Why Gas-Rich Nations Don't Always Prosper: The Surprising Role of Trade and Governance

Natural gas can generate enormous economic value without delivering corresponding improvements in people's lives. Across producing countries, the relationship between gas wealth and human development is far from uniform. Some economies convert resource earnings into broader prosperity, while others experience unfavorable development outcomes despite their substantial natural endowments.

A new study, "For Whom Is Gas a Curse? Heterogeneous Effects of Natural Gas Dependence on Human Development: A Combined Approach of Double Machine Learning with Causal Forests," published in the journal Gases, investigates this divide. Authored by Luis Alfredo Ramírez-Vargas, Ana Lorena Jiménez-Preciado, José Álvarez-García and Francisco Venegas-Martínez, the research examines natural gas dependence across 65 producing countries, revealing how economic structures and institutional conditions shape development outcomes.

The key finding challenges conventional interpretations of natural resource wealth. Gas dependence has no statistically significant average effect on human development, yet the apparent neutrality conceals considerable differences between countries. Nearly 55% of country-year observations show negative estimated effects, while the remainder show positive outcomes.

The findings reposition a longstanding development debate. Rather than asking whether natural gas is inherently beneficial or harmful, the researchers investigate which countries experience adverse consequences, which benefit, and what distinguishes their economic circumstances.

The Resource Curse Has a Hidden Divide

Natural resource abundance has historically presented economists with a contradiction. Extractive industries can generate export earnings, public revenues and investment opportunities, but resource-rich economies do not necessarily achieve stronger development outcomes. Economic concentration, commodity-price volatility and institutional weaknesses have frequently been examined as explanations for this disconnect.

Natural gas introduces additional complexities. Its production and distribution require substantial infrastructure investment, including pipelines, liquefaction facilities and storage systems. Long-term commercial arrangements and the distinction between domestic consumption and international exports can influence how producing economies capture and distribute resource earnings.

Despite these distinctive characteristics, previous research has often examined natural gas alongside oil and other extractive resources. The new study isolates gas dependence and shifts the outcome of interest beyond economic growth to human development, encompassing health, education and living standards.

The researchers assembled a dataset covering 65 gas-producing countries, drawing on World Bank indicators, United Nations Development Programme statistics and international energy data. Although the broader dataset extends to 2025, the estimates cover 1996–2021 because comparable gas-rent information is unavailable for subsequent years. The principal analysis includes 1,285 country-year observations from 57 countries.

Using conventional statistical models alongside Double Machine Learning and causal forests, the researchers examine both average relationships and variations between countries. These techniques allow them to capture complex patterns that standard linear models may overlook, although unobserved influences remain a limitation.

The conventional analysis produces an almost negligible relationship between gas rents and human development. Double Machine Learning similarly estimates an insignificant average effect of approximately +0.0007 on the Human Development Index for a one-percentage-point increase in gas rents relative to GDP.

A different picture emerges when the analysis moves beyond averages. The causal forest estimates indicate negative effects in 55.2% of country-year observations and positive effects in 44.8%. These contrasting results help explain why earlier research has struggled to establish whether resource dependence consistently damages or improves development.

When Gas Dependence Becomes a Development Trap

The strongest warning concerns economies where gas revenues occupy a substantial share of national economic activity. The researchers identify eight countries characterized by relatively high gas dependence and negative estimated development effects. Their average gas rents amount to 4.82% of GDP, considerably exceeding those of more diversified producers.

A separate analysis identifies a turning point at approximately 4.81% of GDP, beyond which the estimated relationship between gas rents and human development becomes negative. This represents a statistical pattern rather than a universal threshold, but it reinforces concerns about economies becoming excessively reliant on a concentrated revenue source.

Geographic comparisons expose substantial differences. The Middle East and North Africa records a significantly negative regional estimate, while OECD producers show a positive relationship. These contrasting experiences suggest that the economic consequences of gas wealth depend heavily on the wider circumstances in which extraction occurs.

