The Impact of Slumping Oil Price on the Situation of Tanker Shipping along the Maritime Silk Road

: Nearly 70% of the world’s maritime crude oil transportation relies on the Maritime Silk Road (MSR). In order to deeply explore the impact of slumping oil price on the shipping situation of tanker along the MSR, this paper establishes the relationship between monthly ship and oil price through Autoregressive Distributed Lag model. Distributions of cargo ﬂow before and after the oil price slumped are compared to explore the changing law of tanker shipping situation. The study ﬁnds: (1) The correlation between the cargo ﬂow situation of the tanker seaborne export and oil price, where the export cargo ﬂow correlation is stronger than that of the import cargo ﬂow. (2) The MSR tanker shipping situation is lagging (3 months) behind the impact of oil price. The lag e ﬀ ect in Europe, North Asia and East Asia is strong while that in Southeast Asia and South Asia is weak. (3) After the oil price slumped, the tanker shipping cargo ﬂow increased less during the crude oil export stage, and the increase in the crude oil shipping trade after the transfer period was larger. The research results can provide a scientiﬁc basis for improving the decision-making ability of the crude oil shipping market and formulating maritime operations management measures. to in Speciﬁcally, Conceptualization,


Introduction
The 21st Century Maritime Silk Road (MSR), originated from the ancient Maritime Silk Road, is an important channel to promote the smooth trade between Asia, Europe and Africa. It is an emerging trade route connecting China to the world under the changing situations of global politics and trade patterns. At present, the coverage of MSR continues to expand, the maritime trade along its route accounts for more than 35% of the global merchandise trade about 70% of the world's maritime crude oil transport [1].
Petroleum energy is an important fossil energy, and its unbalanced distribution and supporting role in economic development determines its importance in the world economy and trade [2]. As an important factor affecting global economic activities, oil price has an obvious impact on the maritime trade, especially on the maritime crude oil trade. Oil price fluctuations directly stimulate oil demand [3], and then affect the change of tanker shipping situation. In particular, the global oil consumption powers, such as the United States [4], China [5], Japan [6], South Korea [7], etc., have more significant changes in strategic oil reserves and consumption. At the same time, shipping power relies heavily on fossil fuel, making freight rates and transport costs more vulnerable to oil price shocks [8,9]. For the shipping industry, fuel costs account for a large proportion of ship operating costs. Therefore, it is of practical The oil price shock in the second half of 2014 was the most significant macroeconomic shock in recent years. To explore the changes of crude oil shipping trade before and after the oil price slumped, we extract the records of arrival and departure of ships based on the AIS data of tankers from 1 January 2014 to 31 March 2015. By combining with the global port index data published by National Geospatial-Intelligence Agency and taking each arrival and departure of each ship as an origindestination (OD) data, we get 427,267 OD data of tankers. As Figure 2 shows, OD data of tankers intuitively reflect the distribution of crude oil trade between ports along the MSR. Additionally, the maritime trade in East Asia, the Strait of Malacca, the Persian Gulf and the Mediterranean Sea are relatively intensive. Based on OD data of tankers, taking the arrival and departures of a port as the ship frequency in the port, and then superimposing the ship frequency in ports belonging to the same region, the monthly number of tankers in each region and the whole along MSR is finally obtained, which can be used to quantitatively express the frequency change of tanker shipping situation in the study region. The monthly oil price data is derived from the U.S. Energy Information Administration (EIA). The monthly oil price and the MSR overall tanker change during the study period are shown in Figure 3.  The oil price shock in the second half of 2014 was the most significant macroeconomic shock in recent years. To explore the changes of crude oil shipping trade before and after the oil price slumped, we extract the records of arrival and departure of ships based on the AIS data of tankers from 1 January 2014 to 31 March 2015. By combining with the global port index data published by National Geospatial-Intelligence Agency and taking each arrival and departure of each ship as an origin-destination (OD) data, we get 427,267 OD data of tankers. As Figure 2 shows, OD data of tankers intuitively reflect the distribution of crude oil trade between ports along the MSR. Additionally, the maritime trade in East Asia, the Strait of Malacca, the Persian Gulf and the Mediterranean Sea are relatively intensive. Based on OD data of tankers, taking the arrival and departures of a port as the ship frequency in the port, and then superimposing the ship frequency in ports belonging to the same region, the monthly number of tankers in each region and the whole along MSR is finally obtained, which can be used to quantitatively express the frequency change of tanker shipping situation in the study region. The monthly oil price data is derived from the U.S. Energy Information Administration (EIA). The monthly oil price and the MSR overall tanker change during the study period are shown in Figure 3. The oil price shock in the second half of 2014 was the most significant macroeconomic shock in recent years. To explore the changes of crude oil shipping trade before and after the oil price slumped, we extract the records of arrival and departure of ships based on the AIS data of tankers from 1 January 2014 to 31 March 2015. By combining with the global port index data published by National Geospatial-Intelligence Agency and taking each arrival and departure of each ship as an origindestination (OD) data, we get 427,267 OD data of tankers. As Figure 2 shows, OD data of tankers intuitively reflect the distribution of crude oil trade between ports along the MSR. Additionally, the maritime trade in East Asia, the Strait of Malacca, the Persian Gulf and the Mediterranean Sea are relatively intensive. Based on OD data of tankers, taking the arrival and departures of a port as the ship frequency in the port, and then superimposing the ship frequency in ports belonging to the same region, the monthly number of tankers in each region and the whole along MSR is finally obtained, which can be used to quantitatively express the frequency change of tanker shipping situation in the study region. The monthly oil price data is derived from the U.S. Energy Information Administration (EIA). The monthly oil price and the MSR overall tanker change during the study period are shown in Figure 3.

