Resource Wealth: A “Curse” for Labor Rights?

Natural Resource Wealth and Labor Rights

While scholars have examined various manifestations of the resource curse, two dynamics—voracity effects and Dutch disease—are particularly influential in explaining its economic impact. Voracity effects delineate how states and political elites “emphasize discretionary fiscal distribution” of resource revenues (Tornell & Lane, 1999: 22; see also Vadlamannati & de Soysa, 2016; Snyder & Bhavnani, 2005), resulting in rent-seeking behaviors and patronage spending. Resource wealth can also create the phenomenon known as the Dutch disease as resource production crowds out the production of tradeable goods and encourages expansion in non-tradeable sectors such as construction and services (Corden & Neary, 1982; Ross, 2008). These phenomena are ultimately linked to multiple problems, including a lack of genuine savings (Hamilton & Clemens, 1999), a misallocation of productive capital between economic sectors (Gamu et al., 2015), and a growing shadow economy (Blanton & Peksen, 2021).

Bringing this to bear on worker rights, resource-intensive industries are problematic from a labor perspective; there are numerous cases of labor rights abuses in this sector, including hazardous working conditions, the use of forced labor, and outright violence against employees (Amnesty International, 2020; Human Rights Watch, 2019). Moreover, the nature of extractive industries themselves, particularly their fixed location as well as the volatility of commodity prices, can serve to magnify the intensity of labor-related conflicts, which can encourage the militancy of labor groups as well as violent intervention on the part of the state. As evidenced by the 2012 violence in platinum mines in South Africa, in which 34 striking workers were killed by local security forces, such repression can occur even within democratic societies that otherwise demonstrate a “robust commitment to core labor standards” (Evans & Sil, 2020: 993). More broadly, cross-national studies have found resource-intensive foreign direct investment (FDI), and the increased presence of resource-intensive firms in general, to be associated with greater levels of human rights abuse (Vadlamannati et al., 2021). We posit that these problems have spillovers for society as a whole, due to the “insulation” resource wealth provides states from domestic and international pressures to protect labor rights.

Domestically, extant work suggests two ways in which resource wealth can insulate states from demands to protect worker rights. First, resource rents enable states to bear the economic costs of repression. Along rationalist lines, the repression of human rights, including labor rights, creates both potential costs and benefits to states, and states weigh these factors in deciding how to respond to societal demands for increased rights (e.g., protests or strikes). Economically, a primary cost of repression can be foregone tax revenues as “citizens who experience repression are simply less able to contribute, since the abuse reduces their own earnings and therefore the amount of income on which they pay taxes” (DeMeritt & Young, 2013: 102; see also Conrad & DeMeritt, 2013). To the extent that rentier states rely on resource rents rather than their citizenry as a source for generating revenue, they are insulated from the economic costs of repression and thus more likely to employ such measures.

In addition to freeing states from the potential economic costs of repression, resource revenues provide a means to fund the expansion of security forces, which can enable states to protect their own political power against domestic threats (Ishak, 2019). Worker movements can ostensibly threaten the existing order and short-term business interests. As such, labor disputes in resource-intensive industries can quickly be viewed as a “high stakes” conflict (Evans & Sil, 2020: 998) and security forces can provide a means to suppress labor rights. Indeed, studies have found that resource-wealthy states are willing and able to repress the personal integrity rights of their populace (DeMeritt & Young, 2013) as well as the broader array of “private liberties” that are associated with freer regimes (Wigley, 2018).

For our purposes, there are numerous instances of rentier states directly restricting labor rights. Resource-wealthy states such as Kazakhstan and Algeria, for instance, rank among the “world’s ten worst countries for workers” due in large part to widespread labor rights abuses. Such abuses can range from capricious administrative hurdles such as “arbitrary and excessive criteria” for union recognition to more overt confrontations, including the dissolving of unions as well as harassment and even physical attacks on union leaders (International Trade Union Confederation, 2019: 22; ITUC 2019).

