UNIVERSITÀ DI PISA & SCUOLA SUPERIORE SANT’ANNA
Department of Economics and ManagementMaster of Science in Economics
The impact of the global oil market on the standards of living in
Russia as a vector for the transition to the green economy
Advisor: Prof. Simone D’Alessandro Candidate: Kseniia Chernova
Introduction ... 5
Chapter 1 Theoretical aspects of the global oil market ... 6
1.1Evolution of the global oil market ... 6
1.2 Features of the oil sector in different countries ... 11
Chapter 2 Background of the oil dependency ... 18
2.1 Problems and features of the Russian oil sector ... 18
2.2 Human development index (HDI) ... 22
2.3 Incarceration rate ... 24
Chapter 3 Econometrical model ... 27
3.1 Methodology ... 27
3.2. Model ... 28
3.3 Model quality assessment ... 32
3.3.1Classical assumption tests for the national level ... 32
3.3.2 Robustness of the model for the district level ... 36
Chapter 4 Results of the model ... 39
4.1 Review of the results on the country level model ... 39
4.2 Review of the results on the district level model ... 44
4.3 Comparison of the country results with the district results ... 48
Chapter 5 The transition to the green economy ... 51
5.1 Russian economy diversification ... 51
5.2 Alternative energy sources ... 52
Conclusion ... 56
Bibliography ... 58
Appendix ... 61
A Central district ... 61
B Far Eastern Federal District ... 61
C North Western Federal District ... 62
D North Caucasian Federal District ... 62
E Siberian Federal District ... 63
G Ural Federal District ... 64 H Volga Federal District ... 64
Nowadays there is a lot of discussions about raw material dependence of the Russian economy, especially oil dependence, and a deterioration in the standard of living of the population with a decline in crude oil prices on the global market. Maintaining and improving living standards, amid steadily declining oil prices in recent years, is a very difficult task for such large oil-exporting countries as Russia.
In this, we implement an econometric analysis to examine the influence of two factors on the human development index: crude oil prices and incarceration rates. Due to the large area of the research object (the area of the Russian Federation is 17.1 million square kilometres), we carried out an analysis at the country level and at the federal districts scale. To determine the robustness of the model we run several tests from classical assumptions and panel data analyses. For the research, data were used for the period 1999-2017.
In this paper we found that world oil prices have a positive correlation with the indicator of living standards of Russia (human development index) that is equal to 4.67%. As for the incarceration rate, we noticed a negative correlation of -22.72%. At the district level, the highest dependence of living standards on changes in crude oil prices was obtained in the Siberian District (4.83%). And on the issue of crime, the Far Eastern Federal District presented a high degree of dependence of the standard of living on the level of prisoning (-27.03%)
Keywords: standard of living, oil price volatility, human development index, Russian economy, oil dependence.
Russia is the largest country in the world with an area of 17.1 million square kilometres and a population of about 147.7 million people. Also, it is a very important and big exporter of crude oil and gas and the revenues from this sector take a very big part of Reserve Fund (01/05/2020: 168 billion $) that proves the fact that Russia is still being an oil-dependent country. On the other hand, the living standard is very low and according to the Federal State Statistic Service of Russia (Rosstat) nowadays there are 21 mln of people (13% of population) living in poverty1. Moreover, the period of
lockdown during the current pandemic crisis has shown that around 60% of all population do not have any savings. These facts led us to analyse the relationship between the standard of living of the population and changes in oil prices on the world market. This analysis will make it possible to evaluate the degree of this influence and empirically evaluate the real effect of the race for the world championship in energy trading on the social and economic well-being of the citizens of the Russian Federation.
In particular, we aim at assessing empirically the influence of oil dependency of Russia on the standard living of population. To this purpose, we propose the following steps:
1. Analysing the data of the global oil market, its features, and problems.
2. On the example of Russia, showing the impact of oil dependency on the living standards by estimating a regression function.
3. Evaluating the possibilities of the transition of the Russian economy to the green economy to escape so-called “resource course”.
Chapter 1 Theoretical aspects of the global oil market
1.1 Evolution of the global oil market
Nowadays oil is a strategically important, unique resource that seems impossible to imagine the world without it. The very first mention of this resource was in the 6th century BC (Before the birth of Christ) in China, when people did not know yet the true purpose and features of the oil, and used the “black gold” as a medicine, in the construction and road business, and even for embalming the dead. At that period, it was obtained through the bamboo pipes. And with each new generation, people realized that the oil can be used also as a fuel and heating agent. Therefore, the methods of oil extraction every year had been improved, so that it was possible to pump it in huge volumes.2
For the first time the oil was produced in an industrial way in the middle of the 19th century in Azerbaijan. But the main events at that time unfolded in the United States. The so-called "founding fathers" of the oil business, New York lawyer George Bissell and the head of the bank in New Haven, James Townsend, launched an extensive oil company. After the high investments they made in to the researching project of this natural resource, in 16 April of 1855 Professor Silliman reported that crude oil could be distilled into various hydrocarbon fractions and that one of the fractions was a very high-quality illuminating oil.3 So the owner of the company
increased the necessary funds and formed the Pennsylvania Rock Oil Company.
Silliman’s study was the turning point in the establishment of the petroleum industry. But at that time, people still did not fully understood that this resource will be soon not only a factor driving the progress, but also that reason of full-scale military operations.
2 https://www.ig.com/uk/commodities/oil/history-of-crude-oil-price#information-banner-dismiss. 3 http://archives.yalealumnimagazine.com/issues/2005_11/old_yale.html.
Only several years later it was invented by Edwin L. Drake the “salt drilling” method to extract the oil out. And immediately the whole financial world already knew about it. Many rich people immediately wanted to invest in this area and to conduct exploration work in other regions and lands. Thus, the years 1860-1870 can be called the period of oil transformation into a key and strategically important resource for all of humanity. Then, oil prices were at a little more than $2 per barrel, and transportation costs as a percentage of the cost of oil were very high (58-60%).4
This fact was one of the main success factors for “Standard oil”, which held a monopoly position in oil refining and bought up all oil pipelines in the upstream regions. John Rockefeller, the director of the company, in the first years after the formation of his company controlled around 10% of the global oil market. Moreover, at its best period “Standard Oil” controlled 85-90% of oil refineries. A key feature of the Rockefeller’ oil trading company is that he agreed with transport companies to provide all kinds of discounts on transportation. Oil consumption grew by amid a rapid increase in the population and industrial production in the USA (the latter's share in global industry grew from 7.2% in 1860 to 23.6% in 1900). From 1873-1875 to 1899 the use of oil for lighting needs increased from 1.6 million barrels to 12.7 million barrels, and to produce lubricants - from 0.2 million barrels to 2.4 million barrels.5
In the same years, in the Russian Empire the oil industry also had developed. However, there were the problem of the long-distance transportation. Then, by 2 brothers Nobel and Rothschild it was invented and launched the world's first oil tanker Zoroaster, which was able to transport huge volumes of oil across the Caspian Sea. As the result, by the early 1880s, the oil company from Baku produced more than 75% of the amount of kerosene in the Russian market, which almost completely replaced overseas oil from the domestic market.