Spearman Rank Correlation Analysis
Spearman rank correlation analysis is a nonparametric statistical correlation analysis method, which uses Spearman rank correlation coefficient s r to measure the rank correlation strength between variables. It is often used to measure the non-linear monotonic relationship [44]. In order to analyze the correlation between ship frequency and oil price in different regions, this paper uses Spearman rank correlation to carry out the analysis test after judging that two variables have a nonlinear relationship (which does not satisfy the hypothetical condition of Pearson), and also due to the small sample size. In practical calculation, the difference of rank is used to calculate the value of s r .
Assuming that the original data i x and i y have been arranged in descending order and that '  (1) In the case of the same rank before and after ordering, the linear correlation coefficients of Pearson between ranks should be calculated according to Equation (2), where x and y is the mean of the sample variables.

ARDL Lag Model
Autoregression is a regression of a variable t Q with its own lag term, referred to as the Autoregressive Model. When a variable is estimated by the AR model and is also affected by the current and lag values of other variables t P , the model is the autoregressive distribution lag (ARDL)

Spearman Rank Correlation Analysis
Spearman rank correlation analysis is a nonparametric statistical correlation analysis method, which uses Spearman rank correlation coefficient r s to measure the rank correlation strength between variables. It is often used to measure the non-linear monotonic relationship [44]. In order to analyze the correlation between ship frequency and oil price in different regions, this paper uses Spearman rank correlation to carry out the analysis test after judging that two variables have a non-linear relationship (which does not satisfy the hypothetical condition of Pearson), and also due to the small sample size. In practical calculation, the difference of rank is used to calculate the value of r s . Assuming that the original data x i and y i have been arranged in descending order and that x i and y i are the positions of the original data x i and y i after the arrangement, x i and y i are called the ranks of the variables x i and y i , and d i = x i − y i is the difference of ranks of x i and y i . In the case of different ranks before and after sorting, r s can be calculated by the Equation (1): In the case of the same rank before and after ordering, the linear correlation coefficients of Pearson between ranks should be calculated according to Equation (2), where x and y is the mean of the sample variables.

ARDL Lag Model
Autoregression is a regression of a variable Q t with its own lag term, referred to as the Autoregressive Model. When a variable is estimated by the AR model and is also affected by the current and lag values of other variables P t , the model is the autoregressive distribution lag (ARDL) model for the event analysis of hysteresis effects [45][46][47]. This paper uses a traditional linear ARDL model constructed by Pesaran et al., which is suitable for the small sample data in this paper [48]. In order to reduce the volatility of the original data and the difference between the statistical data and avoid heteroscedasticity, we have carried out logarithmic transformations of the oil price and the ship Ordinary Least Squares (OLS) Euler algorithm is used to estimate the parameters of the model. At the same time, the fitting effect and lag length k of the model are determined according to Akaike Information Criterion (AIC) and Schwarz Criterion (SC). Where Q t is the statistical value of the t-month tanker ship, ln Q t is the natural logarithm of the ship frequency variable of t-month, that is, the dependent variable; ln Q t−i and ln P t−i are the ship frequency variables of the i-order lag, and the natural logarithm of the oil price variable of the i-order lag, that are, the explanatory variables; α i and β i are the lag model coefficients of ship frequency and the oil price, respectively; k is the maximum lag order of the model; c is a constant term; ε t is a random term.

Nonlinear Correlation
As is shown in Figure 4, based on the monthly variations of the overall tanker flow along the MSR and the distribution of oil price during the same period, and (a), (b) and (c) correspond to the import and export cargo flows, import cargo flows and export cargo flows, respectively. It can be seen from Figure 3 that the variables of ship frequency and oil price are not continuous. We can see from the Figure 4 that the relationship between oil price and ship frequency is non-linear and does not satisfy the hypothetical condition of Pearson. For the above reasons, in this paper, the Spearman rank correlation analysis method is suitable to be used to calculate the correlation between ship frequency and the oil price. model for the event analysis of hysteresis effects [45][46][47]. This paper uses a traditional linear ARDL model constructed by Pesaran et al., which is suitable for the small sample data in this paper [48]. In order to reduce the volatility of the original data and the difference between the statistical data and avoid heteroscedasticity, we have carried out logarithmic transformations of the oil price and the ship frequency variables, which do not affect the cointegration relationship of the data fitting. The equation expression for the ARDL model of oil price and ship frequency is: Ordinary Least Squares (OLS) Euler algorithm is used to estimate the parameters of the model. At the same time, the fitting effect and lag length k of the model are determined according to Akaike Information Criterion (AIC) and Schwarz Criterion (SC). Where t Q is the statistical value of the t -month tanker ship, t Q ln is the natural logarithm of the ship frequency variable of t -month, that is, the dependent variable; k is the maximum lag order of the model; c is a constant term; t ε is a random term.

