Introduction

As society progresses, the economy develops, and technological advancements have given rise to a series of financial instruments serving various aspects of our lives, making significant contributions to the nation’s economic take-off and people’s well-being. The organic integration of the Internet+ and artificial intelligence has accelerated the application of science and technology financial services in various social fields. At the same time, it has also brought certain risks and challenges. How to assess the security and guidance of current science and technology financial policies is a difficult question that lies before us. Looking back at the implementation history of China’s science and technology finance, it can be divided into three periods: the nascent period (1985–2008), the initial stage (2009–2014), and the rapid development period (2015–2024). During the nascent period, the overall characteristic of China’s financial technology was the "technology-first" + "government-led" model, where the government was the main driver of financial support for technology, primarily through increased fiscal spending in the field of science and technology and proposing a series of science and technology financial measures such as guiding financial institutions to increase investment in high-tech industries and enhancing fiscal support for technological innovation. In the initial stage, the overall characteristic was the "financial empowerment" plus "regional pilot" model, which for the first time clearly defined science and technology finance, pointing out that innovations in various financial tools such as credit, investment, and insurance promote technological development, committed to broadening the funding sources for technology enterprises, constructing a multi-level capital market, and formulating a series of science and technology financial policies conducive to "technological empowerment". During the rapid development period, the overall characteristic of science and technology finance was "integration of science and finance" plus "multi-stakeholder participation, multi-level market, diversified tools". In this period, the focus of science and technology financial policies shifted from constructing to perfecting the science and technology financial system, with corresponding policies being more regional, targeted, and innovative. Looking at the three important periods of China’s science and technology finance, the evolution of policies has shown a transformation from abstract to concrete, from discrete to comprehensive, and from fragmented to integrated. In this process, the Chinese government or departments have issued a series of guiding policies on science and technology finance, clarifying specific contents, such as in 1993 when the State Council issued the "Decision of the State Council on Financial System Reform". In 2014 and 2016, the content of financial technology was clarified in the form of government reports and the "13th Five-Year" Science and Technology Innovation Plan, and in 2019, the first top-level document "Financial Technology Industry Development Plan (2019–2021)" was released. Following this, six national ministries approved a one-year financial technology application pilot in 10 places including Beijing, Shanghai, and Sichuan. The pilot cities subsequently introduced financial technology policies with the aim of promoting the development of science and technology finance to improve quality and efficiency. According to statistics from the China Banking and Insurance Regulatory Commission, as of July 28, 2023, China has established over 1000 science and technology branches and specialized science and technology financial institutions. According to data released by the People’s Bank of China, as of the end of June 2023, the balance of medium and long-term loans in the high-tech manufacturing industry reached 2.5 trillion yuan, with a year-on-year increase of 41.5%, maintaining a high growth rate of over 30% for three consecutive years, and the role and influence of science and technology finance have gradually become apparent.

Up to now, the discussions on science and technology finance among policy makers, academia, and industry have mostly focused on how science and technology finance supports technology enterprises and innovation. However, there is still a lack of research findings on the evaluation of the effectiveness of science and technology financial policies. In light of this, it is very urgent to study the effects of financial technology-related policies, which plays a crucial role in guiding China’s science and technology financial market. Currently, science and technology financial policies cover a wide range of content: this includes policy effectiveness, whether policy supervision is in place, the consistency of related policies, the level of policy issuance, knowledge, property rights, financial talent, financial intermediaries, and more. Based on the collation of science and technology financial policy documents, this paper explores whether the execution effects of science and technology financial policies are reasonable and can have policy guidance. It conducts scientific quantitative analysis and evaluation research, which is of great practical significance for the steady advancement of China’s current science and technology financial policies, solving the problems of unbalanced and insufficient development of financial technology, and improving the policy and regulatory standards system.

The main sections of the paper are arranged as follows: the first part is literature review, the second part is materials and methods, the third part is quantitative evaluation of science and technology financial innovation policies, the fourth part is discussion, and the fifth part is research conclusions and recommendations.

