Former Soviet Union oil production and GDP decline: Granger causality and the multi-cycle Hubbert curve

Available online at www.sciencedirect.com

Energy Economics 30 (2008) 271 – 289 www.elsevier.com/locate/eneco

Former Soviet Union oil production and GDP decline: Granger causality and the multi-cycle Hubbert curve Douglas B. Reynolds a,⁎, Marek Kolodziej b a

Oil and Energy Economics, School of Management, University of Alaska Fairbanks, United States b Center for Energy and Environmental Studies, Boston University, United States Received 6 April 2005; accepted 15 May 2006 Available online 30 June 2006

Abstract This paper discusses the transition of the Former Soviet Union (FSU) within the context of a 1987–1996 Soviet and FSU oil production decline. The conventional explanation of the break-up is that economic inefficiencies and the Cold War defense build-up caused it. Another possible explanation, one that is examined at length here, is that declining oil production was a contributing factor. The econometric analysis suggests that the fall in Soviet and former Soviet GDP in the 1980s and 1990s did not Granger cause the decline in oil production, but the decline in oil production did Granger cause the fall in GDP. Coal and natural gas to GDP relationships show alternative Granger causalities that we would expect to see with oil, but do not. A multi-cycle Hubbert trend that may be used to forecast future FSU oil production is also shown. © 2006 Elsevier B.V. All rights reserved. PACS: Q410; Q480; P230 Keywords: Oil production; Soviet transition; Resource scarcity; Hubbert curve

1. Introduction The transition of the former Soviet Union (FSU) from a communist country to a set of capitalist and predominantly democratic countries has to be considered one of the most astonishing events of the 20th century, yet there is surprisingly little econometric analysis surrounding it. Two important sets of analyses are what can be called a “Cold War” scenario, see Schweizer (1994), and a ⁎ Corresponding author. E-mail address: [email protected] (D.B. Reynolds). 0140-9883/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2006.05.021

272

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

“conventional” scenario given by Kotz and Weir (1997). The Cold War scenario suggests that the military expenditures of the Soviet Union took resources away from the rest of the economy causing the entire economy to become “bankrupt” and creating its downfall. The conventional scenario highlights the inefficiencies within the Soviet planned economy as the cause of the fall in Soviet GDP. According to this scenario, the inevitable prolonged recession, along with other political factors, caused the break-up of the Soviet Union. While both scenarios show good reasoning, there is a lack of econometric analysis. Part of the problem with any analysis though has been that few Soviet statistics are available to use, and what is available is often inaccurate. An alternative explanation of the fall of the Soviet and former Soviet GDP during the transition that may add to the above scenarios has to do with oil. The model developed here looks at the relationship between oil production and GDP in order to determine if an inefficient economy was the cause of the decline in oil production or whether a decline in oil production was in fact a precursor of the fall in Soviet GDP, which could suggest that it was a contributing factor to the transition. Considering how important oil production has been in the western industrialized countries where high oil prices have consistently caused macroeconomic problems, see for example Hamilton (1996), Hoooker (1996), Nordhaus (1980), Mork et al. (1994), Rowenberg (1980), Humphrey and Stanislaw (1979), Jones et al. (2004), and Rowenberg (1980), among others, it is reasonable to assume that oil must have affected the Soviet Union's economy. Therefore, in this paper we are only concerned with two variables: oil production and GDP. We test whether there is a relationship between the two. Since statistics on GDP and oil production are available, it is possible to look specifically at the Granger causality between the Soviet and former Soviet GDP and its energy production. However, since oil prices were regulated within the planned economy of the Soviet Union, price variations were illegal up until the break-up. This means we cannot get a relationship between prices and economic performance. However, we can use a proxy for the real price of energy which is the quantities of oil, natural gas, and coal produced and consumed by the economy. Therefore, we test whether the fall in Soviet GDP Granger caused the oil production decline or whether in fact an oil production decline Granger caused the fall in GDP. If the economy experienced a contraction due to economic inefficiency only, then a fall in GDP should Granger cause an oil production decline. However, if falling oil production had an effect on the economy, then it may have in fact Granger caused the GDP decline. We also test Granger causality between coal and GDP and natural gas and GDP. If in fact a decline in GDP was the cause of reduced oil production, then we would expect similar results for coal and natural gas production as well. Therefore, we compare all three energy relationships with GDP and with each other to get a complete energy to GDP picture of the waning days of the Soviet Union. The results of our tests show that the oil production decline Granger caused the fall in GDP, but that the reverse is not true. Similarly, Granger causality runs from GDP to coal and from oil to coal, but reverse causality does not hold. Interestingly though, natural gas has no Granger causality with GDP, which puts into question why the Soviet decline could have been affected by oil, but not gas. This suggests that oil could have been a factor behind GDP and coal production declines in the FSU. Also, since coal causality goes in the opposite direction (that is, GDP changes Granger cause coal changes), which is what we would have expected to happen with oil, then oil looks to be a factor. Furthermore, it is suggested that FSU oil production follows a multi-cycle Hubbert curve. That is, FSU oil production peaked in 1988 and declined, but it started increasing again as soon as the oil industry was privatized. The full deregulation and sell-off of FSU oil assets was not completed until 1996, after which oil production was on the rise. The multi-cycle Hubbert curve shows a forecast that FSU oil production will peak at 12 million barrels a day in 2009.

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

273

2. The two scenarios of former Soviet Union GDP decline In this section we look at the two widely acknowledged scenarios of the Former Soviet Union's decline. The first scenario is the conventional scenario given by numerous authors including Kotz and Weir (1997). The second is the Cold War scenario given by Schweizer (1994). We summarize them below. 2.1. The conventional scenario The classic economic explanation for the decline and break-up of the Soviet Union is given by Kotz and Weir. They argue that it was economic stagnation that precipitated political changes towards democracy. Other variations of the economics of the Soviet transition can be found in Krueger (1993), Kafouros (1996), Ellman and Kontorovich (1992), and Osband (1992). In what we call the Conventional Scenario, the sequence of factors is as follows: 1. Even though a planned economic system is relatively inefficient, it is still able to obtain high rates of growth by merely moving from an agrarian economy to an industrial one. During industrialization, it is a relatively simple feat to borrow the best technology from other industrialized countries and start using it. However, as a greater percent of the workforce becomes factory workers rather than farmers, there is less potential for economic growth based on industrialization alone. Further growth depends on innovation. Thus, the Soviets were faced with the transition from extensive to intensive growth, i.e. growth based on technology, not mere industrialization alone. Not only the Soviets, but many other societies also faced this problem. See Ellman (1986). Unfortunately for the Soviets, their economic system was especially weak at innovation. In general, growth based on innovation rather than transformation is slower, particularly for a planned economic system. By the mid 1970s the Soviet Union started stagnating. 2. As stagnation in the Soviet economy continued, the people and even the party leaders began to become impatient. They wanted faster growth. Not only was the desire for growth caused by the party wanting more income for its members and more power for national hegemony but also to quell discontent in the population. The perceived need for more economic growth allowed Gorbachev to rise to power in 1985 on a platform of reform. Gorbachev followed through with his platform by instituting perestroika, glasnost, and greater economic freedom. 3. The new economic and political openness created even greater economic inefficiency due to the double direction in which these policies went. Some segments of the economy were freed (e.g. wages), while others were still being planned or regulated (e.g. prices), see Osband (1992). 4. The greater freedom perestroika engendered created a changing society that had gradually gotten used to more freedom, free speech, and thought. Soviet society was changing quickly. The Soviet People were recognizing that they were behind the West, and they were more boldly expressing their discontent, see Dallin (1992). 5. The partially free, partially planned hodgepodge of economic policies made the economy worse rather than better, so the economy declined even further. 6. As the economy declined still more, greater political discontent and fractionalization within and outside of the party occurred. The discontent induced an even more dichotomous policy: trying to further liberalize the economy in some sectors while at the same time exercising tighter control in others.

