Performance of international oil companies and national oil companies: a comparative analysis

However, Wolf and Pollitt (2008) write that NOCs' performance could be improved by some partial privatization. It happened in countries where partial privatizations were supported by governments' measures to increase competitive position for the NOCs and keep those performance improvements, like it happened in Norway, where the government controlled the industry by having shares in two oil national oil companies - Statoil and Norsk Hydro, and by so-called “State Direct Financial Interest” - SDFI, which was though used by Statoil on behalf of the government. So, when Statoil was partly privatized in 2001, Norwegian government also sold some SDFI assets to Statoil thus to increase its competitive position on international market (Wolf 2009).

For the upstream segment, the most vital issue is the ownership of the ground beneath the surface (subsoil). For example, the US government allows private ownership for subsoil (Mommer 2002). If the subsoil is publicly owned, government could either give a monopoly to one certain company, or to introduce a licensing system to let several companies participate. Allocation mechanism and fiscal terms are important characteristics of a licensing mechanism since they define access and state part of profits. Government can naturally use the licensing mechanism as a tool for controlling and building industry structure - just by making decisions about the frequency and magnitude of the licensing agreements (auction or negotiated deal), by establishing certain financial incentives or by setting the rules like obligatory government involvement. If the government is willing to take part in the oil industry directly - either as the monopolist or as a license provider, it is possible to do that in the form of corporation or public bureau (Horn 1995). State-owned enterprises are definitely met more often in the oil industry - because of the fact that consumption is easy to measure, marginal costs cannot be neglected and demand price elasticity is high enough (Peltzman 1989) - which provides the opportunity to introduce private shareholders through the privatization (or partial privatization). Even without being privatized, NOCs can attract private partners either as service providers or as joint-venture partners at the exploration and production levels (Wolf 2009).

However, the difference between IOCs and NOCs is not always obvious because some companies that used to be NOCs (like BP or Total) went through full or partial privatization, so that after that they work somewhere similar to the companies that were always private. Also there are companies like Statoil or Petrobras that despite being partially privatized sometimes demonstrate work in governments' interests (Jaffe and Soligo 2007).

All of the above mentioned government structures, apart from subsoil ownership, could also be applied to downstream segment. Since taxes in both upstream and downstream are very high and contain sector-specific taxes, it is also an important mechanism of rent redistribution and government participation - even in the relationships of state and its NOCs, since taxes significantly define the degree of the NOCs' independence (Wolf 2009). In other words, since state-owned and privately-owned oil companies are not mutually exclusive, no contradictions arise if they work together. Oil and gas projects are very capital intensive, have long accomplishing time and are indeed risky (Stevens 2005). Figure below represents the rise of the global capital expenditure spending of publicly traded companies in 2010-2014. It is seen that for those 5 years capital expenditure were growing and the most of the capital expenditure were spent on the exploration and production activities - more than 70% each year throughout 2010-2014.

Figure 4 Global 2010-2014 capital expenditure spending trends of public companies, US$ billions.

Source: AlixPartners, 2015.

Specifically in the upstream oil companies usually arrange a partnership with each other to share risks and financing requirements and to strengthen each other's skills. Since the industry become mature, the reasons for risk diversification could get even more obvious. Although there is no threat of running out of hydrocarbons in the near future (Lynch 2004; Watkins 2006; Greene, Hopson and Li 2006), the most of the onshore and shallow-water offshore fields are exhausting and new projects, like deep-water offshore or remote fields with severe climate and absence of infrastructure, will become more technically challenging and expensive to operate. NOCs - IOCs relationships often were unfriendly, and some of those problems could easily be followed back to the nationalization discussions in the 1970s. More recently, even when OPEC' members - NOCs suggested allowing foreign participation in upstream projects, the IOCs frequently were going down due to unsatisfactory returns. Considering future challenges, a much more tight cooperation could be not just preferred, but required (Marcel 2006).

1.4 Previous studies on comparison of international and national oil companies

Comparison of public and private firms can be influenced by structural differences of the companies, such as operational activity, non-commercial goals, or the oil density, which all must be taken into account. There were two basic research designs made in order to find whether private or state-owned oil companies have better performance:

* research comparing samples of state-owned companies with samples of (different) private companies;

* research targeted at privatization over time, made as case study, single-industry or single-country study or cross-industry cross-country study (Wolf 2009).

Both designs are closely tied: static supremacy of private companies is required to the success of privatization, but it is not enough, because privatization is a dynamic process and it could contain significant changes apart from the ownership, like political, regulatory and organizational changes (Villalonga 2000).

