I am grateful to the editors of Politics & Gender for organizing this exchange. I agree with many of the points raised by Teri Caraway, Mounira Charrad, and Alice Kang, but disagree sharply with Pippa Norris.

The central argument in my 2008 article, “Oil, Islam, and Women” is that in many developing countries, oil wealth tends to keep women out of the formal labor force and, consequently, reduce their political influence (Ross Reference Ross2008). The argument is based on three claims: that in low- and middle-income countries, women commonly enter the labor force by taking jobs in low-wage, export-oriented factories; that holding these jobs can empower women socially and politically; and that when a country has substantial petroleum (or other mineral) wealth, it makes these kinds of industries unprofitable, and more broadly discourages women from working outside the home. As a result, petroleum wealth in low- and middle-income countries tends to shut down an important route toward gender equality.Footnote ¹

While the argument applies to all regions of the developing world, it is most salient for the Middle East and North Africa, where women have made less progress toward joining the labor force, and gaining political office, than in any other region. Some observers fault the region's Islamic traditions (e.g., Inglehart and Norris Reference Inglehart and Norris2003). My article suggests that oil wealth does a better job of explaining a) why the Middle East is different from other regions and b) why the status of women varies so dramatically among Middle Eastern countries.

Caraway's essay focuses on the issue of female employment, and weaves together a discussion of her own research with questions about mine. While we broadly agree that export-oriented low-wage manufacturing is an important source of jobs for women in the developing world, she points out that sometimes domestically oriented factories also rely on female labor—noting the case of the Indonesian textile industry in the 1970s.

A large number of studies—at both the country level and the cross-national level—find that export-oriented companies are more likely to hire women than domestically oriented ones (Braunstein and Brenner Reference Braunstein and Brenner2007; Ozler Reference Ozler2000; Standing Reference Standing1999). Although the Indonesian case is important, it runs against the global pattern.

Caraway also points out that even when women hold a large number of jobs in manufacturing, “most women work in other parts of the economy, often in informal sector jobs or in agriculture” (p. 569). She also criticizes my focus on jobs in the formal sector, and outside of agriculture, as “not justified.”

I focus on jobs in the formal, nonagricultural economy because they seem to have more far-reaching consequences for women. Many women in developing states work in agriculture, but often these jobs are part of the home unit and fail to promote either economic independence or political voice (Matland Reference Matland1998; Oakes and Almquist Reference Oakes and Almquist1993). True, jobs in commercial agriculture may be more consequential if they provide women with their own paychecks (see, for example, Anderson and Eswaran Reference Anderson and Eswaran2005). But data on female work in the agricultural sector usually fails to distinguish between subsistence labor in the home unit and wage labor; to avoid conflating them, I focus on nonagricultural work. In any case, my model implies that oil will also reduce opportunities for women in commercial agriculture, just as it does in manufacturing—by making these sectors unprofitable, thanks to the Dutch Disease.

I agree with most of Caraway's other points: If oil-based economies diversify, women will find more jobs; we need to better explain variations in the demand for female labor among ostensibly similar industries; ideas about gender roles matter; and generalizations about women's work should be empirically valid. We disagree about the number of women who work in retail, an issue that is hard to resolve with existing data sources.

Kang's essay points out that gender quotas have a powerful effect on the number of women elected or appointed to legislatures. While it confirms that oil production reduces female representation, it also suggests that gender quotas can help mitigate the effects of oil.

Kang's analysis has both theoretical and policy importance: It contributes to the growing literature that demonstrates the significance of political institutions in general—and gender quotas in particular—in bringing women into public office (Baldez Reference Baldez2004; Bhavnani Reference Bhavnani2009; Tripp and Kang Reference Tripp and Kang2008). And it highlights the power of quotas to overcome gender disparities even in countries with highly unfavorable socioeconomic conditions.

Still, gender quotas do not appear out of nowhere, and we only have a preliminary understanding of the factors that bring them about (Krook Reference Krook2006; Paxton, Hughes, and Painter n.d.). It seems plausible that quotas are more likely to be enacted—and when enacted, to be effective—when women play a larger role in the labor force, and are more mobilized politically. If this is true, oil wealth could reduce the likelihood that gender quotas will be implemented, just as it retards female empowerment more broadly. As Kang points out, oil-producing states have been significantly less likely to adopt gender quotas.

