The COVID-19 shocks on the stock markets of oil exploration and production enterprises

Abstract

Using daily data from January 1, 2020 to March 31, 2021, this research explores COVID-19 shocks on the stock market of 15 representative oil exploration and production enterprises from 7 countries. We measure the COVID-19 epidemic from two levels, government response stringency index and number of confirmed cases, and employ stock prices and stock market returns to reflect the stock market. Our research results confirm that both the government response stringency index and the number of confirmed cases have a significantly negative influence on stock prices. We further find that the negative reaction of the stock market to the government response stringency index is greater than that from confirmed cases. Finally, we conclude that the government response stringency index have a significantly positive effect on stock market returns of oil exploration and production enterprises. Similar findings arise from analyzing specific enterprises. Overall, our conclusions provide some useful information for the decision-making of oil exploration and production enterprises’ investors and policy makers.

1. Introduction

The severe acute respiratory syndrome coronavirus (COVID-19) epidemic erupted at the end of December 2019 and rapidly spread to more than 200 countries, with the United States, Brazil, India, and Russia having suffered serious epidemics. On March 11, 2020, COVID-19 was designated as a true pandemic by the World Health Organization (WHO) and by August 2, 2020, more than 17.66 million confirmed cases and 680,000 deaths accrued worldwide. In addition to medical measures such as treatment and vaccine development, governments have adopted non-medical policies such as city lockdowns, isolation of patients, school and factory closures, travel cancellations, the prohibition of gatherings, and so on [1]. Although these policies did contribute to prevent the rapid spread of the COVID-19 epidemic, they unfortunately caused tremendous damage to global economic activities and prompted the worst recession since the Great Depression.1 The COVID-19 epidemic at the same time brought great panic and pressure to financial markets [[2], [3], [4]]. As the stock market is a barometer of economic development [5], the panic mood of investors in various countries increased during the pandemic.2 Previous studies have shown some major emergencies have caused extreme turbulence in stock markets, such as wars [6], infectious diseases [[7], [8], [9]], and natural disasters [[10], [11], [12]]. However, due to the spread, speed, high mortality rate, and uncertain control time of the COVID-19 epidemic, its shocks to the stock markets of various industries has been far-reaching [13].

Unlike previous studies' focus of the COVID-19 epidemic's influence on stock markets as a whole, this paper targets its impact on the stock market of oil exploration and production enterprises. Oil, as the most widely used energy in the world, is called “the blood of industry” and is an essential basic material in the process of economic development and modernization of all countries [[14], [15], [16]]. The huge fluctuations in oil prices have had a clear impact on the economy and stocks of the oil sector. The interruption of demand relative to excess supply during the epidemic period made the benefits of oil exploration enterprises uncertain, which affected their stock prices and stock returns.

Under the impact of the COVID-19 pandemic, the measures taken by governments to prevent and control the epidemic, such as city lockdowns, workplace closures, and closing public transport, caused widespread production to stop [[17], [18]]. While rigid expenditures such as rent, staff wages, and financial expenses remain unchanged, downstream oil firms and industries faced the risk of cash flow interruption and reduced demand for oil, which caused an overall deterioration of the oil production and supply chain environment. Furthermore, it affected the cash flow of oil exploration and production enterprises, making their income uncertain and influencing their industry's stock market. Therefore, it is of crucial significance to clarify the COVID-19 epidemic shocks to the stock market of oil exploration and production enterprises through empirical research.

This research thus discusses the impact of the COVID-19 epidemic on the stock markets of oil exploration and production enterprises. On the one hand, according to the results herein, a government can evaluate the severity of an epidemic's impact on the oil exploration and production market and adjust the breadth of its oil reserve policy in time during an epidemic period. On the other hand, the government can form reverse feedback information of its epidemic prevention and control policy and adjust the tightness of the epidemic prevention policy according to the degree of the epidemic's effect on the oil market. This strand of research can also help oil exploration and production enterprises adjust their expectations for the future stability of the oil market according to the dynamic changes of the epidemic situation and fine-tune their short-term and long-term oil development plans to reduce the cost of production risk during the epidemic period.

The COVID-19 shocks on oil exploration and production enterprises' stock markets can theoretically be explained from the following three angles. First, its shocks influenced the supply and demand relationship in the oil market, resulting in turbulent oil prices, which in turn affected the stocks of oil exploration and production enterprises. From the supply perspective, oil exploration, production, and processing all have the characteristics of being labor-intensive. Measures taken by governments in response to the COVID-19 epidemic, such as workplace closures, restricted the normal market workforce and affected the production of many enterprises. At the same time, measures such as the closure of public transport, restrictions on internal movement, and suspensions of international travel negatively influenced the transportation of oil, increasing the cost of oil storage for enterprises and thus affecting oil supply. From the perspective of oil demand, the slowdown in global economic growth has put heavy pressure on the energy industry, especially oil as it is the most representative and frequently traded commodity. Government responses to the COVID-19 epidemic have led to shrinking industries, falling transport usage, and a halt in air traffic. Oil demand even dropped to the lowest level in the last twenty years [19]. Therefore, it is not difficult to infer that the COVID-19 epidemic has affected the supply and demand relationship of oil and caused turbulence therein. Under this situation, most of the world's oil exploration and production enterprises have faced bankruptcy.3 Some oil producers have even applied for bankruptcy protection, thus impacting sector performance even more [20] and leading to a decline in oil exploration and production enterprises' stock prices.

Second, according to the theory of emotion, the impacts from news on the different emotions of investors and financial markets vary widely [12]. Generally speaking, negative news is more likely to cause stock market volatility than positive news [21]. In the case of COVID-19, the media has paid close attention to the development trend of the epidemic and sometimes spread negative news to investors, especially in the oil market, which caused fear and panic in the oil markets due to bearish oil market earnings (Mailal, 2011 [22]; and increased volatility and uncertainty in the oil stock market. Based on the real option theory, when facing uncertainty in the energy market, investors have the right to postpone investment, change the capital distribution of the oil market, and re-organize their asset allocation and portfolio to avoid investment risk [23]. Hence, this ultimately leads to drops in energy stock prices.

Governments eventually adopted a series of containment and closure policies, such as blockade and travel restrictions. Although the implementation of these policies brought short-term adverse economic effects, these measures can reduce the infection rate of the COVID-19 epidemic [24], saving more lives that could have been infected or died by the epidemic, and contribute to the prevention and control of the epidemic (Narayan, Phan and Liu 2020b). In this way, with the further control of the epidemic, investors’ confidence in the future economic recovery and development and the improvement of corporate performance will increase, and stock market returns should subsequently rise [25].

The influence of the COVID-19 epidemic on the social economies and financial markets has also attracted the attention of many scholars [[26], [27]]. Some have discussed the COVID-19 epidemic impact on the stock market [13,[25], [28], [29], [30], [31], [32], [33], [34], [35], [36]], but these studies mainly focus on the global stock market or a country's stock market as a whole, with few studies targeting a single industry. Other scholars have studied changes in oil demand [37,38,39], oil prices (Narayan et al., 2020 [40]; Gil-Alana and Monge; 2020; Sui et al., 2020; [41]), energy enterprises [42] as well as the relationship between oil markets and economic variables (Jeris and Nath, 2020) during the COVID-19 epidemic. In addition, Akrofi and Antwi [19] discuss the measures taken by government energy departments in Africa in response to the COVID-19 epidemic. From the perspective of avoiding investment risks, Gharib et al. [43,44] test and confirm that investment in the oil market is risky during the period of COVID-19 epidemic, while gold can be considered as a reliable safe-haven asset.

