allowing for SBs in the slope and level. This result reflects that, in the presence of CSD, SBs, or economic shocks, the series in each model moves together in the long run.

Table 4.

Westerlund and Edgerton (2008) With a Structural Break, Regime Shift With a Trend.

Model	$Z_{ϕ} (N)$	$Z_{τ} (N)$
Model 1	3.018***(0.001)	−3.099***(0.001)
Model 2	−4.811**(0.010)	−4.811***(0.000)
Model 3	−3.269***(0.001)	−4.086***(0.000)
Model 4	−3.018***(0.001)	−2.673***(0.004)

Note. The LM-based test statistics $Z_{ϕ} (N)$ and $Z_{τ} (N)$ are normally distributed. The number of common factors is determined using the information criterion proposed by Bai and Ng (2004), and the maximum number is set to 5. () = probability value.

***

and **Indicate significance at the 1% and 5%, respectively.

All equations were estimated using the PMG and MG estimators. The Hausman test corroborated the findings of these estimation techniques. If the null hypothesis is not rejected, the PMG estimator is used, whereas if the null hypothesis is rejected, the MG estimator is used. Furthermore, the PMG estimator is the best under the null hypothesis, whereas the MG estimator is best under the alternative hypothesis. The Hausman tests support the PMG estimator as the most effective estimator for China’s GDE–IECLG nexus modeling. Table 5 shows that PMG is the efficacious estimator for all models, regardless of whether the model is non-linear (asymmetric) or linear (symmetric). Correspondingly, only findings from the preferred estimator (PMG) are presented and examined in this study.

Table 5.

Panel Regression Results Without a Higher Lag Order.

Variable	(a)Model with LR	(b)Model with LE	(c)Model with GDE	(d)Model with PDOS
A: Models without asymmetry
GDP	0.603***(0.183)	0.141***(0.0291)	0.904***(0.319)	0.741***(0.300)
$Δ$ GDP	0.542***(0.179)	0.365***(0.116)	0.753***(0.165)	0.712***(0.170)
GDE			0.324***(0.087)
$Δ$ GDE			0.213***(0.061)
PGOE	0.704***(0.114)	0.083***(0.0087)	0.682***(0.108)	0.623***(0.113)
$Δ$ PGOE	0.682***(0.143)	0.620***(0.149)	0.663***(0.102)	0.513***(0.183)
PDOS				0.318***(0.098)
$Δ$ PDOS				0.284**(0.061)
HER&DE	0.704***(0.114)	0.083***(0.0087)	0.682***(0.108)	0.623***(0.113)
$Δ$ HER&DE	0.682***(0.143)	0.620***(0.149)	0.663***(0.102)	0.513***(0.183)
LR	0.349**(0.136)
$Δ$ LR	0.196***(0.049)
LE		0.141***(0.0291)
$Δ$ LE		0.113**(0.051)
Constant	0.349**(0.136)	0.141***(0.0291)	0.324***(0.087)	0.318***(0.098)
Co-integration	−0.722***(0.077	−0.875***(0.080	−0.816***(0.083)	−0.754***(0.111)
Hausman test
B: Models with asymmetry
	0.563***(0.096)	0.758***(0.086)	0.578***(0.094)	0.233***(0.079)
$Δ GD P^{+}$	0.215***(0.098)	0.578***(0.094)	0.507***(0.087)	0.194**(0.079)
$GD P^{-}$	−0.120**(0.059)	−0.507***(0.086)	−0.581***(0.094)	−0.773***(0.115)
$Δ GD P^{-}$	−0.217***(0.088)	−0.507***(0.087)	−0.494***(0.085)	−0.439***(0.111)
$GD E^{+}$			0.422**(0.199)
$Δ GD E^{+}$			0.273***(0.098)
$GD E^{-}$			−0.783**(0.293)
$Δ GD E^{-}$			−0.265***(0.098)
$PGO E^{+}$	0.558***(0.083)	0.478***(0.078)	0.731***(0.057)	0.653***(0.060)
$Δ PGO E^{+}$	0.234**(0.095)	0.118**(0.055)	0.159**(0.063)	0.192***(0.061)
$PGO E^{-}$	−0.513***(0.086)	−0.820***(0.067)	−0.124***(0.035)	−0.318***(0.049)
$Δ PGO E^{-}$	−0.246**(0.094)	−0.159**(0.063)	−0.195***(0.061)	−0.126***(0.035)
$PDO S^{+}$				0.526***(0.087)
$Δ PDO S^{+}$				0.195***(0.063)
$PDO S^{-}$				−0.786***(0.076)
$Δ PDO S^{-}$				−0.615***(0.083)
$HER & D E^{+}$	0.579***(0.089)	0.568***(0.089)	0.656***(0.091)	0.686***(0.175)
$Δ HER & D E^{+}$	0.246**(0.105)	0.212**(0.102)	0.389***(0.107)	0.443**(0.169)
$HER & D E^{-}$	−0.263***(0.094)	−0.291***(0.095)	−0.212**(0.102)	−0.297***(0.095)
$Δ HER & D E^{-}$	−0.212**(0.096)	−0.297***(0.095)	−0.389***(0.107)	−0.212**(0.101)
$L R^{+}$	0.567***(0.095)
$Δ L R^{+}$	0.168***(0.082)
$L R^{-}$	−0.778***(0.082)
$Δ L R^{-}$	−0.186***(0.065)
$L E^{+}$		0.619****(0.092)
$Δ L E^{+}$		0.232**(0.095)
$L E^{-}$		−0.609***(0.093)
$Δ L E^{-}$		−0.256**(0.101)
Constant	0.322**(0.157)	0.370**(0.252)	0.398***(0.084)	0.386***(0.082)
Co-integration	−0.271***(0.073)	−0.202***(0.045)	−0.239***(0.080)	−0.279***(0.079)
Hausman test	0.142	0.491	8.531	0.740