While a considerable amount of studies give evidence supporting private ownership, Villalonga (2000) argues that many studies are not trustworthy as the comparisons are weakened by methodological difficulties. So, choosing suitable measurement variables is one problem, when another problem is that there are interconnected and non-separable results of ownership, competition and regulation (Vickers and Yarrow 1988). Moreover, suitable groups of companies for the comparison are hard to find - a lot of countries and industries have only a small number of or even no suitable firms with both types of ownership. Also, ownership itself might be sophisticated - due to the existing system that has both political and performance objectives (Megginson and Netter 2001).

Since private international oil companies face difficulties in gaining access to big oil and gas reserves, NOCs' control over oil resources is likely to grow. So it is important to know if NOCs would behave the same as IOCs both in times peace and calamity and crises and shocks happening in the global oil market (Hartley and Medlock 2008).

In order to describe systematic behavior variations of a NOC and an IOC, Hartley and Medlock (2008) built a model of NOC working and developing that demonstrates NOC trying to maximize revenue inflow from the production of such an exhaustible resource as oil at the same time being influenced by the side goals of politicians. NOCs' behavior reflects all of the events happened with companies at any moment of time and under any circumstances. However, NOCs average performance in the long run is likely to demonstrate effects of being owned and controlled by the state government rather than shareholders (Hartley and Medlock 2008).

Their model shows two basic features of national oil company:

* NOC must give evidence of extracting an exhaustible resource change comparing current and future periods.

* NOC, comparing with IOC, has different owners, or principals, who have different goals, so that the previously mentioned principal-agent paradigm is critical to justification of the extended number of objectives of NOC, comparing to IOC (Hartley and Medlock 2008).

They proposed a general goal for the NOC that can demonstrate a combination of the goals of managers and politicians. Specifically, the NOC must compare profitability (giving money to the Treasury) and benefits to consumers and suppliers of various resources, for example, labor (Hartley and Medlock 2008).

Boardman and Vining (1989) looked at the financial performance and efficiency of the 500 largest non-American production companies in 1983. They discovered that state-owned and mixed (having both government and private shareholders) companies are less profitable and efficient than their private competitors and those companies with mixed ownership do not outperform state-owned enterprises. Another research of Boardman and Vining (1992) devoted to studying Canadian state-owned and private companies, shows the same results as well, but opposite to their first study, firms with mixed ownership are found to be more profitable than state-owned enterprises. Dewenter and Malatesta (2001), using the same design as Boardman and Vining (1989), compare the differences in profitability, labor intensity, and debt levels of private and public enterprises of the Fortune Magazine 500 largest international companies reported in 1975, 1985 and 1995. Looking at the firm's location, industry, size and business-cycle influence, they provide serious evidence that private firms are much more profitable, less labor-intensive and demonstrate lower levels of financial leverage.

Opposite to those results, Caves and Christensen (1980) and Martin and Parker (1995) suggest that there is no obvious supremacy of private companies. Instead, those researches claim that the key factor of company's efficiency is market competition and that therefore public and private companies are equally efficient if operating under competitive conditions (Wolf 2009).

There are not so many comprehensive empirical analyses, mostly because of absence of data on national oil companies, which are usually characterized as nontransparent (a significant number of NOCs are not publicly listed, thus, there is no available information on their performance). However, Al-Obaidan and Scully (1992) explore the efficiency and performance of 44 international private and state-owned oil companies (observed between 1976 and 1982), using Aigner-Chu frontier, stochastic frontier analysis (SFA) and Gamma frontier analysis. Looking at multinationalism and operational integration, they show that state-owned companies are just 61% to 65% as efficient as private companies.

Two other researches are based on the analysis of the Petroleum Intelligence Weekly ranking of the biggest international oil and gas companies. Eller, Hartley and Medlock (2011) use nonparametric data envelopment analysis (DEA) as well as parametric SFA on a sample of 78 companies for the period 2002-2004, testing the theoretical predictions developed in Hartley and Medlock (2008). They use revenue as output and number of employees, oil reserves and gas reserves as inputs to calculate an average DEA technical efficiency score, which for national oil companies is equal to 0.28, compared to an average score for the five largest private enterprises of 0.73 and a sample average of 0.45 for the other firms. Victor (2007) also analyses the relative efficiency of national oil companies and private oil companies in turning reserves into products and revenues, using roughly 90 companies' data from the year 2004, but uses a univariate linear regression for the analysis. She demonstrates that the largest private oil firms are almost one-third better than NOCs at turning reserves into natural form of output (products), and tend to produce much more revenue per unit of output (Wolf 2009).

Jaffe and Soligo (2007) were exploring how IOCs and NOCs behave during the period of high oil prices and whether they invest a lot in oil exploration or not. They found out that five biggest international oil companies (ExxonMobil, Royal Dutch Shell, BP, Chevron and ConocoPhillips) had oil production levels reducing since mid-1990s. At the same time, shares of national oil companies that were actively participating and investing in oil exploration abroad were growing faster than shares of the biggest international oil companies. Also, consolidations of big IOCs in the 1990s did not directly resulted in successful development of big and serious oil projects and costs reduction of these projects.

