Gender Relations, Family Systems and Economic Development

Gender Relations, Family Systems and Economic Development

Gender Relations, Family Systems and Economic Development: Explaining the Reversal of Fortune in EurAsia Sarah Carmichael, Alexandra M. de Pleijt and...

519KB Sizes 1 Downloads 7 Views

Gender Relations, Family Systems and Economic Development: Explaining the Reversal of Fortune in EurAsia

Sarah Carmichael, Alexandra M. de Pleijt and Jan Luiten van Zanden Utrecht University Email corresponding author: < [email protected] > Preliminary version, please do not quote Abstract This paper argues that gender relations matter for economic development, and in particular help to explain growth trajectories in EurAsia between the Neolithic Revolution and the present. Firstly, we offer a set of hypotheses drawn from the literature about the links between gender relations and economic development. Secondly, we approach gender relations via the classification and measurement of historical family systems, and offer a set of global maps of the institutions concerning marriage, inheritance and family formation that determine the degree of agency that women enjoyed at the micro level. Thirdly, we use these concepts to explain the genesis of the EurAsian pattern in family systems and gender relations as a by-product of the spread of agriculture and the process of ancient state formation that followed the Neolithic Revolution 10,000 years ago. Finally, we link these patterns in family systems and female agency to economic growth after 1300; we empirically demonstrate that high female agency was conducive to growth between 1500 and 1800 and was also positively correlated to growth during the Great Divergence between 1800 and 2000. The ‘reversal of fortune’ that happened within EurAsia between 1000 and 2000, whereby the ancient centers of state formation and urbanization in the Middle East, India and China were overtaken by regions at the margin of the continent (Western Europe, Japan, Korea), can be explained by the spatial patterns in gender relations and family systems found there (and reconstructed here). Introduction: hypotheses linking female agency to development There is now a well-established literature in development studies (inspired by the work of Amartya Sen) arguing that female agency is conducive to economic growth and institutional development (Eswaran 2014; Klasen 2002; Klasen and Lamanna 2009; FAO 2011; Teignier and Cuberes, 2014). The World Bank 2012 report ‘Gender, Equality and Development’ summarized this literature and contended that strengthening the autonomy of women was ‘smart economics’. Agency here is defined as the ability to make decisions and to undertake action in a given environment to achieve a desired outcome. In 2014 a further World Bank report, “Voice and Agency – Empowering women and girls for shared prosperity”, highlighted that although gaps between men and women in many dimensions have narrowed, systematic differences continue to persist, which they ascribe to differences in agency (World Bank 2014). Gender inequality has consistently been shown to inhibit economic growth and development (Klasen 2002; Klasen and Lamanna 2009; FAO 2011: McKinsey 2011). This