The findings were based on five categories: i) the non-asymmetries (Table 5A) and asymmetries (Table 5B) LR–IECLG nexus; ii) the non-asymmetries (Table 5A) and with asymmetries (Table 5B) LE–IECLG nexus; iii) the non-asymmetries (Table 5A) and with asymmetries (Table 5B) nexus between GDE and IECLG associations; iv) the non- asymmetries (Table 5A) and with asymmetries (Table 5B) PDOS–IECLG nexus; v) the forecast performance of both linear and non-linear specifications over a range of forecast horizons and periods (see Table 6).

Table 6.

Out-of-Sample Forecast Performance Using 50% of the Whole Sample.

Category	Model	RMSE			Campbell-Thompson test
Category	Model	h = 5	h = 10	h = 20	h = 5	h = 10	h = 20
Model with LR	Symmetry	6.150	7.740	7.430	0.100	0.130	0.056
	Asymmetry	5.990	6.409	6.651	-	-	-
Model with LE	Symmetry	5.801	6.385	6.488	−0.060	−0.059	−0.047
	Asymmetry	6.980	6.670	6.309	-	-	-

Short- and Long-Run Estimation

Linear Relationships: Factors Affecting Local Government Innovation

After identifying the same order of integration, this study figures out how the response of LR, LE, GDE, and PDOS to changes in IECLG changes in the long and short term. Table 5 presents the panel regression results with linear and non-linear coefficients without a higher lag order. First, the results showed that PGOE, LE, LR, GDE, HER&DE, GDP, and PDOS are essential determinants of IECLG in both the short and long run. Accordingly, a 1% increase in GDP, GDE, PGOE, PDOS, HER&DE, LR, and LE led to an increase in IECLG by 0.60%, 0.32%, 0.70%, 0.32%, 0.70%, 0.35%, and 0.14%, respectively in the long run. Furthermore, the findings also indicated that an increase in GDP, GDE, PGOE, PDOS, HER&DE, LR, and LE led to an increase in IECLG by 0.54%, 0.21%, 0.70%, 0.28%, 0.68%, 0.19%, and 0.11%, respectively in the short run. First, the results imply that increasing GDP increases aggregate demand, production, and net exports. Furthermore, a rise in economic activities allows the government to collect more taxes. Higher taxes and revenues also lead to an increased allocation of funds to enhance government innovation. Increased funding for the provincial government will allow it to implement innovative solutions to long-standing issues, ultimately improving the quality of public services for all residents. In other words, increasing GDP increases tax revenue and enables the government to transform a novel idea into a strategy or method while enhancing existing solutions.