274

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

7. Eventually this bad economy, necessitating ill-suited policies within the framework of a command system, made the situation even worse. Finally the sinking economy produced enough discontent that the political system broke down and the Soviet Union collapsed. Then as the USSR collapsed, so too did oil production. See Reinsh et al. (1992) for a full explanation. 2.2. The Cold War scenario An alternative set of factors for the collapse of the Soviet Union has to do with the Cold War between Russia and the United States. In this model put forward by Schweizer (1994), the fall of the Soviet Union, while not totally caused by a U.S. Cold War military buildup, was hurried along by it. In addition, some U.S. policies caused the Soviets to expend even more of their internal resources on their Cold War strategy, causing the rest of the economy to become bankrupt. This sequence of factors is as follows: 1. Financial and intelligence support was given to the Solidarity movement in Poland. This support was intended to foster greater discontent and force the Soviet Union to expend greater military resources to quell the restive population or to spend more money to help Poland's economy. 2. Financial support was given to the Afghan resistance against the Soviets. This support too was intended to cause the Soviet Union to pay a higher cost in terms of military resources and a greater number of killed or wounded which would create internal discontent. 3. Starting in 1985, Saudi Arabia increased its oil production from 2 to 6 million barrels per day, on U.S. orders. See Schweizer (1994). Oil prices collapsed from approximately $ 30 a barrel in November 1985 to $ 12 in March 1986. This oil market glut reduced Soviet oil earnings. 4. The reduction in oil prices caused the Soviets to have a current account deficit. According to Schweizer, a $ 1 decrease in the price of oil reduced Soviet foreign currency earnings by $ 1 billion per year. This made it difficult for the Soviets to import foreign technology and to support the creditworthiness of Warsaw Pact countries. 5. The United States imposed an embargo on all technology transfers to the Soviet Union. The Soviets needed American technology because they did not have their own turbo-drill oil equipment, which they needed in order to keep oil production from declining. 6. As Soviet oil production fell, internal pressure was put on the industry to engage in overexploitation of existing oil fields. The Soviets followed “perverse” extraction practices and overproduced oil which consequently damaged many oil reservoirs. As Dienes (2004) points out, this overproduction only hastened the post-peak oil decline. 7. The most well-known strategy of the United States is the defense spending increase of the 1980s including the Strategic Defense Initiative (SDI) as well as a conventional and strategic arms buildup. One was to scare the Soviet Union into engaging the world in a more constructive democratic fashion. The second was to force the Soviet Union to increase its own military spending. See Brown (1992). As defense received a bigger slice of the Soviet pie, there were fewer resources to invest in the economic capital base, including oil-related infrastructure, and eventually an economic decline followed. The conventional scenario and the Cold War scenario give a starting point to discuss why there was a collapse in the Soviet economy. The Conventional Scenario emphasizes that there were internal inefficiencies which induced political discontent and break-up. The Cold War scenario

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

275

emphasizes that the Cold War spending on arms within the Soviet Union took away resources from the rest of the economy. An additional factor in the Cold War scenario is that the United States imposed an embargo on the sale of oil industry technology to the Soviet Union that hurt the Soviet oil sector. With the Cold War buildup also taking away resources from the oil sector, both causes could have created an eventual reduction in oil production. This enhanced the Soviet Union's general economic problems. 3. The oil scenario In this section, we delineate the so-called “oil” scenario where the fall of the Soviet GDP is related to oil. This scenario is not intended to explain the entire cause of the GDP collapse, but rather look at the cause and effect relationship between the Soviet economy and its oil and energy industries. The reason to look at energy is because much economic growth can be tied to energy use and we expect it would have been the same for the Soviet Economy. One reason to look at a pure oil scenario is because military buildups in the past have not always caused economic decline, but rather economic growth. For example most wars in which the United States was engaged in have actually caused GDP to increase because of a demand-side stimulus. For example during World War II, the United States piled up tremendous amounts of debt, yet the higher level of output that the economy experienced got the country out of the Great Depression. The same was in fact true for the Soviet Union during World War II. The war brought the Soviet economy out of a depression. Other wars that brought on demand-side stimuli to our economy were the Korean War, the Vietnam War, the Gulf War, and even the Cold War military buildup of the 1980s itself. All of these wars helped the United States go from a recessionary to an expansionary economy. Therefore it is likely that the Soviet war in Afghanistan, or even the Cold War buildup in Russia, could have helped the Soviet economy to expand, rather than cause it to contract. This is a classic Keynesian macroeconomic stimulus policy. On the other hand, long-run effects of extra military spending could result in less investment and fewer goods and services for consumers. That would be a supply-side decline. As long as external debt does not exceed the Soviet Union's capability to pay the interest, then military spending could be sustainable, because it is simply a substitution between more consumer goods and more government goods. Interestingly, Reagan's military buildup in the 1980s was exactly the Keynesian fiscal stimulus that helped the United States get out of a severe recession in the early 1980s. Therefore what has to be shown in the Cold War scenario is that the Soviet Union's economic distortions were severe enough to cause the Soviet economy to act differently than how the U.S. economy acted in exactly the same situation. If a demand-side stimulus worked for America, it should have also worked for the Soviet Union unless a planned economy worked substantially differently than a free market economy. This brings us back to the oil scenario as a possible cause of the decline, since oil too was a part of the U.S. Cold War policy. 3.1. Oil production collapse Since the Soviet Union was a leading oil exporter, one U.S. strategy was to hurt their oil export revenues which would then damage the Soviet economy. One way to do that was to put a downward pressure on oil prices. This was why the United States asked Saudi Arabia to increase its oil production, in order to cause the Soviet oil revenue to decline. With greater Saudi oil supply, oil prices would decline, and so too Soviet oil revenues, making it difficult for the Soviet economy to buy necessary technology from abroad.