Internationally, the fixed location of many natural resources, as well as the relatively inelastic demand for these goods, enables resource-rich states to ignore global pressures to improve labor rights. As is the case with other facets of human rights, there are international pressures for the protection of labor rights. The ILO “boasts an elaborate set of committees and procedures…to monitor state compliance” with its conventions (Koliev & Lebovic, 2018: 437) and labor rights conditions are increasingly a part of trade expansion measures such as preferential trade agreements (Hafner-Burton et al., 2019; Wang, 2020). Less formally, a growing body of literature has covered how a variety of actors, including social movements and non-governmental organizations (NGOs), exert pressures through widely publicizing or “shaming” states or corporations for violating human rights (i.e., Lebovic & Voeten, 2006; Murdie & Peksen, 2015). Such unwanted attention can impose considerable economic cost, as shaming has been shown to have negative effects on corporate profitability as well as the ability of states to attract foreign investment (e.g., Barry et al, 2013; Bartley and Child 2011).

However, the demand structure for natural resources enables rentier states to be much less responsive to such pressures. Specifically, demand for natural resources such as oil and natural gas are relatively inelastic, as the short-term substitutability of these resources is practically non-existent while short-term demand is largely constant. This gives sellers a modicum of leverage over potential buyers and renders them less dependent on the need to make any type of “costly commitments” to access global markets or attract FDI (Ross and Voeten 2016: 87).

There is evidence that such insulation has substantial effects on the policies of rentier states. First, they are less prone to engage in some of the leading types of international cooperation, including membership in intergovernmental organizations (IGOs), adherence to rulings of international judicial bodies, and the arbitration of investment disputes. Simply put, these states can afford to remain “at odds with international norms and institutions” (Ross and Voeten 2016: 85). Such insulation is also apparent in the way in which rentier states deal with prospective investors, as they are less concerned about providing a risk-free investment climate in which contracts are honored and states do not interfere in business operations. In short, in addition to insulating countries from the need to commit to cooperative measures with other countries or IGOs, resource wealth also “insulates the government from behavior that could drive away foreign investors” (Jensen & Johnston, 2011: 664).

Taken as a whole, such “insulation” bodes poorly for labor rights, as resource-rich states are less likely to be responsive to domestic or international pressures related to these rights. We thus postulate that:

H₁: Resource wealth has a negative impact upon worker rights.

Indirect Effects of Resource Wealth through Civil Society and Bureaucratic Capacity

Our first hypothesis delineates how resource wealth can more directly manifest itself regarding labor rights abuses. Yet extant literature reveals a multitude of other economic and political effects of resource wealth. We posit that some of these consequences of resource wealth, particularly its adverse effects on civil society and bureaucratic effectiveness, might in turn influence labor rights. We thus posit that in addition to its direct impact upon labor rights, resource wealth’s adverse effects on civil society and bureaucratic effectiveness constitute two indirect pathways through which resource wealth further undermines labor rights.

First, extant literature leads us to expect that resource wealth negatively effects civil society in general as well as labor organizations in particular. Ideally, an active civil society can serve to hold states and corporations accountable and thus mitigate some of the negative impacts associated with resource wealth. For example, Haslam (2016) examined how private business associations in the Chilean mining sector served to decrease rent-seeking behavior on the part of the state, while Falleti and Riofrancos (2018) traced how participatory institutions involving indigenous groups in Ecuador were able to both ameliorate prospective environmental damage by natural gas companies as well as ensure revenue-sharing with local communities.

However, broader economic and political factors associated with the resource curse serve to undermine the formation of such groups. Economically, the Dutch disease undermines societal mobilization for labor rights. Increased resource wealth triggers a rise in the real exchange rate, as well as a transformation of the economy from traded sectors such as manufacturing to nontraded sectors such as construction and services (Corden & Neary, 1982; Ross, 2008). This creates higher wage levels within the nontraded sectors, as resource rents and high exchange rates increase the spending power of employees within these sectors. These dynamics delay and stunt industrialization, economic reform (Mazaheri, 2016), and in some cases, women’s rights (Ross, 2008). ¹ It also has implications for civil society, particularly labor organizations, as the unduly high prevailing wage rates discourage wage-related labor mobilization within the nontraded sectors, essentially obviating one of the primary rationales behind collective bargaining organizations.

