Analysis of Factors Affecting Freight Demand Based on Input-Output Model

Freight demand is a highly variable process over economic and industrial structure, and accurate freight demand forecasting is the basis of transportation planning. In order to clarify the influencing factors of freight volume so as to analyze and predict the change trend of freight volume accurately, this paper analyzes the impact of changes in economic, industrial structure, and complete consumption coefficients on freight demand, through constructing an input-output model for transportation value analysis and forecasting freight volume by fitting data of transportation value and freight traffic. Studies have shown that the growth in economic aggregate is the main reason for the increase in the value of transportation, and the change in the complete consumption coefficient is the main reason for the increase in freight traffic.


Introduction
Freight demand forecasting is the basis of transportation planning. e wrong forecast of freight demand may lead to the advanced construction of transportation and waste of resources, such as Japan in the 1980s, or cause the delayed construction and the loss of freight transportation congestion, such as Iraq in the 1970s. erefore, freight demand forecasting has aroused the interest of the transportation sector and many scholars. is paper attempts to describe and analyze the driving factors that affect China's freight demand in order to better study the future development trend of freight.
In the past, researchers believed that there was a linear relationship between freight transportation demand and GDP [1][2][3]. e nature of the industry, such as the proportions of primary, secondary, and tertiary industries, has different effects on freight intensity [4]. Alises and Vassallo [5] show that the economic elasticity of freight transportation is gradually decreasing. However, Zhang et al. [6] believe that economic factors are still the main factors affecting freight volume. Wang et al. [7] believe that freight volume is affected by macroeconomics, industrial structure, and supply capacity. e research of Sun et al. [8] shows that resource and population distribution and investment in fixed assets are the main factors influencing railway freight volume. In addition, supply chain management strategies will also affect freight volume, such as inventory management strategies [9][10][11][12][13] and replenishment strategies [14]. Previous studies have analyzed the influencing factors of freight volume from many aspects, but because of the lack of comparative analysis of influencing factors, the importance and relevance of influencing factors cannot be reflected.
Many research models are established, such as developing and applying transportation market shares models [15], transportation supply chain models [16], spatial accounting models [17], and multiple regression models [18]. However, such methods are found difficult to predict the growth of freight volume accurately due to the complexity of the economic system and the rapid development of the service economy.
Input-output (IO) analysis method was proposed by American economist Leontief in 1925. e input-output model can analyze and investigate the quantitative dependence relationship between the freight transportation sector and other sectors in the national economy [19,20]. e IO model captures the nature of interindustry interaction, has relatively low data requirements, and is easy to implement [21]. Based on the expression of intersectoral multiplier effects, the IO model allows us to describe the effects on the transport system caused by shocks in the economic system, both from a theoretical and an application point of view. is paper considers an aggregate level of freight transportation, correlates the freight volume with the operation of the economic system, and then constructs the IO model. e model is structured on two levels: the first level forecasts the output value of the production department according to the exogenous demand and the second forecasts the tonnage of goods transported according to the relationship between the value of freight demand and freight volume in the economic system. rough the elaboration of time series of input-output tables for China, the model is validated and used to formulate forecasts for different future scenarios and then the influencing factors of freight traffic are analyzed.
is paper tries to answer the following two questions: (1) the relationship between economic development and transportation demand and (2) influencing factors of freight volume growth. e novel contribution of this paper is as follows: (1) the national freight volume was predicted by the input-output model. In order to reflect the impact of changes in industrial structure on freight volume better, this paper subdivides the industrial structure into 42 sectors, and the impact of 42 sectors on freight volume was analyzed, which is more specific than the previous division of the three industries; (2) comparative analysis of the influencing factors of freight value and freight volume was carried out through the complete decomposition model, and the influence degree of economic aggregate, industrial structure, and complete consumption coefficient on freight value and freight volume was obtained, so as to obtain the changing trend of freight volume accurately.

Literature Review
e literature related to this paper can be divided into three parts: (1) factors affecting freight demand, (2) input-output model, and (3) freight volume forecast.