276

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

The first question that has to be asked about this policy is: were there other reasons for Saudi Arabia to increase its oil production in 1985?, consider world demand for oil up until 1985. After the oil shocks of the 1970s, the world's economies went through a drastic change in order to reduce their demand for oil. This strategy proved successful as world demand for oil began to decline rapidly in the early 1980s resulting in “demand destruction.” The only way OPEC was able to support high oil prices during the early 1980s was by drastically reducing its output to match the world's continuing reductions in demand. However, it was up to the Saudis to keep OPEC's output lower because everyone else in OPEC was unable to agree on production cuts. Thus the real driver behind the Saudi change in oil production strategy was the world reduction in demand. At the time of the increase in Saudi Arabian oil supplies, there was a lot of infighting among OPEC members. Saudi Arabia in particular wanted to have other oil producers cut their production more while the other members wanted the Saudis to do all the cutting. Seeing that the other cartel members were not cooperating in a repeated game Nash equilibrium, one reason for Saudi Arabia to increase its production was to punish other OPEC producers for not reducing their output. A second reason for Saudi Arabia to increase output was simply to increase its revenue. Saudi output went from 2 million barrels per day at $30 per barrel or $60 million dollars per day in late 1985, to 6 million barrels per day at $12 per barrel or $72 million per day in 1986. That is a 20% increase in revenue. Also, there may have been “unofficial” oil production where the Saudis were trading oil for goods, so the revenue could have been even higher. Since Saudi oil production costs were so low, most of the extra revenue was pure profit. But the story does not end there. Since Saudi Arabia was engaged in a repeated game situation, they were using the production increase to punish the other OPEC members with lower oil prices and force them to make new agreements for oil price increases. Therefore subsequent agreements by the rest of OPEC to cut output due to the Saudis' strategic increase also increased Saudi Arabia's revenue. This was then a big reason for Saudi Arabia to increase its output. In addition, Saudi Arabia itself had a lot of investments in western economies from their oil money. Consequently, they may have decided that if they increased their oil production and lowered oil prices while still increasing oil revenue, they could increase the value of their investments in the West too. So there were two ways for Saudi Arabia to make more revenue with its single decision to increase oil production: one by increasing oil revenue directly and two, by increasing the value of its investments in the West. Interestingly, the United States had other reasons for wanting the Saudis to increase oil output besides eroding the Soviet economy. Elections for congress were only one year away in 1985 and the Presidential election was three years away. Therefore it was in the President's best interest to create a robust, growing U.S. economy to help keep his political party in power. One way to do that was to have lower oil prices. Therefore one big reason that the Reagan Administration wanted to lower oil prices was to improve the condition of the American economy itself. The final reason then for Saudi Arabia to increase its output was to lower oil prices and thus lower oil revenue for the Soviet Union. The Saudi royal family was openly anti-Communist, both because of its “special relationship” with the United States, and due to the fact that the Soviet Union, a country dominated by atheism, posed a threat to the Muslim world. Nevertheless, no matter why the oil prices went down, it put pressure on the Soviet economy. Since Soviet oil export revenues went to the general government budget, the loss of foreign currency reserves had to have had an effect on the Soviet economy.

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

277

3.2. Technology and oil production The United States engaged in a technology embargo from 1981 to 1983. See Gustafson (1989). This factor directly affected oil production. Recently declassified (January 2001) CIA memorandum (ER 77-10147), CIA (1977), confirms the view at the time that the Soviet Union was going to reach a peak in oil production soon. Published in March 1977, the document entitled “The Impending Soviet Oil Crisis” describes problems of over water flooding in oil fields, and predicts an imminent Soviet oil peak “no later than the early 1980s.” Although the peak did not occur until 1987 at 12.6 million barrels per day, it was not much different from the production level of 12.5 million barrels per day in 1983. Therefore the CIA assessment of the peak of somewhere around 12 million barrels per day proved accurate, although instead of analyzing the peak in production using the Hubbert curve framework (see for example Cleveland and Kaufmann, 1991 and CIA, 1988), the CIA relied on an alternative approach of reserves to production (R/P) ratios. The steady decline of reserves vis-à-vis production was telling: During the 1961–65 plan period, only 1.3 million barrels per day […] of capacity had to be replaced. In 1971–75, 5.1 million barrels per day […] of replacement capacity was required because of rapid depletion. ( p. 3). The CIA report also addressed the issue of insufficient technological development: In the 1950s, when wells in the Urals–Volga region began to stop flowing naturally, the Soviets were forced to begin pumping. At that time, however, pumping equipment was in short supply. To forestall a slowdown in the growth of oil output, the Soviets adopted the practice of massive water injection within and along the edges of each field. […] In this case, however, […] special high-capacity submersible pumps are needed […] and the Soviets recognize that the only pumps adequate to deal with their lifting problem are made in the United States. […] 1,000 pumps from the United States have a higher total lifting capacity than 11,000 pumps of domestic origin. (pp. 6–7). Therefore, in light of such struggles going on within the Soviet oil industry, the United States decision to impose an embargo may have exacerbated the water flooding problems there. However, what has to be proven is if the embargo was effective or not. For example during the technology embargo, the Soviet Union needed natural gas compressors for its natural gas pipeline to Europe. Since they could not get compressors from the United States, the Soviets embarked on an intense research and development program to build their own compressors. They managed to design and build the compressors that they needed for their natural gas industry. There may have been innovation in other areas too. For example without advanced U.S. oil pumping equipment, the Soviets could just use more manpower and a greater number of less efficient pumps to get the same oil production output as the U.S. pumps would have provided for the oil industry. Another problem with proving that the embargo was effective has to do with the fact that embargoes in general are often ineffective. It is easy for a firm in say Nigeria to buy some oil equipment from America and then turn around and resell the equipment to the Soviet Union. In fact the embargo imposed on Iraq before the war was often ineffective since a number of countries were found to have by-passed the U.N. embargo and traded with Iraq. Since embargoes are never airtight, it is unclear how much equipment actually got through to the Soviet Union or did not. Also, embargoes are hard to maintain because the companies that produce the banned items can license the production offshore and still make money on selling the equipment to the forbidden buyer. Or more likely, the U.S. companies probably