Literature review

Looking back at the research achievements in science and technology finance, Schumpeter [1] was one of the first to propose that the new economy requires the reintegration of a series of newly formed production economic elements and external conditions into the existing production and operation system to stimulate economic development, which then leads to financial innovation. With the development of the social economy [2], technological innovation has become the new engine of social development. Combining it with financial innovation can further promote the development of emerging industries. However, market failures may occur during the development of emerging industries. At this time, timely government intervention in the market economy, macroeconomic regulation, and the introduction of corresponding industry regulatory policies to improve "general policies + specific policies" measures in response to market failures can not only solve the problems of limited growth of a large number of small and medium-sized enterprises in the post-pandemic era and optimize the national innovation environment but also achieve market supply and demand balance and improve the efficiency of resource allocation [3]. Reviewing the science and technology financial policies introduced since the reform and opening up, it has been found that these policies have been analyzed from different perspectives [4], and research methods such as co-word analysis and social network mapping have been used to study the policy theme changes in science and technology finance [5]. Research findings also indicate that there are differences in the implementation effects of science and technology financial policies on enterprises of heterogeneous nature and market types [6].

With the rapid development of China’s economy, various regions across the country have successively introduced related science and technology financial policies. A considerable amount of research has been conducted by scholars on how to measure the implementation effects of these policies. For instance, the level of development of science and technology finance in Guangdong Province has consistently been at the forefront nationally, but the problem of uneven development within the province is severe [7]. The provincial capital, Guangzhou, places great emphasis on the integration of science and technology with finance, achieving leapfrog development, thereby leading the way in the level of science and technology financial development. In contrast, some other areas have less than ideal development in financial technology [8]. In some provinces, due to a lack of emphasis on the integration of science and technology with finance, not only are there disparities in economic development levels between provinces, but significant differences also exist within the regions of each province. Local governments should analyze specific problems in detail and formulate a series of policies suitable for the local development of science and technology finance. Taking Jiangsu Province as an example, the provincial government has played different roles in financing roadshows for science and technology finance projects, thus alleviating the financing difficulties of technology-based SMEs and creating broad development space for them [9]. Besides, although some regions have a relatively complete system of science and technology financial policies, the synergy between these policies is not strong, such as in Shanghai [10]. Some areas even experience a general situation where coordinated development and policy impact efficiency are both low, such as in Hebei Province [11]. So, how to solve the problem of policy impact efficiency? Pang and Zhang [12] believe that by strengthening the coordination horizontally and vertically among departments, the efficiency of policy transmission from the provincial level to the municipal level can be improved. Given the differences in development among China’s provinces, those with lower development efficiency can formulate local policies based on regional development advantages and find paths that are conducive to their own efficient development of science and technology finance [13].

For a long time, the geographical position of the eastern region has been superior to that of the western region, and the eastern coastal areas, with their numerous ports, facilitate trade with countries around the world. Not only do the eastern coastal cities have advantageous geographical locations, but they also place emphasis on the development of financial technology, becoming economically developed regions [14]; whereas in the western region, cities such as Chengdu and Chongqing, despite having introduced corresponding policies for technological innovation to support technology enterprises and build innovation platforms, still exhibit low efficiency in science and technology finance [15]. In contrast, some western provinces with geographical environments slightly inferior to Chengdu and Chongqing, such as Jiangxi and Guizhou, have low levels of development in science and technology finance, yet the growth rate of their financial efficiency is fast. Huang and Qiu [16] evaluated the efficiency of China’s science and technology finance, suggesting that science and technology finance overall does not significantly enhance technological innovation and efficiency. With the strengthening of centrality in the capital network, macro-orientation, financial support, and policies for technical and financial patent protection will significantly improve the innovation performance of enterprises [17]. Science and technology financial policies can be used to promote enterprises in pilot cities, enhance innovation efforts, and improve innovation performance, accelerating the transformation and upgrading of industrial structures [18–20]. In line with the development requirements of China’s "13th Five-Year Plan," the financial industry should first meet the development needs of real enterprises and accelerate the construction of new science and technology financial service platforms [21]. As China’s economic development enters a new normal, practicing green finance and the concept of low-carbon development, science and technology finance has accelerated the process of high-quality economic development in China [22–25], improving the investment efficiency of corporate green innovation [26] and the driving effect of green development [27]. Meanwhile, the promoting effect of green technology innovation in science and technology finance exhibits spatial heterogeneity, with the eastern region showing an increasing marginal effect, and the central region displaying an inverted "U" characteristic [28], which is conducive to the risk resistance level of enterprises [29], providing strong support for the development of new productive forces [30]. However, some scholars believe that the excessive financialization of science and technology finance may have a negative impact on the real economy [31].