Nonlinear Correlation
As is shown in Figure 4, based on the monthly variations of the overall tanker flow along the MSR and the distribution of oil price during the same period, and (a), (b) and (c) correspond to the import and export cargo flows, import cargo flows and export cargo flows, respectively. It can be seen from Figure 3 that the variables of ship frequency and oil price are not continuous. We can see from the Figure 4 that the relationship between oil price and ship frequency is non-linear and does not satisfy the hypothetical condition of Pearson. For the above reasons, in this paper, the Spearman rank correlation analysis method is suitable to be used to calculate the correlation between ship frequency and the oil price.
(a) Import and export flow (b) Import flow (c) Export flow The results of the Spearman rank correlation analysis are shown in Table 1. There is a significant negative correlation between the total import and export volume of the tanker, the import cargo flow and the export cargo flow along the MSR, that is, when the oil price slumped, the frequency of tanker ships increases, and the shipping situation rose accordingly. At the same time, the negative correlation between oil price and export cargo flow situation is stronger in magnitude than that of import and export, and the correlation between oil price and import cargo flow situation is slightly weaker than that of import and export. The results of the Spearman rank correlation analysis are shown in Table 1. There is a significant negative correlation between the total import and export volume of the tanker, the import cargo flow and the export cargo flow along the MSR, that is, when the oil price slumped, the frequency of tanker ships increases, and the shipping situation rose accordingly. At the same time, the negative correlation between oil price and export cargo flow situation is stronger in magnitude than that of import and export, and the correlation between oil price and import cargo flow situation is slightly weaker than that of import and export.
In the areas along the MSR, the absolute value of the correlation coefficient between ship frequency and oil price in most regions is greater than 0.5, which indicates that the oil price slump has a significant impact on the shipping situation along the MSR. As is shown in Figure 5, the degree of correlation shows a "weak-strong-weak" distribution from the north to the south, and a "weak-medium-strong" class distribution from the west to the east. Among the eight regions, the important transshipment area for maritime crude oil trade, ship frequency in Southeast Asia has the strongest negative correlation with oil price, with a correlation coefficient of −0.871; ship frequency in North Africa has the weakest negative correlation with oil price, with a correlation coefficient of −0.432. The correlation of the Asian region (East Asia, Southeast Asia, South Asia) is greater, with the correlation of Europe and North Asia, sub-Saharan Africa are second. As well, West Asia, North Africa and Oceania have a weaker correlation.  In the areas along the MSR, the absolute value of the correlation coefficient between ship frequency and oil price in most regions is greater than 0.5, which indicates that the oil price slump has a significant impact on the shipping situation along the MSR. As is shown in Figure 5, the degree of correlation shows a "weak-strong-weak" distribution from the north to the south, and a "weakmedium-strong" class distribution from the west to the east. Among the eight regions, the important transshipment area for maritime crude oil trade, ship frequency in Southeast Asia has the strongest negative correlation with oil price, with a correlation coefficient of −0.871; ship frequency in North Africa has the weakest negative correlation with oil price, with a correlation coefficient of −0.432. The correlation of the Asian region (East Asia, Southeast Asia, South Asia) is greater, with the correlation of Europe and North Asia, sub-Saharan Africa are second. As well, West Asia, North Africa and Oceania have a weaker correlation. As far as the import and export cargo flow of oil tankers are concerned, the negative correlation between ship frequency change of export cargo flow and oil price along the MSR is generally stronger than that of import cargo flow. As is shown in Figure 6, the regions with a strong negative correlation of tanker import cargo flow are mainly concentrated in Southeast Asia, East Asia and sub-Saharan Africa; the regions with a strong negative correlation of oil tanker export cargo flow mainly include Southeast Asia, East Asia, Europe and North Asia. Besides, countries with large demand for petroleum resources are mainly distributed in East Asia and Europe. The frequency of maritime crude oil trade in the ports of these countries is higher. In addition to trade with crude oil export ports and transit ports, crude oil transport between countries and within countries is more frequent. Ports in Southeast Asia are important hubs linking East Asia with West Asia, Europe and Africa. In addition to meeting their own demand for crude oil imports, they also undertake the transshipment function of the current maritime crude oil trade. Therefore, the slumping oil price has a more obvious impact on the tanker shipping situation in these areas. Comparing the correlation of import and export cargo flow, we can see that the degree of correlation for both import and export in most regions is consistent: for example, the correlation between tanker number and oil price in Southeast Asia is the strongest, the correlation coefficients are −0.846 and −0.896, respectively; the correlation coefficients in Oceania are relatively weaker, and the correlation coefficients are −0.664 and −0.611, As far as the import and export cargo flow of oil tankers are concerned, the negative correlation between ship frequency change of export cargo flow and oil price along the MSR is generally stronger than that of import cargo flow. As is shown in Figure 6, the regions with a strong negative correlation of tanker import cargo flow are mainly concentrated in Southeast Asia, East Asia and sub-Saharan Africa; the regions with a strong negative correlation of oil tanker export cargo flow mainly include Southeast Asia, East Asia, Europe and North Asia. Besides, countries with large demand for petroleum resources are mainly distributed in East Asia and Europe. The frequency of maritime crude oil trade in the ports of these countries is higher. In addition to trade with crude oil export ports and transit ports, crude oil transport between countries and within countries is more frequent. Ports in Southeast Asia are important hubs linking East Asia with West Asia, Europe and Africa. In addition to meeting their own demand for crude oil imports, they also undertake the transshipment function of the current maritime crude oil trade. Therefore, the slumping oil price has a more obvious impact on the tanker shipping situation in these areas. Comparing the correlation of import and export cargo flow, we can see that the degree of correlation for both import and export in most regions is consistent: for example, the correlation between tanker number and oil price in Southeast Asia is the strongest, the correlation coefficients are −0.846 and −0.896, respectively; the correlation coefficients in Oceania are relatively weaker, and the correlation coefficients are −0.664 and −0.611, respectively. In addition, we find that the correlation of import and export cargo flow in some regions is quite different: for example, the correlation of export flows in West Asia is weaker than that of import flows, with correlation coefficients of −0.432 and −0.707, respectively; the correlation of export flows in North Africa is significantly greater than that of import flows, with correlation coefficients of −0.603 and −0.357, respectively. From the results, it can be found that the oil tanker trade in the crude oil export areas is mostly large-scale tankers. With the influence of economic policy and other factors, there is no large growth rate of the shipping situation of ports in these areas after the oil price slumped. At the same time, the situation of crude oil trade in the areas with more transit ports and import ports has a closer relationship with the change of oil price. respectively. In addition, we find that the correlation of import and export cargo flow in some regions is quite different: for example, the correlation of export flows in West Asia is weaker than that of import flows, with correlation coefficients of −0.432 and −0.707, respectively; the correlation of export flows in North Africa is significantly greater than that of import flows, with correlation coefficients of −0.603 and −0.357, respectively. From the results, it can be found that the oil tanker trade in the crude oil export areas is mostly large-scale tankers. With the influence of economic policy and other factors, there is no large growth rate of the shipping situation of ports in these areas after the oil price slumped. At the same time, the situation of crude oil trade in the areas with more transit ports and import ports has a closer relationship with the change of oil price.
(a) Import flow (b) Export flow Figure 6. Correlation of tanker import and export.