278

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

complained loudly to Congress and the President that they were losing money. This political pressure then may have ended the embargo. Once the embargo was over though, the Soviets were free to buy as much equipment as they needed. The question is, did a three-year embargo really change oil production that much or not? We know that investment in oil infrastructure in general was rising in the Soviet Union from 1981 to 1985, Gustafson (1989), and that the new drills and pumps for oil development were being used more and more during the 1980s. Exploration drilling increased from 5.2 million meters in 1978 to 7.8 million meters in 1987 in the Former Soviet Union. The number of submersible pumps increased from 1981 to 1985, although there were only up to 1500 gas lift units as of 1985 when the plan was for 4500 such units. But Gorbachev did act to provide more investment in oil and gas development, which is probably why oil production increased after 1985. In the end the U.S. embargo may have caused Soviet oil production to peak earlier than it would have, but any hypothesis on it would be difficult to falsify. The final factor in the oil scenario is the question of whether excessive water flooding created an earlier peak in oil production. It is clear that the Soviet Union used water flooding extensively well before the 1980s. The question is, did the technology embargo intensify the water flooding or not, since the equipment that America was selling to Russia involved using water flooding? One of the differences that American technology made was with the efficiency of getting the oil and water out of the ground. When you water flood, you can start extracting both water and oil simultaneously, and they are hard to separate. Without the U.S. equipment, Russia could still extract the oil but it was harder to separate from the water and they would then need more labor and capital to compensate for the technology loss. However if the Soviet Union had had the American equipment, they may have been able to use less water flooding and so not lose as much oil. So it is possible that the embargo exacerbated the problem. 4. Energy and GDP Granger causality In this section we come to the question that most interests us: investigating whether a fall in GDP caused the Soviet Union's oil production decline. If the Conventional Scenario is correct and the fall and break-up of the Soviet Union occurred because of economic inefficiencies not including oil, then it should be the case that the fall of GDP Granger caused the oil production decline. That is, plummeting GDP should cause a decline in oil production either because of the general economic chaos that ensued causing the oil industry problems with getting labor and capital where it was needed, or because of a lack of internal demand for oil. If that is so, then we should see the decline in GDP Granger cause the decline in oil. On the other hand, if the fall in oil production Granger caused the GDP decline, then oil may have had something to do with the Soviet collapse. If falling oil production is causally prior to the decline in GDP, then we should see that oil Granger causes GDP. More generally, we want to check if energy affected the Soviet economy or if the economy affected energy. For example if the fall in GDP affected oil production, then surely the GDP decline would also cause falling coal and natural gas production as well. If there were plenty of oil, coal, and natural gas in the Soviet Union, then the only reason for a decline in these energy production rates would be a poorly performing economy so that production and distribution could not take place or because of weak demand. On the other hand, if falling oil production Granger caused a GDP decline, then it is possible that natural gas and coal could have caused this too. We also want to test the relationship between coal and oil as well as between natural gas and oil to see if they affected each other during the fall. The data from the Soviet and former Soviet economy came from a number of sources. The GNP data for the years 1928–1987 come from Gomulka and Schaffer (1990). The 1988–1991 GNP data

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

279

come from the United Nations (2004). The 1992–2003 GNP data come from the CIS Statistics Committee (2004). The 1918–1984 oil data come from DeGolyer and MacNaughton (2000 and 2003). The 1985–2004 oil data come from BP (2004). The coal and natural gas statistics come from EIA (2004). We assume GNP growth rates are close to GDP growth rates. 4.1. Oil and GDP Granger causality tests suggest a causality from oil to GDP rather than the other way round. In Table 1, we see that over a long history, oil production growth Granger causes GDP growth in the Soviet Union and Former Soviet Union, and that we reject the null hypothesis that oil does not affect GDP at the 5% level. On the other hand, we fail to reject the hypothesis that GDP growth does not Granger cause oil production growth. What is interesting is that over a shorter time frame shown in Table 2, from 1985 to 2002, we get similar results. From 1985 to 2002, we reject the hypothesis that oil growth does not Granger cause GDP growth at the 7% level. The reverse does not hold. There is no evidence that GDP Granger causes oil production. In fact GDP cannot be shown to Granger cause oil production for a one year, a two year, and a three year time lag. Therefore, the conclusion is robust to the selection of the sample period and the lag length. What this data says is that the Expanded Conventional Scenario for the fall of the Soviet Union has nothing at all to do with the decline of Soviet oil production. The oil production decline preceded the fall in GDP. Another way of saying this is that the fall in oil production happened independently of changes in GDP. Certainly the Soviet system itself could be blamed for oil problems, but that could be true at any time during the entire history of the Soviet Union. Most analysts say that the Soviet economy was inefficient in the 1980s causing production to fall and thus scarcity did not cause their production decline. Yet the Soviet Union certainly had increasing oil production in the 1960s and 1970s. So it cannot be that their inefficient system in and of itself caused both increasing production in the 1960s as well as decreasing production in the 1980s. Scarcity had to be a factor. Since it appears that oil production Granger causes GDP and that GDP does not Granger cause oil production, then we suggest that oil production declined due to scarcity within a planned system. In other words, one factor in the Soviet oil production decline certainly could have been the lack of open markets and the planned economic system that could not apply the best technology, capital, labor, and management to the oil fields. However, just like oil production in the United States declined due to scarcity even though the United States had all of the latest technology and open markets, so too could have Soviet oil production declined due to scarcity within their specific system. Under one consistently communist system, the Soviet decline in oil production was bound to happen just as a decline in American oil production was bound to happen under its consistently free market system. Still, since 1996 former Soviet oil production has been increasing. This is due to the opening

Table 1 Causality between Soviet and Former Soviet GDP growth and oil growth Pairwise Granger causality tests Sample: 1970 2003 using 2 year lag Null hypothesis:

F-statistic

Probability

GDPGROWTH does not Granger cause OILGROWTH OILGROWTH does not Granger cause GDPGROWTH

0.04541 3.48213

0.95569 0.04634

280

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

Table 2 Granger causality between energy growth and Soviet and Former Soviet GDP growth and other causalities Pairwise Granger causality tests Sample: 1985 2002 using 1 year lag Null hypothesis:

F-statistic

Probability

Row 1

GDPGROWTH does not Granger cause OILGROWTH OILGROWTH does not Granger cause GDPGROWTH

0.14775 4.03113

0.70802 0.06987

Row 2

COALGROWTH does not Granger cause OILGROWTH OILGROWTH does not Granger cause COALGROWTH

0.11795 8.30796

0.73638 0.01206

Row 3

GASGROWTH does not Granger cause OILGROWTH OILGROWTH does not Granger cause GASGROWTH

2.15643 3.86588

0.16575 0.07100

Row 4

COALGROWTH does not Granger cause GDPGROWTH GDPGROWTH does not Granger cause COALGROWTH

2.94496 7.14817

0.11414 0.02165

Row 5

GASGROWTH does not Granger cause GDPGROWTH GDPGROWTH does not Granger cause GASGROWTH

0.21718 3.17788

0.65118 0.10497

Row 6

GASGROWTH does not Granger cause COALGROWTH COALGROWTH does not Granger cause GASGROWTH

