Workshop on the Impact of Oil Boom in the Caspian Basin Paris, 2 June 2006 Matthias Luecke, Natalia Trofimenko (The Kiel Institute for the World Economy)
Redistribution of Oil Revenues in Azerbaijan
In September 1994, Azerbaijan has signed a $7.8 billion contract with 11 international companies to develop Azeri, Chirag and Gunashi (ACG) fields with estimated reserves of 5.4 billion barrels of oil. The oil production under this agreement started in 1997. Between 1995 and 2002 oil production has increased from 9 to 15.3 million tons and the exports of oil products have tripled increasing the government bonus and profit oil receipts from 37 to over 200 millions of US dollars. Using data from LSMS 1995 and Household Budget Surveys 2002 and exploiting this exogenous increase in the resource-based revenues, we compare changes in the composition of income of households in oil-rich and oil/poor regions, which allows us tell whether all of the benefits have been appropriated by the oil producing regions or have been shared with resource-poor parts of the country. The empirical analysis suggests significant changes in redistributive public expenditure policies. While over ¾ of government revenues in both years have been generated by oil-producing regions, in 1995 most of the revenues stayed within the producing regions and urban areas. The amount of state benefits received by the household was proportional to the resource-endowment of the region in which it was located, with highest benefits received by households in Baku and other oil rich regions. There was also a strong pro-urban bias with the amount of benefits higher in urban areas. In 2002 the level of state support was lowest for the residents of Baku and one can see a strong pro-rural bias even in the resource-rich areas of the country. Overall, the dependency on government, measured as the share of state benefits in total income, has decreased over the period from 28 to 10 percent.
Introduction According to the latest forecasts from the Economist Intelligence Unit, the surging exports of oil and gas will help raise Azerbaijan’s GDP by nearly 30 percent in 2006 and make it the fastest growing country in the world (Economist, February 25th – March 3rd 2006). The challenge imposed on the country will be to translate this temporary resources-driven stimulus to the economy into sustainable growth. Way too often resource-rich countries develop deficient political systems that descend into a spiral of embezzlement, cronyism, and wasted government expenditures, with large income disparities fuelling radical political or religious movements. And way too often non-oil sectors become uncompetitive and shrink greatly because export revenues are spent quickly rather than saved to meet long-term
expenditure commitments or provide a capital stock for future generations after reserves are exhausted. Recognizing that the sensible management of resource rents will play a crucial role in the future development of the Caspian Region, researchers at the Kiel Institute for the World Economy and at the Kiel University Institute for Political Science have launched a project investigating the aforementioned challenges and evaluating alternative strategies for the use of the oil revenues in Azerbaijan and Kazakhstan. This paper is an offspring of the project and focuses on the redistribution of oil revenues in Azerbaijan. Our main goals are (1) to appraise the extent to which the benefits from the oil boom are retained in the oil-producing regions or transferred to the rest of the country; (2) to appraise the extent to which the benefits from the oil boom are retained by the groups directly involved in the extracting sector or transferred to the rest of the population and to its poorest members in particular. Apart from the obvious economic consequences for income inequality across and within regions, efficient redistribution of oil revenues has important political implications. The surging exports of oil can simultaneously increase the discontent in oil-producing regions and the rest of the country, as oil-producing regions wish to keep a higher share of oil revenues whereas the rest of the country insists on a higher redistribution share (Nigeria: Ikean et al., 1998). Alternatively, the struggle can occur between the oil-producing regions and the central government, as the bulk of the resource rents leaks to the capital city, leaving oil-producing regions in poverty (Russia: Dienes, 2002; Kazakhstan: Najman et al., 2006). Our study complements the existing literature by providing a case in which the capital city and surrounding areas are also home to the majority of oil-extracting and oil-processing businesses (hence, the benefits of the burgeoning extracting sectors go directly to the already prosperous region, potentially magnifying income inequality even further). The main value added of the paper, however, is the study’s quasi-experimental estimation framework and its focus on micro-level effects of the resource-based growth. Employing a quasi-experimental framework, we take advantage of the variation in the oil-based revenues generated by a spike in oil production following the cease-fire with Armenia in 1994. Although oil field exploitation in Azerbaijan dates back to the 1870s, the ten year conflict between Azerbaijan and Armenia, as well as the overall economic collapse following the independence from the former Soviet Union, stalled oil production. In
September 1994, however, Azerbaijan and eleven international companies signed a contract to
develop Azeri, Chirag and Gunashi fields with estimated reserves of 5,4 billion barrels of oil and the oil production has been growing at an average 10.2% since 1997.