Lag Analysis
From Figure 3, it can be seen that the change of oil tanker situation has obvious time lag compared with oil price. In order to further explore the relationship between the MSR's overall tanker shipping situation and oil price, we adopt the ARDL model, and use months as the time scale to perform the model fitting calculation with the first-order, second-order, and third-order lag periods, respectively. The results are shown in    Table 2 shows the corresponding results of the tanker fitting index calculations using different lag periods. According to the criteria of AIC and SC, the optimal delay period of the model is first determined as the third-order lag. The explanatory variables include the first-order, second-order, third-order lag term of ship frequency, as well as the current term of oil price, the first-order, secondorder and third-order lag term of oil price. Then the variables with less reliability of fitting coefficients are eliminated by the t-test, that is, the third-order lag term of oil price

Lag Analysis
From Figure 3, it can be seen that the change of oil tanker situation has obvious time lag compared with oil price. In order to further explore the relationship between the MSR's overall tanker shipping situation and oil price, we adopt the ARDL model, and use months as the time scale to perform the model fitting calculation with the first-order, second-order, and third-order lag periods, respectively. The results are shown in  respectively. In addition, we find that the correlation of import and export cargo flow in some regions is quite different: for example, the correlation of export flows in West Asia is weaker than that of import flows, with correlation coefficients of −0.432 and −0.707, respectively; the correlation of export flows in North Africa is significantly greater than that of import flows, with correlation coefficients of −0.603 and −0.357, respectively. From the results, it can be found that the oil tanker trade in the crude oil export areas is mostly large-scale tankers. With the influence of economic policy and other factors, there is no large growth rate of the shipping situation of ports in these areas after the oil price slumped. At the same time, the situation of crude oil trade in the areas with more transit ports and import ports has a closer relationship with the change of oil price.
(a) Import flow (b) Export flow Figure 6. Correlation of tanker import and export.

