Author: Erik Meyersson, SITE.
In this policy brief, I discuss how state-of-the-art econometric techniques can be used to shed light on the causal effects of Islamic rule on women’s rights. A central empirical challenge is that the identity of a politician is endogenous to voter characteristics, which in the case of Islamic political participation is particularly important due to the prevalence of banning such parties in many Muslim countries. Using a research design called Regression Discontinuity, I show that despite a negative association between Islamic rule and female participation in education in Turkey, the causal effect of an Islamic party on women’s rights is positive. In the case of Turkey, this represents the Islamic political movement’s advantage over secular alternatives in overcoming barriers to female participation in voluntary education institutions among the poor and pious.
Is Political Islam Unfit for Democracy?
As many Muslim countries have implemented democratic elections, an often-quoted concern is that Islamic political control will adversely affect women’s living standards (The Economist 2011, The New York Times 2011). If the expansion of political religious freedoms endangers gender equality, this implies democratic institutions could result in adverse development consequences for large numbers of the population.
At the root of this concern is that Islamic parties tend to represent poor and religiously conservative constituencies with correspondingly low women’s rights. A common feature of Muslim democracies is therefore that the politicians with the most conservative views on women are elected in areas where women are most vulnerable.
In the 1990s, secular regimes in Algeria, Tunisia, and Turkey, amongst others, severely restricted participation by political Islamists, going against a principle of open democratic participation. These actions found support among many Western countries, predominantly through fears of Islamists seeking to implement sharia law, which would further restrict women’s rights in countries where these were already very low.
Experiences in several countries have served to cement a stylized view of political Islamists as inimical to political development in general and adversely affecting women’s rights in particular. In the Palestinian territories, the 2006 elections swept to victory Hamas, an organization classified as a terrorist organization by the US, EU, and Japan. Since the coming to power of Hamas, Gazan society has reportedly become more socially conservative (The Guardian, 2009).
In Algeria, an overwhelming electoral victory by the Islamic Salvation Front (FIS) in 1991 was annulled by the military a year later, triggering a prolonged Islamist insurgency. Throughout its armed campaign, militant Islamists deliberately targeted women who did not adopt a conservative lifestyle, such as those who held jobs or refused to wear a veil.
These are but two examples where political Islamists were associated with adverse development outcomes. Yet whereas these have resulted in a strong discourse against Islamists, none are based on research evaluating the causal effects of Islamists holding political office in a democratic system. The debate on whether Islamists are detrimental to development is thus conducted in an environment devoid of research examining its causal effects.
The Causal Effects of Islamists – Evidence from Turkey
Inherent in any empirical study evaluating the consequences of Islamic rule is the endogeneity of policy outcomes and preferences. A correlation between an outcome and a particular politician may reflect the causal effect of the politician as much as the causal effect of voter preferences. If observed development differences between Islamist and non-Islamist are due to the former, regulating participation of Islamist parties implies banning parties; if the differences are due to the latter, any regulation essentially implies banning voters.
In order to identify the causal impact of having an Islamic politician in political control, a research design allowing near-random assignment of political control is essential. This issue is investigated in a recent paper by Meyersson (2013).
Events in Turkey during the 1990s provide a rare opportunity to implement such a research design. In 1994, the Welfare party, a radical Islamist political party, got elected to several mayor seats in the country’s circa 3,000 municipalities. This represents the first time an explicitly Islamist party gained and were allowed to retain control of important local government positions through democratic elections (Kepel 2002).
A useful measure of female participation is completion of high school, since this is today the earliest form of non-mandatory education in Turkey. A naive regression analysis comparing female high school completion in 2000 based on whether the municipality received an Islamic or secular mayor in 1994 shows Islamic-controlled municipalities having significantly lower female educational participation levels. But due to the identification problems mentioned earlier, it is unclear whether this estimate has a causal interpretation.
Local elections in the country are determined by a plurality rule, meaning regardless the voting constellation across different competing party candidates, whoever wins the most votes becomes the mayor. This allows the implementation of a Regression Discontinuity (RD) design (Imbens and Lemieux, 2008).
The RD design allows estimating causal parameters when a “forcing variable” determines treatment (having an Islamic mayor elected) and the probability of treatment jumps discontinuously when the value of the forcing variable reaches a certain threshold. Under the assumption that observations around a narrow bandwidth around this threshold do not systematically differ from each other except in treatment assignment, this design can provide a causal effect of the treatment representative for the observations around the threshold.
When comparing municipalities by whether the Islamic candidate barely won or lost, the results reveal a significant positive effect of having an Islamic mayor of around a 20 percent increase in relative terms. This positive treatment effect is relatively larger in municipalities that are both poorer and more religiously conservative. Corresponding effects for men were absent.
Thus, in places where pre-determined characteristics are similar – i.e. in municipalities with the same preferences for an Islamic mayor – the estimates of receiving an Islamic mayor are of the opposite sign as that resulting from a naive regression analysis. This positive effect of local Islamic rule on female participation in education is robust to a host of specification checks and is persistent up to 17 years after the treatment.
This striking result goes against the stylized view of political Islamists as invariably detrimental to women’s rights. So how can this positive effect have come about? The answer lays to a large extent in the way the Turkish education system, in particular high school, is designed. Turkish general high school (lise, in Turkish) is voluntary and participation is conditional on several constraints, one being the ban on wearing a headscarf.
