Frontiers in Business, Economics and Management ISSN: 2766-824X | Vol. 7, No. 1, 2023 156 The Analysis of the Influencing Factors of Virtual Currency Price Based on Multiple Regression Method Xinyan Liu, Xinyue Zhang Anhui University of Finance and Economics, Bengbu 233030, China Abstract: In view of the influencing factors of bitcoin price, this paper firstly summarizes the relevant data of bitcoin price and annual cumulative supply in the past ten years from the Block Chain website, and selects several explanatory variables including actual bitcoin supply, market macroeconomic level and the number of users. Then, Eviews9 software is used to establish a multiple linear regression model. The econometric test methods such as multicollinearity, heteroscedasticity and autocorrelation were carried out to test and modify the model. The results show that the supply and demand factors of bitcoin itself, internal factors and macroeconomic factors all have a certain impact on the price of bitcoin. The results show that the price of bitcoin is mainly affected by three factors: the supply quantity, the global average annual gross product and the cumulative number of users, among which the global average annual gross product is the largest. Keywords: Virtual currency, Multiple linear regression, Bitcoin price, China market forecast, Eviews9. 1. Introduction With the development of datalization, modern money presents the characteristics of virtualization. The sovereignty of the general currency is in the national center, determined by the central bank; The sovereignty of virtual currency is in the distributed individual nodes and decided by individuals. Its essence is the individual-centered information economy. The value conversion of general currency is completed in the currency market. And the value conversion of virtual currency is completed in the virtual currency market. The value exchange between general currency and virtual currency is completed through the overall exchange of the two markets. Under special conditions, there are immature individual market exchange relations. In addition, general currency and virtual currency are in different markets, and their exchange mechanisms are different. In recent years, all kinds of virtual currencies have developed rapidly, and the digital currency represented by Bitcoin has achieved rapid development[1], has attracted wide attention from all parties. Since its inception in 2008, Bitcoin has become the fastest growing virtual currency in the world, with trading volume rising. However, due to the influence of policies, economic situation and many uncertain factors, the trading price of Bitcoin presents the characteristics of nonlinear, non-stationary and high volatility, which presents a great difficulty for the prediction of bitcoin price[2]. Bitcoin is a currency that lacks intrinsic value. There is no law governing the value of bitcoin. The risks of being a digital currency can be related to a number of factors[3].Through investigation and research of many factors affecting the price fluctuation of bitcoin, this paper focuses on the analysis of bitcoin currency supply, market macroeconomic level, the number of bitcoin users and other factors. Through multiple linear regression, an appropriate model is established to analyze the impact of various factors on the price fluctuations of bitcoin, and the influencing factors are regulated to achieve effective control of the price of bitcoin, and ultimately to help stabilize the virtual currency financial market. 2. Collection of Factors Affecting the Price of Bitcoin 2.1. Bitcoin supply quantity It is generally accepted that the supply and demand of an element is an important factor in the elasticity of commodity prices. The supply of a factor depends, first of all, on its productive capacity[4].The supply quantity of Chinese banknote should be roughly comparable with the GNP. On this basis, we can adjust appropriately according to environment. Unlike traditional currencies, bitcoin has been officially declared to have a maximum supply of 21 million coins since its birth in 2009.In 2009, the birth of bitcoin, block reward 50 bitcoins, at the rate of mining 50 bitcoins every 10 minutes increased year by year, when the total number of bitcoins increased to 10.5 million, block reward halved to 25, when the total amount reached 15.75 million, block reward again halved to 12.5, after the total number will be permanently limited to about 21 million. As Bitcoin nears its maximum supply, consumer demand for it increases. Increased demand and limited supply could push the price of bitcoin higher. The gradual increase in the value of bitcoin mining and the limited supply also create a lot of uncertainty about the price of the digital asset. 