Although these studies have analyzed the influence of the COVID-19 epidemic on oil from multiple perspectives, it is still unclear how stock market of oil exploration and production enterprises reacted to the epidemic by considering oil sources, energy, and the stock market together. At the same time, we note that the proxy indicator for measuring the epidemic shocks in most existing literature mainly focus on the number of confirmed cases or deaths, which only reflect the development trend of the epidemic situation intuitively, and there may be underreported confirmed and deaths in countries. By comparison, government responses to the epidemic situation and the prevention and control measures adopted give more connotation to the judgment over the COVID-19 epidemic shocks' severity. In particular, the reduction in oil prices depends on government prevention and control measures, such as restricting travel and closing factories. Therefore, it is necessary to discuss the impact of the epidemic on the stock prices of oil enterprises according to different proxy variables. At the same time, as Elliott et al. [45] mentioned, we find that most of the existing studies on the COVID-19 epidemic’ impact use Difference in Difference (DID), traditional time series, and panel data model. However, these methods tend to ignore endogeneity problems and cross-sectional dependence in the variables, resulting in a decrease in the effectiveness of the researches.

To overcome the shortcomings of the above research, this paper discusses the COVID-19 epidemic shocks on the opening and closing prices of 12 oil exploration and production enterprises from 5 countries in the world from January 1, 2020 to March 31, 2021. To describe the COVID-19 epidemic shocks more comprehensively, we select the number of daily new confirmed cases of COVID-19 per million people (CASE) like in previous studies and also choose the government response stringency index (GRSI) in consideration of the role governments play in controlling the pandemic.4 We also use the stock prices and stock returns of oil exploration and production enterprises to denote the stock market's response to the epidemic.

We first use the cross-section dependence (CSD) test of Pesaran [46] and Pesaran [47] to test whether CSD exists in this paper.5 The CSD test solves the cross-sectional dependence problem in the variables and also deals with the endogeneity problem in the DID method seen in the previous literature. Once it is confirmed that there is CSD in the variable, the Cross-sectionally Dependent Augmented Dickey Fuller (CADF) test proposed by Pesaran [48] is adopted to test the stability of the variable. Considering that the univariate variable time series may undergo structural changes, we next utilize the minimum Lagrange multiplier (LM) unit root test model with structural break proposed by Refs. [49,50] to test the stability of the variable and execute the endogenous multiple structural break test of Bai and Perron [51] to confirm the existence of structural break points. Finally, to clarify the exact COVID-19 shocks on the whole or specific oil exploration and production enterprises' stock market, we employ the newly developed panel data mean group (MG) estimation proposed by Chudik and Pesaran [52]. Overall, whether in the full samples or for specific enterprise, the research results confirm that the COVID-19 epidemic negatively influences oil exploration and production enterprises’ stock prices, and the government response stringency index has a positive impact on the stock returns of oil exploration and production enterprises.

The remaining structure of the article is arranged as follows. Section 2 introduces the research methods. Section 3 explains the data source and statistical description of the data. Section 4 presents the empirical test results and analyzes them. The last section is the conclusion and some policy suggestions.

2. Methodology

The panel model has many advantages, such as a large information dataset, reducing collinearity between variables, making more accurate predictions of individual results, and improving the effectiveness of model estimation [53]. However, previous studies have confirmed that cross-section dependence problems may often exist in panel models. Therefore, in this paper we exploit some new developed methods to explore COVID-19 shocks on oil exploration and production enterprises’ stock prices.

• (1)Cross-sectional dependence

The first-generation panel unit root test assumes that each variable is independent on the cross-section. If the existence of CSD is ignored, then the empirical results will be biased [54]. Based on this, before examining the COVID-19 shocks on oil exploration and production enterprises’ stock prices, this paper first follows Pesaran [46] and Pesaran [47] to check whether there is CSD in the panel dataset.

We define the model as follows:

$$y_{it}={\alpha }_i+{\beta }_{it}{x}_{it}+u_{it}$$ (1)

Here, i = 1, 2, 3 … …N, as N represents the sample size; t = 1, 2 ……T, where T represents the time period; ${\beta }_{it}$ represents the K × 1 parameter vector for the explanatory variable denoted by $x_{it}$; α_i represents the individual constant parameter; and ${\mu }_{it}$ represents the regression residual. The statistic of Pesaran [46] can be calculated as follows:

$$CD=\sqrt{\frac{2T}{N(N-1)}}{\sum }_{i=1}^{N-1}{\sum }_{j=i+1}^N{\rho }_{ij}$$ (2)

Here, ${\rho }_{ij}$ represents the product correlation errors of countries i and j.

Pesaran [47] puts forward the weak cross-sectional dependence test, which can be applied with a small sample and heterogenous slope:

$$\stackrel{\wedge }{{\rho }_{ij}}=\stackrel{\wedge }{{\rho }_{ji}}=\frac{\sum _{\text{t}=1}^T\stackrel{\wedge }{u_{it}}\stackrel{\wedge }{u_{jt}}}{{(\sum _{\text{t}=1}^T{\stackrel{\wedge }{u_{it}}}^2)}^{1/2}{(\sum _{\text{t}=1}^T{\stackrel{\wedge }{u_{jt}}}^2)}^{1/2}}$$ (3)

Here, u represents the estimation residual.

• (2)Panel unit root test

Once the existence of CSD in panel members is confirmed, the traditional panel unit root test proposed by Maddala and Wu (1999) and Levin et al. (2002) is invalid. This paper takes the CADF test, which is newly developed by Pesaran [48]; to judge the variables’ stationarity. This method is suitable for CSD of variables. The model for the unit root test of Pesaran [48] is defined as:

$$△y_{it}={\delta }_i+{\beta }_i{y}_{it-1}+{\phi }_i{\bar{y}}_{it-1}+\sum _{j=1}^{\rho }{\lambda }_{ij}△y_{it-1}+\sum _{j=0}^{\rho }{\nu }_{ij}△{\bar{y}}_{t-j}+d_{it}+{\epsilon }_{it}$$ (4)

Here, ${\bar{y}}_t=\frac{1}{N}{\sum }_{i=1}^N{y}_{it}$, which represents the unobserved common factor; ${\beta }_i$ stands for the coefficient of first lag; ${\phi }_i$, ${\lambda }_{ij}$, and ${\nu }_{ij}$ stand for the individual specific effect, individual linear trend, and common time effect for all individuals, respectively; and ${\epsilon }_{it}$ is the error term. According to Pesaran [48]; the stationarity test depends on the t-value of ${\beta }_i$, either separately or jointly.

Although the CADF method considers the existence of variables’ CSD, as the univariate variable time series is subjected to a major shock, it will often undergo structural changes and may lead to a misidentification of the time series by the unit root test. Therefore, to reduce the probability of rejection of an alternative hypothesis in the unit root test and obtain more accurate stationarity results, in the revised manuscript we add the minimum Lagrange multiplier (LM) unit root test model with structural break, as proposed by Refs. [49,50]. We then employ the endogenous multiple structural break test of [51]; hereafter BP) to confirm whether the sequence has a structural break and the location of the structural breakpoint.