0.42716 0.50354

0.52478 0.49048

of markets and the addition of better technology and management to the former Soviet oil fields, which the Soviets did not have under communism. Furthermore, the oil production decline itself may have caused a collapse in GDP. Since oil production declined at about the same time as the fall and breakup of the Soviet Empire, then there seems to be a cause-and-effect relationship between oil and GDP. If we rule out the GDP-to-oil causality and if we can establish that oil decline and growth Granger caused GDP decline and growth, then it is possible that the oil production decline contributed to the fall of the Soviet Union. According to Khartukov and Starostina (2005), the reason for the oil decline was a lack of demand in Eastern Europe and within the Soviet republics, but again Granger causality shows that oil decline preceded GDP decline and therefore preceded a demand decline. The same would be true for Eastern Europe which broke apart after the peak in oil production. Also Balabanov and Dietz (1991) show that in 1989, the Soviets announced that trade with CMEA (the Council for Mutual Economic Assistance) countries would be conducted in hard currencies and that oil exports to CMEA would be cut by 10% in 1990 suggesting cause and effect for the decline in oil exports started in the Russian oil sector, not in Eastern Europe. 4.2. Coal and GDP The relationship between coal and GDP is more complicated. First of all, we know that Soviet and Former Soviet coal reserves are vast, since the Soviet Union owned half of the world's total reserves of coal and had proven reserves to last for four hundred years at their current rates of production. What is more, the Soviets possibly had as much as twenty times their proven reserves that were probable to exist (CIA, 1985). Even today, according to the BP Statistical Review of World Energy 2005, Russia owns 17% of proved global reserves. Currently, with total proved reserves of 157 billion tonnes and production of 280 million tonnes, their reserves have an R/P ratio of 561. So there was never any scarcity of coal

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

281

reserves. That means that the only reason for a fall in coal production would be a decline in demand, or problems with production and distribution. In Table 2 row four, we see that indeed GDP does Granger cause coal decline, but that coal does not look to Granger cause GDP decline. One of the differences between coal and oil is the demand for these energy resources outside of the Soviet Union. The coal market is a very competitive worldwide market where no one producer has ever had market power. Oil however is a more valuable commodity than coal, and is traded in a less competitive market. Because the world oil market is somewhat less competitive than the world coal market, the potential for the Soviet Union to export oil was greater than its potential to export coal. In fact the Soviet Union only exported 4% of its coal production in 1980 (CIA, 1985) and most of that went to Eastern Europe or other allies. By contrast, oil represented over 50% of Soviet hard currency earnings (CIA, 1985), and 25% of its oil production was exported. What this means is that there was always worldwide demand for Soviet oil production no matter how much they produced, but there was not much export demand for coal. Thus coal was only produced for internal markets. This is why when Soviet GDP declined, coal production went down too. Coal was an internal market demand driven commodity. Notice however in Table 2 row two that oil growth seems to Granger cause coal growth, but coal growth does not Granger cause oil growth. This could happen for two reasons. One is that when there is less oil, it is hard to run coal machinery and coal transportation systems. So with less oil available, the Soviet Union could not actually produce much coal. The other reason for this is that if the oil decline was the cause of the GDP decline, then the fall in GDP in turn caused the coal production decline. That is, oil indirectly caused the collapse of coal production by first causing the GDP to decline and then when GDP plummeted, there was lower demand for coal. 4.3. Gas and GDP The relationship between natural gas and GDP is also complex. Again Russia and the former Soviet Union owned over one fourth of the world's natural gas reserves. In the 1980s the Soviet Union had an estimated sixty eight years of proven natural gas reserves, based on current production (CIA, 1985). By contrast, Soviet oil reserves were officially only enough to last for fifteen years (CIA, 1985). The USSR exported less than 10% of its gas production which created about 15% of its hard currency earnings. With so much natural gas available, there was never any scarcity problem with natural gas, and the Granger causality shows this. In Table 2 row five, we fail to reject the hypothesis that gas growth did not Granger cause GDP growth and we can even accept the hypothesis that GDP growth did not Granger cause natural gas growth. What was happening is that natural gas production is a high-fixed-cost, low-variable-cost commodity. Since much of the fixed costs were already paid for with the building of fixed capital, including pipelines and wells, during the Soviet high growth period, then the region only had to maintain relatively low variable costs to keep natural gas production on line. A lot of gas was put into pipelines before and after the fall of the Soviet Union, so that even as GDP declined it was relatively cheap at the margin to continue to supply natural gas. Thus natural gas shows no real change during the fall of the Soviet Union. Its production continued independently of political and economic changes. Internal demand may have actually stayed relatively constant because internal gas prices were allowed to fall due to the low maintenance costs which gave the industry leeway to change prices. External demand would have remained robust too allowing for exports.

282

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

This is in sharp contrast to oil production. Oil production fell because oil wells themselves produced less oil and there was actually a need to develop new wells and whole new oil fields to get more oil. But since natural gas fields were not near any scarcity limits, then they could just be maintained at relatively little cost to keep production steady. So consequently natural gas production did not follow the GDP changes closely. One interesting note however is that it looks as though oil growth Granger caused natural gas growth but not vice versa. This is probably due to post Soviet oil production growth. After the fall of the Soviet Union and with new technology and investment, former Soviet oil production started to rise, and this investment in oil probably had an ancillary effect on natural gas development and production since much oil is found along with natural gas. 4.4. Simple growth model We consider a simple growth model to see if oil production is indeed a significant variable for economic growth. According to the theoretical model in Carruth et al. (1995) and in Phelps (1994) it is the level of prices that matter in oil price-GDP relationships because firms (and economic activity) react to different levels of prices. Since the Soviet Union regulated prices, we are left with energy quantities instead of prices. We compare real GDP growth rates to energy growth rates in order to be consistent with most of the literature on oil price–macroeconomy interactions and because differencing of the data takes care of the unit roots. In this model we have: GDPPgrowtht ¼ a1 þ a2 OilPgrowtht þ a3 CoalPgrowtht þ a4 NGPgrowtht þ e

d

d

d

where α1, α2, α3, α4 = empirical parameters Oilt = Oil production growth in the current year Coalt = Coal production growth in the current year NGt = Natural gas production growth in the current year The results are shown in Table 3. We check for a unit root in the residual using the augmented Dickey–Fuller test for a threeyear lag (see Davidson and Mackinnon, 1993). The ADF test statistic was − 6.32, while the Engle–Granger test statistic at the 1% level was − 2.58 for a model without a constant and − 3.14 for a model with a constant. Therefore the hypothesis of a unit root was rejected.