Policy evaluation not only has a direct promoting effect on policy formulation, implementation, and feedback adjustment [32], but also has a significant impact on the understanding of global trends and the establishment and adjustment of national science and technology financial strategies [33]. Policy evaluation originated with Suchman’s [34] five categories of evaluation, followed by Poland’s [35] "Three E’s" evaluation framework and Wollmann’s [36] classical policy evaluation. With the continuous exploration of policy evaluation methods, many methods of policy assessment have emerged. Shen et al. [37] used the difference-in-differences approach to empirically analyze the policy effects of the energy rights trading system, and applied the Analytic Hierarchy Process (AHP) to evaluate China’s military-civil fusion policies [38]. Feng et al. [39] constructed a BP neural network-based model for predicting carbon emissions in the electricity sector; some scholars used the Delphi method and Fuzzy Analytic Hierarchy Process (FAHP) to assess the main factors and sub-factors of green finance [40]; analyzed the risks related to China’s green infrastructure financing and prioritized them [41], as well as evaluating the environmental, social, and governance (ESG) factors and policy choices in China’s green finance investment decisions [42]. Currently, many evaluation methods have certain flaws, mainly reflected in the inevitable subjectivity and lower precision. The PMC index model method, however, obtains original data through text mining, which can largely avoid subjectivity and improve the precision of policy evaluation. The PMC method is based on the Omnia Mobilis (everything is in motion) assumption proposed by Ruiz Estrada et al. [43] when studying the application in Cartesian space, and has constructed the PMC index model, innovating the single policy evaluation method. This method judges the consistency of policies from a multidimensional perspective and directly observes the strengths and weaknesses of policy texts by constructing the concave and convex shapes of the PMC surface [44, 45]. Since then, this method has been widely applied to policy evaluation research: for example, in the construction industry [46]; nursing insurance policies [47] and farmland protection policies [48]; China’s pork industry [49]; the efficiency of green development in the Yangtze River Economic Belt [50]; China’s watershed ecological compensation policies [51] and the effectiveness evaluation of provincial public funding policies for private colleges in China [52]. The PMC method has been used to study the effectiveness of related policies, providing a good evaluation of the consistency and deficiencies of different policies during the specific implementation process.

In summary, scholars have made rich comparisons of the results from the perspective of the role of science and technology financial policies in promoting economic development, as well as the changes in the policy themes of science and technology finance, with a focus on elucidating the impact of science and technology finance on regional economic development and revealing the problems of unevenness and low efficiency in policy execution across different regions. With the robust development of China’s economy, the Chinese government or relevant departments have introduced a series of policies emphasizing the importance of science and technology finance, making the evaluation of the effectiveness of these policies particularly urgent. There are many existing evaluation methods, each with its own shortcomings. The Analytic Hierarchy Process (AHP) analyzes the correlation of a limited number of variables in a time series, which is limited by the number of variables and neglects the influence of some factors that are small yet actually present. The subjectivity of the Fuzzy Comprehensive Evaluation and Grey Comprehensive Evaluation methods is strong, and their resolution is poor [53].

Although the PMC method has been widely used in evaluating policies in many fields, its application in the evaluation of China’s science and technology financial policies is still a blank area. This paper attempts to use the PMC method to evaluate the effectiveness and consistency of science and technology financial policies, thereby analyzing whether the policy itself has any shortcomings in its guiding effectiveness. In view of this, the marginal contribution of this paper lies in: first, constructing for the first time a PMC-AE index model to quantitatively evaluate the science and technology financial policies issued by the central government since the "13th Five-Year Plan" of China, by extracting high-frequency words to construct 9 primary variables and 34 secondary variables, and establishing a multiple input-output table; second, judging the policy execution effects based on the intuitive concave and convex shapes of the PMC surface, and evaluating the consistency and effectiveness of science and technology financial policies; third, proposing five reasonable suggestions for the central government functional departments formulating science and technology financial policies. This is conducive to the Chinese government’s revision of science and technology financial policies, improving the policies, and enhancing the execution effects of science and technology financial policies.