Second, increasing government administration expenditures could enhance local government innovation in China. A feasible explanation is that Chinese organizations have pursued innovation to improve their performance and adaptability to changing environments for achieving different goals, for example, service delivery, IT integration, and infrastructure development (Brudney & Selden, 1995; Munro, 2015; Newman et al., 2001). Even though such adoption has been costly, the local governments have significantly benefited from innovation outcomes (Bartlett & Dibben, 2002; Yigitcanlar et al., 2021). Recognizing the innovation cost as a priori, Chinese local governments have invested in administrative and technical infrastructure to enable better adoption and dissemination of innovation at the regional level. This finding provides sufficient empirical evidence that government administration expenditures have been crucial in generating superior innovation outcomes (Bingham, 1978; Gabris et al., 2001). This finding is consistent with some prior works for China (Zhao, 2012). However, it contradicts previous views (e.g., Bai & Li, 2011) that government spending, support, and R&D have adversely affected local government innovation in China.

Third, the findings suggest that the positive shocks to the degree of openness in regional economies and societies matter and influence how local governments implement innovations. Local government experts explain that innovation (the mechanism by which new objects, services, ideas, and products are created, developed, or adopted) might be new for adoption units. Local government innovation models for regions recognize the impact of states in proximity, as many local governments often mimic their neighboring states or regions when facing policy issues. The current finding validates that the degree of openness from a local government perspective, especially in social and regional economies, is pivotal and could affect communication and the creation of new policy concepts, procedures, rules, objects, and practices, thereby influencing the behaviors in which local governments implement innovation (Rogers, 2010; Walker, 2006).

Fourth, the results suggest that higher education research and development expenditure is important in local government innovation in China. Regionally, China has significantly invested in improving the quality and research and development output among higher education institutions to create world-class universities. For this purpose, the central and local introduced many policies to offer incentives and rewards for scientific output, patents, and innovations across the country (Andone, 2021). Globally, higher education R&D expenditures have experienced considerable restriction and transformation in institutional practices, controls, autonomy, funding, and focus of research focus (Do & Mai, 2022). In China, the disbursement of financial funds based on the National Science Foundation Projects, National Social Science Foundation Projects, and provincial projects has been crucial in increasing Chinese higher education institutions’ scientific and innovation output in the past few decades. These initiatives have resulted in new methods and processes for transforming novel ideas into strategies or methods while improving current methods.

Fifth, the findings suggested that local revenue and expenditure improve government innovation in China. This finding signals a more wide-ranging fiscal and economic capacity within the local governments in China, particularly in facilitating the implementation of local government innovation. However, the factors influencing local government innovation are numerous and highly complex, while the behavior of local government innovation changes in space and time. As an administrative institution, it is safe to conclude that the internal resources and external atmosphere could affect the local government’s innovation patterns and behaviors.

Non-linear Relationships: Factors Affecting Local Government Innovation

The non-linear relationship among GDP, GDE, PGOE, PDOS, HER&DE, LR, LE, and IECLG was estimated using the panel NARDL technique, as depicted in Table 5B. First, the results showed that positive shocks to GDP have led to increased government innovation in both the short and long run. The findings indicated that an increase in economic activities (e.g., production and consumption) increases GDP during boom periods. Additionally, in response to positive GDP shocks, the government boosts the tax rate to increase revenue. Increased government income also increases resource availability for enhancing government innovation. A boost in resources for the regional government will enable it to apply unique solutions to long-standing problems, eventually boosting the overall quality of public services for all residents. A positive GDP shock boosts government revenue and allows the government to develop a creative idea into a method or approach and enhance current solutions. Besides, the results indicated that the negative shock to GDP decreases government innovation in both the short- and long-run. The findings demonstrated that reducing economic activities such as production and consumption during recessions reduces GDP. In response to adverse GDP shocks, the government lowers the tax rate. Reduced government income also means fewer resources available to improve government innovation. Reduced resources for the regional government will make it unable to apply novel solutions to long-standing issues, ultimately lowering the overall quality of public services for all residents. A negative GDP shock reduces government revenue and the government’s capacity to create and convert an innovative idea into a method or approach.

