Article Type : Research Article
Authors : Sheng Wang
Keywords : Digital finance; Model design; Financial digitalization; Fintech
Digital finance belongs to the pioneering
field of finance theory in the 21st century. It demonstrates unique
characteristics and drives the traditional financial theory into financial
digitalization. Some researchers in China have recently been engaging with this
new field to discover a new era in finance, namely digital finance. The main
purpose of this paper is to evaluate the model and characteristics of digital
finance explored in China as of date based on econometric theory and
methodologies. The principal results are problems derived from literature in
digital finance. It would be of importance, significance and implication for
the learning world and society to conclude that the technical characteristics
of digital finance can be summarized as a centralized digital financial system
developed based on blockchain technology; its core foundation must be fiat
currency digitization like E-CNY. Furthermore, decentralized Bitcoin is
impossible to be the cornerstone of digital finance, nothing more than just a
driving force to the development of digital finance.
This is a comprehensive review paper on the current
state of research about digital finance in China. Here the digital
finance-related papers referenced by this paper largely covered the latest
research status in this field. The existing ones summarized here are digital
finance and the related problems derived from the references. It is in the
future that research, the cornerstone of a series of follow-up work, will be
conducted. This may be the academic value of the current work. In order to
study the status of digital finance, there must be a holistic cognition of it.
There will be seven main parts to be investigated in the rest of this paper,
like digitalization of finance, the concept of digital finance, digital
financial technology features, risk characteristics of digital finance, digital
finance with the power in interpretation, digital finance research paradigm and
its problems, as well as conclusion and discussion. Now, they are stated
separately as follows.
The “14th Five-Year Plan” for the Development of
Digital Economy was issued by State Council of the PRC by the end of 2021,
Fintech Development Plan (2022-2025) was issued by People's Bank of China and
guidance on digital transformation for Banking and Insurance was issued by
China Banking and Insurance Regulatory Commission-clear objectives and
requirements were proposed for the digital transformation of financial
institutions in those documents [1]. Financial digitalization for the
traditional financial systems has been in the transformation process, and all
works for that process are at the beginning, all knowledge on the topics are
being constantly enriched on the way. Some research on the process of financial
digitalization was summarized to “the framework of three-dimensional analysis”,
forming a triangular structure with financial institutions (goods),
digitalization corporations (fields) and clients (peoples), and forming a
stable structure of business, fund and data interrelation between two of them,
based upon new regulatory framework established by “finance belongs to finance,
technology belongs to technology, data belongs to credit information” [2]. In
fact, the reformation of financial digitalization doesn’t begin at this moment,
it has been reforming since the past decades, for example, SWIFT would be one
of the best samples. The Society for Worldwide Interbank Financial
Telecommunication, or SWIFT, was founded in Belgium in 1973, with 239 banks as
members from 15 countries. Since 1977, SWIFT had 518 institutions as members
from 22 countries, and dealt with over 10 million messages in the same year. So
far, SWIFT has members over 11,000 institutions from over 200 countries and
districts, which has enormous business financial data transmitting and balance
within the system.
What the
digital finance is
The two digital financial inclusion indexes in two
different areas illuminate clarity the levels of development of inclusive
finance [20]. From the perspective of the convenience of using digital finance,
many scholars tend to accept the digital financial inclusion index compiled by
Peking University, as the alternative of digital finance [21-29]. In the
remainder of this paper, the digital finance the digital financial inclusion
index compiled by Peking University, which seems to reflect the characteristics
of financial technology.