Lag Analysis
From Figure 3, it can be seen that the change of oil tanker situation has obvious time lag compared with oil price. In order to further explore the relationship between the MSR's overall tanker shipping situation and oil price, we adopt the ARDL model, and use months as the time scale to perform the model fitting calculation with the first-order, second-order, and third-order lag periods, respectively. The results are shown in    Table 2 shows the corresponding results of the tanker fitting index calculations using different lag periods. According to the criteria of AIC and SC, the optimal delay period of the model is first determined as the third-order lag. The explanatory variables include the first-order, second-order, third-order lag term of ship frequency, as well as the current term of oil price, the first-order, secondorder and third-order lag term of oil price. Then the variables with less reliability of fitting coefficients are eliminated by the t-test, that is, the third-order lag term of oil price   Table 2 shows the corresponding results of the tanker fitting index calculations using different lag periods. According to the criteria of AIC and SC, the optimal delay period of the model is first determined as the third-order lag. The explanatory variables include the first-order, second-order, third-order lag term of ship frequency, as well as the current term of oil price, the first-order, second-order and third-order lag term of oil price. Then the variables with less reliability of fitting coefficients are eliminated by the t-test, that is, the third-order lag term of oil price ln P t−3 and the first-order lag item ln Q t−1 . The adjusted tanker ARDL model makes all the explanatory variables highly significant. It can be seen from Table 2 that the adjusted model goodness-of-fit, Adjusted-R 2 , is 0.871; the fitting effect of tanker has not decreased. Finally, the expression of the ARDL lag relation of the whole MSR tanker is as follows: ln Q t = 39.815 − 2.022 * ln P t + 0.908 * ln P t−1 − 0.778 * ln P t−2 −1.232 * ln Q t−2 − 0.636 * ln Q t−2 (4) From the fitting results of ARDL model of oil tanker, it can be seen that the months with large residual are concentrated in June-October. Oil price began to fall in June 2014, and tankers gathered to rise in October, reaching a sudden peak after a sharp fall in oil price. Then, the impact of the continuous decline in oil price on the shipping situation gradually diminished over time. Oil price have rebounded since December, and tanker shipments have gradually stabilized. The fitting results of ARDL model of tanker verify that the oil price slump has a significant time lag on the impact of oil tanker shipping situation, and we found that the optimal lag period for the change of tanker situation affected by the slumping oil price is three months. At the same time, the short-term impact of oil price has a certain timeliness.

Lag Affects Regional Differences
According to the ARDL model fitting results of the whole tanker along the MSR, we bring the tanker frequencies in each study area into Equation (4) and calculate the residual of the monthly fitting value and statistical value of the model. The average (AVG) and mean square error (MSE) results of the residual values in each area are shown in Figure 8. The horizontal axis represents the areas along MSR: EA (East Asia), E&NA (Europe and North Asia), ESA (Southeast Asia), WA (West Asia), SSA (Sub-Saharan Africa), OA (Oceania), NA (North Africa), and SA (South Asia). The residual distribution along MSR is staggered, and the absolute values of AVG in North Africa and South Asia is greater than 0.5. The absolute values of AVG in other regions is less than 0.2. From the AVG, there is a big gap between the change of tanker shipping situation in each region and that of MSR as a whole. Among them, the AVG of East Asia, Southeast Asia, South Asia, and Oceania is above 0. Compared with the overall change of oil tanker shipping situation along the MSR, the increase of oil tanker shipping situation in these areas is higher than the overall average level. The AVG of Europe and North Asia, West Asia, sub-Saharan Africa, and North Africa is less than 0, and the increase of oil tanker shipping situation is lower than the overall average level along the MSR. In addition, the maritime shipping situation in North Africa and South Asia varies more significantly. In terms of the MSE, the distribution of residual in Southeast Asia, West Asia, North Africa, and South Asia fluctuates greatly, and the influence of oil price lag on the tanker shipping situation is relatively weak. The error distribution in Europe and North Asia, East Asia, Sub-Saharan Africa, and Oceania fluctuates slightly, and the influence of oil price lag on the tanker shipping situation is relatively strong. As Figure 9 shows, from the point of view of spatial distribution, the influence of oil price lag on both sides of the north and south is more obvious. The surrounding areas of the Malacca Strait, the North Indian Ocean, the Hormuz Strait, the Suez Canal, the Persian Gulf, and other important channels have a weaker lag impact on their shipping situation due to the slumping oil price. In regions with more oil importing countries, such as Europe, East Asia, and East Africa, the lag of oil price has a stronger impact on the tanker shipping situation. regions with more oil importing countries, such as Europe, East Asia, and East Africa, the lag of oil price has a stronger impact on the tanker shipping situation.

