The price and income elasticities of natural gas demand in Azerbaijan: Is there room to export more?

The study examines how income and price affect natural gas demand in Azerbaijan, a country where natural gas constitutes 69% of primary energy consumption and about 94% of electricity generation. Given Azerbaijan’s commitments to expand gas exports to Europe via the Southern Gas Corridor and domestic goals to reach high-income status after 2025, there is a potential conflict between rising domestic demand and export commitments. The research question asks: What are the long-run income and price elasticities of natural gas demand in Azerbaijan, and do these elasticities imply room to export more gas without jeopardizing domestic energy and electricity security? The paper addresses a literature gap for Azerbaijan, where price data are not readily available and past sample sizes hindered advanced estimation. By assembling a novel natural gas price series and applying multiple econometric techniques including STSM and Autometrics-based Gets, the study aims to provide robust, policy-relevant elasticity estimates to inform climate, energy, and export policy during the energy transition.

The literature on price and income elasticities of natural gas demand in resource-rich and developing contexts shows wide heterogeneity across countries, sectors, periods, and methods. Key findings include: (1) Pakistan (STSM): income elasticities <1 for industry (0.62) and residential (0.45), with price inelastic demand (−0.19, −0.13) (Javid et al., 2022). (2) OECD-Europe (STSM): long-run income 1.19, price −0.16 (Dilaver et al., 2014). (3) China: provincial and sectoral studies often find income-elastic demand (≈1.23–1.27) and variable price elasticity; some periods show insignificant or positive price responses linked to reforms (Dong et al., 2019); residents’ own-price inelasticity near −0.22 (Zhang et al., 2018); FGLS for 62 cities shows income elasticity ≈1.235 and price elasticity −0.779 (Yu et al., 2014). (4) Egypt (ARDL): long-run household income elasticity negative (−0.65) suggesting substitution toward alternatives like solar water heaters; long-run price elasticity positive (0.36) with short-run price −0.15 (Farag and Zaki, 2021). (5) Cross-country evidence: average long-run elasticities around price −1.25 and income ≥1 (Burke and Yang, 2016), implying sizable impacts of subsidy reform in below-cost pricing regimes. (6) Northeastern US: inelastic long-run demand across sectors (Gautam and Paudel, 2018). (7) Europe (2005–2020, ARDL): very low short- and long-run elasticities (income 0.09–0.14; price about −0.18 to −0.14), attributed to efficiency policies and updated data (Erias and Iglesias, 2022). (8) Ukraine: short-run price inelasticity around −0.16 due to limited substitution (Alberini et al., 2020). Overall, elasticities are context-dependent, with many studies finding natural gas as a necessity with inelastic price responses. This underscores the importance of country-specific estimates for Azerbaijan’s regulated, gas-dependent market.

Data and variables: Annual data 1993–2021. Dependent variable: natural gas consumption per capita (final consumption in cubic meters from BP Statistical Review 2022, converted to per capita using World Bank population). Key regressors: GDP per capita (constant 2015 US$, World Bank) as income; weighted real natural gas price constructed from regulated nominal tariffs (residential and industrial shares comprising ~77% of final consumption in 2021), deflated by CPI (World Bank, rebased to 2015). Price data are compiled from public sources (post-2007) and archives for earlier years; tariffs are set by Azerbaijan’s Tariff Council rather than market forces. Descriptives (per authors): Ngas mean 1050.11, SD 265.51; Income mean 3427.60, SD 1796.01; Price mean 0.34, SD 1.36. Econometric approach: (1) Test for unit roots using ADF and Phillips–Perron (PP) with up to 2 lags selected by Schwarz criterion. (2) Establish cointegration using Engle–Granger (tau and Z statistics) and ARDL bounds testing (Pesaran–Shin–Smith). (3) Estimate long-run elasticities via multiple complementary methods for robustness: General-to-Specific (Gets) with Autometrics multi-path search and saturation dummies, ARDL (bounds testing framework), Structural Time Series Modeling (STSM) with stochastic trends (maximum lag 2, selected by Schwarz and diagnostics), Fully Modified OLS (FMOLS), Dynamic OLS (DOLS; max lag 2 by Schwarz), and Canonical Cointegrating Regression (CCR). Core specification: ngas_t = a1 + a2·income_t + a3·price_t + unobserved components + u_t, where unobserved components capture factors such as technology and capital utilization. Model diagnostics (reported for Gets) include tests for autocorrelation (Godfrey), heteroskedasticity (White), ARCH (Engle), normality (Doornik–Hansen), and specification (RESET). All diagnostics support the adequacy of the preferred models.