While different regions exhibit vastly different revenue and expenditure structures, the same is true for the same type of regions in different times. Using the exogenous shock to oil production as a plausibly exogenous variation in revenues of the resource-rich regions, we compare the changes the composition of households’ income1 between 1995 and 2002. The empirical analysis uses household surveys, supplemented by government budget data and suggests significant changes in redistributive public expenditure policies. While over ¾ of government revenues in both years have been generated by oil-producing regions, in 1995 most of the revenues stayed within the producing regions and urban areas. The amount of state benefits received by the household was proportional to the resource-endowment of the region in which it was located, with highest benefits received by households in Baku and other oil rich regions. There was also a strong pro-urban bias with the amount of benefits higher in urban areas. In 2002 the level of state support was lowest for the residents of Baku and one can see a strong pro-rural bias even in the resource-rich areas of the country. Overall, the dependency on government, measured as the share of state benefits in total income, has decreased over the period from 28 to 10 percent.
1
At this stage, due to data inconsistencies, only share of state support in total income has been considered.
Brief History of Oil Production in Azerbaijan and the Aftermath of the Latest Oil Boom Azerbaijan is the oldest known oil producing region in the world and the environs of its capital, Baku, are the site of the first purposefully drilled oil well (1848-1849). Since the first oil field exploitation in 1872, the country experienced several oil booms and descents: stalled after joining the emerging Soviet Union in 1920, its oil production peaked during the World War II and relapsed in the post-war years when Russia shifted support to the VolgaUral oil fields. The leading oil producer during the war (Azerbaijan accounted for
approximately 70% of the USSR’s petroleum output), Azerbaijan was producing 60% of Soviet oil extraction machinery and spare parts, but less than 2% of the oil when the USSR disintegrated.
The ten year conflict between Azerbaijan and Armenia, “the bloodiest and most intractable disputes to emerge from the break-up of the Soviet Union”, and the overall economic collapse following the independence from the former Soviet Union (the country’s GDP contract by almost 60% from 1990 to 1995) stalled oil production. Following the ceasefire with Armenia in May, 1994, in September of the same year Azerbaijan has signed a $7.8 billion 30 year contract with 11 international companies to develop Azeri, Chirag and Gunashi (ACG) fields with estimated reserves of 5,4 billion barrels of oil. This consortium originally featured oil companies from Azerbaijan (SOCAR), the USA (Amoco, Unocal and
Pennzoil); United Kingdom (British Petroleum and Ramco); Norway (Statoil); Russia (LUKoil), Turkey (Turkish Petroleum-TPAO) and Saudi Arabia (Delta Nimir). Since then Exxon (USA) and ltochu (Japan) also joined the club. The oil production under this
agreement started in 1997 and has been growing at an average of 10,2% since 1997. The ACG contract has paved the way for a dozen more production sharing agreements (see Attachment) that have committed a total investment of more than $60 billion in Azerbaijan’s oil development. Oil production is dominated by Baku, Absheron (Absheron, Khizi, Sumgait) and Ganja-Gazakh (Aghstafa, Dashkesen, Gedebey, Goranboy, Khanlar, Gazakh, Samukh, Shamkir, Tovuz, Ganja and Naftalan). Baku is a home to a large part of the country’s productive resources, including welldeveloped extraction and processing industries. Absheron is abundant with natural resources and its economy is dominated by oil and gas production, petro-chemical and chemical industry, ferrous and non-ferrous metallurgy, and power industry. Ganja-Gazakh takes a second place in the republic by its economic importance. center of processing industries. This region contributes 12-13 percent of total industrial production. The economy is dominated by extraction and processing industries. Some oil and gas are also present in Aran (Aghjabedi, Aghdash, Beylan, Barda, Bilasuvar, Goychay, Hajigabul, Imishli, Kurdamir, Nefchala, Saatli, Sabirabad, Salyan, Ujar, Zardab, Ali-Bazramli, Mingechevir, Yevlakh. Most of the government revenues are generated by the oil producing regions (Table 4). Baku alone accounts for almost 70 percent of the budget revenues (it also accounts for approximately 30 percent of expenditures).