• (3)Panel long-run estimates

We finally use MG proposed by Pesaran [55] to estimate the long-term estimation parameters of GRSI and stock market as well as CASE and stock market. At the same time, to understand the influence of factors in the long-term estimation, we also use the MG method to estimate the long-term parameters of the impacts of GRSI and CASE on the stock prices and returns of 15 oil exploration and production enterprises. The model is calculated as:

$$\Delta CLOS{E}_{it}={\psi }_{\text{0}}+{\psi }_1\Delta GRS{I}_{it}+d_2(w_t)+{\mu }_{it}$$ (5)

$$\Delta RETUR{N}_{it}={\xi }_{\text{0}}+{\xi }_1\Delta GRS{I}_{it}+d_1(w_t)+{\mu }_{it}$$ (6)

$$\Delta CLOS{E}_{it}={\rho }_{\text{0}}+{\rho }_1\Delta CAS{E}_{it}+d_4(w_t)+{\mu }_{it}$$ (7)

$$\Delta RETUR{N}_{it}={\varphi }_{\text{0}}+{\varphi }_1\Delta CAS{E}_{it}+d_3(w_t)+{\mu }_{it}$$ (8)

Here, the term $w_t$ represents any excluded idiosyncratic process that evolves over time. We calculate stock returns (RETURN) as follows:

$$RETUR{N}_{it}=(CLOS{E}_{\text{it}}-CLOS{E}_{i(t-1)})/CLOS{E}_{i(t-1)}$$ (9)

3. Data description

This paper aims to investigate COVID-19 shocks on the stock prices of 15 representative oil exploration and production enterprises in 5 countries from January 1, 2020 to March 31, 2021. As mentioned earlier, we use CASE and GRSI as proxy indicators of the severity of the COVID-19 epidemic. Data on the number of confirmed cases per million people in each sample country are from https://ourworldindata.org/grapher/new-covid-cases-per-million. The higher the values of CASE are, the greater is the severity of the COVID-19 epidemic. GRSI reflects the information of different policies adopted by governments in response to the pandemic, collected by the Oxford COVID-19 Government Response Tracker (OxCGRT). The data of GRSI are from https://ourworldindata.org/grapher/covid-stringency-index. The variable is composed of nine indicators from the aspects of containment and closure and health system, including school closures, workplace closures, cancelling public events, restrictions on gatherings, closure of public transport, public information campaigns, stay at home mandates, restrictions on internal movement, and international travel controls. GRSI is the weighted average value of these nine indicators, and the value is from 0 to 100. A higher value means the more severe the government's policy toward the pandemic is and, to some extent, the more serious the epidemic situation is.

The 15 representative oil exploration and production enterprises are chosen following the guidance provided by Our World in Data.6 The 15 enterprises include Beach Energy (BCHEY), Cairn Energy (CRNCY), Enquest Lon (ENQUF), Genel Energy (GEGYY), Harbour Energy (HBRIY), Hurricane Energy (HRCXF), Inpex (IPXHY), Kosmos Energy (KOS), LEKOIL (LEKOF), Nostrum Oil & Gas (NSTRY), Pantheow Resrcs (PTHRF), Pharos Energy (SOCLF), Tullow Oil (TUWOY), Vaalco Energy Inc. (EGY), and Vermilion Energy (VET), which are listed in the seven sample countries of Australia, Canada, Japan, Netherlands, Nigeria, United Kingdom, and the United States. The reasons why we choose these 15 enterprises are as follows. On the one hand, these sample enterprises are the top stocks among international oil and gas exploration enterprises as reported by the website of Our World in Data. Their comprehensive strength in terms of exploration, development, and production of crude oil and natural gas assets are at the top level in the world. On the other hand, the stock exchanges these sample enterprises are listed on, such as London Stock Exchange, American Stock Exchange, Tokyo Stock Exchange, and Toronto Stock Exchange, are among the world's leading international financial centers in terms of total market value, trading volume, and liquidity. To sum up, we believe that the stocks of these enterprises are strongly representative of the oil exploration and production sector during the COVID-19 epidemic. We collect the daily closing prices (CLOSE) of these oil exploration and production enterprises from https://www.macrotrends.net/stocks/industry/222/international-e-p.

Fig. 1 describes the trend chart of CLOSE of 15 oil exploration and production enterprises and the trend charts of CASE and GRSI of countries to which these enterprises belong. Figure 2 depicts the stock market returns of the 15 oil exploration and production enterprises and the trend charts of CASE and GRSI. It is obvious that each country's CASE shows a trend of first rising and then falling and then rising and falling, while GRSI presents a stable rising trend. On the contrary, the CLOSE charts of most oil exploration and production enterprises show a trend of first declining and then rising and then declining and then rising, while RETURN presents a stable rising trend.

We analyze the trend charts of COVID-19 epidemic and the stock market from four stages. The first period is from January 2020 to March 2020. In the middle of January 2020, with the outbreak of COVID-19, CASE shows a rising trend, and the governments actively introduced epidemic prevention and control measures, and so GRSI has a rising trend. At the same time, national economies shrank during this period, especially due to the more stringent epidemic prevention and control measures such as factory closures and travel bans, resulting in a further decline in oil demand and a drop in oil prices. As a result, the stock prices of oil exploration and production enterprises showed a downward trend. However, with the governments' responses to COVID-19, investors’ confidence in them controlling the epidemic and in an economic recovery was enhanced. Therefore, in Figure 2, with the rise of GRSI in this period, RETURN gradually increases from a negative value.

The second stage is from April 2020 to July 2020. Government responses to the epidemic played a significant role in the control of the epidemic. At this stage, the number of cases gradually decreased. However, considering the incubation period of the virus, and the global epidemic situation was still relatively severe, GRSI remains at a stable high level despite a slight decline. During this period, in order to hedge against the negative impact of the epidemic and stabilize the liquidity crisis, governments successively issued a variety of positive fiscal policies and loose monetary policies, which helped stock markets to rebound. Therefore, CLOSE shows an upward trend, and RETURN is in a positive and stable trend.

At the end of July 2020, the epidemic in many countries such as the United Kingdom and Canada rebounded, and the probability of virus transmission became greater than in the winter. Therefore, the number of confirmed cases increased, and GRSI remains stable at a high level.

The fourth stage is from January to March 2021. During this period, the epidemic was under control, CASE shows a downward trend, CLOSE rises, and GRSI and RETURN maintain a stable trend. To further clarify the COVID-19 shocks on oil exploration and production enterprises’ stock prices, we shall conduct a quantitative analysis in the next section.

Table 2 presents the data descriptive statistics of each variable during January 1, 2020 to March 31, 2021. For CASE, the maximum value (MAX) is 1004.507, while the minimum value (MIN) is 0, and the standard deviation (SD) is 182.868. The MAX and MIN of GRI are 83.96 and 0, respectively, and SD is 25.187. These huge differences show the severe situation of the COVID-19 epidemic in different countries where the sample enterprises are located, and the responses of governments to the epidemic situation also varied a lot. In the early stage, the epidemic may not have looked serious, and government responses were relatively slow. However, with the spread of the epidemic and the number of confirmed cases increasing, their responses to the epidemic gradually intensified.

We also see huge differences between MAX (35.72) and MIN (0.01) of CLOSE as well as and MAX (17) and MIN (−1.21) of RETURN. This illustrates large differences and fluctuations in the stock markets of energy exploration and production enterprises, which likely were caused by the impact of the epidemic. Therefore, it is necessary to explore the effect of the COVID-19 epidemic on the stock markets of oil exploration and production enterprises.