- Unit roots and cointegration: ADF/PP indicate all series are I(1). Engle–Granger cointegration holds (Tau = −4.127, p = 0.04; Z = −21.949, p = 0.03). ARDL bounds F-statistic = 34.51 exceeds upper critical bounds at conventional levels, confirming a long-run relationship between natural gas demand, income, and price. - Long-run elasticities (Table 6): Across methods, income elasticity ≈ 0.77–0.82 and price elasticity ≈ −0.09 to −0.16. Specific estimates: Gets (income 0.791; price −0.110), ARDL (0.816; −0.118), STSM (0.772; −0.100), DOLS (0.730; −0.161), FMOLS (0.794; −0.093), CCR (0.779; −0.088). Coefficients have expected signs and are statistically significant (levels vary by method). - Interpretation: Natural gas is a normal, necessity good in Azerbaijan: a 1% increase in income raises per capita consumption by about 0.8% in the long run; a 1% price increase reduces demand by about 0.1%, indicating highly price-inelastic demand. The findings are consistent with evidence from other contexts (e.g., OECD-Europe, Pakistan) showing small long-run price elasticities (≈ −0.2). - Policy relevance: Price-based conservation is limited in the short to medium run due to inelastic demand and few close substitutes, especially for households and industry. Export expansion is more plausibly achieved by freeing gas from power generation (which uses the largest share) via accelerated deployment of renewables.

The research question concerns whether Azerbaijan can expand natural gas exports without jeopardizing domestic supply, given its heavy gas dependence for electricity (about 94%) and the commitment to raise exports to Europe. The confirmed cointegration and robust long-run elasticities show that domestic demand will rise substantially with income growth (elasticity ≈ 0.8), while demand responds only weakly to price increases (elasticity ≈ −0.1). Thus, relying on higher domestic gas prices to curb demand is unlikely to create significant exportable surplus in the short run and may harm industrial competitiveness and household welfare. The results suggest that the most effective strategy to create room for exports is to substitute natural gas in power generation with renewable energy sources, thereby decarbonizing electricity while preserving energy and electricity security. In the longer term, gradual pricing reforms and development of close substitutes (renewables, electrification technologies) may enhance responsiveness, but transition policies must be carefully sequenced to avoid adverse macroeconomic and social impacts. The findings align with international evidence on inelastic natural gas demand in regulated or subsidy-affected markets and add country-specific estimates for Azerbaijan to inform energy transition, subsidy reform, and export policy debates.

The paper compiles a novel regulated price series and applies multiple econometric methods to estimate Azerbaijan’s natural gas demand elasticities. All approaches converge on a long-run income elasticity around 0.8 and a price elasticity around −0.1, classifying natural gas as a normal necessity good. Given price inelasticity, raising tariffs alone will not deliver substantial efficiency gains or exportable volumes in the near term. Policymakers should prioritize substituting gas in power generation with renewables to free volumes for export while enhancing electricity and energy security and supporting decarbonization. As Azerbaijan pursues high-income status, income-driven demand growth will otherwise constrain export potential. Future research could: (i) estimate substitution elasticities between renewables and natural gas in electricity generation; (ii) employ macroeconometric or DSGE frameworks to assess general equilibrium effects and policy trade-offs; (iii) explore sectoral heterogeneity and the impacts of pricing reforms and carbon policies (e.g., carbon/methane taxes, ETS).

- Single-equation, partial-equilibrium demand model may omit broader macroeconomic feedbacks; general equilibrium or macroeconometric models (e.g., DSGE) could provide richer system-wide insights. - Regulated natural gas prices required compilation from scattered public sources and archives; despite careful construction (deflation by CPI and weighting by sector shares), measurement error cannot be fully ruled out. - Annual frequency and sample size (1993–2021) may limit detection of short-run dynamics and sector-specific behaviors; sectoral disaggregation was not modeled econometrically. - Unobserved components (e.g., technology, capital utilization) are proxied via stochastic trends or saturation dummies rather than explicit observables.