At the country level, Bahrain records an estimated effect of −0.0033, followed by the United Arab Emirates at −0.0025 and Malaysia at −0.0023. The United States, United Kingdom and Canada occupy the positive end of the distribution, with respective estimates of +0.0034, +0.0024 and +0.0019.

These estimates describe modeled changes in human development associated with greater gas dependence, not overall judgments about national prosperity. Their significance lies in demonstrating that the same resource can accompany sharply different development trajectories.

The researchers identify four groups of producing economies based on estimated effects and national characteristics. One group combines relatively high gas dependence with unfavorable outcomes, while another comprises five countries with strong governance, lower dependence and positive estimated effects.

For the latter group, gas rents average only 0.45% of GDP, compared with 4.82% in the more dependent group. The contrast indicates that the economic weight of gas extraction may be as important as the volume of resources available.

Trade Exposure and Governance Reveal an Unexpected Pattern

Greater trade exposure is commonly associated with access to markets, investment and technology, yet the researchers find that gas-producing economies with higher trade-to-GDP ratios tend to experience less favorable estimated development effects.

Trade openness exhibits the strongest correlation with these effects, at −0.57. Further analysis indicates that the estimated effect of gas rents declines as trade represents a larger share of economic activity, remaining negative among particularly trade-intensive producers.

The researchers offer several possible explanations. Smaller export-oriented economies may face greater exposure to international price fluctuations, concentrated export structures and pressures on other tradable industries. Larger economies that consume more gas domestically may instead capture benefits through industrial production and energy use.

However, the available data cannot establish which mechanism dominates. Importantly, a high trade-to-GDP ratio does not necessarily indicate a diversified economy or liberal trade policies. The findings concern external economic exposure, not evidence that international trade itself damages development.

Governance presents another important distinction. A simple comparison reveals almost no direct relationship between institutional quality and estimated gas effects. Once trade exposure and national income are considered, however, stronger governance becomes positively associated with more favorable outcomes.

The association is especially apparent among countries with very strong institutions. Government effectiveness and the rule of law also show positive direct relationships with human development, reinforcing the relevance of institutional performance.

Nevertheless, conventional models produce statistically insignificant interactions between governance and gas rents. The researchers consequently emphasize the more flexible causal forest findings while acknowledging that these associations do not demonstrate that governance improvements alone cause better outcomes.

Additional comparisons distinguish liquefied natural gas exporters, pipeline exporters and domestic-market producers. LNG exporters show more negative estimated effects on average, with approximately two-thirds of their country-year observations falling below zero. Yet differences between producer categories are not statistically significant, preventing firm conclusions about export arrangements.

The Development Challenge Beyond Gas Revenues

The findings carry particular relevance for developing economies where natural gas supports public finances and export earnings. Heavy dependence can leave national development strategies exposed to changes in international markets, especially when opportunities outside extractive industries remain limited.

The researchers identify export diversification, effective institutions and reduced fiscal exposure to external price fluctuations as important policy priorities. Their analysis supports treating gas revenues as potential resources for improving human capital rather than assuming extraction will automatically generate broader welfare gains.

The findings also intersect with the global energy transition. Natural gas has been characterized as a possible bridge toward lower-carbon energy systems, although its contribution to decarbonization remains contested. For producing countries, investment and revenue decisions must therefore be considered alongside uncertainty surrounding future energy demand and market conditions.

Important research limitations prevent overly definitive conclusions. The estimates end in 2021 and cannot capture subsequent gas-market developments. The analysis also lacks detailed information on contract structures, export composition, infrastructure arrangements and the allocation of gas revenues.

The researchers acknowledge that machine-learning techniques cannot eliminate unobserved influences. Their human-development measure may also conceal differences between health, education and income outcomes. Future investigations could use detailed export information, subnational evidence and infrastructure-level data to identify how resource revenues translate into development.