Second, the findings indicated that positive shocks to GDE lead to increased government innovation in both the short- and long run. The results imply that positive shocks to GDE during the boom period enhance and improve government innovation. In addition, the findings also showed that adverse shocks to GDE lead to a decrease in government innovation in both the short and long run. A recession period is characterized by reduced consumption, investment, government expenditures, and international trade, resulting in reduced funds allocation for local government. Adopting and implementing government innovations require funding for technical and administrative infrastructures. However, the government cannot provide sufficient funds to enhance government innovation during a recession. Thus, the lower the GDE, the less likely it is to innovate. Third, the findings indicated that positive shocks to PGOE and PDOS increase government innovation in the short and long run. The results also showed that negative shocks to PGOE and PDOS lead to decreased government innovation in the short and long run. According to Table 5, negative shocks to PGOE and PDOS on government innovation are greater than the impact of positive shocks to PGOE and PDOS on government innovation. Besides, the findings also showed that the relationship between positive and negative shocks to PGOE and PDOS and government innovation are pro-cyclical.

Fourth, the results indicated that positive shocks to HER&DE lead to an increase in government innovation in both the short- and long-run. The results also showed that negative shocks to HER&DE lead to decreased government innovation in both the short- and long-run. Positive shocks to HER&DE have a more substantial influence on government innovation than negative shocks to HER&DE. Moreover, the results suggest a pro-cyclical link between positive and negative shocks to HER&DE and government innovation. Fifth, the findings indicated that positive shocks to local revenue and expenditure increase government innovation in the short and long run. The findings also indicated that negative shocks to local revenue and expenditure lead to a drop in government innovation in both the short and long run. Positive shocks to local revenue and local expenditure have a more decisive influence on government innovation than negative shocks to local revenue and local expenditure. Besides, the results suggest a pro-cyclical link between positive and negative shocks to local revenue, local expenditure, and government innovation.

Forecasting Performance Estimation: Asymmetric and Symmetric Panel ARDL

As a final step, the performance of the forecast of both the asymmetric and symmetric panel ARDL versions was evaluated to rationalize the need for estimating the effect of asymmetries in the nexus of PGOE, HER&DE, GDP, IECLG, LE, GDE, and GDE. This predictive analysis involved the following steps. First, multiple forecast periods (50% of the total observations) were considered to assess the robustness. Second, the predicting performance of the asymmetric and symmetric models was examined using the Campbell and Thompson (2008) method and the root-mean-square error (RMSE) method. The former method is computed as $1 - (\frac{MS E_{1}}{MS E_{0}})$ . The term MSE₀ and MSE₁ represent the mean square root (MSE) of predictions from the restricted (symmetric) and unrestricted (asymmetric) models, respectively. The model demonstrating a lower value of RMSE is expected to exhibit higher predicted performance. While a positive value of Campbell and Thompson's (2008) test suggests that the unrestricted model is superior to the restricted model, the negative statistical values mean the opposite. An out-of-sample prediction result was also generated for h (out-of-sample prediction horizon) =5, 10, and 20. Table 6 shows the out-of-sample prediction results for 50%.

Conclusion and Policy Implications

This paper examines the linear and non-linear association among GDP, GDE, PGOE, PDOS, HER&DE, LR, LE, and IECLG. This paper employed the third-generation unit root test for estimating the order of integration, which validated the first order of integration. Furthermore, this paper used the Westerlund (2007) co-integration test and Westerlund and Edgerton (2008) with a structural break, regime shift, and trend to investigate the long-term relationship between GDP, GDE, PGOE, PDOS, HER&DE, LR, LE, and IECLG, which confirms the existence of long-run connections between the chosen variables. ARDL and NARDL techniques were employed to estimate the linear and non-linear coefficients. According to the estimated results, GDP, GDE, PGOE, PDOS, HER&DE, LR, and LE positively impacted government innovation. The findings also indicated that the positive shocks to GDP, GDE, PGOE, PDOS, HER&DE, LR, and LE cause an increase in government innovation. Besides, the findings confirmed the negative relationship between negative shocks to GDP, GDE, PGOE, PDOS, HER&DE, LR, LE, and government innovation.

The following policy suggestions are proposed based on current findings. First, China should revise its government and private funding policies for the local administration to sustain government innovation during both expansion and contraction periods. Higher consumer income, purchasing power, and aggregate demand during boom periods facilitate the government to impose new taxes. This additional tax can be provided to the local administration to support their new knowledge creation regarding facilitating the public more efficiently. The government may face criticism from crucial stakeholders for imposing and collecting taxes from consumers and producers when consumers have low income, purchasing power, and aggregate demand during recessions. In this scenario, the government needs more funds for innovation. The government should encourage the banking sector to provide an easy loan with a low-interest rate to the local administration to support their innovation activities during recession times.