Materials and methods

Mechanism of action for science and technology financial policy

Once the science and technology financial policy is enacted and implemented, it will have significant impacts on the economy, society, technology, politics, and the environment. During the implementation of the science and technology financial policy, it is influenced by factors such as the market, regulation, laws and regulations, timeliness, and talent. This also reflects a series of issues. By analyzing and judging the effectiveness of policy implementation, constructive feedback can be provided to the relevant departments. This is conducive to the further improvement of science and technology financial policies and promotes national economic development, thereby forming a complete cycle chain, as shown in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

https://doi.org/10.1371/journal.pone.0307529.g001

Statistical analysis

PMC-AE index model.

The difference between the PMC-AE index model and the PMC index model lies in the method of index calculation. The former uses acoustic emission technology for parameter fusion, which has advantages over the latter’s linear fusion. An Auto-Encoder is an unsupervised learning model. Based on the backpropagation algorithm and optimization methods (such as gradient descent), it uses the input data X itself as supervision to guide the neural network to try to learn a mapping relationship, thereby obtaining a reconstructed output Y. The principle is that it first compresses the input into a latent space representation, forming a hidden layer C (with spatial representation characteristics), and then reconstructs this compressed spatial representation into output, similar to the principle of the commonly seen Principal Component Analysis (PCA) method. During the encoding process, the Auto-Encoder can represent both linear and nonlinear transformations, while PCA can only perform linear transformations. The vector of its hidden layer has the effect of dimensionality reduction. Essentially, an autoencoder is a data compression algorithm, with both compression and decompression algorithms implemented through neural networks. Building an autoencoder requires two parts: an encoder and a decoder. The encoder compresses the input into a latent space representation, creating a hidden layer (or multiple hidden layers) that contains a low-dimensional vector of the input data’s meaning. It can be represented by the function f(x). The decoder reconstructs the latent space representation back into output, that is, reconstructs the input data through the low-dimensional vector of the hidden layer. It can be represented by the function g(x), and both the encoding function f(x) and the decoding function g(x) are neural network models. As shown in Fig 2, the encoding process of the original data X from the input layer X to the hidden layer C is: (1)

The decoding process from the hidden layer C to the output layer Y is: (2)

Then, the optimization objective function of the algorithm is written as: (3) Where dist is the distance measurement function between the two, usually represented by MSE (Mean Squared Error).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

https://doi.org/10.1371/journal.pone.0307529.g002

In the aforementioned formula, X = (x₁, x₂, x₃ ⋯, x_n)^T represents the established multi-dimensional indicators for evaluating science and technology financial policies, while Y = (y₁, y₂, y₃ ⋯, y_n)^T denotes the corresponding values of the output layer nodes. The activation functions for the hidden layer and output layer, denoted by f and g respectively, are typically Sigmoid, Tanh, or Softplus functions, among others, and may be identical or distinct. C = (c₁, c₂, c₃ ⋯, c_m)^T signifies the values of the hidden layer nodes; W and W^T represent the weight matrices between the input and hidden layers and between the hidden and output layers, respectively. The dimensions of these matrices are determined by the number of neurons in the preceding layer (rows) and the subsequent layer (columns). The bias vectors b = (b₁, b₂, b₃ ⋯, b_m)^T and B = (B₁, B₂, B₃ ⋯, B_n)^T pertain to the input-to-hidden and hidden-to-output layer transitions, with dimensions corresponding to the number of nodes in the subsequent layer neurons. The objective of Autoencoder (AE) training is to minimize the discrepancy between Y and X [54].

When the dimensionality of the original data is excessively high, or when a lower-dimensional representation is required, the neural network architecture can be modified by increasing the number of layers and progressively reducing the number of neural unit nodes. Consequently, after training, X is transformed into C through a nonlinear combination, C is then transformed into Y through another nonlinear combination, with the aim of achieving Y = X. Hence, C can be regarded as the nonlinear embodiment of X, and Y is derived from the decoding of C. Thus, C can serve as the composite score of policy texts after the integration of various indicators [55].