2.2. Macroeconomic level of the market The macroeconomic level of the market will more or less affect the consumption and investment of consumers. As a digital currency, Bitcoin does not participate in the actual commodity trading at present. Many people also dig or buy Bitcoin from the perspective of investment[5]And from the perspective of the investee, the macroeconomic level of the market also affects the price of bitcoin.In this regression analysis, we selected the global annual average gross domestic product (Per Capital GDP) as the index to measure, the macroeconomic level of the market. 2.3. Number of users The number of users represents the number of bitcoin wallets in the world, which can also reflect the demand of 157 Bitcoin users and show the liking and acceptance degree of the public for this virtual currency. 3. Empirical Analysis 3.1. Research Ideas According to our hypothesis, when an economic factor is affected by many other variables[6], we use multiple regression model to analyze the degree of influence of variables on its factors. The basic principle of multiple regression analysis is to use the least square method to model the relationship between multiple independent variables[7- 10]. 3.2. Research methods - multiple linear regression model (1) Classical Hypothesis Hypothesis 1: The expectation of the random error term is ),...,2,1,0(0)( i niE  Suppose 2: the variance of the random error term is a constant: ),...,2,1,0()( 2 niVar i  Suppose 3: the random error terms are independent of each other )(0),( jiCov ii  Assume 4 that the random error terms are not correlated with explanatory variables ),...,2,1)(,...,2,1(0),( njnixCov jji  .Usually assumed to be a non-random variable, this assumption automatically holds, at this time the explanatory variable is called exogenous explanatory. Hypothesis 5: There is no multicollinearity in the model, that is, there is no linear relationship between explanatory variables, or the observed values of explanatory variables are linearly independent. The purpose of this assumption is to avoid the linear representation of one explanatory variable nxxx ,,, 21  with other explanatory variables, so as to obtain unique results for n ,,, 21  the estimated values of the parameters. Hypothesis 6: The random error  term follows the multivariate normal distribution, ),0(~ 2 nIN  .This hypothesis actually includes hypothesis 1, hypothesis 2, hypothesis 3, hypothesis 4. 3.3. Model Setup According to the research object, bitcoin price ( ) is selected as the explained variable, and bitcoin supply ( ), market macroeconomic level ( ) and the number of users ( ) are selected as the explanatory variables. The general form of establishing multiple linear regression model is as follows:   332211 XXXYt Eviews9 was used to bring the collected and processed variable data into the multiple linear regression model established above, and the least square method was used to perform regression analysis on the data to obtain the OLS regression results, as shown in Table 1. Table 1. OLS regression results Variable Coefficient Std. Error t-Statistic Prob. C -10549.96 2335.907 -4.516430 0.0040 X1 -0.000163 0.000108 -1.511985 0.1813 X2 1.076284 0.220101 4.889962 0.0027 X3 3.37E-05 5.87E-05 0.573886 0.5869 R-squared 0.982932 — Mean dependent var 5505.849 Adjusted R-squared 0.974398 — S.D. dependent var 5800.528 S.E. of regression 928.1207 — Akaike info criterion 16.79337 Sum squared resid 5168448 — Schwartz criterion 16.91441 Log like lihood -79.96687 — Hainan-Quinn criter 16.66060 F-statistic 115.1784 — Dubbin-Watson stat 2.035716 Prob(F-statistic) 0.000011 — — — Preliminary model results are 3 5 21 1037.3076284.1000163.0-96.10549- XXXY  The multiple determinability coefficient is 982932.0 )( 11 2 2 2        yy e TSS RSS TSS RSSTSS TSS ESS R i i Where the multiplicity determinable coefficient represents the percentage of the explanatory variables jointly explained in the total sum of squares of deviation for kiii xxx ,...,, 21 .The larger the multiple determinable coefficients, the better the model fits the observed values, and the stronger the explanatory variables' ability to explain the explained variables. The t test of the coefficient of explanatory variable in the model is not significant, the coefficient sign of is opposite to the expected, the economic significance is unreasonable, and the model may have multicollinearity. 