Despite the merits of this work, some limitations offer avenues for future research. First, this study explores the connection between GDP, GDE, PGOE, PDOS, HER&DE, LR, LE, and IECLG in China. However, given the importance of GDP, GDE, PGOE, PDOS, HER&DE, LR, LE, and IECLG, comparative or single-country studies may be conducted for local governments in other emerging economies, including African nations, Asian countries, European states, OECD nations, G7 countries, and BRICS states.

Footnotes

We would like to extend our sincere gratitude to Ms. Zhang Niyan for her valuable support and insights during the original submission. Her contributions have been instrumental in shaping the direction of this research,and we appreciate his willingness to share her expertise.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: The following funding sources support this work: i) the Humanities and Social Science Fund from the Ministry of Education of China (Grant No. 18YJC630231);ii) the Research Projects on Education and Teaching in Fujian Province Universities (Grant No. FBJY20230232);iii) Jimei University “Four New” Construction Research and Reform Practice Project (Grant No. C150813).

Ethics Approval

The study was approved by the Ethics Review Board of the Collaborative Innovation Center for Security and Development of Western Frontier,Sichuan University and the Technology Ethics Committee of Jimei University,Xiamen,China.

Informed Consent

There are no human subjects in this article and informed consent is not applicable.

ORCID iDs

Shoukat Iqbal Khattak

Manzoor Ahmad

Data Availability Statement

The data are available from the corresponding author upon reasonable request.

References

Andone

(2021). Globalization and internationalization of the education management. The Annals of the University of Oradea. Economic Sciences, 30(2), 336–343.

Arellano

Bond

(1991). Some tests of specification for panel data: Monte Carlo Evidence and an application to employment equations. The Review of Economic Studies, 58(2), 277. https://doi.org/10.2307/2297968

Asunka

B. A.

Anaba

O. A.

Amowine

(2020). Assessing the asymmetric linkages between foreign direct investments and indigenous innovation in developing countries: A non-linear panel auto-regressive distributed lag approach. South African Journal of Economic and Management Sciences, 23(1), 1–12. https://doi.org/10.4102/sajems.v23i1.3496

Bai

(2011). Regional innovation efficiency in China: The role of local government. Innovation, 13(2), 142–153.

Bai

(2004). A panic attack on unit roots and co-integration. Econometrica: Journal of the Econometric Society, 72(4), 1127–1177. https://doi.org/10.1111/j.1468-0262.2004.00528.x

Bartlett

Dibben

(2002). Public sector innovation and entrepreneurship: Case studies from local government. Local Government Studies, 28(4), 107–121.

Bekkers

Edelenbos

Steijn

(2011). Innovation in the public sector. Springer.

Bhatta

(2003). Don't just do something, stand there! Revisiting the issue of risks in innovation in the public sector. The Innovation Journal: The Public Sector Innovation Journal, 8(2), 1–12.

Bibri

S. E.

(2019). Big data science and analytics for smart sustainable urbanism: Unprecedented paradigmatic shifts and practical advancements. Springer International Publishing. https://books.google.com.pk/books?id=uOmaDwAAQBAJ

10.

Bingham

R. D.

(1978). Innovation, bureaucracy, and public policy: A study of innovation adoption by local government. The Western Political Quarterly, 31(2), 178–205.

11.

Blackburne

E. F.

Frank

M. W.

(2007). Estimation of nonstationary heterogeneous panels. The Stata Journal, 7(2), 197–208.

12.

Borins

(2001). Encouraging innovation in the public sector. Journal of Intellectual Capital, 2(3), 310–319. https://doi.org/10.1108/14691930110400128

13.

Brudney

J. L.

Selden

S. C.

(1995). The adoption of innovation by smaller local governments: The case of computer technology. The American Review of Public Administration, 25(1), 71–86.

14.

Buys

(2007). At loggerheads? Agricultural expansion, poverty reduction, and environment in the tropical forests. World Bank Publications.

15.