Voracity effects add a political dimension to these dynamics, as patronage spending creates a largely bifurcated business environment unlikely to mobilize in support of labor rights. Resource wealth incentivizes business elites to “forge bonds based on rent-seeking” with the state (Mazaheri, 2010: 3). This creates a mutually beneficial situation wherein business and government elites collude to preserve the status quo, as states use rents to “‘buy off’ business elites in order to exercise greater decision-making power and acquire their political support” (Mazaheri, 2016: 49; see also Herb, 2014; Moore, 2004; Shambayati, 1994; Crystal, 1995). For their part, the business elite receives preferential treatment from the government, including enhanced access to credit, market access, and freedom from regulatory scrutiny. ²

Non-elites within the business community, particularly small and medium-sized enterprises (SMEs), face a dramatically different environment. Absent access to state largesse, they are “doubly disadvantaged” as they “are forced to compete with rent seeking elites” in the market while dealing with “a government that adopts policies that impose additional barriers to their activities” (Mazaheri, 2010: 4). While the environments faced by elites and non-elites are quite different, a common outcome is a dearth of mobilization, particularly in the area of labor. Non-elite firms lack the political and economic will to push for worker rights. As firms are primarily interested in their survival in a highly disadvantageous environment, they would likely be less concerned with worker rights. To the contrary, they are often forced to seek labor from the less-protected informal sector (Blanton & Peksen, 2021). Though elite businesses ostensibly have the power to push for worker rights, they have a vested interest in maintaining the status quo and are unlikely to push for increased regulation over their own industries.

Ostensibly, labor unions—a key part of civil society—are vital in mobilizing to support worker rights. The characteristics of the resource sector itself, particularly its geographic concentration and capital-intensive nature, are certainly amenable to the formation of strong unions (Shafer, 1994). Moreover, case studies show that these unions are often willing to employ militant actions on behalf of their employees (Bergquist 1986; Evans, 2022; Evans and Sils 2020). Yet the benefits of such advocacy may not extend to the broader workforce. Resource revenues can enable states to essentially co-opt unions in a manner similar to business elites, producing a “labor aristocracy” in which the “unions exercise their political clout to protect their privileged position” (Karl, 1997: 56). Thus, powerful unions can exist in a regime that is otherwise not conducive to the protection of labor rights.

Ultimately, this environment reinforces a dualist economic system by creating a cadre of well-paid elites alongside a workforce that has little in the way of political power or legal protections (Karl, 1997, 2004; see also Assaad, 2014). In addition to economic costs such as reduced entrepreneurship (Mazaheri, 2010, 2016) and increased informal economic activity (Blanton & Peksen, 2021), these dynamics have implications for labor rights. By using resource rents to reward a small group of elites, state patronage undermines broader societal mobilization as powerful groups are co-opted by the state while outsiders lack the will and capacity to effectively organize for change. Mobilization for worker rights is thus undermined as the state uses “its largesse to prevent the formation of social groups that are independent of the state and may be inclined to demand political rights” (Ross, 2001: 334). ³

As these patronage networks deepen, they “make it harder for independent civil-society groups to take root” (Ross, 2012: 5). Reliance on patronage spending—rather than broad-based investment in productive capital—creates a situation where “an independent middle class fails to develop, and technocratic and entrepreneurial talent remains captive of state largesse in terms of employment and advancement opportunities” (Dietz et al., 2007: 38; see also Chaudhry, 2015). Such a “captive” workforce lacks either the incentive or the power to mobilize and form independent civil society groups such as labor organizations.

This discussion leads to our next hypothesis:

H_2A: Resource wealth negatively affects labor rights through its adverse effect on civil society.