idea that women might be an engine for economic development has risen to prominence in development studies over the past 20 years, largely as a result of the work of Ester Boserup (1970). The line of reasoning received further support in 1992 from the then chief economist of the World Bank, Lawrence Summers, who argued that investments in the education of girls might just be the highest return type of investment possible in developing countries (Summers 1992). A number of quantitative studies have built on these ideas to explore the empirical relationship between gender inequality and economic growth and development (Klasen 1999; Teignier and Cuberes 2014). The overall result has been to show that gender inequality is detrimental to growth. Teignier and Cuberes, for instance, using a model whereby women are inhibited from contributing to the labour force in various ways, find an implied income per capita loss of 27% for Middle Eastern and North African countries and a 10% loss for Europe (Teignier and Cuberes 2014), an economically significant result. However research has also demonstrated that the effects of achieving gender equality extend beyond simply economic gain. Improving the position of women has been shown to have positive effects on a wide range of development outcomes: children’s educational attainment (Currie and Moretti 2003; King et al. 1986; Schultz 1988; Strauss and Thomas 1995), the quality of government, particularly by reducing corruption (Dollar, Fishman and Gatti 2001), reduced infant mortality (Dollar and Gatti 1999; Eswaran 2014), improved household efficiency (King and Hill 1997) and reduced fertility (Rosenzweig and Schultz 1982). Unfortunately these ideas have not played a significant role in the debate on the long term development paths of countries and regions that has been the focus of much new research by economists and economic historians. A major source of inspiration for the latter has been the development of new institutional economics, with seminal publications by North (1981, 1990), North, Weingast and Wallis (2009) and Acemoglu and Robinson (2012). North stresses the importance of rules that constrain the behaviour of those in power . The power of the executive has to be limited to ensure property rights of citizens and hence create the right incentives to invest. Institutions should, in his view, create a level playing field and guarantee a certain balance of power between ruler and ruled. Such institutions limit the freedom of the powerful, but enhance the agency of the great majority of the population. This basic idea has become highly popular in NIE; it is, for example, at the core of the writings about ‘inclusive’ versus ‘extractive’ institutions by Acemoglu and Robinson (2012). Similarly, the introduction of ‘rule of law’ for elites is considered to be the first step towards an ‘open access regime’ by North et.al. (2012). Many studies have analysed the importance of democratic institutions for economic development, in the more distant past and contemporaneously (Rodrik et.al. 2004; Przeworski et.al 2000). But NIE seems to be primarily, if not almost exclusively, interested in the constraints on political power. Other sources of power – such as ‘patriarchy’ – have not received the same attention, although gender inequality may have similar effects in undermining the ‘level playing field’ of women. Gender inequality seriously limits the agency of approximately half of humankind, who can therefore not work, invest and innovate as they

would under free conditions. Patriarchy undermines female property rights and investment incentives in the same way as unconstrained sovereignty may destroy incentives for innovation and investment for (male) entrepreneurs. For instance, the lack of well-defined property rights for women has been highlighted as a substantial barrier to efficient agricultural production in sub-Saharan Africa (see Doepke, Tertilt, and Voena 2012 for a review). It follows that we can formulate what we have called the gendered North hypothesis, arguing that constraints on power holders at the micro level – on the power of the patriarch or the parents – will improve incentives and property rights of women (and young men) and therefore the quality of decision making at that level. A special case of this is related to the switch from ‘quantity’ to ‘quality’ of offspring, introduced by Gary Becker and his associates (Becker 1960; Becker and Lewis 1973; Becker and Tomes 1976; see also Schultz 1961). The idea of the quantity-quality trade-off is that parents face constraints in terms of time and money, and therefore have to make a choice between the number of children and the amount of time and resources they can invest in them. If they choose to have fewer children, they can increase investment in the human capital of those fewer children. This trade-off is in particular driven by the opportunity cost of childrearing for women, as they are the ones who bear most of the costs of having and rearing children. Thus, the higher the level of female education, the larger the costs will be of having more children, in terms of their productivity and the opportunity costs of their time (Becker 1965). And the stronger the bargaining position of women in the household, the more these considerations will affect the parental decision making process. The gendered Becker hypothesis states that increasing the bargaining position of women will further the process of switching from quantity to quality of offspring, and enhance levels of human capital formation of the next generation (and in that way stimulate economic growth). A third hypothesis relates female agency to processes of democratization. Emmanuel Todd (1985, 1987) has developed ideas about the relationship between family systems and the existence of broad, global differences in (political) institutions and ideologies. The underlying intuition is that children learn how to deal with power in the families in which they are raised, which has important implications for the way in which power, at the level of the polity, will be used or abused. The patriarchal household will teach other lessons to children than the bargaining household in which all have a say. This Todd hypothesis will help to explain why certain societies experience, during the process of economic development, a relatively unproblematic change towards democracy, whereas in other parts of the world this change has been difficult and incomplete.