Features of
digital financial technology
According to Ministry of Industry and Information Technology of the People's Republic of China, office of the Central Commission on Network Security and Informationization about instruction of accelerating the application of block chain technology and Industry Development, issued on May, 2021, the comprehensive strength of China's block chain industry will reach the world's advanced level by the year 2025, and the industry shall begin to take shape. The comprehensive strength of China's block chain industry shall have continued to improve, the scale of the industry shall have further expanded, by the year 2030. White Paper on the Development of Block chain Enterprises in China during year 2020-2021 was issued by Research Institute for China Electronics and Information Industry on November 2021, and it declared that the number of enterprises associated with the block chain should have exceeded 70,000, and strengthen “to prudent development of financial technology, to accelerate the digital transformation of financial institutions” [30]. Even though digital and internet finances are the concrete embodiment of financial technology of fintech, the greatest essential feature between them would be the block chain involvement and the paradigm revolution induced. The inner connection among digital identity, digital finance was established by the block chain technology, which has six features of credible, secure, privacy protected, socially responsible, intelligent and efficient [31]. Digital finance is the important cornerstone of the digital finance theory. Digital currency is deeply integrated with applications by new technology like block chain, cryptography, artificial intelligence, and traditional finance theory. Digital currency represents the latest developments of fintech [32]. Digital finance reflects the core concept of joint design, joint innovation and joint optimization of financial systems including content, business, mode, product, process, mechanism, supervision etc., and informational system with respect to technology. The next digital financial information infrastructure measures, system architecture and computing paradigm based on trusted Big Data and artificial intelligence would be bringing about revolutionary change in finance [33]. The relationship between digital technology and the real economy has varied from auxiliary, supporting, empowering development to integration, native and ubiquitous. Block chain, as one of digital technologies, under new technology innovation and Industry transformation, is playing an increasingly important role in those such as, cross-border payments, supply chain finance, agriculture finance, financial Inclusion, smart cities, agriculture, farmer and rural area, livelihood, interbank information sharing for settlement payments, foreign exchange transaction and trade finance, etc. [34]. All in all, the technical characteristics of digital finance can be summarized as a centralized digital financial system developed based on block chain technology, its core foundation must be the fiat currency digitization like E-CNY. The various technical features described above eventually come back to the establishment and improvement within the scope of a centralized digital financial system. The decentralized Bitcoin is impossible to be the cornerstone of digital finance, nothing more than just a driving force to the development of digital finance.
Digital
financial risk characteristics
The risks that arise from digital finance in practice
are mainly from scenarios’ applying the indirect financial risks which arising
from the illegal appropriation of information. As for the direct financial
risk, yes, as referred to here, the digital financial systemic risk, including
system architecture such as software support risks, hardware Information
capacity saturation risk, organizational and operational risks. The data
operational risk at all levels, as network users continue to expand, gradually
emerges [35]. Based on the data from the China Internet Information Centre, it
was displayed that Chinese internet network user scale was up to 988 million,
86.4% of these users’ payment using the network, as of December 2020, among
those internet users, including a large number of enterprise and individual
users. About enterprise users, enterprise risk taking levels such as corporate
earnings fluctuate, policy behaviour, survival state, attitude indicators, etc.
All of these indicators show that the development of digital finance
significantly improves the ability of companies to take risks [36]. As for
individual users including household users, the development of digital finance
significantly reduces the probability that families would fall into poverty in
the future [37]. However, digital finance as a fintech, which faces the risk
management issues mainly manifested in technology manages risk and laws and
regulations lag behind and regulate risks [38,39]. The dilemma facing the
development of digital finance is described by that the financial ethics and
awareness are weak, the relevant legal system is not sound, the social credit
reporting system is not perfect, the way of regulation is not scientific and
the level of technology is low [40]. The block chain network is about passing
credit, a network of trust and value, and it has embedded algorithms and
machine credits, which transforms financial risks like credit risk and section
operational risks etc. into algorithmic and technical risks [41]. Technology
manages risk is mainly embodied in some cases, complicated financial data
risks, concealment and diffusion characteristics, financial data security risks
like the cryptocurrency and web anonymity technology easy to steal data, to
tamper with data, to sell data, to compromise data, to pollute the data and to
make data attacks, etc. [42]. Since the first year of Internet finance in 2013,
That 5IABCDE is the representative of
various types of digital technologies and financial services continue to
integrate and innovate have formed innovative service methods of digital
finance in together, combining telecom with networked payments [43,44]. The
laws and regulations lag behind and regulate risks have caused various risks,
such as, data monopoly risk, damage to the rights and interests of consumers,
hindering the entry of competitors, data-driven merger and acquisition,
undermining the openness and transparency of digital financial markets,
privacy, business secret, social security, being difficult to define financial
data responsibilities and rights, execution being stuck in a dilemma,
cross-border regulatory risks being increased, “long-arm jurisdiction
“increasing data management risk, etc. [45]. Although stated above in aspects
of digital financial risk characteristics, this did not include the
representation of all digital financial risk characteristics. The biggest
financial evolution of the 21st century reflects the inevitable trend of the
development of the times. This prologue has only just begun. The connotation
and extension of financial risk must continue to evolve and change by the
development of digital finance. This shall be the norm of digital finance risk
characteristics that accompany the development of digital finance.