Factors Affecting Freight Demand.
Robert et al. [22] have identified and evaluated the factors of freight demand. eir research shows that population, economic activities, fuel prices, environment, and policies are the main influencing factors, among which GDP and GDP per capita are commonly used indicators of economic activity. Fite et al. [23] conducted regression analysis on 107 indexes related to freight volume and believed that the producer commodity price index of construction materials and equipment (PCPI-CM&E) was the most relevant parameter. Agnolucci and Bonilla [24] conducted a study on the relationship between freight volume and GDP in the UK from 1956 to 2003. eir research found that the decoupling of freight volume and GDP accelerated, and the price and income elasticity also decreased to 18% and 65%. Wang et al. [25] proposed a hierarchical model. e model shows that the demand for truck freight can be estimated by truck traffic, population, number of companies, and income. Short et al. [26] studied the relationship between Sweden's economic activities and freight volume. e study found that, in the short and medium term, changes in imports and exports led to large fluctuations in freight demand; in the long run, freight demand and GDP are coupled, and there is no sign of decoupling. Wijeweera et al. [27] studied the impact of freight prices, international trade, and business cycles on Australia's rail freight demand. eir study shows that the fluctuation of freight rates and the Australian dollar was the most important factor affecting Australian rail freight. Alises and Vassallo [28] studied the impact of economic growth, industrial structure, and road transport intensity on the demand for road freight. e results show that, overall, the growth of total road transport demand is mainly driven by economic activities. Patil and Sahu [29] used regression and time series models to estimate the freight demand of Mumbai ports. eir research concluded that GDP and crude oil production are the most important factors affecting freight. Wang et al. [30] analyzed the relationship between freight demand and economic development. ey believe that China's overall economic development is decoupled from freight development, and the intensity of transportation is declining. Khan and Khan [31] analyzed the demand for rail freight in Pakistan. e study showed that GDP and freight are the two most important determinants of rail freight demand. Table 1 shows the comparison between some studies and this paper.

Input-Output Model.
ere are three main types of input-output models: inter-regional model-IRIO [54,55], multiple model-MRIO [56], and multiregional econometrics model [57,58]. Izard et al. [59] proposed an inter-regional input-output model, which introduced a trade coefficient, which represents the proportion of product m produced and used in region j used in the production and use of product n in the region. Because the proportion of trade is difficult to estimate, Moses [60] simplified the calculation of the trade coefficient, considering only the trade flow and no longer the destination. e wrong estimation of the trade coefficient may cause a large deviation in the forecast of freight traffic, which requires multiple corrections [61]. Miller [62] gave a detailed application explanation of the input-output model and constructed a fitting model of transportation and economy. Voigtlaender [63] applied a dynamic input-output model to forecast freight demand in the United States. Rey [64] discussed the application of econometrics in the inputoutput table model and studied the alternative methods and models of the input-output model. Havenga and Simpson [65] used an economic input-output (I-O) model as a platform, supplemented by actual data, developed supply and demand data classified by space and sector, and converted the supply and demand of South African freight into a freight flow through a gravity model. e above literature has conducted a lot of research on the application of inputoutput models from the perspective of theory and practice and provided theoretical and methodological support for the research of this paper.  [68], and Regan and Garrido [69]. ere are multiple measurement standards for freight traffic, and the common measurement standards are tons, ton-kilometers, and transportation costs (transportation costs or prices paid for transportation services). e accuracy of forecasting freight volume depends on the classification and aggregation of data and the estimated model [70,71]. Among the influencing factors of freight demand, the most important influencing factors are consumer demand, production structure, and trade mode [72,73]. Castro-Neto et al. [74] used the online support vector regression algorithm to predict the traffic flow of the road with good prediction accuracy. Chen [75] constructed a comprehensive transportation network, organically combined multiple transportation modes, and improved the accuracy of travel choice model prediction. Ahn et al. [76] combined Bayesian classifier and vector regression to predict the traffic demand of expressways and predicted and analyzed the expressway freight volume of Korea. Garrido and Mahmassani [77] developed a multinomial probit (MNP) model, which predicts freight volume based on time and space changes in transportation. Studies have shown that the model is more accurate. Pompigna and Mauro [78] used Italian economic data from 2000 to 2014 as the basis and used the macro-input-output method to analyze freight demand and forecasted Italian freight volume in 2027. e above research uses a variety of models and methods to forecast freight traffic. Generally speaking, national freight volume forecasting models are the integration of macroeconomic models.