We are going to test these ideas in a case study of development paths in EurAsia between the Neolithic Revolution (NR) and the present. In this version we will focus on the link between female agency and economic growth, and will not discuss connections with human capital formation and democratization. We will in particular try to explain the ‘reversal of fortune’ that occurred in this part of the world between 1000 and 2000. In his seminal book on the causes and consequences of the Neolithic Revolution, Jared Diamond (1997) discusses the cumulative, self-reinforcing character of technological change. He argues that this is the

principal reason why EurAsia, the largest continent with the biggest population, after pioneering the Neolithic Revolution, has dominated technological change since. That it was the first region to move to sedentary agriculture – and hence developed complex societies, cities and states first – gave the continent a head start over Sub-Saharan Africa, the Americas, and Australia – but also the ease of communication across the Continent (due to its East-West orientation) – and finally the cumulative character of technological change explain why EurAsia became the core of the world economy until very recently (and is arguably, with the rise of China, reclaiming this position after a ‘temporary’ loss to North America). There is indeed a lot of evidence confirming the path dependent, cumulative nature of technological and economic change (eg. Comin, Easterly and Gong 2010). In the year 1000 the core regions of EurAsia were the ‘old’ centers of the Neolithic Revolution – the Middle East (then dominated by the flowering of the Arab world), China (under the Sung arguably the most developed part of the world economy) and, perhaps less obviously, Northern India (which was in an interlude between the Gupta Empire and the Moghul Empire). Since times immemorial, the economic and urban backbone of EurAsia was the band between the eastern shores of the Mediterranean (including Greece and Egypt) and the great river valleys of China – connecting the three early centers of the Neolithic revolution in the Middle East (Fertile Crescent), the Indus Valley and the Yellow River Valley with long distance trade (such as the famous Silk route). Even in 1500 – and some would perhaps argue, 1750 – this was the basic pattern of the EurAsian economic and urban system. Until 1000, perhaps 1500, the map of EurAsia confirmed Jared Diamond’s expectations: based on their headstart, the centers of the Neolithic Revolution by and large maintained their leading position. Afterwards things changed fundamentally: the Industrial Revolution did not originate in Baghdad or Kaifeng, and the first nation to ‘catch up’ with the leaders, was not found near Harappa, the oldest center of the Indus Valley civilization, but was Japan. It was the periphery of EurAsia – North-Western Europe, Japan – that pioneered the industrialization of the post 1800 period, completely overturning the economic map of the continent. At present, the old backbone of Eurasia consists of countries with below-average income levels, the exceptions being the oil producing countries of the Middle East who do not really thank their ranking to endogenous advances in technology and institutions (Olsson and Paik 2013, 2015). But the band stretching from Egypt and Turkey in the west to China in the east was, in 1950 for example (before the ‘oil boom’), and still is, a region of on average low GDP per capita levels, although the recent catching up of China and India is weakening this pattern. As Olsson and Paik (2013) have demonstrated, between 1500 and 2000 a ‘reversal of fortune’ occurred, during which some of the ‘marginal’ regions of Eurasia developed very rapidly, whereas the core stagnated. This was a different ‘reversal of fortune’ from the one analyzed by Acemoglu et.al. (2002) when researching the effects of European colonization on global inequality; their focus was on the institutions introduced by European expansion after 1500, we will focus on the institutions that emerged much earlier, in the period of ancient state formation between c 3500 BC and 0 AD.