Typology of Redistribution of Oil Revenues Which revenues stem from oil? Are all redistributed? Talk about national oil fund. How is the rest redistributed?
A subject of contention on today’s political agenda is to what extent oil-rich regions and workers in oil industry should share their affluence with the rest of the economy.
Data and Identification Strategy In this paper, we argue that the exogenous shock to the oil production in Azerbaijan lead to a causal exogenous shock to its revenues, which altered the income levels and living conditions of oil-rich regions and workers employed in the petroleum sector and supporting industries. We start by testing the following hypotheses: H0: Wages in oil rich areas stayed the same relative to wages in oil poor areas. H1: Wages in oil rich areas increased relative to wages in oil poor areas. Positive and monotonicaly increasing coefficients on 1995-2002 dummy variables in Table 3 show that wages increased across all parts of the country. Even in oil-poor regions wages tripled2 during 1995-2002. However, the growth in wages between 1995-2002 was proportional to the region’s endowment with oil. Comparing 1995 and 2002 wages across various region, Baku had the highest increase in wages between those years: wages increased almost 8-fold3. Of course, part of this differential may result from different rates of inflation in Baku and peripheries, however, even in oil-rich areas outside Baku, the increase in wages between 1995 and 2002 was from 22 to 41 percent higher than in oil poor regions. The overall increase in wages should have reduced household’s dependence on state benefits, thus, decreasing the amount of received benefits. Indeed, the share of state benefits in total income decreased from 28 to 10 percent over the period4, with the drop in dependency proportional to the region’s endowment with resources: 9 percentage points decrease for resource-poor regions, 13 percentage points decrease for regions with limited resources, 17 percentage points decrease in oil rich areas outside of Baku and 30 percentage points decrease in Baku (see Table below).
2 3
Computed as 100*(exp(1.46)-1)=330.596 Computed as 100*(exp(2.499)-1)-330.596=787.65 4 Mean of dependent variable in table 7.
Baku Other Oil Rich Regions Some Oil No Oil
Mean 0.375 0.267 0.253 0.203
1995 Std Dev 0.368 0.358 0.354 0.332
Mean 0.080 0.097 0.127 0.114
2002 Std Dev 0.103 0.098 0.119 0.114
Difference in Means -0.295 -0.170 -0.125 -0.088
Under the assumption of benevolent government, a disproportionate increase in wages in oil-rich regions would induce the government to channel part of the additional oil revenues to peripheral regions. Using the 1995 and 2002 household samples we test the second hypothesis: H0: The state support stayed the same across oil rich and poor regions. H1: The state support changes in favor of oil poor and rural areas. As indicated in Tables 5 and 6, in 1995 the government was disproportionately favoring the households residing in oil rich and urban areas. In 2002 the picture is
dramatically different: state benefits go to oil poor regions and, in particular, to rural areas. The benefits to rural households are 6 times higher5 than the benefits to urban households within the oil poor regions. These preliminary conclusions are drawn on a simple difference-in-difference analysis comparing wages and state benefits across regions differently endowed with oil in pre- and post- oil boom periods. The only additional control variable at this stage is household size. Adding controls for other determinants of wages and state benefits is one our agenda. Also, real wages have increased across all sectors (see Table 1). Biggest increases are in mining and non-tradable sectors (financial intermediation, real estate, construction, hotels and restaurants and trade). We will use info on individuals (so far dealed with households) and their employment sector to support this industry/level analysis and test the hypotheses: H0: Wages in mining and non-tradable sectors stayed the same relative to other occupations.