The construction of the PMC-AE index model mainly consists of the following four basic steps: (1) Classification of variables, calculation, and determination of model parameters. (2) Establishment of a multi-parameter input-output table for science and technology financial innovation policies, and assignment using text refinement techniques. (3) Fusion of multiple parameters using the autoencoder technology of neural networks to calculate the PMC-AE index. (4) Drawing of the PMC-AE index surface to evaluate the merits of science and technology financial innovation policies. The principle of constructing the PMC-AE index model is shown in Fig 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

https://doi.org/10.1371/journal.pone.0307529.g003

Data acquisition and preparation.

This article takes the beginning year of China’s "Thirteenth Five-Year Plan" as the base point, mainly selecting relevant policies on science and technology financial innovation issued by the Central Committee of the Communist Party of China, the State Council, and the innovation demonstration zones. The policy data sources include official websites such as the State Council, the People’s Bank of China, Shanghai, Guangzhou, and the Shenzhen Special Economic Zone. By collecting, organizing, and screening, we ensure the accuracy of the policy texts.

To ensure that the content of policy information is consistent with the theme of science and technology financial innovation, the following principles were adhered to when organizing and selecting policy texts: First, choose policies closely related to science and technology financial innovation, excluding those that only mention science and technology financial innovation. Secondly, the types of texts are general document management systems that have been published, involving outlines, plans, proposals, opinions, suggestions, notices, etc. To ensure the timeliness of policy research, 16 policy documents related to science and technology financial innovation were ultimately selected to construct a database of policies on science and technology financial innovation (see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Science and technology financial policies issued by the central government from 2016 to 2022.

https://doi.org/10.1371/journal.pone.0307529.t001

Policy content keyword.

In recent years, many citation analysis software tools such as CiteSpace and VOSviewer have emerged domestically and internationally. These tools can construct a comprehensive knowledge unit of literature, informing us of the key research areas in a discipline. However, such citation software has a problem of being "too broad" in determining research content. ROSTCM, on the other hand, can analyze the key research content of a field in more detail by capturing literature abstracts. The software is powerful and easy to operate, and the text mining process is shown in Fig 4. Based on the research results of previous scholars [56], and in combination with the mechanism of action of science and technology financial policies mentioned above, this paper has established 9 primary indicators that are broad and universal. The setting of secondary indicators is through comprehensive reference to relevant domestic and international literature [57], and in conjunction with the specific policy content of the relevant policy documents in this paper. Using ROSTCM6 software, the content of 16 policies is imported into the text mining database for word segmentation and screening. After effectively removing interference words such as "implement," "promote," and "improve," the high-frequency words are extracted (see Table 2). The document set is segmented, followed by word frequency statistics, which are displayed from high to low frequency, and a social network graph is established (see Fig 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4.

https://doi.org/10.1371/journal.pone.0307529.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5.

https://doi.org/10.1371/journal.pone.0307529.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. High frequency vocabulary table.

https://doi.org/10.1371/journal.pone.0307529.t002

To explore the operational level of policy tools, based on cybernetics and within the framework of "standard information behavior," this paper studies policy tools [58]. According to the mechanism of action of science and technology financial policies, different keywords are selected from the basic elements of policy orientation to analyze the basic elements of policy tools. Then, through comparison and classification, the main axis and selection codes of policy tools are formed.

Firstly, looking at the distribution of high-frequency words, financial and market vocabulary appears more than 300 times, indicating that the focus of the policy is very clear, reflecting the core idea of the financial marketization trend; the frequency of words such as enterprise, innovation, and technology exceeds 200 times, indicating that the documents focus on execution efficiency, while the remaining frequencies mainly reflect the scope of policy application.

Secondly, based on the Omnia mobilis hypothesis and combining the research findings of Ruiz Estrada et al. [43], related literature, and the results of vocabulary processing from text statistics, it is found that policy texts are centered on "innovation," with finance and technology as the core, deriving keywords such as talent, institutions, markets, and enterprises. Based on the aforementioned research, 9 primary variables and 34 secondary variables are set. The primary variables include: policy timeliness (X1); policy nature (X2); policy content (X3); policy field (X4); target of action (X5); perspective of action (X6); policy system (X7); issuing institution (X8); effectiveness level (X9).