Wolf's (2009) research, as well as Eller, Hartley and Medlock (2011), is based on the data from the Petroleum Intelligence Weekly "World's 50 Top Oil Companies". The research includes multivariate regression analyses with various dependent variables and two different estimators. He uses a panel model with company-specific intercepts and total estimator that does not count company-specific heterogeneity. He states that the company-specific intercept in the fixed effects estimator recognizes every time-invariant variable which influence dependent variable. Consequently companies keeping the same owners during the observance period are considered the same way despite the share of state or private investors' ownership. However, because total estimator lets estimate ownership's influence, it is impossible to control company-specific unobserved variables. Wolf's (2009) and Eller, Hartley and Medlock's (2011) SFA analyses are multivariate panel regression analyses that have special structure of the error terms. The easiest type of SFA supposes that error terms have time-invariant firm-specific components, taken from nonnegative distribution that shows deviations from the efficiency frontier (Hartley and Medlock 2013).

As Hartley and Medlock (2013) mention, there is also element of the error that shows measurement error and is supposed to have a symmetric distribution. On the other hand, it is supposed by the standard random effects panel estimator that error terms are symmetric and efficiency deviations are neglected. Wolf's (2009) study applies more structure on the equation in SFA, right side variables in SFA represent production function. However, all the equations that Wolf (2009) looks at do not have any structure interpretation, thus suitable estimating equation is hard to define. When extra assumptions are irrelevant, applying more structure might change the inferences. This is why non-parametric DEA should be used. DEA does not make specific assumptions about the basic production function (Hartley and Medlock 2013).

Hartley and Medlock (2013) had to use balanced panel (every company must have every used variable for every year), similar to Eller, Hartley and Medlock's (2011) research and different from Wolf's (2009) research, which affects the length of time period used.

A more detailed investigation of ownership change within the oil and gas industry is Wolf and Pollitt (2008), a time-series analysis of the performance and efficiency impact of all available privatizations since 1977. They explore 60 privatizations of 28 NOCs from 20 countries in 1977-2004. A considerable part of those privatizations were made in form of sequence offerings because states' shares were decreased through multiple actions. They were comparing mean results three years prior to shares were sold with mean results' three years after the shares were sold. They discovered that privatization lead to greater profitability, better operating efficiency, increase in production and decrease in employees' number. Wolf and Politt (2008) saw that first step of privatizations increased average performance on all indicators, but statistical significance of the result was achieved only in return on sales or assets and employee per unit of assets at the 10% level. Performance trend was positive and significant in all indicators. However, performance trend after privatization was not that positive as it was before privatization for seven out of ten of them, but significance presented only in return on sales or assets). Generally, privatization leads to improvement of oil companies' performance which starts only when expecting the following sale of shares and slows down after the shares are sold (Hartley and Medlock 2013).

1.5 Research hypothesis statement

To conclude, ownership structure of the oil and gas producing companies, or more specifically, the fact that national oil companies are controlled by the government and international oil companies are not, bring serious differences into the behavior of the international and national oil companies.

Because of state control, national oil companies could be either private or publicly traded. International oil companies usually are publicly traded because it helps them to attract additional funds for developing capital intensive projects all over the world. Besides, going public works for the company's image and transparency, thus increasing the attractiveness and bringing new partners (Draho 2004).

Also, IOC, which is not controlled by the government, has a main goal of maximizing revenue and company's value - the most important things to private shareholders. At the same time, having state government as a major shareholder influences the activity in a certain way. State, while pursuing its non-commercial goals, uses the NOC and its resources for helping achieving those goals. NOC is used as an instrument of achieving foreign policy goals, for example, for creating alliances with another national oil companies from the countries - strategic partners or for manipulating the market to make some countries perform in a certain way. Also, NOC keeps an excessive level of employment thus making real state commitments in the sphere of providing jobs and reducing unemployment for the nation. Besides, NOC invests a lot in various domestic projects, helping the government to develop the infrastructure within the country, which in turn also gives many people opportunities to get a job as well as stimulates the economic development of the state. Also, state makes NOC to supply the domestic market with low (subsidized) prices for the oil, gas and various oil products as well as paying taxes and rents to the country's budget thus contributing to the nation's wealth redistribution (Pirog 2007; Hartley and Medlock 2008).