Resource wealth might also undermine labor rights through its negative impact on bureaucratic effectiveness. Bureaucracies play a key role in the front-line protection of societal rights (Cingranelli and Filippov 2020). Labor inspectors are the most common “street-level bureaucrats” in charge of protecting labor rights, through the monitoring of labor conditions and enforcement of labor regulations. They “have the power to set standards, incentivize behavior and compel firms to undertake improvements” as well as “the unique power to legitimize the claims of workers and worker organizations to a broader society” (Amengual and Fine 2017: 132). Along these lines, the lack of bureaucratic capacity in this area, particularly the shortage of qualified professional inspectors, is a key problem underlying the “global reality of poor labor standards” (Amengual and Fine 2017: 130). Labor inspectors are not the only actors involved in the “sub-politics of regulatory enforcement” (Coslovsky et al., 2011) as labor and corporate leaders, and particularly prosecutors, play key roles. Yet the effective enforcement of labor rights cannot happen without “empowering domestic agencies with a mandate to prevent abuses” (Hobbes, 2015: paragraph 57).

The empowerment of such professionalized and independent regulatory agencies requires political will as well as the investment of resources on behalf of the state. However, fiscal policies associated with voracity effects undermine state capacity in this area and thus hamper the enforcement of labor rights. On the taxation side, as resource revenues are a convenient tax handle (source of state revenues), the government has little need to raise funds by taxing its citizenry. This can be problematic as it reduces the need for the state to defer to citizen preferences (Bates & Donald Lien, 1985; Paler, 2013) and for citizens to demand accountability from the state (Ross, 2001).

Spending patterns associated with voracity effects are particularly important for our purposes, as they serve to undermine the bureaucratic capacity of rentier states. Specifically, the emphasis on patronage spending, which goes primarily to enrich a relatively small group, leaves rentier states with little incentive to build effective bureaucracies (Besley and Persson 2010). To the contrary, as effective state agencies imply a degree of transparency and broad-based accountability, as well as potential constraints in carrying out patronage spending priorities, elites may view impartial and professionalized bureaucracies as an impediment to rent-seeking opportunities (Ross, 2001; Vadlamannati & de Soysa, 2016). In cases where these vested interests are particularly entrenched, rentier states might take on these perceived impediments and simply “use state bureaucracy…to buy off or beat down political opposition” who are seeking such accountability (Anthonsen et al., 2012: 151).

This bodes poorly for the protection of labor rights. An effective regulatory regime requires an investment on the part of the state, including a cadre of well-trained regulators and administrators, as well as state agents who are sufficiently independent to hold employers—a group that is often politically powerful—accountable. Thus, to the extent that resource wealth undermines the ability and willingness of a state to invest in effective and independent bureaucracies, the protection of worker rights is likely to suffer. This leads us to our next hypothesis:

H_2B: Resource wealth negatively affects labor rights through its adverse effect on bureaucratic capacity.

Outcome Variables

Labor rights data are from the Work R dataset (Barry et al., 2022), available for 1994–2010. Given the prevalence of decoupling in the labor rights literature—that is, upholding labor standards in law but failing to enforce them in practice—we follow analytic convention (i.e., Mosley and Uno 2007; Ye, 2020) by including two outcome variables, Labor Rights Practices (de facto) and Labor Rights Laws (de jure). Both are index variables that account for labor conditions in practice and in law, respectively. They are ordinal measures, ranging from zero to 14 with higher scores denoting better de facto or de jure labor rights. The de facto labor rights variable measures the extent to which the following labor rights are protected in practice: minimum age of employment, minimum wage, occupational safety and health, reasonable limitations on working hours, forced or compulsory labor, the right of association, and the right to collective bargaining. The de jure labor rights variable indicates the existence of the same labor rights in law. Data are derived from the US State Department Country Reports on Human Rights Practices. The list of labor rights is grounded in the core International Labor Rights Organization conventions, the International Covenant on Civil and Political Rights (particularly Part III Article 22), and the International Covenant on Economic, Social and Cultural Rights (particularly Part III, Articles 7 and 8).