Table 3 FSU growth as a function of energy growth Variable

Estimated coefficient

Standard error

P value

CONSTANT OILGROWTH COALGROWTH NGGROWTH R-SQUARE ADJUSTED R-SQUARE DURBIN–WATSON DICKEY–FULER on Residual

−0.294761 0.799583 0.168721 0.003458 0.833493 0.788083 2.663633 −6.32

1.014230 0.158311 0.152282 0.244899 – – – –

0.7767 0.0004 0.2915 0.9890 – – – 0.01

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

283

4.5. Discussion The Expanded Conventional Scenario for why the Soviet Union fell looks to be logical and well thought out. The only problem with the logic is why didn't oil production decline after the GDP fell? Why did oil production Granger cause the GDP decline? With coal production the Expanded Conventional Scenario looks true. First GDP falls due to an inefficient Soviet economy and then, because demand for coal fell, and because coal production and distribution may have been affected by the collapsing economy, coal production itself declined even though the reserves of coal could last hundreds of years at current production levels. Natural gas production on the other hand did not fall with GDP and therefore was not Granger caused by GDP because natural gas infrastructure that was already set up was very cheap to maintain at the margin and so gas production could continue through the tumultuous Soviet disintegration. But oil production followed neither the coal production history nor the natural gas production history. Oil production not only fell, but it did so before the decline in GDP and thus it Granger caused GDP. But diminishing GDP did not Granger cause a fall in oil production. In addition, official oil reserves were not very high with only fifteen years of reserves specified. Given the Soviet inclination to exaggerate economic statistics, these reserves could have been less and if they were any less at all then scarcity of reserves could have easily been the cause of the Soviet oil production decline just as scarcity of oil has caused U.S. oil production to reach a peak in 1970. See Cleveland and Kaufmann (1991). If Soviet oil production reached a peak within the planned system and with the available technology of the day, then we know that oil did have an effect on the economy. In the West, oil prices negatively and substantially affected the macroeconomy, so why should not the same be true for the Soviet Union even if they imposed price ceilings? If prices are controlled, then it is no longer prices that affect the economy, but the actual output of oil that affects it. The Soviet Union was faced with two choices when its oil peaked and started declining in 1988. One was to find more oil to produce under its current system of government and industry which it could not do. The other was to completely change its system of government, and let private industry own and run natural resource and business assets and free up markets. It chose the latter because scarcity forced it to. 5. FSU multi-cycle Hubbert curve The fall of the Soviet Union was Granger caused by an oil production decline that happened under a planned economic system. However, post-1996 oil production increased again. What happened after the fall of the Soviet Union was that markets and property rights changed. Before 1991 Russian oil production was owned, controlled, and priced by the Soviet government. But even after the USSR's demise, oil was controlled until about 1995 when government assets were finally sold and freer markets were established. So before 1995 there were neither markets nor well-defined property rights for Russian oil producers. This means that given the inefficient system of the day, scarcity within that closed economic system caused the oil decline and Soviet collapse. Then after 1995, with the sale of Russian assets and the establishment of property rights, new investment and the application of Western technology into Russian oil production was finally possible. Then a second cycle of oil production began. This was a new increase in oil production that looks to be following a multi-cycle Hubbert curve. See Hubbert (1962,1982). Again this was not due to new

284

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

technology or due to higher prices, but rather due to property right changes and freer markets that allowed the investment into new technology and the better management of that technology. The privatization of the Russian economy occurred in two distinct stages. The first stage took place between late 1992 and 1995–1996. See Jermakowicz et al. (1994), Dienes (2004), and Reznik (2003). As a result of a decree signed by President Yeltsin in July 1992, all large government enterprises were expected to become privatized before October 1. Anatoliy Chubays, who directed the “voucher” privatization process, distributed shares to 144 million Russians. Although privatization began in 1992, it was delayed for the oil, natural gas, electrical utility and telecommunications companies until 1994. The direct sale of energy shares for cash took place in 1995, and was pretty much complete by the first half of 1996. As Dienes (2004) explains, there was a clash of corporate ideologies permeating the Russian oil business. This is illustrated by the business practices of Surgutneftegaz and Lukoil, which invest a lot in technology and exploration. While this strategy is certainly beneficial in the long run, it may cause problems if the company should need external financing because its market capitalization is low due to its disregard of delivering adequate profits. On the other side of the extreme are Yukos, Sibneft and TNK-BP, which follow the textbook case of the economic concept of short-run profit maximization, with a total disregard of the future. This is not very surprising, since most of the companies following this business model, especially Yukos and Sibneft, are run not by oilmen, but by financiers as a result of the banks' substantial share of stock in those companies following the “loans for shares” program. These companies have been growing astoundingly in market value, but their growth may be very short-lived as they deplete their resource base. It seems that ironically, the most pro-Western of the oil companies (i.e., Yukos, Sibneft and TNK-BP) in many respects resemble the state enterprises of the Communist era, which were expected to deliver immediate performance under the pressure of another five-year plan. In the end, with or without technological improvement, Russia's oil production will peak at some point. To recall an old oil-industry truism, you can only pump what you have discovered. Russian oil discovery peaked in the 1960s, and even present-day technology has not reversed that. After all, U.S. discoveries peaked in the 1930s and even though it is the world's technological leader and the most free-market-oriented country, that did not prevent the United States from peaking in 1970, in accordance with M. King Hubbert's prediction. Actually, it seems that Hubbert's predictions seem most vindicated in market economies with robust economic growth, because production grows as fast as geological factors allow. Russia peaked prematurely in 1987, since it lacked the technology to extract the already discovered petroleum. Hence, instead of a single peak, we are witnessing a second cycle. Dienes (2004) and other researchers worry that in addition to the invariable geological factors, the management practices, tax structure and political uncertainty can exacerbate the already precarious future of Russian oil production. The usual way to model a Hubbert-type curve is to use a difference equation whereby the rate of the change in production rate in year t is dependent on the production rate in year t − 1 (Pesaran and Samiei, 1995; Cleveland and Kaufmann, 2001) or the reserves in year t − 1 (Moroney and Berg, 1999). An alternative approach is to model production in year t based on cumulative production in year t (Al-Jarri et al., 1999; Pickering, 2002). The Al Jarri and Startzman model is not set up as a cumulative production model that is dependent on time as Hubbert's curve is. Rather it subsumes the time variable to create a relationship of the current production rate as a function of cumulative production.

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

285

The standard Hubbert model, Hubbert (1962), specifies a logistic model as CQP ¼

URR 1 þ e−aðt−t0 Þ

where CQP is the cumulative production, URR is ultimately recoverable reserves, t0 is the year of peak production and “a” is a parameter. The current level of production is the derivative of cumulative production with respect to time, QP = dCQP / dt, or QP ¼

a⁎URR⁎e−aðt−t0 Þ ½1 þ e−aðt−t0 Þ 2

Solve for − a(t − t0) in CQP and plug in to QP URR −1 CQP
 URR −aðt−t0 Þ ¼ ln −1 CQP

e−aðt−t0 Þ ¼

QP ¼

a⁎URR⁎e

−aðt−t0 Þ

½1 þ e−aðt−t0 Þ 2

  a⁎URR⁎ URR −1 CQP ¼ h i2 1 þ URR −1 CQP

2

QP ¼

a⁎ URR CQP −a⁎URR URR2 CQP2

¼ a⁎CQP−

a ⁎CQP2 URR

a as b2. Re-label a as b1 and − URR This becomes QP ¼ b1 ⁎CQP þ b2 ⁎CQP2 Estimate this econometrically and obtain ultimately recoverable reserves by solving URR ¼ −

b1 : b2

Here we are not concerned with solving for URR. What we are concerned with is that the rate of production is a function of cumulative production. We use this model but with changes. In particular, we will add dummy variables to look at the effect of institutional changes. The model is the following: QP ¼ b0 þ b1 ⁎ðCQPÞ þ b2 ⁎ðCQP2 Þ þ b3 ⁎DUM96 þ b4 ⁎DUM96⁎ðCQPÞ The results are shown in Table 4. Since the Durbin–Watson statistic is low and the Breusch– Godfrey LM test indicates autocorrelation, an AR(1) and MA(1) are used. Regression results with the MA(1) term are included. The Breusch–Godfrey LM F statistic is reported, which shows serial correlation. The Breusch–Godfrey statistic is better to use than the Durbin–Watson statistic, because