This is why I did not include a variable measuring gender quotas in my statistical model explaining female political representation. Quotas would be appropriate to include if they are unrelated to the other causal variables in the model, like income, Islamic populations, and female labor force participation. But if quotas are the result of some of these factors, their inclusion in the model could bias downward the size and statistical significance of the other explanatory variables through a “post treatment” effect. In any case, as Kang notes in her essay, my results still hold even when gender quotas are included in the model.

I find little to argue with in Charrad's insightful contribution. She suggests that oil wealth helps explain female subordination in the Middle East, but that other factors also matter and may be even more important; that across much of the region, patriarchy is rooted in kin-based networks, which may help explain the “Middle East effect” I measure with a dummy variable; that oil wealth may interact with these kinship networks in complex ways that we poorly understand; that we should be wary of facile generalizations about women's movements in the region; and that important changes can come from top-down government initiatives, like the 1956 reforms in Tunisia's family laws, imposed by Habib Bourguiba shortly after independence.

I agree with all of these points. By focusing on political reforms initiated from below, I am not denying that reforms can also come from above; in fact, my article highlights the 1956 Tunisian reforms, citing Charrad as a source (p. 119). Enlightened leadership is always welcome; the question I try to address is, What can be done in its absence?

My article directly challenges the research of Norris (and her longtime collaborator, Ronald Inglehart). So it is only fitting that she challenge mine. Before replying to her comments, it may be useful to delineate our areas of disagreement. In their 2003 book Rising Tide, Inglehart and Norris use data from the World Values Survey—covering some 70 countries—to explain global patterns of gender discrimination. They argue that a process of “societal modernization” can largely explain why attitudes toward gender equality change over time and vary among countries. But they note that modernization does not move at the same pace in all countries:

Despite global progress towards female empowerment, they note,

According to Inglehart and Norris, women have made little progress in these states because of their Muslim culture; they find that “an Islamic religious heritage is one of the most powerful barriers to the rising tide of gender equality” (p. 71).

The World Values Survey is one of the major social science achievements in the field of comparative politics, and Inglehart and Norris do a formidable job analyzing a mountain of data. Yet I find their arguments about women and Islam unconvincing for three reasons.

First, it is not clear how religious traditions can explain changes in gender attitudes. It seems self-evident that a country's religious and cultural heritage help explain the origins of attitudes toward gender equality, and that these traditional values leave an enduring mark on norms and beliefs. But virtually all countries have strong patriarchal traditions; the important question is why these traditions persist in some countries but erode in others. How can fixed religious traditions explain why beliefs—and gender practices—change over time at different rates in different places?

Like many others, I think differences in female participation in the workforce can help explain differences in the rate of female empowerment (e.g., Burns, Schlozman, and Verba Reference Burns, Schlozman and Verba2001; Iversen and Rosenbluth Reference Iversen and Rosenbluth2008). My contribution is to show that differences in oil wealth can help explain differences in female labor force participation.

Second, we disagree over the role of popular attitudes in explaining gender inequality. Norris is correct that my article “does not consider any direct evidence concerning attitudes and values” (p. 558), but this does not mean I conclude that they are unimportant—only that it is hard to identify their role. I have no doubt that popular values have an effect on female labor force participation and female political representation. But popular values are also affected by them, which makes it hard to know how much impact these attitudes really have (Blaydes and Linzer Reference Blaydes and Linzer2008; Seguino Reference Seguino2007; Thornton, Alwin, and Camburn Reference Thornton, Alwin and Camburn1983). It is no surprise that gender attitudes are correlated with female empowerment, as Inglehart and Norris show in their book; but it is not obvious that the former is affecting the latter.

Finally, they assume—like most other scholars, to be fair—that all kinds of economic growth have the same “modernizing” effects on gender rights. I argue that different kinds of economic activities have different effects on female empowerment: Growth that draws women into the labor force is empowering, but growth that encourages women to remain home—like growth based on petroleum extraction—can be politically debilitating.