The Work R dataset is optimal for our analysis, as it is the most comprehensive data on labor rights. Given the importance of decoupling in the examination of labor rights, it is also useful that it differentiates between labor rights laws and practices. Specifically, one major advantage of the dataset is it allows us to assess the extent to which resource wealth potentially undermines both the legal environment surrounding labor conditions and the enforcement of existing labor rights. ⁴

Explanatory Variables

To examine the impact of resource abundance on labor rights conditions, we use the Resource Wealth per capita variable from the World Development Indicators database (World Bank, 2018). The variable captures the total amount of rents from oil, natural gas, coal, minerals, and forestry for each country year. We log the measure to correct for the skewness of the data. As a common practice in the literature, we operationalize resource rents on a per capita basis rather than as a percentage of gross domestic products (GDPs) as this better accounts for the domestic consumption of resources and the relative significance of resource rents in underdeveloped and conflict-ridden countries (Wigley, 2018). ⁵

We also include a battery of control variables drawn from extant literature. As wealthier economies on average tend to have fewer labor rights abuses, we control for Economic Wealth (the natural log of GDP per capita). The data come from the World Development Indicator database (World Bank, 2018). We also include a measure of Economic Openness. The measure is drawn from the KOF Economic Globalization Index, which captures the size of trade and financial flows and regulations on a 100-point scale with higher scores indicating more economic globalization (Dreher, 2006; Gygli et al., 2019). We control for Financial Crises to account for economic stability in a given country, as crises tend to lead to a surge in labor rights abuses (Blanton et al., 2015). The variable is coded one if a country is facing at least one of the three major financial crises—banking, currency, and debt crises—in a given year, and zero otherwise. The crises data are from Laeven and Valencia (2013).

Turning to the political factors that impact labor rights, as democracies tend to have higher levels of respect for socio-economic and political rights, they likely have better labor rights practices and protections (Mosley and Uno 2007). We use the Polity IV dataset’s Polity variable that varies from −10 to 10 with higher scores pointing to more democratic institutions (Marshall et al., 2010). It is worth noting that we use the Polity measure to be consistent with earlier labor rights research that has also widely used it (e.g., Blanton et al., 2015; Mosley and Uno 2007; Wang, 2020). Moreover, with its emphasis on institutional measures of democratic governance as opposed to civil and political rights, it enables us to make a better distinction from some of the mediators used in our model. That said, there was no change in our main results when we use an alternative democracy score from the Varieties of Democracy database (Coppedge et al., 2019). As left-leaning political parties often favor labor-friendly policies and practices, we control for the ideology of political parties in power using the binary Government Ideology variable drawn from the Database of Political Institutions (Beck et al, 2001).

We also control for Population (in millions) with the expectation being that more populous societies are more vulnerable to violations due to competition over fewer jobs and the increased difficulty of enforcing labor laws and regulations. This measure is drawn from the World Development Indicator database (World Bank, 2018). To account for possible adverse effects of political violence and conflict, we include a measure of Violent Conflict. The variable, drawn from the UCDP/PRIO Armed Conflict Dataset (Gleditsch et al., 2002), is coded 2 for civil conflicts with 1000 or more battlefield deaths, 1 for conflicts with battle-related deaths between 25 and 1000, and zero otherwise. The recent history of each country’s labor rights laws and practices is likely to affect its current labor conditions. To control for such temporal dependence, we include the one-year-lagged labor rights variables in the model. We use the Huber/White sandwich estimator of variance to obtain robust standard errors. ⁶

In testing for the direct impacts of resource wealth on labor rights, for each outcome variable, we report an ordinary least square (OLS) model with random-effects (RE). To account for potential missing variable bias and to ensure that unobserved country-specific factors do not alter our findings, we next present a model with country fixed-effects (FE). ⁷ While our focus is on the impact of resource wealth and labor rights, it cannot be assumed that resource wealth is not endogenous to broader sociopolitical problems (Menaldo, 2016; see also Brooks and Kurtz 2016). To account for these potential concerns, we also include generalized method-of-moments (GMM) models (Asiedu and Lien 2011; Roodman, 2009). The GMM approach “takes the first difference of the data and then uses lagged values of the endogenous variables as instruments” (Asiedu and Lien 2011: 104). For our purposes, GMM is particularly useful as it allows for the incorporation of multiple endogenous variables in the same model without requiring external instruments. Finding appropriate external instruments is a particularly challenging task in our case as we would need instruments applicable to the different natural resources (i.e., oil, gas, minerals, coal, and forestry) to be captured by our explanatory variable.

