Energy as a substitute for factors of production: case study of an oil rich economy

ABSTRACT

Academic literature and policy notes place great emphasis on the relationship between energy carriers and factors of production. We have evidence from a subsidized economy that provides nationwide energy subsidies; with inferences that partly contradict previous findings. We use a panel of Iranian plants from 2004 to 2013. In 2010, the country conducted an energy reform to increase energy productivity. We use translog functions with SUR and GMM estimators, with clustered and robust standard errors. As a result, the elasticity of energy carriers for labor wages is positive, showing that energy is a substitute for labor on average. The substitution pattern of labor is weakened in parallel with the country's 2010 reform, but not lost completely. Regarding capital, a worsening financial accessibility to the economy after UN sanctions hit the economy in 2012 seemed to have an adverse effect on capital and energy relations.

KEYWORDS:

• Energy
• fossil fuels
• translog
• GMM
• manufacturing

Acknowledgments

We are grateful for the valuable comments by Masoud Nili, Seyed Ali Madanizade, Amine Mahmoudzade, Mohammad Vesal, Ali Jadidzade, and other seminar participants at Sharif University (September 2019). All mistakes are ours.

Table D1. Elasticity of share of energy with respect to price of labor and capital by size (SUR), labor<50.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 koe/$2018p (Kilogram oil equivalent per USD, at 2018 constant price and purchasing power parities).

2 Rahmati and Karimirad (2017) show that the extensive and intensive margins of Iran's manufacturing exportation are positively correlated with the oil prices, which could be an indicator of a comparative advantage induced by subsidized energy prices.

3 At the time, several politicians, including Gholamreza Mesbahi Moghadam, clarified that subsidized energy prices had a negative impact on quality of production and international competency, based on the article ‘Iranian subsidy reform plan’ in Wikipedia (https://en.wikipedia.org/wiki/Iranian_subsidy_reform_plan).

4 International Standard of Industrial Classification.

5 To explain macro situation, it is worth mentioning the banks' malfunctioning in 2010 and afterwards, which raised nominal interest rate up to 20%. The banking crisis in Iran started with the introduction of private banks in early noughties, i.e. 2000 to 2010, (the first private bank initiated in 2001), followed by a very high oil revenue specifically between 2007 to 2011. This caused a boom in Iran's petrochemical sector which raised the interest rate during that time period. At the same time, in lack of enough regulatory over the recently opened private banks, they started to offer high interest rates in order to collect as much resources as possible. The boom in oil and petrochemical sector almost crashed by the 2011 UN sanctions, and the banks' Ponzi game was almost stopped by tougher regulatory of the Central Bank. The central banks' tough monitoring reduced the interest rate which was raised before due to the banks' Ponzi game, however, at the same time, the overall financial accessibility was worsened. This partly explains the substitution of capitals with other factors in our dataset after 2010.

6 Costantini et al. (2018) highlight the connection between energy efficiency and employment dynamics and report that the ‘public actions towards energy efficiency may produce positive effects on employment dynamics’.

7 There are many other scholars who study energy market in Iran, including Rahmati et al. (2020) who study the introduction of natural gas pipelines on the consumption behavior of the households, Ghoddusi et al. (2018) who measure the impact of home-foreign price differences on the smuggling of gasoline, and Ghoddusi et al. (2019) who show that the price elasticity of gasoline smuggling is increasing as the price level rises; and scholars who use a variety of the Computable General Equilibrium models to study the impact of energy reform on social welfare, such as Shahmoradi et al. (2011) and Araghi and Barkhordari (2012).

8 Compared to the study by Nili (2010) we use a more recent dataset from 2004 to 2013, including the 2010 energy reform. Therefore, besides the above mentioned technical superiority, our findings may be affected due to the vastly spread of Dutch disease and banks malfunctioning during the course of our study. During 2007–2010 the oil prices passed $100 and the country faced with a very high stream of oil revenue which was a unique experience compared to prior revenues. Notably, when we restrict our data to 2004–2005 and follow the methodology of Nili (2010), we able to replicate similar results as theirs. Their methodology includes aggregation at 4-digits of ISIC code, not using time trends, and restricting the dataset to before 2006.

9 We follow Salehi Esfahani and Yousefi (2018) for data cleaning. Examples of other scholars who use the similar dataset are Karimirad and Rahmati (2016), Yousefi et al. (2020), Rahmati and Pilehvari (2018), Mahmoudzadeh, Nili, and Nili(2018).

10 Link to the CB's database: https://tsd.cbi.ir/

11 The Btu is an energy measurement which refers to the amount of energy required to increase the temperature of a pound of water by 1 degree Fahrenheit. To unitize the measurement of various energy carriers, it is common to report them in terms of Btu. For example, a Kilowatt-hours (kWh) is equivalent to 3412 Btu; meaning that the energy content of a kWh is equal to 3412 thermal units.

12 The dispersion of the self-reported prices are probably because of inaccurate reporting and geographical location. We cannot control neither of those, thus using the average reported price by the MoE.

13 This method consists of a step-by-step construction of the capital for each plant. We start by $k_0$which is approximated by the book value of the plant's capital at the initial year of data, then, $k_1$is the investment value (which is observed in the data) plus non-depreciated capital ($k_0(1-\delta )$). The depreciation rate is calibrated at 10%. If $k_0$is not observed for a plant, the $k_T$(the last observation) is taken as the starting point and the method is applied recursively. See Esfahani and Yousefi (2017) for more details.

14 Romer (2012, pp. 406) mentions that the user cost of capital is equal to the rental rate minus capital gain.

15 A few other alternatives may have been used by other scholars, such as mortgage interest rates, stock market growth rates, and industrial activity growth rates.

16 For example, land inflation was negative between 2009 and 2012. If we use the negative rate, it is equivalent to sell of capital when deflation is expected; however, the transaction costs and incompleteness of capital market doesn't allow firms to sell their capitals intensely and they have to bear with no gain from capital.

17 There are concerns whether homotheticity assumption is invalid for large firms. Nguyen and Reznek (1991) and Nguyen and Lee (2002) report that homotheticity is not violated by firms' size among US manufacturing. As a robustness test, our main findings are tested for two size categories of small and large firms (see Table A2).

18 A function of homogeneous of degree 1 is homothetic. Here, these assumptions is equivalent to parallel isoquants, which means that rate of technical substitution is the fixed for all levels of input.

19 All the variables of this study are real.

20 Though we control for the industries' fixed effects, we technically cannot control for all the about 30,000 dummies for distinct plants.

21 The average for all ISICs is 1.7%.

Additional information

Funding

This study is partially funded by the Institute of Management and Planning Studies.