3.4. Test and correction of the model 1. Multicollinearity test of regression model In this experiment, Eviews9 software was used to test the regression correlation coefficients of all variables in the model, and the correlation values between all explanatory variables were obtained. The results are shown in Table 2.It can be seen from the analysis of the table that the correlation coefficient between each explanatory variable is greater than 0.79, which can be prelim natively judged that there is correlation, possibly serious multicollinearity, and the variance inflation factor is used to test. 158 Table 2. Regression correlation of each variable Correlation coefficient Y X1 X2 X3 Y 1.0000 0.7953 0.9877 0.9562 X1 0.7953 1.0000 0.8482 0.8165 X2 0.9877 0.8482 1.0000 0.9597 X3 0.9562 0.8165 0.9597 1.0000 In this paper, we calculate the variance inflation factor of the explanatory variable to make further explanation. Table 3. Test results of variance inflation factor method Variable Coefficient Variance Uncentered VIF Centered VIF C 5456461 63.3435 NA X1 1.17E-08 20.7056 3.5648 X2 0.0484 160.6981 15.0635 X3 3.44E-09 24.2649 12.6805 Experience has shown that if the variance inflation factor is used, it usually indicates severe multicollinearity between this explanatory variable and the remaining explanatory variables. The variance inflation factor of the explanatory variables tested 10VIF this time is much greater than 10, indicating a serious multicollinearity problem. (2) Correction of multicollinearity In this paper, the stepwise regression method is adopted for multicollinearity correction. After many references, eliminations and corrections, the optimal model obtained is as follows: 20502.117.11394- XY  t = (-11.4723) (17.8742) 7906.1,4881.319,9725.0,9756.0 22  DWFRR The modified coefficient of determination of the model is high, with a value of 0.9756. The results of F-test and T-test are significant, and the test of economic significance is reasonable. The results are shown in Table 4. Table 4. Stepwise regression results Variable Coefficient Std. Error t-Statistic Prob. C -11394.17 993.1916 -11.4723 0.0000 X2 1.0502 0.0588 17.8742 0.0000 (3) Autocorrelation test The partial correlation coefficient test of this operation is shown in Figure 1.According to the bar chart of Partial Correlation coefficients, none of the partial correlation parts of the model exceed the dotted line, indicating that there is no autocorrelation in the graph. Autocorrelation Partial Correlation AC PAC Q-Sta... Prob 1 -0.00... -0.00... 0.0007 0.978 2 -0.27... -0.27... 1.1427 0.565 3 0.240 0.255 2.1288 0.546 4 -0.27... -0.42... 3.6731 0.452 5 -0.32... -0.15... 6.1633 0.291 6 0.056 -0.24... 6.2572 0.395 7 -0.01... -0.00... 6.2623 0.509 8 -0.00... -0.09... 6.2624 0.618 9 0.100 -0.07... 7.4717 0.588 Figure 1. Partial correlation coefficient test Therefore, the final statistical model result is 20502.117.11394- XY  .That is the price of the virtual currency bitcoin is mainly affected by the global average annual GDP (Per Capital GDP), and the price of bitcoin will increase by 1.0520 units for every unit increase in the global average annual GDP 4. Epilogue 4.1. Analysis of Results As a branch of currency development that attracts much attention in today's society, the price of virtual currency is also affected by many aspects. Under the development of a huge market, the price of virtual currency bitcoin also increases or decreases every year. Its future development trend is also a hot topic of discussion in the industry. From the perspective of macroeconomics, this paper selects the cumulative annual supply of bitcoin, the global average annual GDP (Per Capital GDP) and the cumulative number of users as three factors affecting the price of bitcoin. Finally, the result of regression analysis shows that