Campbell

J. Y.

Thompson

S. B.

(2008). Predicting excess stock returns out of sample: Can anything beat the historical average? Review of Financial Studies, 21(4), 1509–1531.

16.

Criado

J. I.

Sandoval-Almazan

Gil-Garcia

J. R.

(2013). Government innovation through social media. Government Information Quarterly, 30(4), 319–326.

17.

Daglio

Gerson

Kitchen

(2014). Building organisational capacity for public sector innovation. In: Background paper prepared for the OECD conference “innovating the public sector: From ideas to impact,” Paris (pp. 12–13).

18.

Deng

Zhang

Ahmad

Draz

M. U.

(2019). Local government competition, environmental regulation intensity and regional innovation performance: An empirical investigation of Chinese provinces. International Journal of Environmental Research and Public Health, 16(12), 2130.

19.

Mai

(2022). Role of the government in the establishment of world-class universities in China. Globalisation, Societies and Education, 20(3), 281–298. https://doi.org/10.1177/14782103221101775

20.

Gabris

G. T.

Golembiewski

R. T.

Ihrke

D. M.

(2001). Leadership credibility, board relations, and administrative innovation at the local government level. Journal of Public Administration Research and Theory, 11(1), 89–108.

21.

Goolsbee

(1998). Does government R&D policy mainly benefit scientists and engineers?National bureau of economic research Cambridge, Mass., USA.

22.

Halvorsen

Hauknes

Miles

Røste

(2005). On innovation in the public sector. In Halvorsen

Hauknes

Miles

(Eds.), On the differences between public and private sector innovation (pp. 2–21). NIFU STEP.

23.

Jordan

Markham

Altman

J. C.

(2020), Linking indigenous communities with regional development: Australia overview, Commissioned Report No. 5, Centre for Aboriginal Economic Policy Research, Australian National University. https://doi.org/10.25911/5f7edceea4190

24.

Jugend

Fiorini

P. D. C.

Armellini

Ferrari

A. G.

(2020). Public support for innovation: A systematic review of the literature and implications for open innovation. Technological Forecasting and Social Change, 156, 119985. https://doi.org/10.1016/j.techfore.2020.119985

25.

Kankanhalli

Zuiderwijk

Tayi

G. K.

(2017). Open innovation in the public sector: A research agenda. Government Information Quarterly, 34(1), 84–89.

26.

Khattak

S. I.

Khan

A. M.

Khan

M. K.

Liu

(2022). Do regional government green innovation preferences promote industrial structure upgradation in China? Econometric assessment based on the environmental regulation threshold effect model. Frontiers in Environmental Science, 10, 1616.

27.

Khattak

S. I.

Khan

M. K.

Sun

Khan

Wang

Niu

(2022). Government innovation support for green development efficiency in China: A regional analysis of key factors based on the dynamic GMM model. Frontiers in Environmental Science, 10, 1713.

28.

Kline

S. J.

Rosenberg

(2010). An overview of innovation. In Rosenberg

(Ed.), Studies on science and the innovation process: Selected works of Nathan Rosenberg (pp. 173–203). Edward Elgar Publishing.

29.

Lerman

L. V.

Gerstlberger

Lima

M. F.

Frank

A. G.

(2021). How governments, universities, and companies contribute to renewable energy development? A municipal innovation policy perspective of the triple helix. Energy Research & Social Science, 71, 101854. https://doi.org/10.1016/j.erss.2020.101854

30.

Liguo

Ahmad

Khan

Haq

Z. U.

Khattak

S. I.

(2023). Evaluating the role of innovation in hybrid electric vehicle-related technologies to promote environmental sustainability in knowledge-based economies. Technology in Society, 74, Article 102283. https://doi.org/10.1016/j.techsoc.2023.102283

31.

Khattak

S. I.

Ahmad

(2021). Measuring the impact of higher education on environmental pollution: New evidence from thirty provinces in China. Environmental and Ecological Statistics, 28(1), 187–217. https://doi.org/10.1007/s10651-020-00480-2

32.

Munro

(2015). Accelerating innovation in local government. Public Money & Management, 35(3), 219–226.

33.

Newman

Raine

Skelcher

(2001). Transforming local government: Innovation and modernization. Public Money & Management, 21(2), 61–68.

34.