The
explanatory power of digital finance
According to the literature here, the explanatory
power of digital finance is primarily measured by the test of the t-statistic
of the parameter, through the empirical research done derived from the
establishment of relevant models. The digital finance explains the distribution
of topics, included but not limited to, financing constraints, enterprise
Innovation, total factor productivity, business growth, green innovation,
regional Innovation, region entrepreneurship, agriculture-related loans,
agricultural innovation, rural economy, entity economy, financial assets,
economic growth, economic inclusive growth, trade finance, banking risk,
impoverished/ poverty. For these topics, a large number of scholars have
adopted an empirical approach to study them. On the issue of financing
constraints, many scholars from different perspectives gave a different
portrayal. For example, to define the financing constraint problem based on
three dimensions, in terms of financing costs, financing structure namely
indirect financing ratio and financing efficiency [46]. There are also
differences in financing behaviour in companies of different sizes. Maybe it's
not a simple linear process in financing constraints for businesses. In fact,
as a measurement of financing constraints for business, the enterprise size is
a nonlinear representation of the logarithm [47]. The larger the enterprise,
its operating costs can occur complex changes, namely, as the size of the
enterprise grows, the high operating costs could decrease, and after reaching
the lowest point, the costs will continue to expand as the size of the
enterprise continues, successively show a rapid or even sharp rise, until to be
bankruptcy. The operating costs of the business are reflected in the
sensitivity to cash flows [48,49]. Corporate financing constraints seriously
affect corporate innovation. A measure of enterprise innovation is not exactly
unified. The general metrics are: the ratio of research or experimental
development expenditures and GDP, the ratio of inputs to outputs in science and
technology innovation, the number of patent applications, the ratio of number
of patents granted and the investment in scientific and technological
innovation, the invention creation yield [50,51]. No matter what kind of
enterprise innovation is measured, including the breakthrough innovation and
incremental innovation, or equity pledge, estimated by parameter t, the
criterion of the t test shows that the developments in digital finance seem to
be contributing significantly to it [52-54]. In order to ease financing
constraints, in turn, to promote enterprise innovation, digital finance
encourages companies to lower the threshold for corporate financing, reduces
valuation approval costs and reduces information asymmetry [55]. The enterprise
innovation has increased dramatically in total factor productivity. To improve
total factor productivity must surely be going to accelerate the growth of
enterprises, and to consolidate the development of a green economy in urban
areas enables regional innovation, namely the ratio of the number of patent
applications for invention to R&D and the green innovation [56,57].
Regional entrepreneurship is achieved through an increase in employment rates
[58]. It is broad and deep for digital finance to impact on all walks of life
in society, which significantly and positively impacts agriculture-related
lending, significantly promotes agricultural innovation [59,60]. It is not
significant for digital finance to improve the rural economy. The role played
by digital finance to improve economic development is difficult to form a
consistent decision based on existing research methods. As known by empirical
examination, digital finance significantly improves energy efficiency,
significantly promotes the development of the real economy, and at the same
time, significantly inhibits the tendency to allocate financial assets,
significantly contributes to economic growth, including inclusive economic
growth, and having injected new impetus into the development of small and
medium-sized enterprises [61,62]. The small and medium-sized enterprises play
an important role in foreign trade, but the trade finance is an important
initiative to make up for the development of small and medium-sized micro
enterprises. Digital finance drives the probability of poverty. Has deepened
the extent of multidimensional poverty, and does not have a statistically
significant impact on consumption among poor households. However, it has a
statistically significant impact on the consumption of non-poor households, of
a Matthew effect [63]. The size of consumer funds and the efficiency with which
they are used can be fed back to the bank's operational risk monitoring,
because consumers' spending money basically comes from their balance of deposits
in the bank. With the development of digital finance, the transfer of consumer
funds is controllable, and the probability of the risk occurring will be
significantly reduced. So, the development of digital finance will
significantly strengthen the risk bearing capacity of banks, the level of
economic development has increased instead has weakened significantly banks'
risk-taking capabilities.