Comparison of Cargo Flow before and after Oil Price Slumped
In order to further understand the impact of the oil price slump on the MSR tanker shipping situation, we selected the time intervals (June and October) before and after the oil price slumped based on the optimal lag period we found and obtained the structure map of the MSR crude oil transport flow, as shown in Figure 10. Nodes represent ports, and the size of nodes represents the frequency of trade within a month; the connection lines between nodes represent the trade between two ports, and the line width represents the frequency of trade on that route. The outermost label represents the port index number of the corresponding port in the WORLD PORT INDEX, for example, ZHOUSHAN port (59960). It can be found at the NGA Maritime Domain website (https://msi.nga.mil/NGAPortal/MSI.portal).
This paper chooses three typical regions from the perspective of crude oil import, export and transit. First, as the gathering place of global oil consumption and import, the ports in East Asia mainly connect the crude oil exporting countries and the transit hub ports of Malacca Strait, such as ZHOUSHAN port in China, CHIBA KO port in Japan, and ULSAN port in Korea. As Figure 10a,b shows, tanker cargo flows in East Asia are mainly concentrated within the region: between ports along Japan, South Korea and East China. In October after the oil price slumped, the frequency of port trade in these areas reached its peak, the frequency of maritime trade in CHIBA KO port increased from 1506 to 2405, and the frequency of ZHOUSHAN port increased from 876 to 106. Even the trade frequency of ports such as KAWASAKI KO port and QINGDAO port increased exponentially. At the same time, the hub ports of these crude oil importing countries have increased significantly with the maritime trade of JURONG port and KEPPEL port in Singapore. Of the crude regions with more oil importing countries, such as Europe, East Asia, and East Africa, the lag of oil price has a stronger impact on the tanker shipping situation.

Comparison of Cargo Flow before and after Oil Price Slumped
In order to further understand the impact of the oil price slump on the MSR tanker shipping situation, we selected the time intervals (June and October) before and after the oil price slumped based on the optimal lag period we found and obtained the structure map of the MSR crude oil transport flow, as shown in Figure 10. Nodes represent ports, and the size of nodes represents the frequency of trade within a month; the connection lines between nodes represent the trade between two ports, and the line width represents the frequency of trade on that route. The outermost label represents the port index number of the corresponding port in the WORLD PORT INDEX, for example, ZHOUSHAN port (59960). It can be found at the NGA Maritime Domain website (https://msi.nga.mil/NGAPortal/MSI.portal).
This paper chooses three typical regions from the perspective of crude oil import, export and transit. First, as the gathering place of global oil consumption and import, the ports in East Asia mainly connect the crude oil exporting countries and the transit hub ports of Malacca Strait, such as ZHOUSHAN port in China, CHIBA KO port in Japan, and ULSAN port in Korea. As Figure 10a,b shows, tanker cargo flows in East Asia are mainly concentrated within the region: between ports along Japan, South Korea and East China. In October after the oil price slumped, the frequency of port trade in these areas reached its peak, the frequency of maritime trade in CHIBA KO port increased from 1506 to 2405, and the frequency of ZHOUSHAN port increased from 876 to 106. Even the trade frequency of ports such as KAWASAKI KO port and QINGDAO port increased exponentially. At the same time, the hub ports of these crude oil importing countries have increased significantly with the maritime trade of JURONG port and KEPPEL port in Singapore. Of the crude

Comparison of Cargo Flow before and after Oil Price Slumped
In order to further understand the impact of the oil price slump on the MSR tanker shipping situation, we selected the time intervals (June and October) before and after the oil price slumped based on the optimal lag period we found and obtained the structure map of the MSR crude oil transport flow, as shown in Figure 10. Nodes represent ports, and the size of nodes represents the frequency of trade within a month; the connection lines between nodes represent the trade between two ports, and the line width represents the frequency of trade on that route. The outermost label represents the port index number of the corresponding port in the WORLD PORT INDEX, for example, ZHOUSHAN port (59960). It can be found at the NGA Maritime Domain website (https://msi.nga.mil/NGAPortal/MSI.portal).
This paper chooses three typical regions from the perspective of crude oil import, export and transit. First, as the gathering place of global oil consumption and import, the ports in East Asia mainly connect the crude oil exporting countries and the transit hub ports of Malacca Strait, such as ZHOUSHAN port in China, CHIBA KO port in Japan, and ULSAN port in Korea. As Figure 10a,b shows, tanker cargo flows in East Asia are mainly concentrated within the region: between ports along Japan, South Korea and East China. In October after the oil price slumped, the frequency of port trade in these areas reached its peak, the frequency of maritime trade in CHIBA KO port increased from 1506 to 2405, and the frequency of ZHOUSHAN port increased from 876 to 106. Even the trade frequency of ports such as KAWASAKI KO port and QINGDAO port increased exponentially. At the same time, the hub ports of these crude oil importing countries have increased significantly with the maritime trade of JURONG port and KEPPEL port in Singapore. Of the crude oil trade routes with ports in East Asia in June, 46 routes have had more than 30 trade frequencies, with the highest being ZHOUSHAN to ZHENHAI port (120 times); in October, after the oil price slumped, routes with frequencies over 30 had risen to 112, with the highest frequency being YOSU port to GWANGYANG port (410 times).
concentrates on the ports of Singapore and Malaysia. The cargo flow of these hub ports is mainly directed at the ports of oil exporters in West Asia and oil importers in East Asia, Europe and Africa. After the oil price slumped, Singapore's port trade frequency increased sharply, JURONG port's trade frequency increased from 3800 to 8689, KEPPEL port's trade frequency increased from 2929 to 8147, and Thailand's MAP TA PHUT port and BANGKOK port's trade frequency also increased exponentially. Among the crude oil trade routes with Southeast Asian ports in June, 31 routes had more than 30 frequencies, with the highest routes being PULAU BUKOM port to JURONG port (594 times), while 39 routes had more than 30 frequencies in October, with the highest still being PULAU BUKOM port to JURONG port (1969 times). Although crude oil transport routes have increased as a whole, the frequency of crude oil transport routes through transshipment hubs is far greater than that of other ports. This phenomenon fully proves that crude oil transport is mainly carried out by transshipment hubs for offshore crude oil trade.