Nie

Wan

Zou

Chang

(2021). Local government competition, development zones and urban green innovation: An empirical study of Chinese cities. Applied Economics Letters, 28(14), 1177–1180. https://doi.org/10.1080/13504851.2021.1942426

35.

OECD. (2018). Embracing innovation in government: Global trends 2018. OECD Paris.

36.

Pavitt

Walker

(1976). Government policies towards industrial innovation: A review. Research Policy, 5(1), 11–97.

37.

Pesaran

M. H.

(2007). A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Economics, 22(2), 265–312. https://doi.org/10.1002/jae.951

38.

Rogers

E. M.

(2010). Diffusion of innovations. Simon and Schuster.

39.

Salisu

A. A.

Isah

K. O.

(2017). Revisiting the oil price and stock market nexus: A non-linear panel ARDL approach. Economic Modelling, 66, 258–271. https://doi.org/10.1016/j.econmod.2017.07.010

40.

Schroeder

U. C.

(2013). The organization of European security governance: Internal and external security in transition. Taylor & Francis. https://books.google.com.pk/books?id=WHisAgAAQBAJ

41.

Jiang

Khattak

S. I.

Ahmad

(2021). Do higher education research and development expenditures affect environmental sustainability? New evidence from Chinese provinces. Environmental Science and Pollution Research, 28(47), 66656–66676. https://doi.org/10.1007/s11356-021-14685-w

42.

Vassallo

J. P.

Banerjee

Zaman

Prabhu

J. C.

(2023). Design thinking and public sector innovation: The divergent effects of risk-taking, cognitive empathy and emotional empathy on individual performance. Research Policy, 52(6), Article 104768.

43.

Walker

R. M.

(2006). Innovation type and diffusion: An empirical analysis of local government. Public Administration, 84(2), 311–335.

44.

Westerlund

(2007). Testing for error correction in panel data. Oxford Bulletin of Economics and Statistics, 69(6), 709–748. https://doi.org/10.1111/j.1468-0084.2007.00477.x

45.

Westerlund

Edgerton

D. L.

(2008). A simple test for co-integration in dependent panels with structural breaks. Oxford Bulletin of Economics and Statistics, 70(5), 665–704. https://doi.org/10.1111/j.1468-0084.2008.00513.x

46.

Westerlund

Kaddoura

(2022). CCE in heterogeneous fixed-T panels. The Econometrics Journal, 25(2), 283–309. https://doi.org/10.1093/ectj/utac012

47.

Yigitcanlar

Corchado

J. M.

Mehmood

R. Y. M.

Mossberger

Desouza

(2021). Responsible urban innovation with local government artificial intelligence (AI): A conceptual framework and research agenda. Journal of Open Innovation Technology Market and Complexity, 7(1), 71.

48.

Youtie

Shapira

(2008). Building an innovation hub: A case study of the transformation of university roles in regional technological and economic development. Research Policy, 37(8), 1188–1204.

49.

Guo

(2019). The Palgrave Handbook of Local Governance in Contemporary China. Springer Singapore. https://books.google.com.pk/books?id=Us6CDwAAQBAJ

50.

Zagler

Dürnecker

(2003). Fiscal policy and economic growth. Journal of Economic Surveys, 17(3), 397–418.

51.

Zhao

(2012). The regional disparities in Chinese provincial government innovation. Innovation, 14(4), 595–604.

52.

Zhu

Bai

Sarkis

(2022). Blockchain technology and supply chains: The paradox of the a theoretical research discourse. Transportation Research Part E Logistics and Transportation Review, 164, Article 102824.

Capturing the Asymmetrical Effect of Macroeconomic Factors on Public-Sector Innovation in Chinese Local Government: A Dynamic Technology Forecasting Approach

Abstract

Plain language summary

Keywords

Introduction

Literature Review

Methodology

Variables and Data Sources

Theoretical Setting and Model Specifications

Results and Discussion

Stationarity and Unit Root Testing

Co-integration Testing

Short- and Long-Run Estimation

Linear Relationships: Factors Affecting Local Government Innovation

Non-linear Relationships: Factors Affecting Local Government Innovation

Forecasting Performance Estimation: Asymmetric and Symmetric Panel ARDL

Conclusion and Policy Implications

Footnotes

Declaration of Conflicting Interests

Funding

Ethics Approval

Informed Consent

ORCID iDs

Data Availability Statement

References