286

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

Table 4 Multi-cycle Hubbert curve for Soviet and former Soviet oil production Variable

Estimated coefficient

Standard error

P value

CONSTANT CUMPROD CUMPROD^2 DUM96 DUM96*CUMPROD AR(1) MA(1) R-SQUARE ADJUSTED R-SQUARE BREUSCH–GODFREY LM F DICKEY–FULLER on Residual

− 179.3851 0.114725 − 7.34E-07 − 19014.77 0.149533 0.922781 0.578822 0.999167 0.999130 8.049760 − 6.428615

98.57609 0.009275 8.95E-08 4492.714 0.035644 0.062337 0.164410 – – – –

0.0710 0.0000 0.0000 0.0000 0.0000 0.0000 0.0006 – – 0.000504 –

it detects types of serial correlation other than AR(1). The standard errors listed in the table are Newey–West standard errors that are autocorrelation-consistent (Newey and West, 1987). We check for a unit root using ADF on the residuals and test for a two year lag. The results show an ADF of −6.43 which means a unit root is rejected at the 1% level using the Engle–Granger critical values. For the regression without the MA(1) term, the Breusch–Godfrey LM F statistic was 51.68, whereas with the MA(1) term, it was 8.04. These values are for two-year lags. While we still have statistically significant serial correlation, the situation is better than when the MA(1) term was absent. However, considering that over 99% of production is explained with only four variables, the serial correlation and the unit root tests will look weak even with the smallest of deviations. Because much of the oil production was under a planned system, it would be hard to find variables that could explain the rest of the variation. Fig. 1 shows the oil production trend as a function of cumulative production using the Al Jarri and Startzman model. If we forecast using the above model, we see that FSU oil production looks to peak in the year 2009. See Fig. 2 where we show a multi cycle Hubbert curve forecast. It is difficult to check such

Fig. 1. Former Soviet Union oil production as function of cumulative production.

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

287

Fig. 2. Former Soviet Union oil production as function of cumulative production-forecast.

a forecast using cross validation because data splitting depends on having a consistent model throughout the whole period considered for the model building set. However, since there is a change in institutions almost at the end of the data set, the model building set would have to include data points after the transition, and that only includes seven points of data. Therefore there are not enough data points for the validation set of data. This is why it is difficult to use data splitting on any multicycle Hubbert model. Nevertheless, the forecast does predict another Hubbert peak in less than ten years, and it could happen sooner. See for example Mosnews.com (2004). 6. Concluding remarks It is not clear whether the Cold War scenario or the conventional scenario explains fully the fall of the Soviet Union since very little statistical evidence can be brought to bare on any given hypothesis. However, the evidence does indicate that an oil production decline did Granger cause a fall in GDP, but not the other way around. If indeed we are to believe in the pure Soviet inefficiency argument as the main cause of the Soviet decline, then a theory must be shown as to why only one year before the Soviet economy began to unfold, its oil production started declining. Yet coal production declined after the fall of the Soviet GDP as we would expect. Natural gas production on the other hand stayed relatively steady during the turmoil due to high fixed costs of production that were already sunk costs and due to low marginal costs of continued production. Both the coal and natural gas production histories would make sense for oil, yet oil shows neither of these characteristics. Instead oil decline happens before the fall in GDP. Over and over again the emphasis in the fall of the Soviet Union, and even the decline in oil production, is on communist inefficiency and a lack of technology. Yet communist inefficiency was in play when the Soviet economy and oil production were all increasing in the 1950s, 1960s and 1970s. In fact the West was actually afraid that communism would win the Cold War in those decades, and even right up to 1990, which was the motivation behind the United States' military buildup in the 1980s. Therefore, communist inefficiency in and of itself could not have been the only cause of the decline of the Soviet Union or of Soviet oil production. There has to have been another factor.

288

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

Resource scarcity had to be a factor. Within a consistently communist system, oil production managed to go up as long as oil reserves were relatively abundant. However, once scarcity increased substantially, the communist system saw declining oil production which in turn could have caused their inefficient economic system to finally decline. It required scarcity and inefficiency together to create the fall, not inefficiency by itself. Nevertheless, it is still not clear in the West whether an efficient, high-tech economy can withstand an oil shortage. Evidence shows that oil scarcity has been more powerful than technology in adversely affecting economic growth in North America, Japan, and Western Europe. See for example Hamilton's (1983) classic paper. As for the Soviet Union, the evidence suggests that there was an oil scarcity problem within its economic system before the fall of Soviet GDP, and because the USSR was a virtually closed system, that oil shortage may have caused its fall and decline. Interestingly, U.S. oil production peaked in 1970, but since the dollar was the global reserve currency in the post-Bretton Woods world, the United States could “import its way out of the problem.” The USSR did not have the “seignorage” option of just sending its own money abroad in exchange for commodities, so once its oil production peaked, the economic system had to decline to accommodate an oil shortage.

References Al-Jarri, Abdulrahman, S., Startzman, Richard A., 1999. U.S. Oil Production and Energy Consumption: A Hubbert Modeling Approach to Forecast Long Term Trends in Various Components of U.S. Energy Consumption with an Emphasis on Domestic Oil Production. In: Moroney, J.R. (Ed.), Advances in the Economics of Energy and Resources. Fuels for the Future, vol. 11. JAI Press, Stamford, Connecticut, pp. 37–58. Balabanov, Todor, Dietz, Raimund, 1991. Eastern and East West Energy Prospects. In: Havlik, Peter (Ed.), Dismantling the Command Economy in Eastern Europe. Westview Press, Boulder CO, pp. 125–137. BP, 2004. Statistical Review of World Energy. Accessed October 12, 2004 from http://www.bp.com/subsection.do? categoryId=95andcontentId=2006480. Brown, Thomas G., 1992. The Soviet Union as a great power: the need for reform. The American Economist 36 (1), 77–84 (spring). Carruth, A.A., Hooker, M.A., Oswald, A.J., 1995. Unemployment, oil prices, and the real interest rate: evidence from Canada and the UK. In: Chritstofides, L.N., Grant, E.K., Swadinsky, R. (Eds.), Aspects of Labor Market Behavior: Essays in Honour of John Vandercamp. University of Toronto Press, Toronto. CIA, Directorate of Intelligence, 1977. The Impending Soviet Oil Crisis. Memorandum (ER 77-10147), Declassified January 2001 (Accessed October 12, 2004 from www.foia.cia.gov). CIA, Directorate of Intelligence, 1988. Revisiting Soviet Economic Performance under Glasnost: Implications for CIA estimates, SOV 88-10068. CIA, Washington DC. September. CIA, United States Central Intelligence Agency, 1985. USSR Energy Atlas. Central Intelligence Agency, Washington DC, pp. 12–28. Cleveland, Cutler J., Kaufman, Robert K., 1991. Forecasting ultimate oil resources and its rate of production: incorporating economic forces in the models of M. King Hubbert. The Energy Journal 12 (2), 17–46. Cleveland, Cutler J., Kaufman, Robert K., 2001. Oil production in the lower 48 states: economic, geological, and institutional determinants. The Energy Journal 21 (1), 27–49. Commonwealth of Independent States (CIS) Statistics Committee, 2004. Volume Indices of Gross Domestic Product. Accessed October 12, 2004 from http://www.cisstat.com/eng/mac-01.htm. Dallin, Alexander, 1992. Causes of the collapse of the USSR. Post-Soviet Affairs 8 (4), 279–302. Davidson, Russel, Mackinnon, James G., 1993. Estimation and Inference in Econometrics. Oxford University Press, Oxford, p. 722. DeGolyer, E.L., MacNaughton, L.W., 2000 and 2003. Twentieth Century Petroleum Statistics. DeGolyer and MacNaughton, Dallas TX. Dienes, L., 2004. Observations on the problematic potential of Russian oil and the complexities of Siberia. Eurasian Geography and Economics 45 (5), 319–345. Ellman, Micheal, 1986. The macro-economic situation in the USSR — retrospect and prospect. Soviet Studies 38 (4), 530–542.