Inglehart and Norris cite five Middle Eastern countries—Kuwait, Qatar, Saudi Arabia, Oman, and the United Arab Emirates—where economic growth has not had the modernizing effects on gender rights that their model predicts; they conclude that a common Islamic culture must be at fault. But they fail to notice that these five countries have something else in common: They are among the world's most oil-dependent countries. Moreover, in other Middle East countries with little or no petroleum wealth—like Syria, Jordan, Morocco, Tunisia, Lebanon, and Turkey—conditions for women have improved much faster. Indeed, my article shows that when oil wealth is accounted for—in a statistical analysis covering 161 countries—both oil and the Mideast region are strongly linked to gender inequalities, but the Muslim religion is not.

Norris's essay has more critiques of my article than I can answer succinctly. Here are some replies to the main charges:

• • The essay suggests that the correlations between petroleum wealth and gender inequality in the Middle East are driven by four “outliers”—Saudi Arabia, Qatar, Kuwait, and the United Arab Emirates (my Figures 3–6). This is incorrect: Even when these countries are dropped, there is still a strong correlation between oil wealth and gender disparities in the Middle East, due to states like Libya, Oman, Bahrain, and at times Iraq.

• • It argues that the Middle East case studies (Algeria, Morocco, and Tunisia) are subject to “alternative interpretations,” but does not state what they are.

• • It states that “it might have been more persuasive to select cases from non-Muslim petroleum- and gas-rich states outside of the Middle East” (p. 555). But one goal of the case studies was to point out the huge variations in female economic and political empowerment in the Middle East—variations that cannot be explained by a shared Islamic heritage.

• • It suggests that oil-rich non-Muslim states, like Russia and Venezuela, do not fit my argument. But they do: women hold just 9.8% of all parliamentary seats in Russia, versus 14.7% in the other former Soviet republics; and 9.7% of seats in Venezuela, compared to 16.7% in the rest of Latin America.

• • It also implies that “energy-rich Norway and Canada (and Scotland?)” (p. 555) do not fit the model. This is not right: The article emphasizes that oil wealth only hurts women in countries where they are excluded from jobs in the service sector—a condition that does not arise in most wealthy Western countries.

• • It faults me for looking at oil and gas wealth but not other kinds of mineral wealth. In fact, petroleum represents more than 90% of all internationally traded mineral wealth; while it would be useful to measure other forms of mineral wealth, the data are much harder to compile and tend to be less accurate.

• • It suggests my statistical model is misspecified, since it does not control for gender quotas. As I explain above, I had good reason to leave them out. And Kang notes in her essay that my results still hold when gender quotas are placed in the model.

• • It argues that “the theoretical link between the value of oil rents per capita and the structure of the labor market may also prove tenuous,” (p. 557) because petroleum industries employ few people—and hence should only have a small impact on job opportunities for women. But this misstates my argument, which is spelled out on pages 109 and 110 of the original article: The number of people employed by the oil industry is irrelevant. Rather, I argue that oil production reduces the number of jobs for women in other parts of the economy—notably export-oriented, low-wage manufacturing—through the Dutch Disease, and discourages women from joining the labor force by raising household incomes. These effects would still occur if the oil industry employed no one at all.

I have great respect for Norris but think that she and Inglehart have overlooked an important source of gender disparities and, as a result, wrongly concluded that Islamic traditions are a persistent impediment to female empowerment. Yet traditional values are subject to rapid change in Muslim-majority countries, just as they are elsewhere, under the right economic conditions. When economic growth draws women into the labor force—like in Morocco, Tunisia, Turkey, and Bangladesh—they have made rapid political progress; when economic growth discourages women from working outside the home—like in many oil-producing states around the world—their progress has been much slower.

Ultimately, our disagreements can only be resolved through further research. Caraway, Charrad, and Kang all suggest research paths that could deepen our understanding of the economic, institutional, and cultural forces that impede or hasten the empowerment of women. I would only add a plea for scholars to pay close attention to the political economy of women's work, and how it might be affected by—among other things—a country's resource endowment.