Assessing the performance of NOCs and IOCs is better by using revenue, since it represents all the activities and efforts companies undertake and their success as well. Hartley and Medlock (2008) highlight that revenue is a key factor and indicator for state-owned and private enterprises. They also claim that since politicians want to increase their support, they would push NOCs to sell gasoline on home country market at subsidized prices so that it is impossible to stay 100% sure that consequences of such a move could be measured by physical production. In addition to this, because of the fact that most of the oil companies produce not only crude oil or gasoline but also other oil products, an obvious way to calculate those outputs would be to use their prices or generally to use revenue as the measure of the output (Eller, Hartley and Medlock 2011).

Generally, NOCs can sell more, since they have dominating position on the home market, thus, generating more revenue. At the same time, NOCs usually have bigger reserves, so they can also produce and sell more than IOCs. Demonstrating the excessive employment rate is possible though company's headcount, which is higher in NOCs due to providing excessive level of employment. And finally, investing a lot in various projects that NOCs have (both commercial and mon-commercial) would mean that they are to have bigger capital expenditure. However, these are the absolute values at a certain moment of time, which do not always demonstrate the behavior of NOCs and IOCs over time. Generally, previous researches are focused on using absolute values of indicators (like revenue) or ratios (like revenue per employee) on a sample containing both national and international companies. Previous researches use either cross-sectional or time-series data (over the long period of time - for several years and in some research for several decades.

Nevertheless, the performance could also be measured over the short period of time, so that the behavior of the companies would demonstrate their goals in the short run. Consequently, in order to explore the behavior of NOCs and IOCs in the short term, this research uses the difference of the indicators' values over one year, thus showing the results achieved over one year of operation and behavior of the companies in the short run. As it was previously mentioned, it is important to know how IOCs and NOCs behave in both stable and prosperous and shock times. This research aims at exploring IOCs and NOCs behavior during the significant drop of oil and natural gas prices happened in 2014, after almost half a decade of high oil prices, focusing on the companies' progress over the year of 2014 through several indicators values' changes. Also, since previous studies explored NOCs which were both publicly traded (like Statoil) and private (like Saudi Aramco), current research focuses on comparison of only publicly traded national and international oil companies.

More specifically, revenue change is defined as the difference of annual revenues in 2014 and 2013. As for the reserves, there are 3 types of them: proved reserves, possible proved reserves, probable reserves and possible reserves. The first type represents the reserves that could be recoverable under existing economic and political conditions using existing technology of reserves with 90% or more confidence. Probable reserves include proven reserves and also those that are not proven but have probability of 50% and more of being technically and commercially producible. Possible reserves include probable reserves and the reserves that have the probability of less than 50% of being technically and commercially producible. Because of the highest probability of being extracted and produced, proven reserves change (2013 to 2014) is chosen for this research. They include oil and gas reserves together, since, as it was earlier mentioned industry players explore and produce both oil and gas (Van Vactor 2010). As for the production, it includes both oil and gas annual production combined. Capital expenditure and headcount change from 2013 to 2014 are defined as the difference of the indicators' values over that year.

However, before starting the comparison of one year's performance of IOCs and NOCs, it is necessary to explore the influence of the chosen performance indicators - reserves change, production change, headcount change and capital expenditure change - on the output indicator of revenue change, in order to analyze whether they are significant for revenue generating and which of those influences revenue change the most during the oil and natural gas prices decrease.

Firstly, the amount of revenue generated by company depends on the amount of sales level and, consequently, on the production level. Next, since the oil and gas are limited resources, the reserves company has are crucial for supporting (or increasing) the production level. Moreover, the company should spend significant funds on capital expenditure for increasing the level of production and adding new reserves, which in turn affect the revenue amount generated. Finally, company needs to hire new people so that the operation expansion could be possible, that, however, affect the operation expansion not as much as the extensive use of machinery does. Supposedly, revenue change is affected by the production change the most because they are directly connected, then the second place takes the reserves change, because the oil and gas are limited and production (so as revenue) depends on the reserves available, the third place takes the capital expenditure, because production and reserves increase could be done through high capex spending and, finally, the headcount change influences the revenue change the least, because the revenue increase is more likely to be achieved by an extensive use of machinery rather than employees.

So, here is the first group of hypotheses:

1a: production change has the highest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.

1b: capital expenditure change has the second highest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.

1c: reserves change has the third highest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.

1d: headcount change has the lowest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.

After estimating the relationships between four indicators' changes and revenue change, the comparison of performance of two groups of companies - NOCs and IOCs is conducted. For this analysis the same performance indicators' changes are chosen: revenue change, production change, proven reserves change, headcount change and capital expenditure change, because they demonstrate how the state control over the NOC affects its behavior and let to compare it with IOC's behavior.