Digital
finance research paradigm and its problems
Non-large data indicates that information disclosure
is not sufficient. Digital finance, or trusted finance can be decomposed as,
big data, model, algorithm, system, process, operation, management, design,
pricing, trade, hedge, risk control, internal control, supervision, etc., where
big data is trustworthy and a trusted model is a support. Quantitative finance
based on data and models will be the backbone of finance. However, the model on
which digital finance is based is not necessarily credible, that is, the best
evidence is that its model has a low fit to the data. According to the
literature listed here, there may be a big discrepancy from the actual
observation for a lot of empirical conclusions, since researchers rarely do
in-depth research on it, and properly explain the reasonable or unreasonable
parts of it. The current research paradigm for digital finance can be
summarized as follows?
This research paradigm is popular today, but the
results of the proof are not ideal. The reasons are at least the following?
One:
There is no rational logical analysis of
the model, to borrow from or the traces of imitation are too obvious, so be it.
Inadvertently, the logical relationship between variables is diluted, it seems
a little more casual;
Two:
The definition of variables is not precise
enough, and compare rough. The reason may be multifaceted, one of the high
probabilities is that variables are selected based on the ease of availability
of the data. As for whether such a variable can better illustrate the problem
itself under study, no one did mind whether subjectively or objectively;
Three:
“Only see the trees but not the forest”,
the results are generally interpreted in mode, regardless of whether the mode
is set appropriately after passing the empirical test, regardless of whether
the explanatory variable of interest is statistically significant. On the
surface, it looks beyond reproach, but in practice it is really ineffective.
The so-called; “the skin does not exist (the rationality of the model), the
hair will be attached (explain the statistical significance of variables)”? The
skin no longer has its proper function, what is the point of studying Mao?
Regrettably, this is how many papers do the like;
Four:
In addition to the so-called robustness
test in form, the empirical conclusions drawn about the results, have not been
explained from a theoretical level, but abruptly stopped. It is difficult to
justify itself;
Five:
Theoretical analysis lacks logical
reasoning, and it is simply to cite the existing literature to elaborate. The
problem is that the cited literature is inherently problematic, the discussion
is very inadequate. In this case, there seems to be a point of “the effect of
spreading false rumours”. This reason may be a good explanation, why so many
writing styles of articles are the same or like! The theoretical basis for
model building is insufficiently elaborated, the addition or subtraction of
variables is arbitrary. This invisibly affects the fit of the model, and it is
also difficult to avoid the risk of model interpretation. This is the case,
why, basically, it reflects the current state of similar research. The risk is
very high to draw conclusions based on models that are not robust. It's like
gambling, in an irrational situation. This is precisely the use of seemingly
rational scientific analysis methods, but to get a true portrayal of irrational
conclusions. It's all irrational to make any policy recommendations based on
such a conclusion, although the original intention was rational. The problem
with model setup is mainly based on the simplification of subjective
speculation, this results in the setup of the model or selection on, its
objectivity is relatively low, the model's ability to interpret samples is very
limited, or the predictive risk of the model is very large. This case is
detached from the academic value and practical significance of using models to
study practical problems, thus derived from that for the sake of the model and
the form of the model is formatted. It is extremely serious for such academic
thinking and research methodology to harm the entire academic community. It is
almost always exercised in the name of academic research to spread false
rumours, the degree of harm cannot be overemphasized.