Discussion
Based on the AIS trajectory data, this paper analyzes the impact of oil price on tanker shipping situation in areas along the MSR from the view of space-time: the correlation between oil tanker shipping situation and oil price is significant in time, and the correlation of the MSR maritime export cargo flow situation is higher than that of import cargo flow situation. The slumping oil price directly stimulates changes in crude oil supply and demand, raising the tanker trade in importing, exporting Second, West Asia is the region with the richest oil reserves, with the largest production and the largest export volume in the world, mainly including Saudi Arabia, the United Arab Emirates, Iraq and other oil exporters. From Figure 10c,d, it can be seen that most ports in West Asia trade crude oil directly with the hub ports of oil importing countries such as East Asia and Europe, and transshipment ports such as KEPPEL port and JURONG port in Singapore. Tanker cargo flows are mainly concentrated in the Persian Gulf and around the Mediterranean Sea. After the oil price slumped, port trade in West Asia has increased slightly, but the increase is far from that in East Asia. In June, the trade frequencies of KHAWR FAKKAN port and FUJAYRAH port in the United Arab Emirates were 1630 and 1555, respectively. By October, the trade frequencies of the two ports were still in the top ranking, with a slight increase to 1717 and 1625. In addition, the trade frequency of AMBARLI port and ISTANBUL port increased significantly. Before the crude oil price slumped, 34 crude oil trade routes in West Asia had more than 20 frequencies; the highest route was KHAWR FAKKAN port to FUJAYRAH port (481 times). After the oil price slumped, 31 routes had more than 20 frequencies; the highest route was still KHAWR FAKKAN port to FUJAYRAH port (544 times).
Thirdly, Southeast Asia, especially the ports around the Strait of Malacca, relies on its geographical advantages to become an important regional hub for the transfer and convergence of global maritime trade. From Figure 10e,f, it can be seen that the distribution of cargo flow of maritime trade in Southeast Asia presents an obvious hub-and-spoke structure, and its crude oil trade mainly concentrates on the ports of Singapore and Malaysia. The cargo flow of these hub ports is mainly directed at the ports of oil exporters in West Asia and oil importers in East Asia, Europe and Africa. After the oil price slumped, Singapore's port trade frequency increased sharply, JURONG port's trade frequency increased from 3800 to 8689, KEPPEL port's trade frequency increased from 2929 to 8147, and Thailand's MAP TA PHUT port and BANGKOK port's trade frequency also increased exponentially. Among the crude oil trade routes with Southeast Asian ports in June, 31 routes had more than 30 frequencies, with the highest routes being PULAU BUKOM port to JURONG port (594 times), while 39 routes had more than 30 frequencies in October, with the highest still being PULAU BUKOM port to JURONG port (1969 times). Although crude oil transport routes have increased as a whole, the frequency of crude oil transport routes through transshipment hubs is far greater than that of other ports. This phenomenon fully proves that crude oil transport is mainly carried out by transshipment hubs for offshore crude oil trade.