4. Empirical results

4.1. Cross-sectional dependence test

Before analyzing the correlation between COVID-19 and oil exploration and production enterprises’ stock prices, we first check whether there is the problem of CSD in all the variables. Table 3 shows the CSD test results of four variables. We can see from the test results of Pesaran [46] and Pesaran [47] that all the P-values for the CSD test are 0, indicating under the two tests that the four variables all reject the null hypothesis, which means there is CSD in the variables and reveals that the stock prices and stock returns of each enterprise and the development of COVID-19 epidemic situation in every country are related to each other.

4.2. Panel unit root test

Considering the existence of CSD for variables, the second-generation CADF method, which is newly developed by Pesaran [48]; is exploited to test the stability of CLOSE, RETURN, CASE, and GRSI. From Table 4, except RETURN, CLOSE, GRSI, and CASE all accept the null hypothesis, which means that these three variables all are non-stationary. We then perform first-order difference processing for CLOSE, RETURN, CASE, and GRSI. In Table 4, the test results confirm at the 1 % significance level that all variables after first-order difference processing reject the null hypothesis - that is, the variables of ΔCLOSE, ΔRETURN, ΔCASE, and ΔGRSI are stationary.

If there is a structural breakpoint in the variable, then it may influence the effectiveness of the unit root test results. Therefore, the LS (2003, 2004) test with structural breaks is then used in this section to test the stability of the variables. Table 5 presents the LS test results with an intercept (Model A) as well as intercept and slope break (Model C). It can be seen that the null hypothesis of Model A is rejected in some variables, while the null hypothesis of Model C is rejected in all variables. Since Model C performs better than Model A and provides more information for mean regression [56], we believe that all variables are stable.

We next use the Bai and Perron [51] test on the structural break for all variables. The results appear in Table 6. It can be seen from Table 6 that the values of UDmax statistic and WDmax statistic of CASE in 7 countries, GRSI in 5 countries, CLOSE in 15 enterprises, and RETURN in 2 enterprises are respectively significantly greater than their critical values, indicating that there are structural breakpoints in these variables. For GRSI, the breakpoints may occur in March 2020, June 2020, August 2020, November 2020, and January 2021; for CASE, the breakpoints occur in March 2020, May 2020, September 2020, November 2020, and January 2021; for CLOSE, the breakpoints occur in March 2020, May 2020, August 2020, November 2020, and January 2021; and for RETURN, the breakpoints of the two enterprises occur in March 2020, June 2020, August 2020, November 2020, and January 2021. In general, the BP test results show that the time series of GRSI and CASE in most countries, as well as CLOSE in enterprises, have structural breakpoints, thus confirming the necessity of the LS test to a certain extent.

4.3. Panel long-run estimation analysis

This paper uses the MG method to further explore the impact of GRSI and CASE on the stock prices and returns of oil exploration and production enterprises. The results are in Table 7. From Table 7, at a 1 % significance level, GRSI has a long-run negative impact on CLOSE. Similarly, in the long term, CASE significantly negatively influences CLOSE. The results indicate that the more people there are who are confirmed with COVID-19, the greater is the probability that the stock prices of oil exploration and production enterprises decrease accordingly. Similarly, increases in the strictness of prevention and control measures by government in response to the COVID-19 epidemic also have a negative shock on stock prices. This is because, on the one hand, the COVID-19 pandemic caused great harm to the world from medical treatments to economic growth [19]. On the premise of ensuring the health of citizens, various government departments are forced to introduce policies that sacrifice economic development, such as closing schools and factories and mandating citizens stay at home. The shutdown of production, tourism, and the aviation industry undoubtedly led to the decline of oil demand in the market as well as an increase in the cost of oil transportation and storage for oil exploration and production enterprises [57]. Overall, this spurred a significant drop in oil prices and reduced revenue for oil exploration and production enterprises, which in turn affected stock prices. On the other hand, the large-scale real economic depression caused by the epidemic has brought potential risk factors to financial and capital markets. In the face of uncertainty, investors have undoubtedly increased their demand for safe-haven assets, leading to a shift in the layout of the oil capital market. The COVID-19 epidemic shocks broke the balance of supply and demand in the oil market and transmitted negative sentiment to investors, which eventually led to the poor performance of oil exploration and production enterprises and a decline in their stock prices.

The remarkable thing in Table 7 is that the negative shocks of GRSI on CLOSE are significantly higher those of CASE on CLOSE. Specifically, for every 1 % increase in GRSI, CLOSE decreases by 13.1 %, and for every 1 % increase in CASE, CLOSE drops by 7.9 %. The potential reasons may lie in the following two aspects. First, CASE only objectively reflects the number of confirmed cases, while GRSI denotes the government's response to the COVID-19 epidemic. This latter indicator includes the health system, containment, and closure measures adopted by the government, which cover more information than CASE. These measures affect a whole country's economic activity, which in turn is transmitted to the capital market.

Second, GRSI reflects epidemic prevention and control measures such as closing factories and public transportation and restricting international travel and movement. These measures directly reduced the demand for oil consumption in industrial production and transportation, spurring oil prices to fall further. Compared with CASE, the information in response to GRSI provides oil investors with evidence on oil price fluctuations caused by the epidemic more directly and comprehensively. Therefore, the GRSI shocks on oil exploration and production enterprises’ stock prices are naturally greater than the shocks of CASE on stock prices in past research, where only CASE is used to measure the COVID-19 epidemic, which undoubtedly weakens the shock of the COVID-19 epidemic upon the stock market.

We further draw an interesting conclusion that the increase of GRSI did not lead to a decrease of RETURN, but rather that an increase of GRSI raised RETURN, but the impact of CASE on RETURN is not significant. The results indicate that a government's policy in response to the COVID-19 epidemic can improve the stock market returns of oil exploration and production enterprises. The implementation of restrictions and closure policies, such as closing schools, workplaces, and public transportation, had a negative impact on economic activities in the short term and affected stock prices. However, the implementation of this series of policies strictly controlled social distancing, reduced the flow of people, and thus decreased the probability of epidemic infection. This sent a positive signal for investors' confidence in governments' control of the epidemic and in the future economic recovery. Therefore, although the COVID-19 epidemic brought about a decline in the stock prices of these enterprises, government policies toward the COVID-19 epidemic were able to slow down the decline of stock prices by enhancing the confidence of investors, thus bringing positive market returns. The studies of Phan and Narayan Phan and Narayan (2020) and Chang et al. [25] also confirm our conclusion.

4.4. Long-run parameters’ estimations for a specific enterprise

Taking into account the heterogeneity of companies is an important factor in exploring the COVID-19 epidemic shocks on the stock markets of oil exploration and production enterprises. Therefore, after discussing the overall impact of the epidemic on the stock markets of oil exploration and production enterprises in the previous section, this section uses the MG method to analyze the impact of the epidemic from the perspective of specific enterprises. Here, we report the estimate results of the shocks of GRSI or CASE on CLOSE and RETURN for each specific listed firm for purposes of a more in-depth investigation.

The results are shown in Tables 8 and 9. As we have speculated, among the stock prices of the 15 oil exploration and production enterprises affected by the COVID-19 epidemic, GRSI significantly negatively affects their stock prices at a 1 % significance level. When using CASE as the proxy indicator, only 2 (LEKOF and PTHRF) of 15 enterprises’ stock prices experience a positive impact, while the stock prices of the remaining 13 enterprises are significantly negatively affected by the COVID-19 epidemic.