Model Description.
e basic composition of the model is (1) Direct consumption coefficient: the direct consumption coefficient refers to the value of the unit's total output of the jth product sector directly consumed by the ith product sector in the production process, usually recorded as a ij (i, j � 1, 2, . . . , n). e formula of a ij is where x ij represents the input amount provided by the ith sector as an intermediate product to the jth sector and X j represents the total investment of the jth sector. (2) Complete consumption coefficient: the complete consumption coefficient is the number of products completely consumed by the jth sector when producing a unit product. It is called the complete consumption coefficient of the jth sector for the first sector, usually denoted as b ij . e complete consumption coefficient is the sum of direct consumption and all indirect consumption, and the calculation formula is b ij � a ij + n k�1 a kj b jk .
(2) e matrix formula is B � A + BA, and the formula for solving the complete consumption matrix can be obtained: where Y i represents the final use amount provided by the ith sector andX i represents the total output of the ith sector.
Its matrix is expressed as AX + Y � X. Column balance: initial investment + intermediate investment � total investment: n i�1 x ij + N j � X j , j � 1, 2, . . . , n, (5) where N j represents the added value of the jth sector.

Model
Assumptions. e basis of the input-output theory is Walras' general equilibrium theory, which involves the following basic assumptions.

Homogeneity Assumption.
Assume that each industrial sector produces only one homogeneous product. at is, the product is completely replaceable in this department but irreplaceable in other departments.

Proportionality Assumption.
e input and output of the department are directly proportional, that is, there is a linear relationship between input and output.

Stability Assumption.
e technology, production process, and management level are relatively stable.

Division of Departments.
is paper uses the inputoutput tables compiled by China from 2002 to 2017 as the basic data. To facilitate model processing and analysis, the 135 departments, 149 departments, and 41 departments in different years have been uniformly adjusted to 42 departments. e specific divisions are shown in Table 2.

Decomposition of the Input-Output Model.
According to the construction principle of the input-output table, the growth of transportation demand can be decomposed into three parts: increase in transportation demand caused by economic growth, industrial relevance and changes in production technology lead to the increase in transportation demand caused by changes in the complete consumption coefficient, and changes in transportation demand caused by the upgrading of industrial structure. Since the input-output table uses the value as the unit of measurement, the formula for calculating the economic value Q of transportation can be expressed as where Z represents transportation intensity and X represents the total output. Taking the derivative of the above formula, we can obtain where b j represents the complete consumption coefficient of each department for transportation, X j represents the total output of each department, and s j represents the proportion of each department's output to the total output. e matrix of the above formula is expressed as Z � B × S, where B is the complete consumption coefficient matrix and S is the structural variable.
Taking the derivative of formulas (3)-(8), we can obtain Putting formulas (8) and (9) into (7), we can obtain where SXzB represents the changes in the transportation economy caused by changes in the complete consumption coefficient, XBzS represents the changes in the transportation economy caused by changes in the industrial structure, and BSzX represents the changes in the transportation economy caused by changes in the total economic scale.
In the time period [0, t], the change of the economic value of transportation is Let B t � B 0 + ΔB, S t � S 0 + ΔS, and X t � X 0 + ΔX, and formula (11) can be expressed as follows: with Q � ZX � b j s j X, the demand for transportation value caused by different departments can be obtained as Let b jt � b j0 + Δb j , s jt � s j0 + Δs j , and X t � X 0 + ΔX; then, in the period [0, t], the changes in the economic value of transportation caused by different sectors are Combining the factors that cause the economic value of transportation into similar items and decomposing them according to the complete consumption coefficient change, industrial structure change, and total output change, the change in the economic value of transportation can be expressed as where b jq , s jq , and X jq , respectively, represent the impact of changes in the complete consumption coefficient, changes in industrial structure, and total output on the economic value of transportation.

Total Output Forecast.
e total output of a department refers to the total value of all goods and services produced by the department in a certain period. To predict the total output of a department, you can first predict the added value of each department and then use the input-output model to calculate the total output of each department.
According to the column balance principle A c X + N � X in the input-output table, the total output of the ith sector can be calculated as Because gray prediction is an exponential growth prediction, its prediction interval is trumpet-shaped, and the accuracy is poor in medium and long-term predictions. erefore, to ensure the accuracy of the prediction, this paper adopts the equal-dimensional gray number recursive dynamic prediction method. e basic principle of this method [79] is that only one value is predicted at a time, and the predicted value is used to replace the first value of the original data sequence, keeping the same dimension and predicting one by one, which speeds up the convergence of the predicted value with high accuracy. From the verification Mathematical Problems in Engineering 5 results, the 7-dimensional prediction overemphasizes the extension of the past trend, and the prediction is too high, the 5-dimensional prediction overemphasizes the extension of the recent trend, and the prediction is lower, and the 6dimensional prediction is more appropriate.