Because the 2014 was the year of significant oil and natural gas prices decrease, both NOCs and IOCs tried to generate as much revenue as possible, IOCs because of keeping their investor attractiveness and share prices, NOCs because of government's extensive plans for the non-commercial goals (decline in oil and natural gas prices leads to generating less revenue and, consequently, paying less taxes). Since the prices decrease was very big, both types of companies supposedly faced revenue decrease, although both IOCs and NOCs supposedly increased the production in attempts not to let revenue fall. In turn, increased production combined with the cancellation of very costly projects resulted in decrease of reserves, so that both IOCs and NOCs supposedly reduced the reserves.

The resources prices decrease made the companies to review their participation in various projects: very costly project had to be suspended, so they gave no return. Also, because of being afraid that prices would continue to fall, both IOCs and NOCs supposedly increased the capital expenditures for some other projects in order to speed up the development of them so that the oil and gas from them became cheaper to extract and thus in future companies would have more resources available. Besides, increase in production was another reason for the increased spending on capital expenditures, which was done by both types of companies as well.

Generating less revenue leads to cutting costs, so that IOCs supposedly decreased the headcount. NOCs supposedly did the same, because it is more important for the government to let NOCs earn more and thus pay more taxes, which then might be spent on supporting various government activities (as well as subsidies for the people) rather than allow NOCs having extensive level of employment, earn less and, consequently, pay less taxes.

So here is the second group of hypotheses:

2a: there has been a decrease in revenue for both IOCs and NOCs during oil and natural gas prices decline.

2b: there has been an increase in production for both IOCs and NOCs during oil and natural gas prices decline.

2c: there has been a decrease in reserves for both IOCs and NOCs during oil and natural gas prices decline.

2d: there has been an increase in capital expenditure for both IOCs and NOCs during oil and natural gas prices decline.

2e: there has been a decrease in headcount for both IOCs and NOCs during oil and natural gas prices decline.

2. EMPIRICAL RESEARCH DESIGN

2.1 Methodology and sample

In order to estimate the influence of the production change, proven reserves change, capital expenditure change and headcount change on the revenue change, the multiple regression analysis is used.

The regression model:

Revenue change = b0 + b1*production change + b2*reserves change + b3*headcount change + b4*capex change.

In order to compare the performance of NOCs and IOCs for the year of operation, using revenue change, production change, proven reserves change, capital expenditure change and headcount change the t-tests are used.

This research considers national oil companies as the ones publicly traded and having state share of 50,1% and more. International oil companies are considered as publicly traded oil companies with government share of 0-50%, thus not being fully controlled by the state, which also operate in several countries (at least two countries of operation).

The companies for the sample are taken from the 2015 Platts Top 250 Global Energy Companies rankings, made by the Platts, a financial division of Mcgraw Hill. The ranking includes 250 world's biggest publicly traded energy industry companies and based on the undisclosed formula that considers such companies performance indicators as asset worth, revenues, profits and return on invested capital. The full list of companies is included in Appendix 1.

Table 6 Sample selection process

Characteristics	Number of companies
Biggest global publicly traded energy companies	250
Companies from the oil and gas industry	115
Companies operating in upstream segment Of them:	46
- NOCs	16
- IOCs (operating in at least 2 countries)	30

2.2 Variables

Revenue change - represents the percentage by which the revenue from the end of 2013 reporting year has changed to the end of 2014 reporting year (increased or decreased or stood the same) during the company's operation for a year. It is measured in percentage and calculated the following way:

Revenue change: (2014 revenue - 2013 revenue) / 2013 revenue * 100.

2013 and 2014 revenue represents the company's annual revenues and is measured in US dollars. If company reports the revenue in other currency, it is converted to the US dollars using the exchange rate on the date on which it is stated in the report (the last day of the year).

For the regression analysis revenue change is used as a dependent variable.

Production change - represents the percentage by which the annual production from the end of 2013 reporting year has changed to the end of 2014 reporting year (increased or decreased or stood the same) during the company's operation for a year. It is measured in percentage and calculated the following way:

Production change: (2014 annual production - 2013 annual production) / 2013 annual production * 100.

2013 and 2014 annual production combines the production of oil, natural gas, oil and products (if company produces any) and is measured in barrels of oil equivalent (BOE) on the date on which it is stated in the report (the last day of the year).

For the regression analysis production change is used as an independent variable.

Reserves change - represents the percentage by which the proven reserves from the end of 2013 reporting year has changed to the end of 2014 reporting year (increased or decreased or stood the same) during the company's operation for a year. It is measured in percentage and calculated the following way:

Reserves change: (2014 proven reserves - 2013 proven reserves) / 2013 proven reserves * 100.

2013 and 2014 proven reserves combines the proven reserves of oil and natural gas and is measured in barrels of oil equivalent (BOE) on the date on which the they are stated in the report (last day of the year).

For the regression analysis reserves change is used as an independent variable.