4 Dolson J. Understanding oil and gas shows and seals in the search for hydrocarbons//Springer
International Publishing Switzerland, Switzerland. 2016. p. 114-123.
But the interest in world domination on the European oil market initiated the competition between these 2 main actors - Nobel / Rothschild and John Rockefeller companies. This struggle included both a decrease in export prices for products by both companies, and the formation of new subsidiary companies and branches in other countries. Also, the Rothschilds together with English company M. Samuel & Co (the old name of the company "Shell"), wanted to enter also on the Asian markets, that was done successfully. In 1895, Standard Oil proposed an agreement to distribute the oil market as following: 75% - American kerosene, 25% - Russian. Unfortunately, this agreement was not signed due to disagreement on the part of Russian government.
Together with the first car invention the demand for the gasoline significantly increased. This event stimulated the further development of the oil industry, expanded the global sales markets. Hence, in the middle of the 20th century the new oil companies appeared in the market: “Petroleos de Venezuela” in Venezuela, and the “gold chain” of oil fields (later Petroleos Mexicanos) was opened in Mexico.
Later, the Second World War period had a twofold effect on the oil corporations. First, some of them (mainly American) experienced an unprecedented growth, because they made an export of their products to the forefront countries. Second, in Russia (at that time the USSR) the leaders of the fascist wanted to capture the natural resources. Therefore, the enterprises located in the dangerous territory moved to the safer regions of the country. That caused the troubles with the fuel production for the cars and other equipment.
However, by the middle of the 20th century, in every industrialized country there were large oil refining companies:
Table 1.1: “Seven Sisters” Group of Companies
Anglo-Iranian Oil Company (now BP) GB
Royal Dutch Shell Netherland
Standard Oil Company of California (now Chevron) USA
Standard Oil Company of New Jersey (Esso, Exxon, now ExxonMobil) USA Standard Oil Company of New York (Socony, Mobil, now part of
Texaco (later merged into Chevron) USA6
Based on the Table 1.1 more than half of these companies are in America. In total “Seven Sisters” controlled more than 85% of the world oil market, hence, we can conclude that World War II seriously affected the oil industry of the Russian economy. While the global oil companies had constant development, the USSR oil sector was in stagnation.
However, after this period the oil industry received a new round of development. In 1950-1960 there were discovered new oil fields in the east countries such as Indonesia, Libya, Saudi Arabia, and Kuwait. It was the reason of the creation, in 1960, of the Organization of Oil Exporters (OPEC). Originally OPEC included Iran, Iraq, Saudi Arabia, Venezuela. But after the discovery of new oil fields, some other countries were also entered in OPEC (Angola, Indonesia, Libya, Algeria, UAE, Qatar, Gabon, Ecuador, and Nigeria). Nowadays OPEC has 13 countries-participants in this organization.
As it stated, OPEC's objective is to co-ordinate and unify petroleum policies among Member Countries, in order to secure fair and stable prices for petroleum producers; an efficient, economic and regular supply of petroleum to consuming nations; and a fair return on capital to those investing in the industry.7
The organization of oil exporters, as it developed, was forced to make important decisions in various situations, which sometimes even led to ethnic conflicts
or, on the contrary, drowned them. For example, as a result of the 18-day war between the Egyptian-Syrian union and Israel, the Western world and OPEC diverged in their views on this conflict (the West supported Israel, OPEC - Egypt and Syria) and OPEC was forced to declare an embargo on oil exports to the West countries including USA. This event was the starting point of the world's first oil crisis - oil prices climbed steadily upward (from $3 to $18 per barrel in 1979). Most Western countries at that time began to experience "oil starvation": there was an acute shortage of gasoline at gas stations and there were constant lines, the government even had to set limits on heating buildings and structures. All this was a proof of only one thing: developed countries were indisputably dependent on “black gold”. The US and Western European countries have realized the fact that OPEC is a serious competitor and it must be reckoned with.
The next oil crisis was in March 1974, after the OPEC embargo was lifted, the full-fledged oil supplies to Western countries started again. The problem was that the volume of sales markets was not enough to cover all the oil produced. And because of the imbalance between supply and demand was the oil price dropped. Hence, only after a long dialogue between all oil-producing countries and with the new pricing system and oil production quotas creation, it was possible to find the optimal solution.
The beginning of the 21st century was marked by new challenges for the oil industry. It is worth mentioning the crisis of 2008, when the cost of a barrel of oil fell from $145 to $40 in just a few months. Surprisingly, oil began its rapid decline just a couple of days after breaking through the historically maximum mark of $150 per barrel. The main factors that caused such a decline in prices were called a decrease in demand in Europe, the bankruptcy of Lehman Brothers, which was an absolute shock for the whole world. It triggered a panic, which also affected the oil market.8
In 2014, several key events took place, that were the reasons of the rapid decline of prices: lifting sanctions from Iran, the shale revolution in the USA, and
uncoordinated actions to prevent the crisis in OPEC countries. Iran has long sought to become an active exporter of oil to foreign markets, and the full-scale lifting of sanctions only contributed to this. The US shale revolution (shale oil production, successful search for solutions to reduce the cost of production and refining) also led to the emergence of another new potential exporter petroleum products to the world market. As a result, supply exceeded demand, and oil prices crawled down. The percentage drawdown depth for the period of this crisis was approximately 45%.
As for the countries of operations of OPEC countries, despite the slowdown in economic growth in Europe and Asia, which, of course, reduced the demand for hydrocarbon raw materials, they did not quickly respond to this inhibitory factor and did not immediately introduce quotas for oil production, which could settle the cost. Actually, they tried to fight for part of the market by further extraction of excess oil, which, in their opinion, would lead to deprivation of the profitability of competitors. But they failed, and they had to take the path of reducing production.
As an intermediate conclusion, we can say that over the entire period of its development, the oil industry has undergone tremendous changes. There have been various periods of both prosperity and decline, but the international community has always tried to support those countries that actively participated in the trade of this resource. With the advent of the Organization of Oil Exporters (OPEC), oil production became clearly regulated, that is, each country entering in it had to follow clearly the set of conditions when crisis prerequisites appear. In cases of the emergence of any conflicts between the countries, they tried to stop them in every possible way, since this could provoke new crises in the world oil sector. As for the main historical problem faced by the oil industry, this is the excess of supply over demand, which leads to a decrease in hydrocarbon prices and the emergence of crisis situations. The world market is facing it from time to time today.