Complete Consumption Coefficient Prediction.
e RAS method, also known as the biproportional scaling method, was proposed by Deming and Stephan [80] in 1940. e basic principle is to first assume that the input structure of the target year and the base year are the same. Under the control of the total output and intermediate use in the target year, use a set of row control vectors and a set of column control vectors to adjust the base year accordingly. Each row element and each column element of the direct consumption coefficient matrix in the input-output table makes the total calculated direct consumption coefficient equal to each control data. e mathematical expression of the RAS method is where A t is the direct consumption coefficient matrix of the target year, A 0 is the direct consumption coefficient matrix of the base year, R is the total row multiplier matrix, and S is the total column multiplier matrix. e total row multiplier matrix R � r k r k−1 , . . . , r 2 r 1 can reflect the degree to which intermediate products are

Numbering
Industry name 1 Agriculture, forestry, animal husbandry, and fishery 2 Coal mining and washing industry 3 Oil and gas extraction industry 4 Metal mining and dressing industry 5 Nonmetallic minerals and other mining and dressing industries 6 Food manufacturing and tobacco processing industry 7 Textile industry 8 Textile, clothing, shoes, hats, leather down, and its products' industry 9 Wood processing and furniture manufacturing 10 Papermaking, printing and cultural, educational, and sporting goods manufacturing 11 Petroleum processing, coking, and nuclear fuel processing industry 12 Chemical industry 13 Nonmetallic mineral products' industry 14 Metal smelting and rolling processing industry 15 Metal products' industry 16 General and special equipment manufacturing industry 17 Transportation equipment manufacturing 18 Electrical machinery and equipment manufacturing 19 Communication equipment, computer, and other electronic equipment manufacturing 20 Instrumentation and cultural office machinery manufacturing 21 Crafts and other manufacturing 22 Scrap 23 Electricity and heat production and supply industry 24 Gas production and supply industry 25 Water production and supply industry 26 Construction industry 27 Transportation and storage industry 28 Postal industry 29 Information transmission, computer service, and software industry 30 Wholesale and retail 31 Accommodation and catering industry 32 Financial industry 33 Real estate 34 Leasing and business services 35 Research and rxperimental development 36 Comprehensive technical service industry 37 Water conservancy, environment, and public facilities management industry 38 Resident services and other services 39 Education 40 Health, social security, and social welfare industries 41 Culture, sports, and entertainment industry 42 Public administration and Social Organization

Economic Forecast.
Assuming that the return to scale of the production function remains unchanged and according to the global economic growth model, it is estimated that China's total GDP as where TFP represents total factor productivity, K represents capital stock, L represents the total labor force, and α represents the output elasticity of capital. According to the calculation of most scholars, α is 0.55. According to the basic laws of world economic development, when an economy enters a capital surplus, the output elasticity of capital will gradually decrease, while the output elasticity of labor will slowly rise. erefore, this paper assumes that the elasticity of capital-output will slowly decline from 0.6 in 2016 to 0.45 in 2035 and further to 0.4 in 2050.
Step 1 (predict the total labor force): the formula is where P t is the labor participation rate in year t, P t−1 is the labor participation rate in the previous year in year t, and α is a constant parameter. e labor participation rate refers to the ratio of the total employed population to the total population. Statistics from the National Bureau of Statistics of China show that, between 1995 and 2015, China's labor participation rate reached an average of about 56%, and it was quite stable. Based on this, it can be assumed that, between 2020-2050, China's labor participation rate will also remain at around 56%. e prediction results are shown in Table 3.
Step 2 (predict the capital stock): the formula is where K growth is the growth rate of capital stock, I t−1 is the amount of capital investment in the previous year, and 6% represents the annual depreciation rate of capital. Since 2010, the growth rate of China's capital stock has been declining. e decline in capital growth may be due to the adjustment of the industrial structure. is paper predicts that, from 2021 to 2030, the average annual growth rate of capital is 7%, from 2031 to 2040, the average annual growth rate of capital is 5%, and from 2041 to 2050, the average annual growth rate of capital is 3%.
Step 3 (forecast total factor productivity): the formula is where 1.3% is the potential growth rate of TFP, CB represents the growth rate of inertial growth in emerging developing countries, and FP represents factors that hinder productivity growth from failure. e formula for inertial growth rate is where 2.33% is a parameter of development inertia, calculated from historical data, and c is the national development speed parameter. e value is 1 for highgrowth countries, 0 for declining countries, and c is between 0 and 1. Economic development itself has inertia, so this paper believes that China's economic growth inertia CB is about 0.3. e hindering factor FB formula is where f is a hindering parameter, and its nature is opposite to c. It is 1 for declining countries and 0 for high-growth countries. Since China is an emerging developing country, the value of f should be between 0 and 0.5. Since 2006, China's total factor productivity has gradually declined. Total factor productivity is composed of human capital, R&D innovation, infrastructure, urbanization rate, and investment rate. Due to the aging of the population, the gradual improvement of infrastructure, and the decline in investment, the growth rate of TFP will continue to decline slowly in the future Trend [37]. erefore, this paper predicts that the TFP growth rate will be 1.5% from 2021 to 2030, 1.4% from 2031 to 4040, and 1.3% from 2041 to 5050. In summary, under the baseline scenario, China's economic development will achieve an average annual growth rate of 4.8% from 2020 to 2035 and an average annual growth rate of 3.4% from 2036 to 2050.