the price of virtual currency bitcoin is mainly affected by the global average annual GDP. 4.2. China's market laws and policies 1. Areas of Criminal law Regulation: Bitcoin transactions have the characteristics of decentralization, anonymity, trans-regional and so on. While ensuring the security of bitcoin transactions, these characteristics also attract many criminals to use bitcoin or carry out illegal actions against bitcoin. Cross-border exchange of virtual currency is used to convert criminal proceeds and earnings into overseas legal currency or 159 property, which has become a new means of money laundering crimes. Some involved economic crimes take virtual currency as an object of crime or as a criminal means, bringing security challenges to Chinese finance[11];However, at present, our legal regulations on Bitcoin are not perfect, and the criminal regulation on bitcoin-related behavior is facing many practical problems. 2. The field of civil and commercial law At present, our country has only two identical normative legal documents with effect on bitcoin. One is the "notice on the prevention of Bitcoin risk" led by the Banking and Insurance Regulatory Commission. It is the first normative document on bitcoin issued in our country, reminding the majority of bitcoin consumers to prevent the risk; Second, at the end of 2017, China's seven major financial regulatory authorities led by the People's Bank of China issued the "Notice on Preventing the risk of coin issuance financing" for the bitcoin market, that the coin issuance financing is suspected of illegal and criminal activities, requiring the closure of all the platform of coin issuance financing and the majority of investors to be vigilant about the risk of coin issuance financing[12].Both documents deal a major blow to Bitcoin-related businesses. At the same time, however, the legal documents do not directly address the question of the legality of bitcoin in the country or whether it is legally protected and whether it is legal for individuals to trade bitcoin with each other. It makes bitcoin in an unregulated "gray area". The domestic financial market is more chaotic and people are having more and more disputes with Bitcoin. References [1] Zheng Bugao, Xu Haidong, Li Shanyan. Analysis on the price impact of digital currency -- A case study of Bitcoin [J]. Price Monthly,2021(8): 10-17. [2] Mo Shibing, Lin Huijun, Chen Yunwei, Liang Yi, Wang Zhenkun. Bitcoin price prediction based on wavelet analysis based on ArMI-Transformer combined model [J]. Modern Information Technology, 202, 6(3):32-35 [3] Wu Xinyu, Wang Xiaona. [J]. The impact of Bitcoin on Volatility of Chinese stock market: An empirical study based on Double Asymmetric GARCH-MIDAS Model [J]. Journal of Chongqing University of Technology (Social Sciences), 2022, 36(08):65-75. [4] Niu Yumin. [J]. Time Finance,2017(27):220-221.] [5] Wu Xinyu, Wang Xiaona. [J]. The impact of Bitcoin on Volatility of Chinese stock market: An empirical study based on Double Asymmetric GARCH-MIDAS Model [J]. Journal of Chongqing University of Technology (Social Sciences), 2022, 36(08):65-75. [6] Zhang Chunli. [J]. Zhejiang Social Sciences,2022(11):52- 59+156-157.] [7] Xie Rui, Du Yue, Zhu Jiaming. Econometric analysis of influencing factors of Anhui housing price based on multiple regression [J]. Journal of Natural Sciences of Harbin Normal University,2022,38(4):36-42.] [8] Liu Wenhui, Gao Wei, Zhu Jiaming. Empirical analysis of influencing factors of China's total agricultural output value based on multiple regression [J]. Journal of Natural Sciences of Harbin Normal University,2022,38(01):14-20. [9] Zhu Jiaming, Chen Yan-qun, Shen Qiong, FANG Yue, Chen Yuan-ning. [J]. Journal of Xi 'an Shiyou University (Social Science Edition),20,29(04):17-23.] [10] Li Menghuan, Yang Penghui, Zhu Jiaming. Analysis of commercial housing price in Hainan Province based on multiple linear regression [J]. Journal of Jingdezhen University, 2019, 34(05):29-34.] [11] Weng Yinyun;Financial Security Challenges brought by virtual currencies from the Criminal Perspective [J]. Shanghai Legal Studies, Vol. 6, 2022 [12] Hua Jie. Application of Blockchain technology and Smart Contract in Intellectual property Rights Confirmation and transaction and its legal regulation [J].Intellectual Property Rights,2018(02):13-19.