Headcount change - represents the percentage by which the headcount from the end of 2013 reporting year has changed to the end of 2014 reporting year (increased or decreased or stood the same) during the company's operation for a year. It is measured in percentage and calculated the following way:

Headcount change: (2014 headcount - 2013 headcount) / 2013 headcount * 100.

2013 and 2014 headcount represents the company's number of employees and is measured in persons on the date on which it is stated in the report (last day of the year).

For the regression analysis capex change is used as an independent variable.

Capital expenditure change (capex change) - represents the percentage by which the capital expenditure from the end of 2013 reporting year has changed to the end of 2014 reporting year (increased or decreased or stood the same) during the company's operation for a year. It is measured in percentage and calculated the following way:

Capex change: (2014 capital expenditure - 2013 capital expenditure) / 2013 capital expenditure * 100.

2013 and 2014 capital expenditure represents the company's annual capital expenditure and is measured in US dollars. If company reports the capital expenditure in other currency, it is converted to the US dollars using the exchange rate on the date on which it is stated in the report (last day of the year).

For the regression analysis capex change is used as an independent variable.

All the indicators' values for 2013 and 2014 are taken from the companies' annual reports.

2.3 Descriptive statistics

Here are presented the descriptive statistics for the regression analysis only. The descriptive statistics for the t-tests are presented in the next paragraph.

Table 7 Descriptive statistics for the regression analysis

Table above represents the descriptive statistics for the regression variables. It is seen that dependent variable has a negative mean of about -7. The mean for reserves change and capex change are also negative, while means for the production change and headcount change are positive.

As it is seen from the table below, no high correlations of variables are spotted.

Table 8 Pearson correlation for the regression analysis variables

3. EMPIRICAL RESEARCH RESULTS

3.1 Results and their interpretation

First of all, the regression analysis is done to explore the degree of influence of production change, reserves change, capital expenditure change and headcount change on the revenue change.

Regression analysis:

Table 9 Results of the regression analysis, dependent variable - revenue change, %

Variable	Coefficient	Significance
Production change, %	0,619	0,028
Reserves change, %	-0,791	0,041
Headcount change, %	0,279	0,103
Capex change,%	0,229	0,041

Table above represents the summarized results of the regression analysis, and more detailed results are included in Appendix 2.

Since the significance of all the independent variables but the headcount change are less than 0,05, it means that production change, reserves change and capex change significantly contribute to predicting revenue change. Unfortunately, the headcount change have significance is more than 0,05, which means that headcount change is insignificant for predicting the revenue change.

T-Tests analyses:

T-Tests analyses were made to compare the performance of the IOCs and NOCs on the 5 chosen variables. For conducting the comparisons, NOCs were given a code of 0, and IOCs - a code of 1. More detailed t-tests results are included in Appendix 3.

Table 10 Results of the t-test for the revenue change, %

It is seen that the mean for the revenue change of the NOCs is -10,57%, while the mean for the revenue change of the IOCs is -6,27%. Because p-value of this t-test is more than 0.05, it can be concluded that there has been no significant difference between the means.

Table 11 Results of the t-test for the production change, %

It is seen that the mean for the revenue change of the NOCs is 3,81%, while the mean for the revenue change of the IOCs is 2,73%. Because p-value of this t-test is more than 0.05 it can be concluded that there has been no significant difference between the means.

Table 12 Results of the t-test for the reserves change, %

It is seen that the mean for the revenue change of the NOCs is -1,89%, while the mean for the revenue change of the IOCs is -1,79%. Because p-value of this t-test is more than 0.05, it can be concluded that there has been no significant difference between the means.

Table 13 Results of the t-test for the headcount change, %

It is seen that the mean for the revenue change of the NOCs is 8,95%, while the mean for the revenue change of the IOCs is -2,3%. Because p-value of this t-test is less than 0.05, it can be concluded that there has a difference between the means.

Table 14 Results of the t-test for the capex change, %

It is seen that the mean for the revenue change of the NOCs is -18,04%, while the mean for the revenue change of the IOCs is 1,55%. Because p-value of this t-test is less than 0.05, it can be concluded that there has a difference between the means.

3.2 Results discussion

The table below represents the results of hypotheses testing from regression and t-tests analyses:

Table 15 Hypotheses testing results

Hypothesis	Result: confirmed or rejected
1a: production change has the highest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.	Rejected
1b: capital expenditure change has the second highest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.	Rejected
1c: reserves change has the third highest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.	Rejected
1d: headcount change has the lowest degree of influence on the revenue change of the oil and gas producing company among production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline.	Rejected
2a: there has been a decrease in revenue for both IOCs and NOCs during oil and natural gas prices decline.	Confirmed
2b: there has been an increase in production for both IOCs and NOCs during oil and natural gas prices decline.	Confirmed
2c: there has been a decrease in reserves for both IOCs and NOCs during oil and natural gas prices decline.	Confirmed
2d: there has been an increase in capital expenditure for both IOCs and NOCs during oil and natural gas prices decline.	Rejected
2e: there has been a decrease in headcount for both IOCs and NOCs during oil and natural gas prices decline.	Rejected

The regression analysis was made to explore how the revenue change is influenced by production change, reserves change, headcount change and capital expenditure change during oil and natural gas prices decline. Based on the regression analysis results, table below summarizes the effects of the independent variables on the revenue change.