Industrial Structure
Forecast. Use the gray system structure prediction method to predict the industrial structure in the input-output table. e calculation steps and methods are briefly described as follows.
Step 1: list the original data. Take the intermediate input data and intermediate output data of 42 industrial sectors in the 2002-2017 input-output table as the modeling sequence.
Step 2: establish a gray dynamic GM (1,1) model for the above data series. e model can reflect N related factors, and the equation of the gray state model is where a represents the coordination coefficient between the variables.

Mathematical Problems in Engineering
Step 3: according to the above GM model group, list the system state equation matrix: or it is written as Predict the main indicator values of each quadrant, and calculate the structure of the intermediate input and intermediate output, respectively.
Step 4: use the Runge-Kutta method to solve the system state equation.
Step 5: perform cumulative reduction on the solution results of the equation to obtain the fitted and predicted values of each factor in the system and analyze the results. If there is a deviation, the coefficient matrix can be adjusted, simulated, and analyzed step by step to achieve a better prediction effect.

Forecast and Analysis of the Economic Value of Transportation.
rough the analysis of total output, industrial structure, and complete consumption coefficient, combined with the basic data of the input-output table from 2002 to 2017, the economic value of transportation in 2035 and 2050 can be predicted. e forecast data is shown in Tables 4 and 5.
By 2035, the transportation value will be 696,538 trillion yuan. From 2017 to 2035, the average annual growth rate of the transportation value will be 7.25%. By 2050, the transportation value will be 1,121,637 billion yuan, and the average annual growth rate of transportation value will be 3.23%.