In Table 9 we see among the stock returns of the 15 oil exploration and production enterprises affected by the COVID-19 epidemic that 10 enterprises' returns are positively affected by GRSI, and 7 of these 10 are significantly affected. CASE has a positive impact on the stock market returns of 11 enterprises, but only 1 has a significant impact. Therefore, the shock of the epidemic on the specific enterprises' stock prices and returns varies with different proxy indicators. However, generally speaking, the MG estimation results of specific enterprises are basically consistent with the results of the whole sample in the previous section. The stock prices of most oil exploration and production enterprises were initially negatively affected by COVID-19, while the stock returns of most enterprises were positively affected by COVID-19, especially when GRSI is used to measure the severity of the epidemic. This conclusion further reveals that the COVID-19 epidemic has negatively shocked the stock prices of most specific oil exploration and production enterprises, and government responses to the epidemic enhanced investors’ confidence and had a positive impact on the stock returns of most enterprises.

5. Conclusion

During the period of the COVID-19 epidemic outbreak, the situation facing the global oil market was extremely grim, and oil prices plummeted at an unprecedented rate, with even some futures contracts experiencing negative prices. In this context, we incorporate the COVID-19 epidemic and the stock prices of oil exploration and production enterprises into a research framework. Using data of 15 oil exploration and production enterprises along with variables of COVID-19 epidemic development from January 1, 2020 to March 31, 2021, this paper examines the shock of the pandemic on their stock prices and returns. On the basis of previous studies using the number of confirmed cases to measure the epidemic situation of COVID-19, this paper also introduces the government response stringency index from the perspective of government intervention. At the same time, we discuss the impact of COVID-19 on the stock market from two aspects of oil exploration and production enterprises: stock prices and stock returns. Our study presents some interesting results.

The results based on the MG estimation test show that GRSI and CASE have a negative impact on the stock prices of oil exploration and production enterprises. In particular, the information of the COVID-19 situation as reflected by GRSI is more comprehensive, and the impact of GRSI on the stock prices of oil exploration and production enterprises is greater than that of CASE. On the other hand, GRSI has a significantly positive impact on the stock returns of oil exploration and production enterprises. Similar conclusions are drawn from the impact of the COVID-19 shock on specific oil exploration and production enterprises’ stock prices and stock returns. This shows that although the epidemic had a negative impact on the economy and stock prices, the strict epidemic prevention and control policies brought confidence in epidemic control and economic recovery for investors in the capital market to a certain extent and even alleviated the negative impact on stock prices.

The research findings have important significance for investors and government decision-making. Previous studies believe that the COVID-19 epidemic had a negative impact on economic development and the stock markets, yet according to the research conclusions herein, we believe that governments should indeed firmly implement epidemic prevention and control policies during an epidemic period. On the one hand, these efforts can reduce the negative impact on oil and stock markets caused by the numbers of confirmed cases and deaths. On the other hand, such efforts can reduce investors' fear over the uncertain impact of the epidemic, enhance their confidence, and contribute to the stability of the financial market. For investors in the oil capital market, they should clearly realize that epidemic prevention and control policies have dual attributes. On the one hand, they do bring short-term stock price fluctuations, but on the other hand, they show a government's determination to deal with the epidemic and restore the economy. Investors could obtain good profit opportunities in the oil capital market by accurately grasping the dynamic change signals of epidemic prevention and control policies.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix.

Fig. 1.

Trend Stock Price, GRSI, and Confirmed Cases. Note: The green curve represent CLOSE of each oil exploration and production enterprise, while the blue curve and the red curve report the trend of GRSI and CASE, respectively. Taking into account the difference in the value range of each variable and in order to show the trend of each variable more accurately, the left ordinate axis represents the value scale of GRSI and CASE, and the right ordinate axis represents the value scale of CLOSE.

Fig. 2.

Trend Stock Return, GRSI, and Confirmed Cases. Note: The black curve represent RETURN of each oil exploration and production enterprise, while the blue curve and the red curve report the trend of GRSI and CASE, respectively. Taking into account the difference in the value range of each variable and in order to show the trend of each variable more accurately, the left ordinate axis represents the value scale of GRSI and CASE, and the right ordinate axis represents the value scale of RETURN.

Table 1.

The literature about the impact of COVID-19 epidemic on stock markets and oil.

Author(s)	COVID-19	Stock market/Oil	Findings
[29]	CASE and DEATH	Stock market in China	Both CASE and DEATH negatively influence stock returns.
[33]	CASE	21 stock indices	CASE has a significantly negative impact on the stock index trend.
[31]	CASE	12 stock markets	COVID-19 epidemic increases the risk of the financial market.
Badar et al. (2020)	CASE and DEATH	64 countries	Stock returns decrease with CASE.
[34]	Panic Index, Global Sentiment Index, and Media Coverage	Indices for the world and the U.S.	The panic caused by media's epidemic news relates to the violent fluctuation of financial markets.
[35]	Government intervention	20 OECD countries	CASE negatively influence stock markets, and governments' intervention measures magnify the negative effect of COVID-19 on stock returns.
[37]	Daily cumulative infected people in China	Oil and electricity of China	The severity of the epidemic directly and indirectly affects the demand for electricity and oil.
Jeris and Nath (2020)	CASE and DEATH	Brent oil	COVID-19 has a long-term impact on the uncertainty of UK economic policy, and Brent crude oil prices negatively influence the uncertainty of UK economic policy equilibrium.
Devpura and Narayan (2020)	CASE and DEATH	Oil price	COVID-19 increases oil prices volatility.
[58]	–	WTI crude oil prices	The shocks of COVID-19 to oil prices are transitory, albeit have long-lasting effects.
[59]	DEAT	WTI crude oil price	COVID-19 epidemic strength significantly impacts oil prices.

Note: CASE represents the number of daily confirmed cases from COVID-19. DEATH represents the number of daily death cases from COVID-19.

Table 2.

Summary of descriptive statistics.

Variable	N	Mean	SD	Min	Median	Max
CLOSE	4663	3.968	6.701	0.01	1.18	35.72
RETURN	4663	0.007	0.274	−1.21	0	17
GRSI	4633	58.544	22.184	0	67.13	87.96
CASE	4633	116.722	182.868	0	27.075	1004.507

Table 3.

Cross-section dependence tests.

	[46]				[47]
Variable	CD-test	p-value	Corr	abs (corr)	CD	P
CLOSE	78.41***	0.000	0.435	0.499	13.701***	0.000
RETURN	13.71***	0.000	0.076	0.097	83.281***	0.000
GRSI	162.12***	0.000	0.899	0.899	158.141***	0.000
CASE	127.36***	0.000	0.706	0.764	125.977***	0.000

Notes: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.

Table 4.

Results from the CADF panel unit root test.

Group	CLOSE	ΔCLOSE	RETURN	ΔRETURN	GRSI	ΔGRSI	CASE	ΔCASE
CADF	−0.77	−13.934***	−14.813***	−18.935***	3.366	−5.414**	−1.231	−9.124***

Notes: The statistic of CADF is t-bar, and the critical value of t-bar for the CADF test of the ALL sample at the 1 %, 5 %, and 10 % levels is −2.770, −2.650, and −2.590, respectively. ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.

Table 5a.

Results from the LM unit root test for CLOSE.