5
Computed as 100*(exp(1.94)-1)=595.875%
H1: Wages in mining and non-tradable sectors increased relative to other occupations. A disproportionate increase in wages of individuals working in mining and nontradable sectors should have diverted the state support to the households whose members may have been hurt by the oil boom. We will test the hypotheses: H0: The amount of state benefits for workers in mining and non-tradable sectors stayed the same relative to other occupations. H1: The amount of state benefits for workers in mining and non-tradable sectors decreased relative to other occupations. and H0: The share of state benefits in total income for workers in mining and non-tradable sectors stayed the same relative to other occupations. H1: The share of state benefits in total income for workers in mining and non-tradable sectors decreased relative to other occupations.
Conclusions
Estimated Reserves
Details
SOCAR
1. Offshore PSA Azeri, Chirag, and Deepwater Gunashi Shah Deniz Lankaran-Talysh Yalama/D-222
5.4 billion barrels of oil 2.5 billion barrels of oil; 14 Tcf of natural gas 700 million barrels of oil 750 million barrels at Yalama field Exports began late 1997. Producing 144,000 bbl/d at Chirag field as of July 2004. First exploration well drilled at Azeri field. First phase of development approved in February 2003. Production to begin in Autumn 2006 First test well (2001) came up dry. Drilling to 4,400 meters failed to find commercial reserves. Lukoil announce it may abandon well but must drill at least one more well.
10% 10% 25% 20%
Absheron Oguz Nakhchivan Kurdashi-Araz-Kirgan Daniz Inam
Project closed. Chevron abandoned first exploratory well in 858 million barrels of oil; up to 100 Tcf of 2001. In November 2003, Chevron and Total paid $40 million in natural gas compensation rather than drill a second well as required under contract. Dry well drilled in April 2001. ExxonMobil announced plans to 290 million barrels of oil and 685 bcf of gas quit the project in April 2002. 750 million barrels of oil 730 million barrels of oil 2.2 billion barrels of oil ExxonMobil drilled one successful well, will drill a second well. First test wells drilled, with poor results. BP suspended drilling of its first appraisal well in Aug. 2001 due to high pressure. New well planned to be completed by 2005 In exploring phase in 2004. Confrontation with Iranian gunboat in July 2001; exploration suspended, pending resolution of Caspian Sea borders between Azerbaijan and Iran. Seismic work being undertaken. Second well at the Yanan-Tava field, part of a concession that also includes Ateshgah and Mugan-Deniz, struck gas, but not enough to be commercial. In June 2003, JAOC announced it would leave Azerbaijan. Exploration D-43, D-44, and D-73 blocks Exploration D-9 and D-38 blocks. Reached final drilling point in September 2004, well likely to be shut down due to abnormally high pressure, and Exxon-Mobil failed to reveal commercial hydrocarbon reserves. Production averaged 2,700 bpd of oil in 2004. Plans exist to raise level to 3,200 bpd Drilled 4 wells 1998-1999. Oil Production averaged 77,000 bpd in 2004. Gas production averaged 1.1 mcf/day for 2004 Production averaged 1,000 bbl/d in 1999.
50% 50% 50% 50% 50%
Araz, Alov, and Sharg
4 billion barrels of oil
40%
Atashgah Lerik, Jenab, Savalan, Dalga Zafar-Mashal 2. Onshore PSA
600 million barrels of oil in Atashgah, Mugandeniz, and Yanan Tava fields
50%
1 billion barrels of oil
50% 50%
1-2 billion barrels of oil, 1.8 tcf gas
Kalamaddin-Mishovdagh 200 million barrels of oil Anshad Petrol AzGeroil Southwest Gobustan Zykh-Govsany Kursangi-Garabagli Muradkhanli-JafarliZardab Padar-Kharami Shirvanoil
219 million barrels at Neftchala, Khilly, Babazanan 140 million barrels at Ramany, Balkhany, and Sabunchi fields
15% 51% 51% 20% 50% 50% 50% 15% 49%
Gas found at onshore East Duvanny field, well tested at a flow 147 million barrels of oil; up to 7 trillion cf rate of 8.6 million cubic feet/day. Azeri government estimates of natural gas oil should be found in offshore portion of block. Rehabilitating fields; produced 1,830 bbl/d in 2000. Contract 66-150 million barrels of oil start date pushed back due to environmental issues. 10 additional wells drilled in 2003 to increase production; fields 730 million barrels of oil producing 6,600 bbl/d in June 2004 1st test well at Muradkhanli shut down in April 2001. CNPC 730 million barrels of oil won a tender to develop the block, although no new PSA has been signed yet. 580-750 million barrels of oil 650 million barrels of oil at Kyurovdag field Seismic work being undertaken. Rehabilitating existing wells since 1997. Oil production averaged 57,000 bpd in 2004. Gas production was 1.5 mcf/d SOCAR moved to take over the concession in December 1999 following BMB's request to suspend operations.