1.2 Features of the oil sector in different countries
Nowadays in the world there are many countries that have a decent part of oil resources. Below we presented the graph with the countries with the highest oil
reserves in the world in 2018. Figure 1 shows that the leader over the past years is Venezuela (17.53% of the global oil reserves). But each of these countries has its own methods of extraction that primarily depends on the location of extraction: whether it is the peatlands, mountain subsoil or deep-sea reserves. On the other hand, the composition of the resource itself also affects its production since it can be either pure or shale oil or bitumen.
Russia ranks 6th in reserves in the total world, but the potential for exploring
new deposits both in Siberia and on the shelf gives a possibility in the future to become a leader in this sector.
Figure 1: Oil reserves by countries (2018), billion barrels.
Source: BP Statistical Review of World Energy 2019 | 68th edition
It is obvious to consider only the countries represented on this graph, since almost a third of the countries of the world possess a most of the oil reserves (68.09%), even in the small volumes. In addition, the countries with the largest oil reserves are the largest exporters as shown in Figures 1 and 2 It is not surprisingly to see also Russia in these lists because it is a country with huge reserves of the most famous mineral resources. The main destination for Russian exports is Europe. More than 70% of the volume of exported oil and gas goes there.
But not only the volume of reserves shows which country will be in leading positions in oil exports. One of the main factors is an accessibility to the main sales markets, because if the producing country is located at a certain territorial distance from the main consumers of oil and oil products, then the transportation costs will be impressively high.
Accordingly, the cost of the product will increase, and this will increase its price for foreign markets. All these factors have a negatively impact of oil competitiveness on the international area. Importing countries will be oriented towards the exporter country which price and costs are lower.
Figure 2: Largest oil exporters (2018), mln barrels per day.
Source: BP Statistical Review of World Energy 2019 | 68th edition
However, the companies that have certain costs could get some benefits from tax deductions. Naturally, in each country the oil multinational corporations are taxed in a different way, but as for Russian Federation, the tax system stimulates companies to extract and process oil with the higher costs to get certain tax benefits. As for the largest countries that are importing oil and oil products, the first places are taken by those countries that have a huge production capability, as well as the territory with many multinational corporations’ branches. That are the industrialized countries
(USA and European Union countries), as well as new industrial countries (Hong Kong, Singapore, Taiwan, South Korea), as well as China, India, and Japan.
According to Figure 3, the China and United States are the country’s leading in oil imports. In 2018 for the USA the biggest exporter was Canada (around 43% from the total imported oil), when for the China the main trading partners were Russia (13.52% of imported oil) and West Africa (13.46% of imported oil).9
Figure 3: The Top Oil Importers 2018, mln barrels per day.
Source: BP Statistical Review of World Energy 2019 | 68th edition
It is necessary to distinguish between the countries from OPEC (organizations of oil exporting countries) and other countries that are engaged in production and export. As we mention in the previous subchapter, Organization of Petroleum Exporting Countries - Special Cartel, created by countries with the impressive reserves of “black gold” with the aim of influencing the world oil market, primarily in the price policy. According to the OPEC Annual Statistical Bulletin (2019), OPEC member countries had around 79.4% (1,189.80 billion barrels) of the global oil reserves in 2018,
with the biggest part (64.5% of the OPEC total) in the Middle East. Moreover, looking at the Table 1.1 we can see that from the top 5 countries with the highest reserves – 4 countries are the members of the OPEC organisation, only Canada (that took the third place) is not included there.
First and probably the most influential country in the oil market is Saudi Arabia. In 2018, the country had 297.7 billion barrels of oil in reserve that is 17,21% of the global amount of oil. The economy of Saudi Arabia is based on the oil industry, it takes around 45% of the country's GDP, 75% of budget revenues and 90% of exports.10
A characteristic feature of the Arabian oil sector is the fact that oil extraction here has the lowest cost in the world ($ 8.98 / bbl.).11Another competitive advantage is that in
the countries of the Middle East there is no tax on the extraction of minerals. This factor is almost the most important in the sale of oil to the markets. Nevertheless, in Saudi Arabia there is a high-income tax, which varies from 50% to 65% depending on the amount of capital of companies operating in the territory of this state. Accordingly, the smaller the amount of capital, the higher the income tax rate. So, in the absence of a tax on mining, companies are still forced to give up more than half of their profits to the state. Over the past 30 years, the industrial sector has developed significantly (production of petrochemical products, fertilizers, steel, building materials, etc.). Although there are huge revenues from oil exports, the standard of living of Saudi Arabia is still being lower than many other developed countries.
Venezuela is the largest oil reserves country in the world now. The South American country, even though it has the largest reserves, is still forced to import oil from other oil-producing regions. The reason is the severity of oil produced in Venezuela. The “black gold” mined in the Orinoco River belt, in the largest mining area in South America, must be diluted with light oil to be exported later. Oil production in Venezuela has a huge impact on the country's economy. It accounts for more than 80% of all export revenues of the state. The budget of the country's oil
companies generates 50% of revenue12. Crude oil produced in Venezuela is exporting
to the USA, South America, the Caribbean, and Europe. Although Venezuela (PDVSA) is the world leader in terms of reserves (303.3 billion barrels), it is far from the volume of production (1.72 million barrels per day). All this is happening because oil deposits are in very remote areas and there is a lack of high-tech equipment for its production. All these factors prove the fact that the cost of extracted Venezuelan oil is quite high ($27.62 per barrel in 2018) and it is not very attractive for export. The main countries that import oil from Venezuela are USA, Mexico and Canada. Transport costs are minimal, and this block of countries has a certain advantage in the purchase of oil. The level of oil production in the country in recent years, however, has been steadily decreasing due to the lack of investment and lack of skilled labour, as well as under the pressure of US sanctions entered against the company PDVSA: if in 2014, oil production was 2.7 million barrels per day, at the beginning of 2016 - 2.6 million, then in 2018 - already 1.3 million barrels per day, by the end of 2018 this figure fell to 1.24 million barrels per day.13
The next country that takes the third place in the list of the countries with the biggest proven reserves of the oil in the world is Canada. The specific of the oil industry in Canada is such that the reserves are located mostly in the sands of Alberta location (97% from the total oil reserves). This type of oil from sands is more expensive than the other kinds since the process of extraction is much more complex because of the non-liquid consistence. So that it needs more technological equipment to extract it without a huge damage to the environment. Unfortunately, nowadays there is not such a high-tech machine to do it, otherwise the Canadian oil would be the most attractive one in the world due to the unique composition of the oil. If in the short-term Canada will find a way to decrease the costs of extraction, then it will increase the position not only in the reserve’s leadership, but also in the export rank. However, at this moment with the constantly decreasing prices per barrel for Canada it is
economically rational to decrease the volume of the extraction to do not sell the oil by the price much less than the cost for the production. In the second quarter of 2020 the oil sector faced the worst period in all history: according to the Oilprice article (21 April 2020), due to the price dropping on the oil market, many oil projects on the Alberta drastically cut the oil production. The total decrease was around 1,5 million barrel per day.14
Another big challenge for the Canadian oil market is the neighbourhood with USA. Since approximately 95% of the total export is going to the America (BP Statistical Review of World Energy, 2019), there is a possibility of border closure and some other sanctioned manipulations that could cause a huge pressure on the economy of Canada.15
14 https://oilprice.com/Energy/Energy-General/Oil-Price-Mayhem-Is-The-Market-Broken.html 15 Guliev I. A., Litvinyuk I. I. Topical issues of oil production, processing, and transportation in Canada
Chapter 2 Background of the oil dependency
2.1 Problems and features of the Russian oil sector
Russia is the richest country in the world in natural resources. It has a huge mineral and raw material base, including reserves of natural gas, coal, iron ore, some non-ferrous and rare metals, apatite, and others. And one of the most important industries after the gas is the oil sector.