Table 16 Independent variables' effects on the dependent variable (revenue change, %)

Variable	Effect on the revenue change
Production change, %	As production changes by 1%, the revenue changes by 0,619%.
Reserves change, %	As reserves change by 1%, the revenue changes by 0,791%.
Headcount change, %	Headcount change does not significantly influence revenue change.
Capex change,%	As capital expenditure changes by 1%, the revenue changes by 0,229%.

The model is significant and has the R Square equal to 0,248, which means that these independent variables can explain 24,8% of the variation in the revenue change. In other words, revenue change is influenced by many factors, and there are some other factors not included in this model that influence the other 75,2% of the revenue change variation, but, obviously, the main factor influencing the revenue change over time for both IOCs and NOCs most likely is the resources price (not included in this model).

At the same time, it is seen that among those four independent variables, reserves change influences revenue change the most, despite being expected to have the third highest degree of influence on the revenue change during oil and natural gas prices decline. Production change turned out to have the second highest degree of influence on the revenue change of the oil and gas producing company, while the capital expenditures has the third highest degree of influence on the revenue change. Headcount change does not significantly influence revenue change, which means that hiring or firing employees does not bring effect to revenue change as the production change, reserves change or capex change do.

Thus, hypotheses 1a, 1b, 1c and 1d are all rejected.

However, despite the fact that headcount change does not have any significant influence on the revenue change, it is still taken for the comparison of IOCs and NOCs performance, since it demonstrates how the governments controlling NOCs behave towards the companies' personnel during the market shocks, such as resources prices decline.

T-tests analyses were made to compare NOCs and IOCs' one year of operating performance results during oil and natural gas prices decline, using revenue change, production change, reserves change, headcount change and capital expenditure change.

The results of the t-test on comparison of the revenue change of NOCs and IOCs showed that there has been a decrease in revenue for both NOCs and IOCs, so that hypothesis 2a is confirmed. It happened because of the significant resources prices decrease, so that companies could not even keep the revenue at the same level. NOCs faced the reduction of revenue by 10,6% and IOCs by 6,3%.

The results of the t-test on comparison of the production change of NOCs and IOCs showed that there has been an increase in production for both NOCs and IOCs, so that hypothesis 2b is confirmed. As it was earlier mentioned, companies tried to avoid revenue decrease by increasing the production: production increase for NOCs was 3,8% and for IOCs was 2,7%.

The results of the t-test on comparison of the reserves change of NOCs and IOCs showed that, there has been a decrease in reserves for both NOCs and IOCs, so that hypothesis 2c is confirmed. It is most likely caused by, as it was mentioned earlier, the increase in production and cancellation of very costly projects. NOCs' reserves decreased by 1,9%, while IOCs' reserves decreased by 0,2%.

The results of the t-test on comparison of the headcount change of NOCs and IOCs showed that there has been an increase in headcount for NOCs by almost 9% and a decrease in headcount for IOCs by 2,3%, so that hypothesis 2d is rejected. Such opposite moves could be explained by the government policy for providing jobs during market shock, thus supporting the nation, so that the suggestions about NOCs reducing headcount are wrong.

The results of the t-test on comparison of the capital expenditures change of NOCs and IOCs showed that there has been a decrease in capex for NOCs by impressive 18% and an increase in capex for IOCs by 1,5%, so that hypothesis 2e is rejected. NOCs dramatically reduced the capex probably because of extracting resources from the already developed oil and gas fields, thus increasing the revenue and money flows to the government. IOCs did the opposite thing: they increased the capex, trying to develop capital intensive projects as much as possible until prices do not decline even more.

Also, such serious reduction in revenue and capital expenditures for NOCs compared to IOCs might be explained by falling exchange rates of national currencies, which were caused by oil and gas prices decrease: usually, countries with NOCs have the oil industry as one of the main and important for the domestic economies, so that national currencies might follow the oil price, as, for example, Russian ruble was falling in 2014 (Bloomberg 2014).

3.3 Managerial implications

Results of this research demonstrate the behavior of both NOCs and IOCs during markets shock, more specifically, oil and natural gas prices decline, which happened in 2014.

The results show how the NOCs and IOCs behave during this market shock and crisis so that it is possible to use these results for the further forecasting and analysis of the NOCs and IOCs own behavior in future in order to get prepared in case market shock happens. It is important because NOCs control the majority of oil and gas resources on the planet, so that their behavior is really important for the global market, since it might affect the global oil and natural gas prices.