West Absheron (Karadag200 million barrels of oil Kergez-Umbaki fields)
Table 1: Real Wages by Sector
Increase 1995 - 2002 274775 64643 482820 255573 33929 71034 462207 917385 280832 1263530 281342 543166 447596 298815
1995 Country Average Agriculture Construction Electrictricity, Gas and Water Supply Fishing Health and Social Services Hotels and Restaurants Financial Intermediation Manufacturing Mining 54384 29208
1996
1997
1998
1999
2000
2001 263995 80152 443652 328345 54493 81871 425906
2002 329158 93851 568899 338765 63137 93934 505046
93249 153272 180848 181086 221606 39413 53645 47201 63963 69081
86079 174441 324477 474185 420846 416663 83192 152519 250294 297496 218552 267932 29208 22900 42839 39413 48919 53645 73417 47201 75233 52332 69833 52103 73395
78396 122924 129849 337239 314858
147187 176673 258417 357356 580442 632626 83192 152519 250294 297496 239742 284272
734749 1064572 307832 364024
83192 152519 250294 297496 618908 814540 1107543 1346722 72288 118414 17719 178511 1945528 226862 548391 151953 335367 331415 690354 480260 384344
Public Administration and 50073 Defence Real Estate Trade Transportation, Storage and Communication
147187 176673 258417 357356 270033 334051 32664 52798 72957 81382 115844 119284
85529 156982 245389 312506 245422 292421
Table 2: Real Wages Relative to the Country Average
Increase 1995 - 2002 -0.25 3.11 -0.37 -0.16 0.70 3.51 -1.33 -0.22 2.70 -0.36 -0.86 0.47 -1.82
1995 Agriculture Construction Electrictricity, Gas and Water Supply Fishing 0.54 2.95 0.97 0.35
1996 0.42 4.43 0.87 0.26 1.24 1.60 2.25 0.86 1.00 0.47 2.44 0.30 2.97
1997 0.35 6.05 0.77 0.21 1.37 1.67 2.10 0.97 1.00 0.47 2.18 0.28 3.36
1998 0.26 10.05 0.63 0.16 1.59 1.73 2.75 0.83 1.00 0.06 20.17 0.23 3.84
1999 0.35 6.58 0.52 0.24 1.33 4.83 1.72 0.41 2.58 0.29 1.51 0.43 2.12
2000 0.31 6.03 0.64 0.19 1.41 4.29 2.01 0.45 2.87 2.39 0.17 0.36 2.45
2001 0.30 5.54 0.74 0.17 1.50 5.20 1.73 0.42 3.60 0.20 2.42 0.28 2.21
2002 0.29 6.06 0.60 0.19 1.49 5.38 2.11 0.34 3.70 0.25 2.08 0.70 0.80
Health and Social Services 0.78 Hotels and Restaurants Financial Intermediation Manufacturing Mining Public Administration and Defence Real Estate Trade Transportation, Storage and Communication 1.87 3.44 0.57 1.00 0.60 2.94 0.22 2.62
Table 3: Changes in Log(Real Wages)
Coefficient Std Error P-Value 0.818 0.446 0.0677 1.511 0.446 0.0008 1.550 0.446 0.0006 1.752 0.446 0.0001 1.686 0.446 0.0002 2.036 0.446 <.0001 2.294 0.446 <.0001 2.499 0.446 <.0001 -0.345 0.258 0.650 0.913 0.831 1.233 1.420 1.548 0.005 0.508 0.855 0.973 1.101 1.266 1.400 1.513 0.248 0.635 0.929 0.875 1.053 1.321 1.463 0.153 0.153 0.153 0.153 0.153 0.153 0.153 0.153 0.130 0.130 0.130 0.130 0.130 0.130 0.130 0.130 0.129 0.128 0.128 0.128 0.128 0.129 0.129 0.0245 0.0925 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.9683 0.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.0547 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