Are the models on which most studies are based are
necessary? If necessary, why don't most research papers care about the
suitability of models? If not, why spend a lot of time building a model and do
regression analysis on it? Maybe a meaningful explanation is true that those
who are used by “me” are true. “Only see the trees and not see the forest”
thinking mode weakens the ability to interpret the variables of the
explanation, even the interpretive function that causes failure. The model's
building argument is inadequate, basically, to borrow someone else's model or
that the original does not move or to make small additions or subtractions to
the original model, etc., to prove the authority of the applied model. At the
same time, “tangerine planted in the south of Huai river is orange, but in the
north of Huai river is ‘bitter orange’”, the results are predictable. Many
studies are based on model sources derived from the same way. That's why a lot
of research is similar, lacking the necessary academic value.
According to the previous study, the following four
main conclusions are summarized, respectively, the statements are as follows:
The model lacks
rationality for its settings. This is primarily based on the model's fit to the
data, not enough to support the credibility of the model. It also shows that
the establishment of the model lacks scientific and rationality
This section is listed that empirical findings lack reliable arguments, their common problem is that the model fits very poorly, the resulting final result interpretation credibility is not enough to support their conclusions, see Table 3 in summary (Table 3).
The model setup lacks
logic, this leads to the occurrence of pseudo-regression
The pseudo-regression due to improper model settings, in a wide variety of academic papers that have been published, is seemingly, not easy to spot. But, from the regression reports of those papers, the evidence is still difficult to hide. Here are some of the specious phenomena that appear in the literature, to be discussed below.
The rationality of the
model structure lacks rigorous scientific arguments, the model reflects the
poor objectivity of the sample
Those are explained from the settings of the following
model. Settings for the baseline model, the explained variable, area technology
innovation and the core explain variables, digital finance, they have a
cross-generational relationship in time, but why is the relationship, between
them, of the same generation missing? What is the theoretical basis for the
settings? One of the reasons given in the text is “to consider that digital
finance may affect the level of technological innovation with the hysteresis
characteristics for the region, and to a certain extent, relieve endogenous
problems of reverse causality” [64]. Such reasons are subjective and
insufficient. In fact, technological innovation is a continuous process, not an
intermittent process, and the missing values for the same period can be
logically represented by 0. So, the scientific and rational nature is debatable
for such a model’s setting. Given the reasons above, the arguments are not too
robust for the conclusion of “digital finance significantly increased the regional
level of technological innovation” in the article. According to empirical
studies, other than that the digital financial upfront conditions have a
significant positive impact on today's bank loans. In a word, during the setup
of the model, why should digital finance lag behind one period? What role will
digital finance play in the same period? For the key factors, if the article
can be demonstrated from a theoretical level, or from an empirical point of
view, characterized by statistical analysis, then, that would solidify the
robustness of the empirical results, and circumvention should be nowhere in the
occurrence of pseudo-regression.
Literature citations
lack scientific rigor, this leads to citations only to my liking, and do not
ask whether the cited text is reasonable and credible.
Not all published papers are logically correct, and
all the conclusions obtained would be inevitable from logical reasoning. In
fact, there is no shortage of rare ones for conclusion fallacy caused by
research methodological errors, among the many published papers. Unfortunately,
there are no fewer authors that have very serious behavioural biases in quoting
in their papers. They just go to look for those related a few words as a point
of view to support their own research, but don't explore that just a few words
are reliable, scientific reasoning, or that the resulting arguments are
supported by lack of factual evidence based on virtual assumptions? There is no
doubt that there is a huge risk of error and omission for quoting without
interpretation pointed out that the conclusions obtained by are inconsistent
with those by for the same problem, but did not argue the difference between
them, namely, why are they so? Why? All unreasonable or who is reasonable and
who is unreasonable? These key elements are not clarified, but just a few
listings [65]. What does this mean? This is not an isolated phenomenon, but
it's the norm of academic writing for years. This model of research shows that
everyone only says their own thing, convergence citation, if not, it should be
charged. Such a research status, the beneficial communication for scholarship
is negative, nor does it fit the scientific logic of the problem [66].
The author would like to thank Kangjin Wang for
proofreading of the paper. Of course, the author would be only responsible for
any mistakes of the paper if available.