Regression Models.
e difficulty of using the inputoutput model to analyze freight volume is to construct the functional relationship between value volume and freight traffic. Judging from the law of value volume and freight volume shown in the industrialization process of developed countries, the freight elasticity tends to gradually decrease, that is, the growth rate of freight volume is lower than that of value volume. e economic significance is that the tertiary industry develops faster in the middle and late stages of industrialization, while the growth rate of the primary and secondary industries that bring more freight demand has slowed, leading to a further slowdown in freight volume growth.
Based on the 2002-2017, input-output table and the statistical data of freight volume from the National Bureau of Statistics of China analyze the functional relationship between transportation value and freight traffic. e statistical data over the years is shown in Table 6.
We started regression analysis with one explanatory variable. Firstly, analyze the correlation between freight value and freight traffic. e Pearson correlation coefficient of the two is 0.977, and the significance test probability is 0.01. e analysis shows that the correlation between the two is significant, and predictive analysis can be performed. Secondly, to predict freight traffic, it is necessary to perform a regression analysis on freight value and freight traffic. Regression analysis refers to the statistical method of quantitative analysis of the interdependence between two or more variables. According to the relationship between the independent variable and the dependent variable, it can be divided into linear regression analysis and nonlinear regression analysis. Nonlinear regression includes power function and logarithmic function.
Given the characteristics of the growth rate of freight traffic, the quadratic function, cubic function, compound function, exponential function, logistic function, and s function are excluded because the predicted values of these models are too high or too low. Based on the relationship between freight value and freight volume from 2002 to 2017, this paper uses the freight value as an independent variable and freight volume as a dependent variable to construct onevariable linear, power, and logarithmic functions. e fitting effect and prediction are shown in Table 7.
Although the linear regression model and power regression model have higher R 2 , the logarithmic model also has good R 2 . In recent decades, the freight economic elasticity of the United States, Germany, and other major developed countries has shown a trend of "∩" shape. Considering that China is about to enter the postindustrialization period and the growth law of freight volume in developed countries, the prediction effect of the logarithmic function is better. Based on the forecast of the logarithmic function (M1), China's freight volume in 2035 will be 64.485 billion tons and the freight volume in 2050 will be 72.518 billion tons. In 2017-2035, freight volume will achieve Note. e 2015 data comes from the National Bureau of Statistics of China, and the rest of the data are predicted by the author. Among them, the birth rate is assumed to be maintained at the level of 2019, and the average life expectancy is 77 years. 8 Mathematical Problems in Engineering an average annual growth rate of 1.65% and freight volume in 2035-2050 will achieve an average annual growth rate of 0.79%. For comparison, China's average annual growth rate of freight volume from 2000 to 2010 was 9.09%, and the average annual growth rate of freight volume from 2010 to 2020 was about 3.8%. Figure 1 shows the changes in the value of each industry's demand for transportation. From the perspective of structural changes in the proportion of transportation demand by various industries, from 2017 to 2050, almost the proportions of the primary and secondary industries have declined, while the proportions of the tertiary industry have almost increased. is shows that, in the postindustrialization period, the development of the service industry was significantly faster than the development of other industries, which also led to a further slowdown in the growth of freight traffic.
From the statistics of the past years, the ranking of the value of transportation demand by each industry is shown in Figure 2.
As can be seen from the above figure, apart from the transportation industry itself, the largest demand for transportation is the construction industry, chemical industry, metal smelting, and rolling processing industry in order. From the comparison of the three industries, the second industry has the largest demand for transportation, with an average demand of 67.56%, followed by the tertiary industry, with an average demand of 29.36%, and, finally, the primary industry, with an average demand of 3.08%.

Time Series Models.
Time series regression is another reasonable method to examine the relationship between time-ordered variables. In recent decades, the freight volume per unit GDP has been declining year by year, which is closely related to the change of industrial structure. e fitting and forecasting trend line is shown in Figure 3 where Y t represents the freight in million tons in year t and GDP 1 represents the difference between the added value of the primary industry this year and the previous year. e remaining variables and parameters are self-explanatory. e fitting results are as follows (M3):     1  2  3  4  5  6  7 8  9  10 11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  41  32  33  34  35  36  37  38  39  40

(27)
According to Model 3, China's freight volume will be 65.02 billion tons in 2035 and 79.18 billion tons in 2050. e errors for univariate time series models vary from about 0.06% to 7.98%, whereas the error for multivariate time series models lies between 0.13% and 7.17% at 95% confidence level. Multivariate time series model shows better prediction effect and further verifies the accuracy of the regression model.

Analysis of the Impact of Changes in the Economic Value of Transportation.
e complete decomposition model is used to decompose the changes in the economic value of transportation, and the effects of changes in the complete consumption coefficient, changes in industrial structure, and changes in economic aggregates on the economic value of transportation are calculated, respectively. e results are shown in Table 8.
From 2017 to 2035, the economic value of transportation increased to 498,865 billion yuan. From the perspective of industrial structure, the secondary industry has the greatest impact on the incremental value of transportation, accounting for 53.42%, the tertiary industry also has a greater impact, accounting for 44.53%, and the primary industry has the least impact, accounting for 2.05%. From the perspective of influencing factors, the increase in total output has the greatest impact on the changes in the economic value of transportation, accounting for 80.88%, the second is the change in the complete consumption coefficient, accounting for 19.14%, and changes in the industrial structure make transportation the economic value increment decreased by 0.01%. e increase in the economic value of transportation from 2017 to 2050 was 923,964 billion yuan. From the perspective of industrial structure, the tertiary industry has the greatest impact on the incremental value of transportation, accounting for 49.28%, the secondary industry also has a greater impact, accounting for 48.77%, and the primary industry has the least impact, accounting for 1.95%. From the perspective of influencing factors, the increase in total output has the greatest impact on the changes in the economic value of transportation, accounting for 86.55%, the second is the change in the complete consumption coefficient, accounting for 13.9%, and changes in the industrial structure make transportation economic value increment decreased by 0.45%.