The oil industry of Russia is the leading branch of the whole Russian industry, which includes the production, transportation and selling of the oil itself and oil products. According to Figure 4, since 1996 we have seen a continuous increase in oil production and export, but only a constant level of consumption. Obviously, the fluctuations in the period 1991 -1995 were caused by the consequences of the collapse of the Soviet Union that is why the production level drastically decreased almost by 3 million barrels per day.
But in 2018 Russian oil companies extracted 11.4 million barrels per day that are 12.3% of the total capacity of oil production in the world. Only Saudi Arabia with 12.3 million barrels per day (13% from the world daily production) and the USA with 15.3 million barrels (16.2%) have the oil extraction level higher than in Russia.
However, in 2020 the oil market had serious challenges. Since Russia is a country-member of OPEC+ organisation, it determined the decline in oil production in Russia by the terms of agreement. According to it, in May-July 2020, Russia must produce no more than 8.5 million barrels per day of oil. Then, by the end of 2020, it will be raised to 8.99 million barrels per day, and in the period from January 1, 2021 to April 30, 2022 - to 9.495 million barrels per day. As a result, the Ministry of Energy of the Russian Federation (Minenergo) reported that in July 2020, the level of Russia's fulfilment of obligations under the OPEC+ agreement was reached by 99%.
Accordingly, oil export from Russia decreased as well and in the period of January-July 2020 it dropped by 8.8%.16
According to the BP (Statistical Review of World Energy, 2019), approximately 56% of the all crude oil exports in Russia in 2018 was transferred to Europe and 26% - to China. This fact illustrates the high dependence of Russian oil incomes from the European oil importers. Therefore, with a sharp decline in oil import in Europe for any political or geopolitical reasons, it can become a serious problem for replenishing the budget of the Russian Federation. 17
Figure 4: Oil production and export in Russia, mln tons / year.
Source: BP Statistical Review of World Energy 2019 | 68th edition
The cost of oil production is one of the most important indicator to choose the fields for development. The price spent on the production of 1 barrel of oil depends on the complexity of the extraction of raw materials and the level of technology used during operation. The issue of extraction of raw materials in some regions is especially acute in the context of falling world oil prices.
17 Mitrova, Tatiana, and Sarah O. Ladislaw. New Russian Oil and Gas Export Strategy. Center for Strategic and International Studies (CSIS), 2016, pp. 39–48.
0,00 2,00 4,00 6,00 8,00 10,00 12,00 14,00 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Production of oil, mln. barrels per day Consumption of oil, mln. barrels per day Export of oil, mln. barrels per day
It is important to mention the location of oil production. It should be noted that the West Siberian Basin (Krasnoyarsk Territory, Khanty-Mansi Autonomous Okrug) became the main region for oil production in 2018 (Minenergo annual report, 2019): it accounted for 58% of all oil production. In second place (23%) was the Volga-Ural basin (it includes Bashkortostan, Tatarstan, Orenburg region), in third - East Siberian (Krasnoyarsk Territory, Irkutsk Region, Yakutia). On the territory of the Timan-Pechora basin (Komi, Khanty-Mansi Autonomous Okrug), 6% of the total volume of production was obtained, 3% was mined from the bowels of the Okhotsk (Sakhalin, Kamchatka, Magadan Region), and 1% - on the lands of the North Caucasian basin (Kalmykia, Adygea, Karachay-Cherkessia, etc.). So, we can conclude that the biggest part of the oil extraction is referred to the Siberian region.
Oil production in Russia is carried out by the following production companies: Rosneft, LUKOIL, Gazprom Neft, Surgutneftegaz, Tatneft and RussNeft:
• Rosneft produced 285.5 million tons of oil in 2018 (+ 1.3% in compared to 2017).
• The result of Gazprom Neft is 92.9 million tons (+ 3.5%).
• LUKOIL's indicators - 85.61 million tonnes - remained at the 2017 level (85.592 million tonnes).
• Surgutneftegaz produced 60.9 million tons, Tatneft's share was 29.53 million tons, and RussNeft's share was 7.6 million tons (Wood Mackenzie, 2019).
In such corporations as Gazprom Neft and Rosneft, the Russian Federation holds a controlling stake. Hence, the influence of government on the oil sector of the economy is very high, and although in recent years the share of oil revenues significantly decreased, this sector is still playing a dominant value in the redistribution of income.
Oil and gas revenues have been taking a very large share of the Russian federal budget at all times. In 2018, the share of revenues from the oil and gas industry was 46.35%, and in 2019 - 39.25%, in the period 2012-2014 this value was greater than 50%
of the entire federal budget. However, these revenues include not only revenues from exports, but also revenues from the mineral extraction tax (MET) on oil and gas, export customs duties on energy carriers and the tax on additional income from the extraction of hydrocarbons (annual report of the Ministry of Finance of the Russian Federation, 2019). This fact explains the independency of the gasoline price in Russia from the global oil market prices, since it depends on the 80% from the tax system of the country (The Federal Tax Service of Russia, 2019). Thus, given the decline in energy prices as well as the OPEC + requirement for a decrease in oil production, it can be assumed that revenues from the oil and gas industry in the structure of the Russian budget in subsequent years will be decreased.
Figure 5: Correlation between oil price and exchange rate in Russia.