Causal Mediation Analysis: Mediators and the Estimation Technique

To assess the hypotheses that civil society and bureaucratic capacity might mediate the impact of resource wealth on labor rights, we run a causal mediation analysis (Imai et al., 2011; Tingley et al., 2014). Substantively, causal mediation analysis allows us to assess both the direct impact of resource wealth on labor rights as well as the extent that resource wealth influences labor rights through its effects on civil society and bureaucratic capacity. Statistically, the causal mediation approach provides information on not only the average total effect (ATE) and the average direct effect (ADE) of resource abundance on labor rights, but also the average causal mediation effect (ACME) of resource wealth through our mediation factors. Whereas ACME specifically captures the indirect effects of resource wealth through the mediators, the ADE accounts for the effects through other possible causal mechanisms. The estimates are calculated using 1000 simulations and with 95% confidence intervals.

Causal mediation analysis is a two-equation model in which the mediator equations control for the covariates of labor rights practices while the other equation predicts the covariates of the mediator variable under consideration. It first fits the model for each mediator and labor rights (outcome variable) and subsequently simulates model parameters based on their sampling distribution. That is, it simulates the likely outcomes based on the simulated scores of the mediators. Following this process, the model calculates the causal mediation effects of our measures on the resource wealth-labor rights relationship. Each of the mediators is included one at a time along with all the explanatory variables used in the direct effect analysis discussed above.

Turning to the specific mediating variables, we employ two different proxies to estimate how civil society might be linked to promotion of labor rights. Our first mediator is a broad-based Civil Society variable. The index, gathered from the Varieties of Democracy database (Coppedge et al., 2019), captures the aggregate amount of influence that non private or public-sector organizations, including labor unions as well as other civil organizations and social movements, have on key political and socio-economic decisions made by policymakers. It ranges from zero to one with higher scores denoting more civil society participation. Our second civil society-related mediator is Labor Organizations, which counts the number of labor rights organizations with international ties within a given country. The data come from Peksen and Blanton (2017) and were derived from the Yearbook of International Organizations. Compared to the civil society measure, the labor organizations variable more clearly focuses on the labor-related aspects of civil society, particularly the prevalence of labor movements connected with transnational advocacy networks. Previous research has found that transnational non-state groups have a role in various human rights outcomes, including physical integrity rights (Murdie and Davis 2012) and women’s rights (Murdie & Peksen, 2015). While the linkages between labor INGOs and labor rights have not been fully explored, the presence of these NGOs has been found to have a significant impact upon labor rights (Koliev & Lebovic, 2018).

Our third mediator, Bureaucratic Capacity, estimates the independence and effectiveness of bureaucratic organizations. As explained earlier, such institutions are particularly important for the front-line enforcement of labor laws and regulations. The index, drawn from the International Crisis Research Group (ICRG) (Knack and Keefer 1998), captures the efficacy of countries’ bureaucratic apparatus in fulfilling their administrative and policy obligations and protecting public interests over private interests. The variable ranges from zero to four with higher scores indicating a stronger bureaucratic capacity.

In the equations predicting the covariates of our mediators, we control for the following explanatory variables: Resource Wealth, Democracy, GDP per Capita, Economic Openness, Financial Crises, Social Globalization, and Violent Conflict. The expectation is that democratic and wealthy countries are more likely to have stronger civil society, organized labor, and bureaucratic establishment. Likewise, more economic and social globalization might spur strong civil society and bureaucratic accountability. Conversely, economic hardships and political instability associated with financial crises and violent conflict might undermine civil society and bureaucratic capacity. The social globalization variable comes from the KOF Social Globalization Index, which accounts for the extent of each country’s social engagement with the rest of the world (Dreher, 2006; Gygli et al., 2019). In the civil society and labor organizations equations, we also control for one-year-lagged values of Labor Rights Practices and Labor Rights variables as they are likely to be associated with the strength of civil society.