Analysis of the Impact of Changes in Economic
Aggregates on Freight Volume. Because of the uncertainty of total economic growth, two scenarios of faster economic development and slower economic development are considered: (1) Scenario of relatively rapid economic development: in this scenario, China's economy will develop faster. It will achieve an average annual growth rate of 5% from 2021-2035 and an average annual growth rate of 3.6% from 2036-2050. Some institutions and scholars are optimistic about China's development, as shown in Table 9.
(2) Scenario of slower economic development: in the context of slower economic development, China's economic development is slightly lower than the baseline scenario. e average annual growth rate of China's economic development from 2021-2035 will drop to 4.6% and the average annual growth rate will drop to 3.2% from 2036-2050.
According to the previous assumptions for the two scenarios, the logarithmic function is used for prediction, and the calculation results and the comparison with the baseline scenario are shown in Table 10.
Judging from the calculation results of the three scenarios, the difference in economic growth rate has little effect on the forecast results of freight traffic. e degree of economic growth rate affecting freight volume represents the elasticity of GDP to freight traffic. e freight elasticity of the United States has gradually decreased since 1990 and is currently about 0.1-0.2. Germany's freight elasticity has gradually decreased since 1990 and is currently around 0.2-0.3. Due to the difficulty of economic recovery in Japan, freight elasticity has been in a negative state since 1990. Judging from the experience of the United States, Germany, and other developed countries, its freight elasticity is between 0.1-0.3, which shows that the impact of economic aggregate growth on freight volume is gradually decreasing.

Analysis of the Impact of Industrial Structure Changes on
Freight Traffic. In the late industrialization and postindustrialization period, the decline in freight volume growth was mainly caused by the upgrading of industrial structure. In this stage of industrialization, the proportion of the service industry will increase from 50% to about 70%-80%. e entire national economy is dominated by the service economy and information economy, so the growth rate of freight volume is limited.
To sort out the upgrading of China's industrial structure, its evolutionary structure over the years and the structure forecast for 2035 and 2050 are shown in Figure 4.
From the perspective of future changes in the total output structure of the industry, the primary industry has the largest decline, from 4.88% in 2017 to 2.2% in 2050, and the secondary industry has all declined to vary degrees.
Among them, the coal industry, petroleum industry, and metal mining accounted for the largest decline, with 77.9%, 45.14%, and 75.74%, respectively. e proportion of the accommodation and catering industry in the tertiary industry dropped slightly by 0.1 percentage point, and the rest increased to varying degrees. Among them, the financial industry rose the most, from 4.18% in 2017 to 8.01% in 2050.
By 2050, China's three industrial structures based on total output will account for 2.2 : 47.66 : 50.14, respectively. e structure evolution of the three industries is shown in Figure 5.
To compare the impact of industrial structure on transportation, consider the change in freight volume when only changing the industrial structure with the total economic aggregate, and the complete consumption coefficient is unchanged. Assuming that the industrial structure of 2017 is maintained in 2035 and 2050 and that the demand for freight and passenger transportation from agriculture and industry is 8 : 2 and the demand for freight and passenger   Mathematical Problems in Engineering 13 transportation from the service industry is 2 : 8, then the forecast value of freight volume in 2035 and 2050 is shown in Table 11. When the industrial structure of 2017 is maintained, the predicted value of freight volume in 2035 and 2050 is 9.49% and 16.22% higher than the baseline value, respectively. is shows that the upgrading of the industrial structure has slowed down the growth rate of freight traffic. In the forecast of freight traffic, the importance of industrial structure exceeds that of economic aggregate.