D.B. Reynolds, M. Kolodziej / Energy Economics 30 (2008) 271–289

289

Ellman, Micheal, Kontorovich, Vladimir, 1992. Over View. In: Ellman, Micheal, Kontorovich, Vladimir (Eds.), The Disintegration of the Soviet Economic System. Routledge, New York, pp. 1–39. EIA, United States Department of Energy (DOE) — Energy Information Administration (EIA), 2004. International Energy Outlook, and Energy Year Book, various pages, Washington DC. Gomulka, S., Schaffer, M., 1990. A New Method of Long-Run Growth Accounting with Applications to the Soviet Economy 1928–87 and the U.S. Economy 1949–78. LSE CEP Paper, vol. 014. London School of Economics, Centre for Economic Performance, London. Gustafson, Thane, 1989. Crisis Amidst Plenty: The Politics of Soviet Energy under Brezhnev and Gorbachev. Princeton University Press, Princeton, NJ. Hamilton, James., 1983. Oil and the macroeconomy since World War II. Journal of Political Economy 91 (2), 228–248. Hamilton, James, 1996. This is what happened to the oil price–macroeconomy relationship. Journal of Monetary Economics 38, 215–220. Hoooker, M.A., 1996. What happed to the oil price–macroeconomy relationship. Journal of Monetary Economics 38, 195–213. Humphrey, William S., Stanislaw, Joseph, 1979. Economic growth and energy consumption in the UK, 1700–1995. In: Stevens, Paul (Ed.), The economics of energy, vol. I (2000). Edward Elgar Publications Limited, Northampton, Massachusetts, U.S.A., p. 3. Hubbert, M.K., 1962. Energy Resources. A Report to the Committee on Natural Resources. Publication 1000-D. Technical Report. National Academy of Sciences, Washington, D.C. Hubbert, M.K., 1982. Techniques of Prediction as Applied to the Production of Oil and Gas. In: Glass, S. (Ed.), Oil and Gas Supply Modeling. Special Publication, vol. 631. National Bureau of Standards, Washington, D.C., pp. 16–141. Jermakowicz, W., Pankow, J., Abramov, A., 1994. Voucher Privatization in Russia: First Results and Experiences. Center for Social and Economic Research, Warsaw, Poland. Jones, Donald W., Leibiz, Paul N., Paik, Inja K., 2004. Oil Price shocks and the Macro Economy: what has been learned since 1996. The Energy Journal 25 (2). Kafouros, Wassily, 1996. Towards a dynamic two sector model of Soviet economic decline. Review of Radical Political Economics 28 (3), 35–49. Khartukov, Eugene, Starostina, Ellen, 2005. Ex-Soviet oil exports: constraints, outlook, and global impact. Energy Politics (VII), 31–55 (fall). Kotz, David M., Weir, Fred, 1997. Revolution From Above, The Demise of the Soviet System. Routledge, London, p. 34. Krueger, Gary, 1993. Goaszakazy and the Soviet Economic Collapse. Comparative Economic Studies 15 (3), 1–18 (fall). Mork, Knut, Anton, 1994. In: Stevens, Paul (Ed.), Business Cycles and the Oil Market. The Economics of Energy, vol. I (2000). Edward Elgar Publications Limited, Northampton, Massachusetts, p. 62. Moroney, John R., Berg, M. Douglas., 1999. An integrated model of oil production. The Energy Journal 20 (1), 105–124. Mosnews.com, 2004. Russia's Oil Exports Reach Maximum, Decline to Start in 2 Years — Official. Created: November 09, 2004 http://www.mosnews.com/money/2004/11/09/oilproduction.shtml. Newey, Whitney, West, Kenneth, 1987. A simple positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55, 703–708. Nordhaus, William D., 1980. Oil and Economic Performance in industrial countries. In: Stevense, Paul (Ed.), The Economics of Energy, vol. I (2000). Edward Elgar Publications Limited, Northampton, Massachusetts, p. 3. Osband, Kent, 1992. Economic crisis in a shortage economy. Journal of Political Economy 100 (4), 673–690. Pesaran, M.H., Samiei, H., 1995. Forecasting Ultimate Resource Recovery. International Journal of Forecasting 11 (4), 543–555. Phelps, E.S., 1994. Structural Slumps. Harvard University Press, Cambridge, Massachusetts. Pickering, Andrew, 2002. The discovery decline phenomenon: microeconometric evidence from the UK Continental Shelf. The Energy Journal 23 (1), 57–71. Reinsh, Anthony E., Lavrovsky, Igor, Considine, Jennifer I., 1992. Oil in the Former Soviet Union: Historical Perspective Long-term outlook. Canadian Energy Research Institute, Alberta, Canada. Reznik, B.A., 2003. Corporate Governance in Russia: The Yukos Case. http://www.worldbank.org/transitionnewsletter/ dec03apr04/pgs23-26.htm, Accessed October 4, 2004. Rowenberg, Nathan, 1980. Historical relations between energy and economic growth. In: Stevense, Paul. (Ed.), The Economics of Energy, vol. I (2000). Edward Elgar Publications Limited, Northampton, Massachusetts, p. 3. Schweizer, P., 1994. Victory: The Reagan Administration's Secret Strategy that Hastened the Collapse of the Soviet Union. The Atlantic Monthly Press, New York. United Nations Statistics Division, 2004. USSR (former): Estimates of GDP at constant 1990 prices in National currency (in Million). Accessed October 12, 2004 from http://unstats.un.org/unsd/snaama/SelectionQuick.asp.