The headscarf ban has been at the center of a conflict between state institutions heavily inspired by secularism, and a substantial portion of the population defining themselves as pious Muslims. Among religiously conservative communities in Turkey, a parent sending an uncovered teenage daughter to high school thus comes at a social cost, which not all parents are willing to bear. A possible explanation for the initial negative association between Islamic rule in 1994 and female high school education in 2000, is precisely such barriers to entry, exacerbated both by a ban on religious symbols in schools, affecting women disproportionately, as well as the prevalence of religiously conservative beliefs among parents.
In 1998, the Welfare party was banned by Turkey’s constitutional court for “anti-secular activities” (ECHR, 2003). In the court documents, one piece of evidence used against the party was its unwillingness to enforce the headscarf ban. Another was its involvement with officially banned Islamic brotherhoods. These brotherhoods, through a particular charity organization called the vakif, are an essential component of Turkish civil society, and invest heavily in education, not just in schools but also in extra-curricular activities such as Qur’an study centers and student dormitories. Many secular critics of the Islamic movement saw these institutions as evidence of Islamification. Yet these add-on facilities, often located near or in conjunction with state schools, may also have proved a key selling point in convincing conservative parents to send their daughters to secular schools (Cowell, 1994).
Through the mayor’s control over urban planning policies, essentially building permits, this provides a clear policy link with which the mayor may affect education outcomes. Analyzing new buildings as well as building permits, the paper shows that local Islamic rule, while not leading to more education building space, did shift the composition of it towards those owned by religious charities. The very facilities, in which critics of the Islamic party saw as a threat to the development of the country, may thus have alleviated one of the key frictions preventing women from educational participation.
The concern over Islamic rule in a democracy goes beyond its potential effects on female education to include the quality and adherence to democracy itself. The above-mentioned study includes an analysis of effects on female participation in local municipal councils fifteen years after the treatment of receiving an Islamic mayor in 1994. Despite being one of very few Muslim countries to have had a female prime minister, women in politics remain largely nonexistent in Turkey. It is therefore noteworthy that a near-randomly assigned Islamic mayor in 1994 resulted in increased female participation in politics in the 2009 municipal councils.
The empirical design employed here provides a solution to an identification problem where elected politicians are endogenous to constituency characteristics. In this case, constituencies prone to supporting Islamists tend to exhibit preexisting traits unconducive to women’s rights, resulting in an adverse correlation between Islamic rule and empowerment outcomes. The main contribution of the paper is to evaluate Islamic rule in isolation from such confounding factors.
The external validity of results from the above-discussed study is constrained by a focus, not just of elections determined by a thin winning margin, but also its reliance on the specific institutional setting existing in Turkey. In order to better answer the question of Islamic rule’s effects on women’s rights more research is needed in other institutional settings.
The implications of these findings are dual in nature. First, it suggests that under specific circumstances, socially conservative politicians can lead to socially progressive outcomes. Policies to regulate participation by Islamic political parties thus need to take into account that these controversial but popular movements may have development-related effects difficult for secular parties to replicate. An example of this is the European court of Human Rights’ ruling on the court base that dissolved the Welfare party. In its final ruling, it stated that the dissolution of the Islamic party was “necessary in a democratic society”. Yet this decision was based on evidence in the form of statements and actions by individual politicians; not on the party’s actual democratic and development-related track record.
Second, Turkey has both direct and indirect barriers to educational participation that, in combination, may provide Islamic parties with a competitive advantage, which in turn boosts their popularity. On the one hand, secular restrictions to participation, such as the headscarf ban and mixed classes, result in conservative parents being unwilling to send their daughters to school. On the other hand, the voluntary nature of high school further gives parents this option. It is therefore not easy to distinguish the positive effects of Islamic rule on the poor and pious from the barriers to entry that restrain them in the first place.
- Cowell, Alan, “Toehold for Islam – A special report; Muslim Party Threatens Turks’ Secular Heritage”, The New York Times, November 30, 1994
- ECHR, “Case of the Refah Partisi (The Welfare Party) and Others v. Turkey,” ECHR Third Section judgment and ECHR Grand Chamber judgment, 13 February 2003, http://echr.coe.int/echr/en/hudoc
- Imbens, Guido and Thomas Lemiuex, 2008, Regression discontinuity designs: A guide to practice”, Journal of Econometrics, Vol. 142
- Kepel, Gilles, Jihad: The Trail of Political Islam, Belknap Press, 2002.
- Meyersson, Erik, 2013, “Islamic rule and the Emancipation of the Poor and Pious”, working paper, http://www.erikmeyersson.com/meyersson_islamicrule_1302.pdf
- The Christian Science Monitor, “Islamists Strike At Women In Algeria War,” June 28, 1995
- The Economist, “The law of man or the law of God?”, September 11th 2003 December 29, 2011.
- The Guardian, Hamas patrols beaches in Gaza to enforce conservative dress code, 18 October 2009, http://www.guardian.co.uk/world/2009/oct/18/hamas-gaza-islamist-dress-code
- The New York Times, “Unfit for Democracy?,” February 26 2011