Series	Model A				Model C
	TB	K	St-1	Bt	TB	K	St-1	Bt	Dt
BCHEY	20,210,216	8	−0.01 (-0.86)	−0.14 (-0.11)	20,200,219	8	−0.12***(-4.41)	−261.01***(-3.17)	175.73***(4.79)
CRNCY	20,200,504	8	−0.03 (-1.30)	−0.21 (-1.09)	20,200,219	8	−0.13***(-4.40)	−267.78***(-3.17)	180.62*** (4.77)
ENQUF	20,200,526	8	−0.08 (-1.60)	−0.01 (-0.34)	20,200,219	8	−0.13***(-4.40)	−268.92***(-3.17)	181.35*** (4.77)
GEGYY	20,200,415	8	−0.05** (−2.45)	−0.05 (-0.09)	20,200,219	8	−0.20*** (−4.78)	−2.53*** (−2.66)	1.30*** (4.33)
HBRIY	20,200,616	8	−0.03*(-1.71)	0.13*** (2.90)	20,200,219	8	−0.13*** (−4.39)	−268.99***(-3.17)	181.46*** (4.76)
HRCXF	20,210,129	8	−0.04***(-3.07)	−0.002 (−0.25)	20,200,219	8	−0.13*** (−4.39)	−268.92***(-3.17)	181.39*** (4.76)
IPXHY	20,200,120	8	−0.01 (-1.21)	−0.09 (-0.27)	20,200,219	8	−0.12*** (−4.41)	−266.91***(-3.19)	179.85*** (4.80)
KOS	20,210,210	8	0.00 (-1.08)	−0.02 (-0.05)	20,200,313	8	−1.59***(-33.59)	3211.01***(32.10)	−3212.96***(-33.57)
LEKOF	20,200,218	8	−0.006 (-0.83)	1.99 (0.03)	20,200,218	8	−1.59***(-33.55)	3210.01***(32.06)	−3211.83***(-33.50)
NSTRY	20,200,226	8	−0.04**(-2.30)	0.92 (0.98)	2,020,727	8	−0.65*** (−9.76)	8.60***(8.06)	8.06*** (−9.70)
PTHRF	20,200,224	8	−0.08**(-2.18)	0.02 (0.2412)	20,210,112	8	−0.97***(-10.14)	0.55*** (6.68)	−0.44*** (−10.25)
SOCLF	20,200,501	8	−0.05***(-2.46)	0.01 (0.49)	20,200,221	8	−0.14*** (−4.62)	−268.51***(-2.85)	171.60*** (4.24)
TUWOY	20,210,117	8	−0.68***(-4.41)	0.02 (0.20)	20,200,219	8	−0.77***(-6.39)	−0.24***(-2.82)	0.13***(5.74)
EGY	20,200,526	8	−0.03 (-1.63)	−0.003 (-0.04)	20,200,219	8	−0.22*** (−5.17)	−0.42*** (−2.85)	0.24***(4.83)
VET	20,200,415	8	−0.05**(-2.45)	−0.06 (−0.11)	20,200,219	8	−0.20***(-4.82)	−2.53***(-2.67)	1.30*** (4.37)

Notes: TB, T1, and T2 are structural breakpoints. St-1 is the t statistic of LM. Bt is the coefficient of intercept term. Dt is the coefficient of the slope term. *** significance at 1 % level; ** significance at 5 % level; * significance at 10 % level.

Table 5b.

Results from the LM unit root test for RETURN.

Series	Model A				Model C
	TB	K	St-1	Bt	TB	K	St-1	Bt	Dt
BCHEY	20,200,708	8	−1.44***(-5.97)	−0.11**(-2.40)	20,200,218	8	−0.12***(-4.41)	−261.01***(-3.17)	175.73***(4.79)
CRNCY	20,200,424	8	−1.07***(-5.35)	0.002 (0.02)	20,200,407	8	−0.57***(-6.15)	0.29***(2.84)	−0.15***(-5.91)
ENQUF	20,201,120	8	−1.18***(-5.54)	−0.07 (-0.83)	20,200,827	8	−1.22***(-5.99)	0.11 (1.35)	−0.05***(-4.18)
GEGYY	20,200,226	8	−0.02 (-1.51)	−0.45 (-0.54)	20,200,218	8	−0.17***(-5.14)	−326.23***(-3.35)	211.20***(4.74)
HBRIY	20,200,220	8	−0.06**(-2.07)	−0.05 (-0.32)	20,200,313	8	−0.44***(-6.01)	−0.37**(-2.47)	0.39***(5.74)
HRCXE	20,200,915	8	−0.23***(-2.68)	0.02 (0.26)	20,200,203	8	−1.2047***(-7.1382)	−0.24**(-2.46)	0.15***(6.31)
IPXHY	20,200,429	8	−1.20***(-6.84)	−0.02 (-0.80)	20,200,507	8	−1.23***(-7.15)	0.05 (1.54)	−0.005 (-1.59)
KOS	20,210,210	8	0.00 (-0.84)	−0.04 (-0.17)	20,200,218	8	−1.59***(-33.55)	3209.95***(32.06)	−3211.79***(-33.50)
LEKOF	20,210,111	8	0.00 (-0.85)	−1.11***(-4.68)	20,200,218	8	−1.59 (-1.59)	3209.99***(32.06)	−3211.83***(-33.50)
NSTRY	20,200,220	8	−0.01 (-1.08)	6.99 (0.10)	20,200,810	8	−0.66***(-9.47)	−9.73***(-7.79)	10.21***(9.46)
PTHRF	20,200,220	8	−0.16**(-2.53)	−0.03 (-0.41)	20,201,203	8	−0.83***(-7.21)	−0.27***(-3.12)	0.26***(7.31)
SOCLF	20,200,220	8	−0.06*(-1.92)	−0.01 (-0.11)	20,200,303	8	−0.39***(-5.11)	−0.13*(-1.67)	0.16***(4.67)
TUWOY	20,200,814	8	−0.40***(-3.21)	0.01 (0.10)	20,200,723	8	−1.54***(-7.12)	−0.22**(-2.08)	0.27***(6.84)
EGY	20,200,320	8	−0.01 (-1.10)	−13.85 (-0.19)	20,210,209	8	−1.20***(-20.86)	2420.63***(19.29)	−2423.74***(-20.82)
VET	20,200,226	8	−0.02 (-1.54)	−0.45 (-0.54)	20,200,218	8	−0.17***(-5.17)	−326.43***(-3.37)	211.18***(4.78)

Notes: TB, T1, and T2 are structural breakpoints. St-1 is the t statistic of LM. Bt is the coefficient of intercept term. Dt is the coefficient of the slope term. *** significance at 1 % level; ** significance at 5 % level; * significance at 10 % level.

Table 5c.

Results from the LM unit root test for GRSI.

Series	Model A				Model C
	TB	K	St-1	Bt	TB	K	St-1	Bt	Dt
AUS	20,210,211	8	−0.01 (-1.12)	−13.55 (-0.19)	20,200,403	8	−1.21***(-20.95)	2419.19***(19.38)	−2421.26***(-20.92)
CAN	20,201,124	8	−0.04 (-1.44)	0.28 (0.03)	20,200,221	8	−0.14***(-4.71)	−283.37***(-2.96)	180.76***(4.35)
GBR	20,200,220	8	−0.01 (-1.16)	1.91 (0.02)	20,210,308	8	−1.14***(-19.22)	2295.21***(17.58)	−2231.95***(-19.18)
JPN	20,200,219	8	−0.04*(-1.94)	−3.99 (-1.63)	2,020,036	8	−0.37***(-6.91)	−12.69***(-4.36)	8.39*** (6.81)
NGA	20,200,327	8	−0.09***(-2.93)	−1.04 (-0.38)	20,200,409	8	−0.22***(-5.16)	−5.47* (−1.77)	4.65*** (4.84)
NLD	20,200,421	8	−0.06*(-1.83)	−4.65 (-0.22)	2,020,029	8	−0.15***(-4.80)	−301.46***(-3.02)	190.30***(4.47)
IPXHY	20,200,408	8	−0.02**(-2.41)	29.85 (1.53)	20,210,208	8	−0.21***(-5.08)	243.09***(2.66)	−250.61***(-4.73)

Notes: TB, T1, and T2 are structural breakpoints. St-1 is the t statistic of LM. Bt is the coefficient of intercept term. Dt is the coefficient of the slope term. *** significance at 1 % level; ** significance at 5 % level; * significance at 10 % level.