Analysis of the Impact of Changes in the Complete Consumption Coefficient on Freight Traffic.
e impact of technological progress on transportation is mainly reflected in three aspects: industrial upgrading has led to a decline in the proportion of traditional industrial added value such as coal, petroleum, and steel and an increase in the proportion of high-end industries such as computers, precision instruments, and equipment manufacturing. e improvement of technology makes the industry's consumption rate of coal, oil, natural gas, and other energy sources gradually drop; the transportation capacity and service level are improved. Reflected in the input-output table, it is mainly reflected in the change of the complete consumption coefficient. e complete consumption factor is affected by two factors. One is the direct consumption coefficient, which represents the relationship between the demand of the industry and another industry and reflects the degree of energy consumption; the second is the indirect consumption coefficient, which represents the direct and indirect demand relationship between the industry and other industries and reflects the depth and breadth of the industrial chain. Figure 6 shows the direct consumption of transportation by various industries.
From the perspective of the direct consumption coefficient, the largest direct consumption of transportation is the transportation, postal, and nonmetallic mining and processing industries. Because the direct consumption coefficient cannot reflect the industry's complete demand for transportation, it is necessary to calculate the complete consumption coefficient. e complete consumption coefficient is shown in Figure 7.  4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29  From the perspective of the full consumption coefficient, the postal, transportation, and construction industries have the greatest demand for transportation. From the perspective of the changes in the complete consumption coefficient and the direct consumption coefficient, except for petroleum, coking products, and nuclear fuel processing industries, the coefficient of variation of the complete consumption coefficient of other industries exceeds that of the direct consumption coefficient.
is indicates that the industry chain will further deepen from now on, which also leads to a further increase in the complete consumption coefficient.
Assuming that the full consumption coefficient of 2017 is maintained in 2035 and 2050, the forecast value of freight volume in 2035 and 2050 is shown in Table 12.

Conclusion
is paper constructs a freight value and freight volume analysis model based on the input-output method and predicts the development trend of China's freight volume and uses a complete decomposition model to analyze the factors affecting freight value and freight volume. e study reached the following conclusions: (1) e growth rate of China's freight volume will gradually decline, with an average annual growth rate of 1.65% from 2017 to 2035 and an average annual growth rate of 0.79% from 2035 to 2050. Research shows that China freight volume in 2035 is 64.458 billion tons and 72.518 billion tons in 2050. (2) From the perspective of industrial structure, the tertiary industry has the greatest impact on the incremental value of transportation, accounting for 49.28%, the secondary industry also has a greater impact, accounting for 48.77%, and the primary industry has the least impact, accounting for 1.95%. From the perspective of influencing factors, the increase in total output has the greatest impact on the changes in the economic value of transportation, accounting for 86.55%, the second is the change in the complete consumption coefficient, accounting for 13.9%, and changes in the industrial structure make transportation economic value decreased by 0.45%. (3) When the industrial structure of 2017 is maintained, the predicted value of freight volume in 2035 and 2050 is 9.49% and 16.22% higher than the baseline value, respectively. is shows that the upgrading of the industrial structure has slowed down the growth rate of freight traffic. When the complete consumption coefficient remains unchanged, the predicted freight volume in 2035 and 2050 will decrease by 40.98% and 18.7%, respectively. is shows that the change in the complete consumption coefficient is the main reason for the change in freight traffic. e change in economic aggregate has a limited impact on freight traffic, and the upgrading of industrial structure will cause a decline in freight traffic. e increase in the full consumption coefficient indicates that the integration between industries is improving, which means that the accuracy of freight forecasting through industry division will decrease.
As this paper is compared with the previous models, the main differences of this paper are as follows: (1) this paper further subdivides the industry, dividing the industry sector into 42 sectors, which is more refined than the previous three industry divisions and can better reflect the relationship between different industries and freight demand; (2) in the past, factor analysis often used methods such as multiple regression and cluster analysis, which could not reflect the importance and relevance of factors.
is paper uses a complete decomposition model to analyze the influencing factors of freight volume, which can effectively analyze the relationship between the influencing factors; (3) this paper combines gray forecasting and input-output analysis, which can analyze the factors of future freight demand changes, which is more practical and referential than the previous analysis of the current status of influencing factors.
Due to the lack of freight data of related industries, this paper predicts the freight volume from the overall freight value and industry GDP, and the accuracy of the forecast needs to be tested. Future research can further analyze the influence relationship between different industries based on this model, use feature analysis to further optimize the industrial structure, and use this model in conjunction with the OD tables for transportation planning research. Further statistics on the freight volume data of related industries and the use of the fusion of multiple models to make predictions can improve the accuracy of freight volume forecasts, which is the direction of future research.

Data Availability
e data used to support the findings of this study are included in the paper.

Conflicts of Interest
e authors declare that they have no conflicts of interest.  16 Mathematical Problems in Engineering