Source: BP Statistical Review of World Energy 2019 | 68th edition
The Russian economy can be attributed to the so-called "commodity curse" or commodity dependence. This is explained by the high share of the oil and gas industry in the structure of GDP over a long period. Statistical data confirm this fact: for the period 2017-2019, mining amounted to 9.8%, 11.9% and 11.3% of GDP, respectively (Rosstat, 2020).18 Moreover, oil prices and Russian currency are correlating over the
last years (Figure 5). It is one of the biggest indicators of the oil dependency since it is
18 Nesvetailova, Anastasia. The Offshore Nexus, Sanctions, and the Russian Crisis. Istituto Affari
Internazionali (IAI), 2015. 0,00 20,00 40,00 60,00 80,00 100,00 120,00 140,00 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
Correlation between oil price and exchange rate in Russia
presenting the negative effect for the whole economy during the periods of oil price fluctuations. Following the tendency of the oil market to the fluctuations and to overall decrease of the oil prices, we may conclude that the rouble costs is going to decrease as well that will cause a negative impact on the economy and on the living standard of the population.
2.2 Human development index (HDI)
The standard of living is the degree of satisfaction of the material and spiritual needs of people with by the goods and services used per unit of time. 19 The concept
of "quality of life" is broader than the standard of living (material security) so it is not right to use them as synonyms. It is because the quality of life also includes such factors as health status, life expectancy, environmental conditions, food, household comfort, social environment, satisfaction of cultural and spiritual needs, psychological comfort, etc.
Establishing the different approaches to determine the standard of living it is important to note that in Russia there are two indicators that are usually used: the minimum cost of living and the minimum consumer budget. Also, average GDP per capita and Gross National Income (GNI) can be chosen as proxies of living standard. But according to the literature, there are many other indicators that can be used. For example, Simon Kuznets (1965) in his research supports the idea of using real GDP per capita as an indicator of well-being.20 Also, there is the genuine progress indicator
(GPI) that takes into account the influence of environmental pollution (John Talberth, Clifford Cobb, and Noah Slattery, 2007).21 In addition, the index of social health,
19 Report No. 3 of the United Nations Research Institute for Social Development (UNRISD) (Geneva,
1996, p. 8).
20 Kuznets, S. (1965), Towards a Theory of Economic Growth, W.W. Norton and Co., New York. 21 Talberth, J., Cobb, C., Slattery, N. (2007), The Genuine Progress Indicator 2006: A Tool for
representing the social well-being based on 16 different factors, can be considered as an indicator to estimate the standard of living (Marque-Luisa Miringoff, 1999).22
Less common indicator is the Human Development Index (HDI) (was created by Mahbub ul Haq in 1990) (Figure 6), which is much broader than GNI or GDP per capita and it considers, in addition to economic indicators, the next components (Human development reports, 2020):
- Life expectancy (estimates longevity),
- The literacy rate of the population (the average number of years spent on education) and school life expectancy,
- Living standards measured in terms of GNI per capita at purchasing power parity (PPP) in US dollars (Baru, S.,1998).
Nowadays there is a lot of discussions about the rationality and correctness of the use of this indicator. Basically, criticism is aimed at the weight of the influence of each of the three factors on the overall value of the HDI indicators. Since when comparing countries, there may be significant differences in terms of some indicators of human development, while remaining similar in some other indicators.23 That is, a
country with a fairly low ranking in terms of economic well-being may rank highly in terms of literacy or life expectancy, or both, compared to other countries in similar economic conditions. But when the analyses are proceeding within one country, this problem is missing.
The value of the index is in the range from 0 to 1, respectively, the value closer to 1 demonstrates a higher standard of living of the population of the considered geographical entity, when closure to 0 – low standard of living. For reference, the last 20 years a leader in the value of the human development index is Norway (only in 2007 and 2008, Iceland was the first place).
22 Miringoff, M. and Miringoff, M.S. (1999), The Social Health of the Nation, Oxford University Press,
New York, NY.
Figure 6: Human Development structure.
Source: UNITED NATIONS DEVELOPMENT PROGRAMME (UNDP), 2020
Following the specific of Russian Federation, we have concluded that the traditional standard of living indicator – GDP per capita - does not describe the whole situation in the country. Therefore, we have seen that the HDI represents more information about the standard of living in Russia taking into the account not only economical, but also social and political aspects of the life.
2.3 Incarceration rate
One of the main factor that has an impact on the standard of living of population is the crime level. The crime rate in the country depends on the various spheres of social development, such as: economy, politics, culture, social sphere, moral and psychological environment, etc. An analysis of the current state of crime in Russia makes it possible to draw a conclusion about the need for a more dynamic development of these areas of public life. At the same time, high values of the state of crime, which is an indicator of the development of society, have an extremely negative impact on the areas listed above. The increase in the number of prisoners as social diseases is a signal not only for the judicial system, but also for the whole economy (Kirchhoff 2010).24
In recent years, the system of social crime prevention has practically disappeared in the country. This greatly weakens the ability to deter less dangerous
24 Kirchhoff, S.M. (2010) “Economic impacts of prison growth”, Washington, DC: Congressional
crimes, which are the "impetus" for serious crimes.25 In addition, there is practically
no system of acquittals in Russia. Only in 2019 the share of acquittals in Russian courts increased for the first time since 2013 (Report of the Judicial Department of the Supreme Court, 2020). In total, 2,256 acquittals were issued in 2019 (0.36%). In 2018, there were 2,083 (0.3%), in 2017 - 2,233 (0.3%). For comparison, in 2002, 51.2% of the accused were acquitted in India, 13.7% in Turkey, and 1.8% in France.26
After analysing statistical data and relying on individual sociological studies, we concluded that the most significant factors in the manifestation of crime in Russia are:
1. High level of poverty of the population. Poverty creates a so-called “marginalized environment” consisting of anti-social elements such as vagabonds, homeless people, street children, etc.
2. Economic inequality of the population. It significantly increases the negative impact of poverty, generates social disharmony in society, leads to an increase in protest moods, motivates a significant part of the violations of the law that have taken place.
3. High unemployment rate. The high level of unemployment is the reason for the ill-being of the population, which provokes an increase in theft and other crimes. In 2018 the unemployment rate reached 4.9% (that is 7.08 million of people). According to the World Bank report (July 2020) in May 2020 the rate of unemployment increased to the level of 6.1%.
4. Corruption. Corruption is one of the most striking reasons for the increase in the crime rate in the country. According to a study carried out by the research institute of the Academy of the Prosecutor General's Office of the Russian Federation, the
25 Korsantia A.A., Maksimenko I.V. Main reasons determination of crime in the Russian Federation //
Law and safety. 2009. No. 2. S. 100-105.
majority of Russian citizens are seriously concerned about the effectiveness of the judicial system and law enforcement agencies in the fight against corruption.