Table 5d.

Results from the LM unit root test for CASE.

Series	Model A				Model C
	TB	K	St-1	Bt	TB	K	St-1	Bt	Dt
AUS	20,200,330	8	−0.09 (-2.49)	−0.87 (-0.26)	20,200,406	8	−0.35***(-6.69)	−11.33***(-2.95)	8.84***(6.62)
CAN	20,200,708	8	−0.06 (-1.76)	−0.12 (-0.01)	20,200,218	8	−0.17***(-5.15)	−326.91***(-3.36)	208.73***(4.75)
GBR	20,201,113	8	−0.07 (-1.58)	−0.03 (-0.001)	20,210,104	8	−0.36***(-5.43)	−46.56**(-2.10)	42.4676***(5.68)
JPN	20,200,219	8	−0.03 (-1.43)	2.11 (0.76)	20,200,318	8	−0.56***(-9.40)	34.27***(8.94)	−20.86***(-9.34)
NGA	20,200,209	8	−0.01 (-1.09)	−13.82 (-0.19)	20,200,407	8	−1.21***(-20.92)	2425.72***(19.35)	−2428.95***(-20.89)
NLD	20,210,210	8	−0.02*(-1.91)	7.20 (0.33)	20,210,211	8	−0.18***(-4.57)	200.01***(2.22)	−193.26***(-4.28)
IPXHY	20,200,409	8	−0.004**(-2.53)	26.96 (1.83)	20,210,122	8	−0.99***(-16.69)	1941.32***(15.01)	−1922.27***(-16.62)

Notes: TB, T1, and T2 are structural breakpoints. St-1 is the t statistic of LM. Bt is the coefficient of intercept term. Dt is the coefficient of the slope term. *** significance at 1 % level; ** significance at 5 % level; * significance at 10 % level.

Table 6a.

Results from Ref. [51]] for GRSI.

Series	Break (1)	Break (2)	Break (3)	Break (4)	Break (5)	UDMax	WDMax
AUS	20,200,318	20,200,522	20,200,729	20,201,027	20,210,108	3.8484	6.2233
CAN	20,200,313	20,200,618	20,200,824	20,201,102	20,210,118	542.22***	542.22***
GBR	20,200,320	20,200,527	20,200,812	20,201,019	20,201,223	113.39***	148.89***
JPN	20,200,127	20,200,521	20,200,728	20,201,001	20,201,120	15.59***	30.96***
NGA	20,200,320	20,200,630	20,200,909	20,201,112	20,210,121	66.94***	104.88***
NLD	20,200,311	20,200,529	20,200,804	20,201,009	20,201,215	808.41***	1061.47***
USA	20,200,313	20,200,612	20,200,909	20,201,112	20,200,121	167.47***	167.47***

Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.

Table 6b.

Results from Ref. [51]] for CASE.

Series	Break (1)	Break (2)	Break (3)	Break (4)	Break (5)	UDMax	WDMax
AUS	20,200,312	20,200,702	20,200,908	20,201,111	20,210,120	24.33***	43.28***
CAN	20,200,325	20,200,601	20,200,908	20,201,111	20,210,125	238.53***	597.06***
GBR	20,200,324	20,200,529	20,200,908	20,201,111	20,210,120	21.16***	52.96***
JPN	20,200,401	20,200,708	20,200,911	20,201,116	20,210,125	37.05***	75.91***
NGA	20,200,319	20,200,526	20,200,730	20,201,005	20,201,209	133.03***	264.12***
NLD	20,200,313	20,200,522	20,200,729	20,201,005	20,210,108	58.41***	79.34***
USA	20,200,325	20,200,625	20,200,903	20,201,109	20,210,121	149.48***	319.60***

Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.

Table 6c.

Results from Ref. [51]] for CLOSE.

Series	Break (1)	Break (2)	Break (3)	Break (4)	Break (5)	UDMax	WDMax
BCHEY	2020/4/29	2020/7/6	2020/9/11	2020/11/16	2021/1/25	82.7979***	165.2055***
CRNCY	2020/3/9	2020/5/15	2020/8/3	2020/11/10	2021/1/19	60.4555***	151.3234***
ENQUF	2020/3/13	2020/5/28	2020/8/13	2020/10/21	2021/1/7	174.2849***	174.2849***
GEGY	2020/4/20	2020/6/24	2020/8/28	2020/11/4	2021/1/12	262.7821***	521.7434***
HBRIY	2020/3/9	2020/6/4	2020/8/18	2020/11/10	2021/1/19	60.4446***	92.7030***
HRCXF	2020/3/9	2020/5/20	2020/9/3	2020/11/10	2021/1/19	61.9432***	104.3951***
IPXHY	2020/3/9	2020/5/13	2020/9/10	2020/11/13	2021/1/22	19.78***	39.17***
KOS	2020/3/10	2020/5/14	2020/8/31	2020/11/12	2021/1/21	28.06***	70.24***
LEKOF	2020/3/10	2020/5/22	2020/8/10	2020/10/14	2020/12/18	30.13***	75.43***
NOG.L	2020/3/9	2020/6/5	2020/8/11	2020/10/19	2021/1/5	8.0354*	15.95***
PTHRF	2020/3/10	2020/5/14	2020/7/21	2020/9/24	2020/12/30	147.34***	176.26***
SOCLF	2020/3/10	2020/5/14	2020/7/21	2020/9/25	2020/12/11	91.84***	182.34***
TUWOY	2020/3/10	2020/6/2	2020/8/20	2020/11/11	2021/1/20	16.74***	41.90***
EGY	2020/3/10	2020/6/3	2020/8/26	2020/11/12	2021/1/21	31.05***	77.72***
VET	2020/3/10	2020/5/14	2020/9/1	2020/11/16	2021/1/25	29.57***	59.54***

Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.

Table 6d.

Results from Ref. [51]] for RETURN.

Series	Break (1)	Break (2)	Break (3)	Break (4)	Break (5)	UDMax	WDMax
BCHEY	2020/3/11	2020/5/15	2020/7/23	2020/11/9	2021/1/15	3.70	9.25
CRNCY	2020/3/18	2020/6/2	2020/8/24	2020/10/28	2021/1/5	3.90	5.48
ENQUF	2020/4/3	2020/6/10	2020/8/21	2020/11/5	2021/1/13	17.56***	17.56***
GEGY	2020/3/25	2020/6/1	2020/8/5	2020/11/4	2021/1/19	3.25	5.01
HBRIY	2020/3/16	2020/6/5	2020/9/1	2020/11/10	2021/1/19	4.22	6.82
HRCXF	2020/3/13	2020/5/19	2020/7/24	2020/10/2	2020/12/8	3.57	5.39
IPXHY	2020/3/17	2020/6/5	2020/8/25	2020/10/30	2021/1/12	10.36**	13.09
KOS	2020/3/19	2020/6/8	2020/8/26	2020/11/6	2021/1/14	7.39	10.65
LEKOF	2020/3/19	2020/5/26	2020/7/30	2020/10/21	2020/12/28	3.16	5.37
NOG.L	2020/3/20	2020/6/10	2020/8/17	2020/10/23	2021/1/4	5.38	5.38
PTHRF	2020/4/13	2020/6/26	2020/9/1	2020/11/5	2021/1/13	5.12	6.73
SOCLF	2020/3/20	2020/6/2	2020/9/4	2020/11/11	2021/1/20	9.61**	9.61
TUWOY	2020/3/16	2020/6/10	2020/8/24	2020/10/30	2021/1/19	17.80***	17.80***
EGY	2020/3/23	2020/6/5	2020/8/12	2020/11/2	2021/1/12	7.68*	7.68
VET	2020/3/18	2020/6/8	2020/8/17	2020/10/28	2021/1/20	8.31*	13.45

Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.

