Islam, Authoritarianism, and Female Empowerment: What Are the Linkages?

Most Islamic countries are governed by authoritarian regimes, and in most Islamic countries women's economic, social, and political rights are significantly less than those of the men of their country. These well-known facts provide the basis for a highly original and important analysis by Steven Fish.  Fish's major contribution is his suggestion that the relationships at issue are not merely a matter of spurious correlation but rather are indicative of a deep underlying causal pattern. In a sophisticated cross-national quantitative analysis, he shows the following:
 1. Even controlling for many other alleged influences on countries' political systems, countries with a largely Islamic religious tradition have significantly more autocratic governments than do non-Islamic countries.
 2. Controlling for level of economic development, the condition of women is significantly worse in Islamic countries than in others.
 3. In a more limited test, even when controlling for economic development and Islamic tradition, a low level of women's rights may produce lower levels of democracy.
 4. More speculatively, Islamic states may be more autocratic so as to repress women's rights more effectively, or autocratic government may permit and even require the repression of a range of human rights, including those of women. This second causal arrow may replace the one just above, or the two may reinforce each other in a feedback loop.

Fish combines his quantitative analysis with a wide-ranging theoretical and factual review of the relationships among Islam, democracy, and female empowerment. We will not significantly add to or modify that part of his contribution. We are, however, skeptical of cultural arguments about why some states are authoritarian. Such arguments, after all, were made about Asian and Latin American cultures, yet democracy now flourishes in many such states. Thus we replicate and modify his quantitative analyses to see whether his findings are robust, and we address the question of complex causation. In doing so we find the following:

1. Islamic countries in general are more likely to be ruled autocratically even when other influences are controlled. But Arab countries in particular are more often ruled autocratically. So, too, are states that have frequently engaged in international military conflicts. However, the effects of culture can change dramatically. Before 1980, countries with large Catholic populations were even less likely than Islamic ones to have democratic governments, but that relationship subsequently turned strongly positive.
 2. Islamic countries generally are more likely than others to repress some rights of women. But the effect is much stronger and more consistent for Arab countries in particular.
 3. The evidence that autocratic governments systematically repress women's rights in particular is at best inconsistent rather than general and distinctive. Furthermore, we find little indication that female empowerment contributes causally to democratic government as measured by specific indicators.
 4. Thus, while Islamic religious tradition and especially characteristics of Arab states or culture do seem to support both autocratic government and the repression of women, our analysis does not support the view that the repression of women is central to maintaining autocratic rule in Arab or other Islamic states or that greater democracy in those countries would greatly improve women's conditions on a broad range of measures. Our analysis of more nuanced equations for female empowerment finds the relationships among regime, Islamic/Arab tradition, and the status of women to be more complex than Fish suggests. In addition, such causal relationships as there are differ depending on the issue-area being considered. Female political rights, economic activity, health, and educational attainment should be treated as distinct measures and not substituted for one another as general proxies for the status of women.

Our contribution, while supporting much of Fish's argument, is thus to greatly weaken major parts of it, especially about any causal role linking women's rights to Islam and democracy. These results leave still largely unanswered the question of why countries with a largely Islamic or Arab tradition tend to be more autocratic and to keep their women so little empowered, as measured by conventional criteria of human rights. They also raise fundamental and more gen eral questions about the role of culture and democracy in enhancing the condition of women.
 Does Islamic Tradition Discourage Democracy?
 We begin with the analysis of determinants of democracy, taking off from Fish's analysis. In numerous regression analyses, he examines the effects of many hypothesized influences, including level of economic development and recent economic growth, sociocultural division, colonial or communist heritage, OPEC membership, and Islamic religious tradition. Of these he finds only three exert a strong and consistent impact: economic development, OPEC membership, and Islamic tradition. We nevertheless suspected that his analysis could be more finely and completely specified.
 The correlation between democracy and economic development is one of the strongest and oldest in the literature. Although the particular nature of any causal relationship is strongly contested among scholars, we are inclined to believe in a causal linkage from economic development to the sustainability and perhaps the initiation of democracy, as well as perhaps a reverse arrow from democracy back to development. Since our purpose is not to enter the debate on the reciprocality of the relationship between development and democ racy, we simply accept-in the company of Fish and many others the plausibility of a causal arrow from development to democracy and include development as one independent variable affecting democracy. We also concur with Fish that the independent effects of recent economic growth, sociocultural division, and colonial or communist heritage are inconsistent and at best weak. Consequently we omit them here.
 For purposes of replication we conduct our analyses on the same set of countries Fish used, and we use his data whenever we use his variables. When we change or add a variable, we discuss the measure and source and provide details in the appendix available on the web as identified in the unnumbered note. Our reservations concern, first, two of the measures that he finds to have strong and robust effects and, second, possible underspecification in the equations.
 OPEC membership is the first of the two measurement decisions with which we quibble. Fish chooses this as a proxy for the theoretical variable of resource abundance, notably its availability to authoritarian actors to buy off large segments of the populace and thus stunt demands for political accountability and its enablement of the state to sustain a large and powerful internal security apparatus. Similarly resource abundance is one of the most commonly and successfully employed variables in examinations of the causes of civil wars.³ But a dummy variable for OPEC membership unnecessarily loses the potential explanatory power of an interval variable of more or less oil, specifically, the ratio of oil and other fuel income to the economy. It is more plausible to hypothesize a continuous relationship; that is, the greater the relative input of fuel income to the total GDP, the greater the ability of an authoritarian regime to sustain itself. Even more relevant may be the ratio of the value of fuel exports to GDP, as that more directly measures the regime's potential to import hard currency consumer goods and security equipment to monitor and suppress dissent.4 The other measure we modify is Fish's dummy variable for Islamic religious tradition. He uses "predominance" of Muslims as the "tipping point" for substantial Islamic influence on politics (p. 7). This is espe cially plausible when considering influences on democracy, since an Islamic majority could in some kinds of democratic systems impose its practices on the entire population. Nevertheless, the binary vari able still discards valuable information about the relative impact of small or large Islamic populations on state policy and is at least equally problematic for our later discussion of influences on female empower ment. Thus, for example, the subjugation of women may be largely the result of overt state action or of less formal social practices that may be enabled by state nonintervention in religious cultures. Especially for the latter, the appropriate hypothesis would be the larger the number of Muslims as a percentage of the total population, the lower the country's overall level of female empowerment-potentially strengthening Fish's argument."
 We identify an additional five possible variables for his equation to explain democracy. First, it is increasingly recognized that a state's political neighborhood matters. In his discussion of conditions pro ducing transitions to democracy, Huntington mentions "snowballing," or demonstration effects, enhanced by international communication,