5. Alcoholism and drug addiction of a significant part of the Russian population. In 2016, 13.5% of the crimes solved were committed by alcoholic intoxication. Moreover, in the structure of especially grave crimes in 23.8% of cases the perpetrators were in alcoholic intoxication. The share of alcoholic crime is even higher for certain categories of criminal acts. According to official statistics, 61% of murders and 50% of attempted murders, 49% of cases of intentional infliction of grievous bodily harm, 57% of rapes, 25% of attacks were committed by offenders under the alcohol (Rosstat, 2017).
Accordingly, we may conclude that the incarceration rate presents the real social situation in the country and it is a very important factor that has an impact on the standard of living of the population. Therefore, in the next chapter we estimate the impact of the crude oil price together with the incarceration rate on the standard of living in Russia that is presented by Human Development Index (HDI).
Chapter 3 Econometrical model
The main objective of this study is to define the influence of oil dependency on the standard living of population in the countries-exporters. The focus is on the Russian Federation as one of the main participants of the global oil market.
In order to evaluate the relationship between price fluctuations in the oil market and the chosen standards of living indicator in Russia, we will estimate a regression model with two independent variables “Crude oil price” and “Incarceration rate” and a dependent variable “Human development index”. Following the size factor of the considering country we will firstly run the econometric model for the whole country and then will compare the obtained results with the results in each federal district of Russia. We believe that it is necessary to do this kind of analyses because there is a remarkable heterogeneity of standard of living of the population across the country, and an analysis of data only at the country level would not be informative (problems of the capitals and the periphery). Therefore, we will divide our research in 2 parts:
• the country level analysis (the statistical database of Russia);
• the district level analysis (the model will be considered for all 8 federal districts separately with their human development indexes).
This approach of analyses will provide a clear evidence of the oil dependency existence and the level of its impact on the standards of living. Moreover, it will describe the social, political, and economic issues associated with oil prices fluctuations.
This study is based on the annual data from 1999 to 2017 from the database of the official governmental statistical sources such as:
- The Russian Federal State Statistics Service (Rosstat27);
- Reports of the Ministry of Finance of the Russian federation;
- Federal Penitentiary Service of Russia statistics;
- Analytical Centre under the Government of Russian Federation;
- Publications and reports of the Ministry of energy of Russian Federation.
According to the literature and statistical sources, we estimate a multiple regression model. The multiple regression model is represented as:
where =1, 2, …, n and 𝜀𝑖 random variable of the error term.
Theoretical concepts of analysis and calculation of the necessary procedures can be facilitated by the matrix describing the regression. Let us designate:
• - is a column matrix, or vector, of values dependent variable with n size; • 𝑋 = ( 1 𝑥11 𝑥12 … 𝑥1𝑝 1 𝑥21 𝑥12 … 𝑥2𝑝 … 1 … 𝑥𝑛1 … … … 𝑥𝑛2 … 𝑥𝑛𝑝
) is matrix of explanatory variables values of the
• is a column matrix or vector of the measurement parameters (p+1);
• is a column matrix, or vector, error (disturbance term) variables of size n.
Therefore, the model (1) can be defined in the matrices form as:
where = and . Changes in the value of the dependent
variable with an increase in the explanatory variable can be judged by the regression coefficient .
To evaluate the parameters of the model, we used the Ordinary Least Squares method. Ordinary Least Squares (OLS) - the basic method of regression analysis identifies the model parameters that minimize the sum of the standard deviations between the observed values and the calculated ones.
In the process of research, we went through many articles estimating the problem of oil dependency impact on the living standards. And we realised that the authors mainly use similar regressors in their models. For instance, Khaleefah Jaber Al Sabah, Ralph Palliam and Athmar Al Salem (2016) identified investment growth, inflation, price per barrel and incarceration rate as the independent variables and real GDP growth rate as an dependent variable for the example of the country Kuwait; also, Theodosios Perifanis and Athanasios Dagoumas (2017) estimated GDP correlation with industrial production index, government expenditure, crude oil prices and unemployment rate for the Russian example. In the article of Helmi Hamdi and Rashid Sbia (2013) as the regressors were chosen the oil revenues, total government spending and economic growth. And from all of them we selected those independent factors, based on which it was possible to get the most significant result of the analyses.
In the beginning our model included next regressors: investment growth, inflation, crude oil price and incarceration rate. However, after the first tests we made, we noticed a high correlation of 2 variables that were erased from the consideration (more details in the Chapter 4). So, for the final model we chose Human development index as a dependent variable, and as independent variables - Crude oil prices and Incarceration rate. The model can be expressed in the next formula:
30 • HDI: human development index,
• COP: crude oil price, • IR: incarceration rate, • i: year, =1, 2…16,
• random component of the model.
However, all variables are in logarithms, implying that their coefficients are also their elasticities. This approach will help to determine the level of impact in the relative terms (percentages) of each independent variable on the Human development index. The equation of the regression model is expressed:
log 𝐻𝐷𝐼𝑖≈ log 𝐶𝑂𝑃𝑖 + log 𝐼𝑅𝑖 (4) Although, for the second part of the analyses we have chosen the Panel data since there is not only time series data but also cross-sectional one. This approach increases the size of the sample under consideration, which provides a more efficient estimation of the parameters of the regression model. Accordingly, it takes into the account every individual object of the analyses over the 18 years and make the result more efficient. Especially on the example of Russia this approach is very useful, since we will consider the country that is divided by 85 regions that are grouped into 8 big districts. Therefore, we will be able to analyse some numbers of regions across 1 specific district that in the end can help us to compare the data between all these 8 districts and make a conclusion about the most dependent from the oil prices districts in all Russia. So that the formula representing the impact of oil dependency on the HDI in each district is expressed:
log 𝐻𝐷𝐼𝑚,𝑖≈ log 𝐶𝑂𝑃𝑚,𝑖+ log 𝐼𝑅𝑚,𝑖 (5)
where m=1, 2, … n regions across 1 district, and i=1, 2, …16 the year of consideration.
According to the whole this thesis we want to consider next hypotheses. Based on the studied sources, we can argue that there are many different theories on the topic of resource dependence. One of the seminal paper facing this issue we recall "Natural
resource abundance and economic growth" (1995), in which the authors J. Sachs and A. Warner point to a negative relationship between the level of a country's natural resources and its economic growth. As an example, the 17th-century Netherlands and Switzerland, poor in natural resources, have outstripped resource-rich Russia and Spain in terms of economic growth.
In an article by E. Papyrakis and R. Gerlagh (2004), the authors argue that resource abundance has a negative effect through such variables as a high level of corruption, an excessively closed economy, a low level of education and investment. T. Friedman and M. Ross (2001) found that the presence of a large amount of natural resources causes a low probability of a democratic regime and freedoms in the country. This phenomenon is called “the basic law of oil policy”.