Table 7.

Results of the MG panel data estimation method.

	CLOSE		RETURN
GRI	−0.131*** (-4.79)		0.040** (2.43)
CASE		−0.079*** (-4.19)		0.057 (1.08)
Trend	0.0007*** (2.64)	0.001*** (2.93)	0.0004 (0.09)	0.004** (2.13)
Cons	1.415*** (4.71)	1.142*** (4.24)	−1.704*** (-2.85)	−0.401 (-1.64)
N	4663	4663	4663	4663
Wald	22.95***	17.54***	5.89**	1.17

Notes: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively. Z statistics are in parentheses.

Table 8.

Firm-specific long-run estimates for CLOSE (MG estimator).

Firm	GRSI	CASE	Trend	Cons
BCHEY	−0.111*** (-10.43)		−0.002** (-1.97)	3.712*** (102)
		−0.067*** (-5.80)	−0.009*** (-8.16)	3.441*** (62.67)
CRNCY	−0.206*** (-19.66)		0.003*** (26.20)	1.912*** (57.05)
		−0.068** (-7.01)	0.003** (13.12)	1.393*** (58.14)
ENQUF	−0.043*** (-13.90)		0.0001*** (3.93)	0.317*** (32.36)
		−0.032*** (-17.81)	0.0004*** (10.28)	0.226*** (50.24)
GEGYY	−0.064*** (-8.58)		0.006*** (6.99)	1.239*** (51.84)
		−0.048*** (-9.87)	0.001*** (9.20)	1.100*** (92.56)
HBRIY	−0.150*** (-16.89)		−0.0003*** (-3.34)	1.015*** (35.76)
		−0.100*** (-17.10)	0.0004*** (3.09)	0.682*** (47.12)
HRCXF	−0.043*** (-23.029)		−0.0003*** (-16.46)	0.318*** (53.00)
		−0.013*** (-6.47)	−0.0004*** (-10.01)	0.208*** (43.55)
IPXHY	−0.157*** (-12.20)		−0.0002** (-2.05)	2.520*** (72.19)
		−0.056* (-3.94)	−0.002 (-1.21)	2.124*** (122.02)
KOS	−0.334*** (-26.88)		0.002*** (13.23)	2.045*** (51.41)
		−0.217*** (-16.46)	0.004 (11.97)	1.498*** (40.92)
LEKOF	−0.006*** (-5.01)		0.0001** (2.38)	0.060*** (14.24)
		0.003 (1.26)	−0.001 (-0.88)	0.041*** (18.85)
NSTRY	−0.101*** (-10.35)		0.001*** (-3.09)	2.666*** (92.50)
		−0.102*** (-6.03)	0.001*** (2.73)	2.562*** (71.34)
PTHRF	−0.029*** (-7.69)		0.001*** (30.93)	0.204*** (17.23)
		0.011*** (4.40)	0.009*** (15.05)	0.114*** (17.75)
SOCLF	−0.105*** (-26.08)		0.0003*** (6.74)	0.589*** (45.90)
		−0.052*** (-13.78)	0.0004*** (5.85)	0.341*** (36.69)
TUWOY	−0.057*** (-16.66)		0.0004*** (9.78)	0.334*** (30.46)
		−0.036*** (-15.05)	0.001*** (11.59)	0.205*** (34.81)
EGY	−0.202*** (-22.10)		0.002*** (20.93)	1.292*** (44.14)
		−0.160*** (-21.74)	0.004*** (23.31)	1.003*** (49.18)
VET	−0.352*** (-21.85)		0.001*** (6.12)	2.932*** (59.13)
		−0.220*** (-10.25)	0.002*** (4.55)	2.222*** (51.02)

Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.

Table 9.

Firm-specific long-run estimates for RETURN (MG estimator).

Firm	GRSI	CASE	Trend	Cons
BCHEY	−0.007 (-0.68)		0.003 (1.20)	−0.095 (0.868)
		0.007 (0.16)	0.002 (1.02)	−0.397 (-0.94)
CRNCY	−0.003 (-0.33)		−0.014 (-0.47)	−6.910** (-2.18)
		−0.003 (-0.33)	0.018 (1.07)	−2.169 (-1.06)
ENQUF	−0.110 (-0.36)		0.064 (0.72)	3.321 (0.23)
		0.020 (0.56)	0.015 (0.19)	1.193 (0.1)
GEGYY	−0.012 (-0.81)		0.005 (1.06)	0.221 (0.31)
		0.0001 (0.01)	0.002 (0.55)	−0.161 (-0.29)
HBRIY	0.071* (1.81)		−0.011 (-0.93)	−2.491*** (-1.33)
		0.0001 (0.13)	0.003 (0.31)	−0.239 (-0.16)
HRCXF	0.015 (0.36)		0.006 (0.50)	−1.727 (-0.87)
		−0.001 (-0.17)	0.010 (0.95)	−1.321 (-0.86)
IPXHY	0.008 (0.39)		0.003 (1.04)	−0.784 (-1.32)
		0.019 (0.86)	0.002 (0.69)	−0.522 (-1.33)
KOS	0.059** (2.19)		−0.002 (-0.33)	−2.867** (-2.09)
		0.005 (1.41)	−0.001 (0.900)	−0.572 (-0.56)
LEKOF	0.077* (1.70)		−0.005 (-0.39)	−2.635 (-0.92)
		0.788 (1.24)	−0.009 (-0.63)	0.833 (0.39)
NSTRY	0.051** (2.43)		−0.010 (-1.53)	−1.304 (-1.34)
		−0.001 (-0.24)	0.002 (0.24)	−0.083 (-0.09)
PTHRF	−0.001 (-0.02)		−0.001 (-0.13)	0.901 (0.63)
		−0.002 (-0.50)	0.001 (0.14)	0.768 (0.70)
SOCLF	0.050** (2.14)		−0.002 (-0.28)	−2.829** (-2.55)
		0.001 (0.24)	0.007 (1.24)	−1.241 (-1.43)
TUWOY	0.061* (1.72)		−0.008 (-0.73)	−2.154 (-1.26)
		0.0003 (0.07)	0.005 (0.50)	−0.223 (-0.17)
EGY	0.020 (0.257)		0.002 (0.49)	−1.343 (-1.52)
		0.006*** (2.66)	−0.004 (-0.82)	−0.361 (-0.55)
VET	0.060*** (2.98)		−0.002 (-0.41)	−3.179*** (-3.24)
		0.010 (1.06)	0.003 (0.50)	−1.068 (-1.36)

Note: ***, **, and * denote statistical significance at the 1 %, 5 %, and 10 % levels, respectively.