Sensitivity Analyses

We also ran alternate models to verify the robustness of the results across different sample selections and measures of our key variables. First, as wealthier countries on average have higher levels of labor rights protections and less reliance on resource rents compared to resource-rich developing countries, we repeated the models in Table 1 with a sample restricted to non-OECD countries. The results from the non-OECD sample confirm that our results are not likely driven by the global sample used in the main analysis (see Appendix Table A3).

While some of the resource curse literature focuses more narrowly on oil revenues, our measure captures five main natural resource types. To make sure that our results are not skewed by the inclusion of oil-rich states or (more specifically) member countries of the Cooperation Council for the Arab States of the Gulf (also referred to as the Gulf Cooperation Council or GCC), we run models that control for oil states as well as membership in the GCC (see Appendix Table A4). The oil state variable is a dichotomous measure that is coded one for states whose oil sales account for at least one third of their total exports, and zero otherwise. The GCC is coded one for six member countries (Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and United Arab Emirates) and zero otherwise. In the models controlling for these two additional variables, the resource wealth variable remains statistically significant, which suggests that the results are not driven by the inclusion of these states in our analysis.

To further explore whether our findings vary with the type of natural resource, we run a disaggregated analysis of resource wealth by individually controlling for gas, oil, minerals, coal, and forestry, as well as models employing a widely used measure of oil and gas revenues (Ross and Mahdavi 2015). We find some variation, as the impact of resource wealth is more significant for gas, oil, and minerals than coal and forestry. This does not negate the possibility that countries that are dependent on coal and forestry rents suffer from labor rights abuses, but that the industry-specific impacts may be greater in countries that are also dependent on rents from gas, oil, or minerals. ¹¹ However, we maintain that it is important to account for the overall resource wealth of each state, as we do in our main analysis, as many countries such as China, Russia and the Democratic Republic of Congo have an abundance of multiple types of natural resources. Thus, the combined effect of all resources, rather than one specific type of resource, could create a less labor-friendly environment. Results using the Ross and Mahdavi data are very similar to our main findings (Appendix Tables A5 and A6).

To ensure that our results are not an artifact of our measure of labor rights, we also run our models using labor rights data from Mosley and Uno (2007) and the CIRI worker rights measure (Cingranelli et al., 2014). Results show that the CIRI measure produces results essentially consistent with findings reported in the main analysis. The Mosley and Uno data, which focus only on collective labor rights—as opposed to both collective and substantive rights—yields somewhat different findings (see Appendix Table A7). We specifically find that though resource wealth negatively covaries with both collective labor rights laws and practices, it has a more significant impact on collective labor rights than practices. We also run models in which forced labor is the outcome variable (Table A10). Results show no significant linkage between resource wealth and this narrower measure of labor rights. While diagnostics (correlation coefficients and variance inflation factors) found no issue with multicollinearity in our models, to ensure that our findings are not a result of any partial correlations that may exist among our independent variables, we ran multiple models with reduced numbers of independent variables (see Appendix Table A8). We find similar results in models with fewer control measures.

Finally, a good deal of resource wealth literature focuses on the specific linkages between resource wealth and democracy (i.e., Ross, 2001). Given the importance of democracy to the literature, we run our main model with an alternate measure of democracy from the Varieties of Democracy (V-Dem) project (Coppedge et al., 2019). Results are similar to those reported in the main model. Finally, we run a mediation analysis to see if resource wealth undermine labor rights through its impact upon democracy (Table A11). Results show that while our resource wealth measure maintains its negative impact upon labor practices, democracy is not significant as a mediator. These findings also show that the impact of resource wealth on labor rights is independent of the impact that resource wealth has on democracy.