Also, the results of this research might be used for the comparison with NOCs and IOCs behavior during stable market situation, thus showing what actions companies from both types undertake in both situations and what is different. It might help to get prepared for the future shocks, which are likely to happen because both oil and natural gas are limited resources are they are depleting as the time goes by.

Also, since NOCs and IOCs sometimes work in cooperation with each other in various projects, the results might help each party to get prepared for the actions of another party (partners) during market shock and thus reduce possible risks that might arise and use it for planning their own future steps in case such situation happens.

At the same time, NOCs and IOC compete on the same market (global market), so that the results of this research might help each party to forecast the actions of another party (competitors) during market crisis and thus again reduce possible risks and use it for planning their own steps in case such situation happens.

3.4 Research limitations

Current research focuses on comparing only publicly traded national and international oil companies. Besides, only companies operating in upstream segment are chosen for the research (so that companies operating only in downstream or midstream are not included in this research), because the variables used for the analysis are connected with exploration and production activities. Data from 46 companies used, 30 of which are IOCs and 16 are NOCs (and IOCs operate in at least 2 countries).

The companies' performance is compared using only five indicators: revenue change, production change, reserves change, capital expenditure change and headcount change. Many other existing performance indicators are not taken for this research.

The research uses methods of regression and t-tests analyses, first one for exploring the influence of four independent variables - production change, reserves change, capital expenditure change and headcount change - on the dependent variable (revenue change). The second method is used for comparison of NOCs and IOCs performance over one year of operation.

The research focuses on companies' performance of only one year - 2014, thus demonstrating behavior of the NOCs and IOCs in the short run. Also, the research focuses only on comparing NOCs and IOCs performance during market shock - decline in oil and natural gas prices occurred in 2014, thus leaving out of scope NOCs and IOCs behavior during stable market situations, like high resources prices.

As for the further research, it is possible to explore NOCs and IOCs behavior during stable market situation and then compare it with the behavior during market crisis. Also, the research might explore the longer period of time (more than one year) and have more companies in the analysis: not only publicly traded but also the private ones. In addition to this, comparison might be done using other methods, other performance indicators and a bigger number of indicators.

REFERENCES

Addison, Velda. 2015. "Forecasting More E&P Spending Cuts." Oil and Gas Investor, 10 September. Accessed 20 February 2016. http://www.oilandgasinvestor.com/forecasting-more-ep-spending-cuts-817831

AlixPartners. 2015."Capital Productivity in the Oil and Gas Industry." Accessed 21 February 2016. http://www.alixpartners.com/en/Publications/AllArticles/tabid/635/articleType/ArticleView/articleId/1559/Capital-Productivity-in-the-Oil-and-Gas-Industry.aspx

Al-Mazeedi, Wael. 1992. "Privatizing the National Oil Companies in the Gulf." Energy Policy 20, no. 10: 983-94.

Al-Obaidan, Abdullah M., and Gerald W. Scully. 1992. "Efficiency Differences between Private and State-owned Enterprises in the International Petroleum Industry." Applied Economics 24, no. 2: 237-46.

Boardman, Anthony E., and Aidan R. Vining. 1989. "Ownership and Performance in Competitive Environments: A Comparison of the Performance of Private, Mixed, and State-owned Enterprises." The Journal of Law & Economics 32, no. 1: 1-33.

Boardman, Anthony E., and Aidan R. Vining. 1992. "Ownership versus Competition: Efficiency in Public Enterprise." Public Choice 73, no. 2: 205-39.

Bozec, Richard, Mohamed Dia, and Gaetan Breton. 2006, "Ownership-efficiency Relationship and the Measurement Selection Bias." Accounting & Finance 46, no. 5: 733-54.

Caves, Douglas W., and Laurits R. Christensen. 1980. "The Relative Efficiency of Public and Private Firms in a Competitive Environment: The Case of Canadian Railroads." The Journal of Political Economy 88, no. 5: 958-76.

CNN Money. 2001. "Phillips, Conoco Set Merger." Accessed 20 February 2016. http://money.cnn.com/2001/11/19/deals/phillips_conoco

Dewenter, Kathryn L., and Paul H. Malatesta. 2001. "State-Owned and Privately Owned Firms: An Empirical Analysis of Profitability, Leverage, and Labor Intensity." American Economic Review 91, no. 1: 320-34.

Draho, Jason. 2004. The IPO Decision: Why and How Companies Go Public. Cheltenham, UK: Edward Elgar Pub.

Eisenhardt, Kathleen M. 1989. "Agency Theory: An Assessment and Review." Academy of Management Review 14, no. 1: 57-74.

...