Also V.M. Polterovich, V.V. Popov and A.S. Tonis (2007) believe that countries with developed economic and political institutions demonstrate positive economic growth in the presence of a large amount of natural resources, and low growth is characteristic of countries with undeveloped democracies (in addition, they may be threatened by further destruction of institutions).
Therefore, literature describes the problem of oil dependency and its negative impact on the economic growth. Since the originality of this work is precisely the analysis of the impact of the oil market on the standard of living of the population, we can consider the value of the human development index because of economic growth. And then the hypotheses proposed below have theoretical support. From this reason we will hypothesize:
H1: There is a positive relationship between HDI and COP;
H2: There is a negative relationship between HDI and IR;
3.3 Quality assessment
To determine the robustness of the model we decided to choose the tests from the classical assumptions for the first part of the analyses (country level) and Hausman test for the panel data analyses (district level).
3.3.1 Classical assumption tests for the national level Multicollinearity
Multicollinearity is the correlation of independent variables, which makes it
difficult to assess and analyse the overall result. When the explanatory variables are correlated with each other, multicollinearity is said to occur.
In practice, multicollinearity can cause overfitting of the model, leading to an incorrect result. In addition, redundant coefficients increase the complexity of the model, which means that its training time increases. The multicollinearity of factors is also bad because the mathematical regression model contains redundant variables, which means:28
- the interpretation of the parameters of multiple regression as the values of the action of factors becomes complicated, the parameters of the regression become meaningless and other variables should be considered;
- parameter estimates are unreliable – there are large standard errors that are changing with the volume of observations, which makes the regression model unsuitable for forecasting.
To test our model for the multicollinearity we chose the variance inflation factor
(VIF) method. VIF is a measure of multicollinearity that allows to estimate the increase
in variance due to the linear dependence of the factor from the others.
where - coefficient of determination of the linear dependence of the factor from the rest. According to the results, we can say that there is a multicollinearity if 5 < VIF < 10 and strong multicollinearity if VIF > 10. That means that our model does not have multicollinearity if the VIF value of every independent variable is less than 5.
Serial correlation (Autocorrelation)29
Autocorrelation is the relationship of successive elements of a time or spatial
data series. In econometric studies, these situations often arise when the variance of the residuals is constant, but their covariance is observed. Autocorrelation of residuals is most often observed when the econometric model is built based on time series. If there is a correlation between successive values of some independent variable, then there will be a correlation between successive residual values. Autocorrelation can be also a result from erroneous specification of the econometric model. In addition, the presence of autocorrelation of residuals may mean that a new explanatory variable needs to be introduced into the model.
The most popular criterion for detecting first-order autocorrelation is the
Durbin-Watson test and the calculation.
The criterion (5) is defined as the ratio of the sum of the squares of the differences of successive residual values to the sum of the squares of the residuals. The value of the Durbin-Watson criterion is indicated along with the coefficient of determination, the values of the t- and F-tests. The following relationship holds between the Durbin-Watson test and the autocorrelation coefficient ( ) of the first-order residuals (Motulsky HJ, Christopoulos A., 2004):
Thus, if there is a complete positive autocorrelation in the residuals and = 1, then d = 0. If there is a complete negative autocorrelation in the residuals, then = -1 and, therefore, d = 4. If there is no autocorrelation of the residuals, then = 0 and d = 2. Therefore, 0 ≤ d ≤ 4.
However, there are several significant restrictions on the application of the Durbin – Watson test.
• it is inapplicable to models that include lagged values as independent variables, i.e., to autoregressive models.
• the methodology for calculating the Durbin - Watson criterion is aimed only at identifying the autocorrelation of first-order residuals. For checking residuals for higher order autocorrelation, other methods should be used.
• the Durbin – Watson test gives reliable results only for large samples.
Therefore, for our model we will also test another criterion - Breusch-Godfrey
, t=1…, n (7)
(where are the regression residuals obtained by the usual least squares method)
The test is based on the following idea: if there is a correlation between neighboring observations, then it is normal to expect that in the equation (7), the coefficient will be significantly different from zero.
The advantage of the Breusch – Godfrey test over the Durbin – Watson test contains a zone of uncertainty for the values of the statistic d. Another advantage of the test is the possibility of generalization: the number of regressors can include not only residuals with a lag of 1, but also with a lag of 2, 3, etc., which makes it possible to reveal a correlation not only between neighboring, but also between more distant observations.
It is important to denote that for both these autocorrelation tests there are the following hypotheses:
(there is not serial correlation),
(there is a serial correlation).
Heteroscedasticity is a characteristic of cross-sectional samples. It consists of the difference in the variances of the regression errors. The presence of heteroscedasticity is a serious problem in the application of the regression analysis, including analysis of variance, as this can lead to the invalidity of statistical tests that assume that errors of the model are uncorrelated and homogeneous, therefore, their variances do not change. Heteroscedasticity is present if the following property is not satisfied:
∀ i=1, …, n (8)
Consequences of heteroscedasticity:
• estimates of standard errors of regression coefficients are biased; • OLS estimates of regression coefficients are ineffective;
• t-statistics of regression coefficients are inadequate.
For the heteroscedasticity check we will use in our model next zero and alternative hypothesis.
As for the testing heteroscedasticity we used Breusch-Pagan test. It is one of the statistical tests for checking the presence of heteroscedasticity of random errors in the regression model. It is used if there is reason to believe that the variance of random
errors may depend on a certain set of variables. Moreover, this test checks the linear dependence of the variance of random errors on a certain set of variables. The test includes several steps that need to be done.
It is assumed that the variance of the random error depends on several independent variables:
After calculating the OLS estimation of the regression coefficients, we need to find residuals and the square of residuals , then to calculate the coefficient of
determination for regression and
then calculate .
As a result, if is greater than the critical value of chi-square statistic for the m degrees of freedom, there is heteroscedasticity. And we need to reject null hypothesis (homoskedasticity) if p-value is below chosen α level.
Once the tests have been completed to assess the significance of our model, we can begin in the next Chapter a detailed analysis of the data obtained.
3.3.2 Robustness of the model for the district level
Panel data is applied when the database contains statistical information about the same set of objects for several consecutive periods of time. When there is no heterogeneity between the sample objects, we can run a regression based on the pooled regression. The formula is the same as for the ordinary least square regression model (Seddigi, H, 2000). The robustness of the model is expressed in the p-value results and the adjusted R square value to see the percentage of the data that our model covered.
There are several models to run estimations with panel data. Therefore, to choose the best model representing the most significant results we considered fixed effect model and random effects model. Panel data models with fixed effects
section FE): an approach based on the introduction of individual effects, which allows to get rid of the influence of an unobservable variable (constant over time) and obtain unbiased estimates of parameters. The specific of this model is that it provides guaranteed unbiased and consistent estimates.