The secondary variables contained within each primary variable are as follows:

1. Policy Timeliness (X1) includes three secondary indicators: (X1:1) Long-term (>5 years); (X1:2) Medium-term (3–5 years); (X1:3) Short-term (1–3 years), which are used to define whether the timeliness of the science and technology financial policy is long-term, short-term, or medium-term.
2. Policy Nature (X2) includes six secondary indicators: (X2:1) Forecast; (X2:2) Suggestion; (X2:3) Identification; (X2:4) Regulation; (X2:5) Description; (X2:6) Guidance. These are used to evaluate whether the policy possesses functions of suggestion, forecast, identification, regulation, description, or guidance.
3. Policy Content (X3) includes six secondary indicators: (X3:1) Finance; (X3:2) Technology; (X3:3) Market; (X3:4) Knowledge; (X3:5) Service Industry; (X3:6) Property Rights. These six indicators are used to evaluate the content included in a science and technology financial policy. Finance refers to financial policies issued by central documents (investment and financing, loans, interest rates, monetary tools, etc.); Technology refers to policies introduced to promote scientific and technological innovation (corporate scientific research, scientific and technological productivity, etc.); Market mainly involves China’s multi-level capital markets; Knowledge refers to systems such as science and technology finance talent and theories; Service Industry refers to the service industries related to finance, such as insurance, accounting, auditing, securities, and other financial services; Property Rights mainly refer to aspects such as the definition of property rights for technology-based enterprises.
4. Policy Field (X4) includes five secondary indicators: (X4:1) Finance; (X4:2) Technology; (X4:3) Talent; (X4:4) Market; (X4:5) Industry, which are used to measure whether a science and technology financial policy involves the fields of finance, technology, talent, market, and industry.
5. Target of Action (X5) includes four secondary indicators: (X5:1) Enterprises; (X5:2) Banks; (X5:3) Government; (X5:4) Other Financial Institutions. These four indicators are used to measure whether a science and technology financial policy involves these entities.
6. Perspective of Action (X6) includes two secondary indicators: (X6:1) Micro; (X6:2) Macro.
7. Policy System (X7) includes four secondary indicators: (X7:1) Clear Objectives; (X7:2) Feasible Scheme; (X7:3) Adequate Basis; (X7:4) Conformity with National Conditions. These four indicators are used to measure whether a science and technology financial policy has clear objectives, an adequate basis, a feasible scheme, and conforms to national conditions.
8. Issuing Institution (X8) includes four secondary indicators: (X8:1) State Council; (X8:2) Ministry of Finance; (X8:3) Central Bank; (X8:4) Securities Regulatory Commission, Banking and Insurance Regulatory Commission, and other financial departments. The data selected in this paper mainly comes from policies enacted at the central level.
9. Effectiveness Level (X9) includes five secondary indicators: (X9:1) Laws and Regulations; (X9:2) Administrative Regulations; (X9:3) Departmental Rules; (X9:4) Normative Documents; (X9:5) Industry Regulations. These five secondary indicators are used to evaluate the effectiveness level of a science and technology financial policy, and the levels of effectiveness decrease in the order listed.

The setup of each level of variables in the index model for the science and technology financial policies is shown in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Variable settings for the PMC-AE index model of science and technology financial policy.

https://doi.org/10.1371/journal.pone.0307529.t003

Setting of secondary variable parameters.

The calculation of the PMC-AE Index Model uses binary for the setting of secondary variable parameters, with efficacy indicators exhibiting progressive and exclusive characteristics. If the content is present in the policy document, it is counted as 1; otherwise, it is counted as 0. The score results of the secondary variable parameters are shown in Table 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Setting of secondary variable parameters.

https://doi.org/10.1371/journal.pone.0307529.t004

Establishment of multiple input-output tables.

Multiple input-output tables are an optional data analysis framework that can store a large amount of data and measure the variables of science and technology finance innovation policy evaluation from multiple dimensions, reflecting the evolution process of a certain policy. These tables consist of several primary variables and an unlimited number of secondary variables. This article combines the indicators of China’s science and technology finance innovation policy to establish multiple input-output tables (see Table 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Multi-input-output tables.

https://doi.org/10.1371/journal.pone.0307529.t005

Calculation of PMC-AE index.