Baku
1995 1996 1997 1998 1999 2000 2001 2002 1995 1996 1997 1998 1999 2000 2001 2002 1995 1996 1997 1998 1999 2000 2001 2002 1996 1997 1998 1999 2000 2001 2002
Other Oil Rich Regions
Somewhat Oil Rich
Oil Poor
Table 4: Regional Distribution of Government Revenues, Expenditures and Transfers from the Centre
log(Revenues) Baku vs. No Oil Other Oil Rich Regions vs. No Oil Some Oil vs. No Oil Year Control 6.23 0.33 1.13 0.11 1.09 0.09 yes *** *** *** log(Expenditures) 3.45 0.02 0.33 0.06 0.16 0.05 yes *** *** *** 0.23 0.06 0.07 0.05 yes *** *** log(Transfers)
R-squared Mean of Dependent Variable Number of Observations
0.486 14.537 496
0.419 16.759 496
0.028 16.561 496
Table 5: Government Support to the Households
Dependent Variable: Baku vs. No Oil Other Oil Rich Regions vs. No Oil Some Oil vs. No Oil Household Size Urban vs. Rural log(Support from State) 1995 2002 3.799 *** -1.39 *** 0.67 0.27 1.55 *** 0.35 *** 0.59 0.24 0.89 * 0.66 0.53 0.34 0.62 *** 0.75 ** 0.08 0.06 0.61 -2.09 *** 0.48 0.22 0.057 3.062 1872 0.062 6.293 7686
R-squared Mean of Dependent Variable Number of Observations
Table 6: Differences in Government Support to the Households Relative to Rural Households in Resource Poor Regions
Dependent Variable: Baku Oil Rich, Urban Oil Rich, Rural Some Oil, Urban Some Oil, Rural No Oil, Urban Household Size log(Support from State) 1995 2002 4.36 *** -3.45 *** 0.59 0.26 2.39 *** -1.31 *** 0.73 0.31 1.21 *** -0.09 0.73 0.33 1.18 * -3.23 *** 0.79 0.51 0.94 2.07 *** 0.61 0.45 0.36 -1.94 *** 1.05 0.32 0.63 *** 0.74 *** 0.08 0.06 0.057 3.062 1872 0.065 6.293 7686
R-squared Mean of Dependent Variable Number of Observations
Table 7: Dependency on Government Support Across Regions
Share of Government Support in Total Income 1995 2002 0.12 *** -0.02 *** 0.03 0.00 0.05 * -0.01 *** 0.03 0.00 0.05 * 0.02 *** 0.03 0.00 0.01 *** -0.01 *** 0.00 0.00 0.07 *** -0.03 *** 0.02 0.00 0.039 0.281 1872 0.049 0.101 7686
Baku vs. No Oil Other Oil Rich Regions vs. No Oil Some Oil vs. No Oil Household Size Urban vs. Rural
R-squared Mean of Dependent Variable Number of Observations
Table 8: Dependency on Government Support Relative to Rural Households in Resource Poor Regions
Share of Government Support in Total Income 1995 2002 0.18 *** -0.05 *** 0.03 0.00 0.12 *** -0.03 *** 0.03 0.00 0.02 -0.03 *** 0.03 0.00 0.11 *** -0.05 *** 0.04 0.01 0.03 0.04 *** 0.03 0.01 0.00 -0.04 *** 0.05 0.00 0.01 *** -0.01 *** 0.00 0.00 0.041 0.281 1872 0.062 0.101 7686
Baku Oil Rich, Urban Oil Rich, Rural Some Oil, Urban Some Oil, Rural No Oil, Urban Household Size
R-squared Mean of Dependent Variable Number of Observations