However, if unobservable factors do not correlate with regressors, to obtain more efficient estimates, it is possible to consider a panel data model with random effects (RE - random effects): it is assumed that missing variables are one of the components of errors and that the individual effects are randomly distributed across the cross-sectional units and in order to capture the individual effects, the regression model is specified with an intercept term representing an overall constant term (Seddighi, 2000).
In addition, to check whether the pooled OLS model is better fitted than random effect model we used Lagrange multiplier test (LM test). The null hypothesis is stated that there is no significant difference between cross-sectional individuals, so that in the case when the test presents the p-value less than 5% it is more efficient to choose the random effect. Moreover, to control whether pooled OLS model is better than fixed effect model we used panel F-test, that has the same approach in interpretation of the results as in LM test.
However, when there is a problem in the comparison of fixed effects and random effects model in our analyses we used the Hausman method to test the estimators. The Hausman test is a test to identify the errors in the specification of an econometric model based on a comparison of two different estimators of the model's parameters. Comparative models must have the following properties:
- under the null hypothesis, both estimation models are consistent for the “true parameters” of the model (those that correspond to the data generation process),
- under an alternative hypothesis, the estimators should have different probability limits.
The Hausman test involves comparing two different estimates of the parameters of a panel data regression model.
1) If the panel estimates of "random effects" and "fixed effects" are consistent under the assumption that the model is set correctly and that the regressors do not depend on "individual effects" then the difference between the estimates of random effects and fixed effects will be small. Thus, the estimation of random effects is more efficient.
2) If the assumption about random effects is incorrect, but otherwise the model is set correctly, then the estimate of the fixed effects remains consistent, but the estimate of the random effects is inconsistent. Consequently, the difference between the estimates of random effects and fixed effects can be large. Thus, the fixed effects method will be more efficient.
Chapter 4 Results of the model
In this chapter we present the main results. As we mentioned before, the work is separated in two parts: country level and district level.
4.1 Review of the results on the country level model
In order to calculate the impact of oil prices on the human development index and check if there is oil dependency, in the beginning we defined a regression model with 4 independent variables that are Investment, Inflation, Crude oil Price and Incarceration rate as we mentioned in the previous chapters. However, after running the correlation analysis (Table 4.1) we realised that there is a high correlation for 2 variables – investment and inflation. Hence, we have found high correlation between investments and incarceration rate (-86.24%), investment and inflation (-54.15%), investment and crude oil price (62.02%), inflation and crude oil price (-56.31%), inflation and incarceration rate (61.22%). These 2 variables have a correlation grater 50%, so that there would be a strong relationship among these regressors so that the effect of one regressor on the dependent variable depends on another regressor too.
Table 4.1. Correlation test for 4 independent variables
Based on the literature, Nancy Morrow-Howell (1994)30 suggests checking the
model with highly correlated variables for the existence of multicollinearity. This helps us to select the main regressors. As an output we have seen that variance inflation factor (VIF) is less than 10 for all 4 considering variables (Table 4.2), therefore there is not multicollinearity and as the next step will be running the OLS model.
30 Nancy Morrow-Howell. The M word: Multicollinearity in multiple regression //Social Work
Table 4.2. Multicollinearity for 4 variables
As it was mentioned in the Chapter 3, we apply a log-log model. The equation of the regression model is log HDI ≈ log (Investment) +log (Inflation) +log (Crude_oil_Price) + log (Incarceration_rate). According to the results presented in Table 4.3 it appears that in the regression model with investment and inflation presences the main factor of this study “Crude_oil_Prices” is insignificant. Even though the whole model has good evidences of significance with p-value very closed to 0 and 𝑅2 = 96.86% and Adjusted =95.96% that means that our model fits the
data very well and it explains the variability of the HDI due to the all regressors. But the value of 𝑅2 is very close to 1 that is explained by the presence of multicollinearity
of some variables.31 Based on that we decided to adjust our model by erasing inflation
and investment from consideration.
Table 4.3 Regression on the relationship between HDI and 4 variables
A new regression model with only two independent variables shows the impact of the oil market prices on the human development index of Russian
31 T. Distefano, M. Tuninetti, F. Laio, L. Ridolfi. Tools for reconstructing the bilateral trade network: a
population with the factor of incarceration rate. Hence, the main formula now is presented as log HDI ≈ log (Crude_oil_Price) + log (Incarceration_rate).
Relatively to the result (Table 4.4) we have seen that the selected regressors are significant and the coefficient of determination shows that the goodness of it is 83.96% (corrected for the number of regressors) that is less than in the model with 4 variables but still has a high value. Moreover, it includes only statistically significant variables that proves that the data is fitted well. Crude oil price as well as Incarceration rate has the p-value less than 1% and the whole model has p-value less than 5%. Therefore, we reject the null hypothesis (all the coefficients of the regressors are equal to zero) and we can conclude that our model is more informative than the one with only intercepts.
Table 4.4 Regression on the relationship between HDI and 2 variables.
Moreover, to determine the robustness of our model we calculated varience inflation factor (VIF) obtaining that for both independent variable the value is 1.15 that is less than 10, so that we can conclude that there is no multicollinearity in the model.
After that it was important to check the model for the serial correlation (autocorrelation) presence. As it was described in the Chapter 3, we run 2 different tests:
• Durbin-Watson test – it has shown that the Durbin-Watson criteria is 1.0148. According to the discription of the test this value shows the positive autocorrelation since the value is not equal to zero, but it refers to the interval between 0 and 2. And with p-value=0.001552 is very close to zero we have to reject null hypothesis and accept the presence of the serial correlation.
Table 4.5 Durbin-Watson test for serial correlation
• Breusch-Godfrey test: the advantage of this test is that there is a possibility to check the autocorrelation not only for the 1st but also for the 2nd order residuals.
This approach is presenting more detailed data and giving more evidences to have confidence in the obtained results because the test is taking into account the dynamic of the generating process of the noise to test for correlation of any order in the residuals. As we have seen in both tests the criterias is between 3 and 4. Following this results together with the p-value that is greater than 5% we should reject null hyphothesis that there is serial correlation. Finally, according to the Breusch-Godfrey test wereject the null Hypothesis of aurocorrelation.
Table 4.6 Breusch-Godfrey test for serial correlation
To sum up, based on the above robustness checks, we keep the results from the Breuch-Godfrey test because it presents clearer and more accurate output. Moreover, since Durbin-Watson test is recommended for the big data analyses, we cannot use it for our model that has only 18 years of consideration and 2 independent variables. One of the most important assumption for the regression model analyses is heteroscedasticy control. The main idea of this step is to determine if there is constant