The first step is to calculate the PMC-AE index. The specific steps are: first, place primary and secondary variables into the multiple input-output table; second, calculate the specific values of the secondary variables; third, calculate the specific values of the primary variables; fourth, calculate the PMC-AE index. The calculation steps and methods are shown in Table 6, where in the third and fourth steps of the formula, i stands for a primary variable, i = 1, 2, 3,…, m; j stands for a secondary variable, j = 1, 2, 3,…, n.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Steps for calculating the PMC-AE index.

https://doi.org/10.1371/journal.pone.0307529.t006

PMC-AE index scoring standards.

Based on Table 6, after assigning and calculating values one by one, since this article has selected 9 primary evaluation indicators, according to the evaluation standards of Ruiz Estrada [45], the calculated PMC-AE index ranges between 0–9. The specific values of the PMC-AE index are then classified into levels (see Table 7).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. PMC-AE index score levels.

https://doi.org/10.1371/journal.pone.0307529.t007

Drawing of PMC-AE surface.

The PMC surface is based on the PMC-AE index, presenting the policy evaluation effect in a graphical form. The purpose of constructing the PMC surface is to reflect the research results more vividly and intuitively. To construct the PMC surface, it is first necessary to build the PMC matrix based on the obtained PMC-AE index (Formula 4). This article sets 9 primary variables, thus forming a third-order matrix. Then, according to the calculation results of Table 6, the PMC surface is drawn, which visually displays the policy evaluation effect.

(4)

Quantitative evaluation of technology finance innovation policies

The fundamental idea behind the PMC index model is not to overlook potential related variables. Subjective selection of policy texts can lead to biases in policy evaluation models. Therefore, when choosing policy samples, it is not necessary to follow a specific pattern, nor is it required to classify policies according to criteria such as issuing institutions or levels of efficiency. The PMC-AE index model is developed through an unsupervised data dimension reduction technique that extracts the data characteristics of the main variables. Regardless of the sample selection, the core variables of the database are preserved. The PMC model can be used to evaluate both national and regional policies. In order to ensure the authoritative and instructive nature of policy text evaluations, this paper, based on the aforementioned considerations, randomly selects five major science and technology finance policies introduced at the central level in recent years for quantitative analysis, which reflects the specific implementation of policy texts. These policies play a major role in better evaluating China’s technology finance innovation policies, and the selected texts are shown in Table 8.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Five science and technology finance policies issued from 2020 to 2022.

https://doi.org/10.1371/journal.pone.0307529.t008

Based on the multiple input-output tables and policy texts, variables are set to obtain the multiple input-output tables for the five technology finance innovation policies, as shown in Table 9.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Multiple input-output tables for the five science and technology finance innovation policies.

https://doi.org/10.1371/journal.pone.0307529.t009

Scores from the five policy texts were used to construct a neural network model using autoencoder technology, and its parameters were learned. The data fusion process includes two stages. The first stage involves merging the scores of the second-level variables for each strategy to obtain the scores for the nine first-level variables of each strategy. In the second stage, the scores of the nine first-level variables are merged to derive the PMC-AE index for the five policies (see Table 10), which are then ranked according to Table 7.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 10. PMC-AE index of the policies.

https://doi.org/10.1371/journal.pone.0307529.t010

Drawing PMC surface charts

PMC surface charts provide a comprehensive analysis of science and technology finance innovation policies based on intuitive visualization results, with the depth of the concavity negatively correlated with policy ratings. The PMC matrix of the science and technology finance policies can be derived from Table 10 (see Table 11). Based on the aforementioned calculated data, the surface charts for the five science and technology finance policies can be obtained, from Figs 6–10.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6.

https://doi.org/10.1371/journal.pone.0307529.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7.

https://doi.org/10.1371/journal.pone.0307529.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8.

https://doi.org/10.1371/journal.pone.0307529.g008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9.

https://doi.org/10.1371/journal.pone.0307529.g009

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10.

https://doi.org/10.1371/journal.pone.0307529.g010

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 11. PMC matrix of the five science and technology finance policies.