Discussion
Based on the AIS trajectory data, this paper analyzes the impact of oil price on tanker shipping situation in areas along the MSR from the view of space-time: the correlation between oil tanker shipping situation and oil price is significant in time, and the correlation of the MSR maritime export cargo flow situation is higher than that of import cargo flow situation. The slumping oil price directly stimulates changes in crude oil supply and demand, raising the tanker trade in importing, exporting and transshipment hubs. At the same time, it is found that there is a three-month lag in the response process of tanker shipping situation to the change of oil price. From the perspective of the international environment, economic globalization has become the trend of world economic and trade development. On the one hand, the rapid development of the global economy has increased the demand for international crude oil, and on the other hand has also driven the development of international shipping demand. Due to the special status and role of international shipping companies and ports in economic globalization, adapting to economic transportation needs, being able to respond quickly and effectively in emergencies, and ensuring the sustainable development of shipping has become an important demand for maritime transportation. Moreover, significant changes in crude oil prices have a serious impact on maritime transport and will also affect shipping efficiency. Therefore, a reasonable grasp of the lag of tanker transport can provide a prejudgment for the operation and management of the transshipment hub, so that reasonable arrangements can be made in advance and transportation congestion and lack of berths can be avoided. It can bring the maritime market back to a new equilibrium. In the meanwhile, it can also provide decision-making basis for oil consuming countries to make strategic policies in oil reserves and fuel management plans, thus ensuring the sustainable development of maritime crude oil trade along the MSR [49].
From the research results, we find that in the region with rapid economic development and energy demand expanding year by year, the shipping situation is obviously affected by the lag of oil price. However, that effect is relatively weak in the port areas which occupy the important hubs and are connected with many maritime trade areas. For example, Southeast Asia has superior natural conditions, especially the ports around the Malacca Strait that have significant geographical location and port condition advantages. Taking KEPPEL port of Singapore as an example, as a regional hub port, it plays an important role in the transfer and connection of maritime crude oil trade. On the current pattern of maritime trade structure, the advantages of the special maritime network in Southeast Asia have greatly improved the anti-interference capability of the change of the shipping situation on oil price. At the same time, the port infrastructure and service level have met the supply, warehousing and other needs of passing vessels [50] to make sure the route is clear. West Asia and North Africa also have important shipping hubs (the Strait of Hormuz and Suez Canal), so the shipping situation in these regions is less affected by the lag of oil price. In addition, in recent years, economic activities, production and consumption is increasing day by day and the maritime trade is also increasing year by year in East Asia, Europe, South Africa, East Africa and other regions. Especially in East Asia, Japan, South Korea and other major oil-consuming countries, oil demand is large, so the shipping situation in the region is vulnerable to the impact of the oil price slump, and the impact of the lag is even more obvious. Moreover, ports in South Asia are constrained by their hinterland traffic and infrastructure conditions, and their maritime trade is declining and less affected by such impact.
Tanker transportation is mainly responsible for the global maritime trade in petroleum and refined oil. The maritime crude oil trade depends on the structure of supply and demand, so tanker transportation has a direct and close relationship with crude oil export and consumption. Port conditions in many oil importing countries do not meet the ultra large crude carrier (ULCC), so crude oil transportation requires transit hub ports and then is shipped to import destinations by small vessels [35]. Comparing the changes of oil tanker cargo flow before and after the oil price slumped, we find that the shipping situation of importing countries and transshipment hub ports are concentrated rapidly. For example, the share of tanker transportation in the East Asia region rise from 93.7% to 95.6%, and that in the Southeast Asian region increased from 88.5% to 93.9%. Therefore, transit port and hub port of oil importing country play an important role in maritime crude oil trade. In summary, we put forward the following suggestions: (1) enhancing the bearing capacity of leading transshipment ports in tanker transportation system, ensuring the efficiency and sustainability of port operation when the shipping situation gathers, and reducing the risk of traffic congestion along the MSR. (2) Invest in the construction of potential ports with prominent geographical advantages in importing countries, such as QINGDAO port and DALIAN port in China, and TAKAMATSU port and KOMATSUSHIMA port in Japan. The construction of facilities and conditions of potential ports should be strengthened to meet the berthing needs of large ships, so as to develop potential ports into new import hub ports, share the transportation pressure of existing leading hub, and improve the smoothness of tanker shipping.
(3) For ports in regions with strong delays, it is necessary to monitor the fluctuation of oil price in time and predict the development trend of crude oil trade rationally and formulate a reasonable port response plan to avoid risks, such as cargo accumulation, ship congestion, and ship delay, so as to ensure the smooth operation of ports and the sustainability of crude oil trade.
Compared with previous studies on maritime crude oil trade, we find that there are obvious differences in the impact of oil price in different regions. Before and after the oil price slumped, we analyzed the changes of oil import, export and transshipment areas in the short term based on the port-scale oil tanker cargo flow. Of course, the change of oil price itself is a dynamic process of long-term fluctuations. Our research mainly focuses on the situation of oil price slump, in the short-term, without considering the relationship between the trend of tanker shipping and oil price rebound and the long-term fluctuation after that. In addition, the ARDL model is essentially a linear relationship, so it will be a future research direction to study whether oil price and shipping have a non-linear causality. Fuel, as one of the shipping costs, has different grades of fuel and different types of cargo ships have different impacts on transport costs. Therefore, future research can also combine container ships, dry bulk carriers and different grades of fuel to explore the impact of oil price on the shipping situation.

Conclusions
This paper establishes the lag relationship between ship frequency and oil price by using the ARDL model and compares the differences between different regions along the MSR affected by the oil price slump and changes of tanker cargo flow distribution. The following conclusions are drawn: (1) There is a significant negative correlation between tanker shipping frequency and oil price in the same period, that is to say, the shipping frequency will increase correspondingly when oil price slumps. The correlation between maritime export cargo flow situation and oil price is stronger than that of import cargo flow situation. (2) The impact of slumping oil price on the overall maritime trade along the MSR is lagging, with the lag period of three months. After the oil price slumped, the shipping situation of crude oil import in East Asian rose substantially, and the lag affected by oil price was obvious. However, Southeast Asia, which occupies important transshipment hub ports, is less affected by the lag of oil price. (3) After the oil price slumped, tanker cargo flow increases slightly at the stage of oil export, but most obviously at the stage of post-transit in maritime trade, especially between transit hubs and importing countries, and between ports within importing countries. (4) The sustainable development of crude oil trade along the MSR can be promoted by strengthening the carrying capacity of transshipment hub ports, increasing the investment in the construction of import potential ports (Qingdao port, TAKAMATSU port, etc.), and optimizing the structure system of the MSR maritime crude oil trade network.