https://doi.org/10.1371/journal.pone.0307529.t011

Discussion

The emergence of technology finance originally stemmed from the national demand for planning and construction of scientific and technological innovation. From the perspective of building an innovative country, the government has, in accordance with the needs of different stages, systematically matched different financial systems with a focus on priority and sequence, committed to achieving the deep integration of financial empowerment of technology and technology finance. A series of policy documents have been formulated to guide the work in various provinces (cities), achieving some results. Through the analysis of 16 documents, we extracted 9 primary variables and 34 secondary variables. Then, we randomly selected 5 national-level texts for evaluation using the PMC-AE model index, which provides a useful reference for assessing the effectiveness of China’s policies on financial innovation in technology. This will promote the improvement of a multi-level technological financial service system and strive to provide strong financial support for the accelerated construction of a modern industrial system.

In the course of our research, we found that China’s policies on financial innovation in technology have played a good guiding role in the specific implementation process. Evaluating five policies, three were rated as excellent and two as good, indicating that the policies were comprehensive and consistent in their effects. In fact, the different combinations of technology finance policies reflect the scientific nature of these policies. The objectives, target subjects, policy content, target goals, and policy system of the policy are clear and have played a significant role in guiding China’s economic development, social progress, and the development of small, medium, and micro enterprises, as well as the financial and technological sectors.

China has set the strategic goal of building an innovative country at the national level, elevating scientific and technological innovation to a national strategic status. It emphasizes the need to guide financial institutions to further optimize products, markets, and service systems according to the different needs of technology enterprises at different development stages, providing a more favorable environment for scientific and technological innovation and a platform for orderly competition in the financial industry. With the successive introduction of science and technology finance policies, the guiding effect of the policies has gradually become apparent. According to data from the People’s Bank of China, by the end of June 2023, the balance of medium and long-term loans in China’s high-tech manufacturing industry reached 2.5 trillion yuan, and the balance of loans for technology-based SMEs was 2.36 trillion yuan. The dividends of policy guidance have gradually emerged, stimulating new momentum for economic growth and achieving a virtuous cycle of finance in the technology industry. Our policy document research shows that X2 (nature of policy), X3 (content of policy), and X6 (target subjects) scored relatively high [53, 57], covering a wide range of content. Overall, China’s policy combination is basically consistent with the model of technological finance innovation and development found in existing research.

Science and technology financial innovation policies are generally aimed at cutting-edge technology innovation projects, which often involve large investments, long cycles, high risk, and significant uncertainty in research and development outcomes. The intrinsic value of technology companies is more reflected in the intangible assets formed by innovation. The development of technology finance urgently needs to strengthen the overall planning and policy coordination at the national level, balance the mechanisms for sharing benefits of innovation across different regions, and require regulatory authorities to view and steadily promote the development of technology finance rationally and objectively, maintaining the sustainable development of the technology finance system. This study indicates that the policy samples scored low in X8 (publishing structure) and X9 (effectiveness level) [56], reflecting a lack of unified scheduling and guidance. Especially, financial institutions in the specific practice process should fully understand the national policy intentions and, under the premise of controllable financial risks, boldly attempt to form a combination of punches with science and technology financial innovation policies.

This study also has some limitations: First, a high PMC-AE index value for a policy model indicates that the policy modeling research has considered a comprehensive set of variables with high consistency, and thus is superior. However, policy modeling is not the same as policy itself, so the appropriateness of transplanting the PMC index model method to policy evaluation is worth discussing. Second, in the target literature, there is ambiguity in the operability of some indicators’ design. For instance, the secondary indicators "finance, technology, market" under the primary indicator "policy content" overlap with the secondary indicators under the primary indicator "policy field" [53]. Despite different focuses in data mining, this can cause interference in the recognition by the autonomous learning module, affecting the accuracy of the evaluation. Third, this paper reviews the policies on technology finance since the 13th Five-Year Plan period, with fewer policies related to the 14th Five-Year Plan period being considered, which may lead to an incomplete analysis of technology finance policies. The use of the PMC model only considers single indicators, lacking a comprehensive grasp and precise quantification of policies. We attempt to address these issues in future similar research by seeking other more advanced methods to avoid deficiencies in research methodology and data processing.

Research findings and insights