Why did the center of gravity of EurAsia move from the central belt between Egypt and China to Western Europe and Japan? We build on the hypothesis originally developed by Friedrich Engels that the rise of sedentary, complex societies, and in particular the ‘urban revolution’ that began after about 3500 BC, resulted in a fundamental change in gender relations. Before the Neolithic Revolution, in hunter-gatherer societies, gender relations were relatively equal, an hypothesis confirmed by recent research (Hansen et.al. 2015 for an overview). The subsequent development of cities, states and hierarchical societies following the Neolithic Revolution resulted in the introduction and spread of more hierarchical family systems, backed up by hierarchical state structures. The family systems with greater autonomy for women only survived in the marginal parts of EurAsia, at great distances from the original centers of the Neolithic Revolution and the urban revolution (Todd 2011). This is a gendered version of the hypothesis developed by Olson and Paik (2013, 2015), who also find a strong negative link between ‘years since transition to agriculture and contemporary levels of income’, a link which they attribute to the long-term impact of hierarchical values and structures arising in ancient societies, which gave rise to autocratic states. Firstly, we test this hypothesis by (1) quantifying the position of women (or reversely, the level of patriarchy) in the family systems of Eurasia (building on work by Emmanuel Todd and Murdock), and by (2) empirically testing for the correlation between the early adoption of the Neolithic Revolution in the mentioned three centers and the features of the family system. We follow Hansen et.al. (2015) and demonstrate that near the centers of the Neolithic Revolution family systems emerged which were relatively patriarchal and allowed less agency for women, whereas at greater distance from these centers family systems (as analyzed by 19th and 20th century anthropologists) were much more ‘female friendly’.1 Next, we set out to quantitatively explain the ‘Little Divergence’ (between 1500 and 1800) and the ‘Great Divergence’ (between 1820 and 1913/1950) by focusing on the link between female agency and economic growth. Because the EurAsian pattern of family systems and female agency emerged already thousands years ago, we can use this pattern as an exogeneous variable explaining economic growth after 1500. We demonstrate that in EurAsia the correlation between GDP per capita and female agency becomes gradually stronger over time, suggesting a strong link with post 1500 economic growth. Moreover, we show that the connection between the timing and spread of the Neolithic Revolution and contemporary patterns in GDP (and female labour force participation?) that has been suggested in the literature, …….

1

We should make explicit here than when we talk about the position of women or female agency we do not construe these in the way that modern measures of gender equality do (i.e. with data on labour force participation, life expectancy, political empowerment etc,). Rather we turn to institutional measures which capture the position of women in the ways families organise themselves across Eurasia.

The long-term consequences of the Neolithic Revolution for female agency The Neolithic Revolution, which began some 10,000 years ago, is one of the crucial turning points in world history, which had a fundamental impact on all aspects of social life. It led, as is well known, to the rise of cities, states and complex societies. There are reasons to believe that it also impacted on gender relations. The literature suggests two ways in which this may have been the case, given the relatively equal gender relations in hunter-gather societies.2 The spread of agriculture probably resulted in a fundamental change in the division of labour between the sexes – men specializing in the production of food, women in reproductive activities (Hansen et.al. 2015). Moreover, as Boserup (1970) has argued, the spread of the plough gave rise to further specialization between men and women, which resulted in growing inequality between the sexes. This hypothesis has been rigorously tested by Alesina, Guiliani and Nunn (2014), who found strong correlations between early adoption of the plough and contemporary attitudes towards women, and in particular their labour force participation. The second way in which the Neolithic Revolution affected the position of women was via the emergence of complex forms of social inequality in the cities and the states that followed its spread, which also strongly affected the position of women. Wright (2007) in a recent analysis of the evolution of gender relations in the first urban societies in Mesopotamia between 6000 and 2000 BC confirmed the hypothesis (which goes back to the ideas developed by Engels in the 19th century) that it was the rise of urban society in the late 4th Millennium which gave rise to patriarchal systems and that the status of women declined during the corresponding process of state formation (Wright 2007). A recent comparative analysis of ancient civilizations summarized the evidence as follows: ‘In early civilizations … inequality was regarded as a normal condition and injustice as a personal misfortune…. Structures based on differential power were pervasive. Every child was born into and socialized by a family that was internally hierarchized in the image of the state. The subordination of children to their parents and, to varying degrees, of wives to their husbands went unquestioned…. Young people were expected to obey older people, especially older men. ‘Father’, ‘king’, and ‘god’ were often synonymous and metaphors for power…. If egalitarian social organization was known to people in early civilizations, it was a feature of small-scale and usually despised societies beyond the pale’ (Trigger 2003: 142). This neatly summarizes the argument developed here. A last mechanism by which the two may be linked is through the process of pushing out individuals or groups who’s ideology no longer matched with that of the newly formed states. This means, in the case of the Neolithic Revolution, that the more individualistic individuals who are less accepting of authoritarian structures leave the centres to take up residence in the margins (Olsson and Paik 2015). The anthropologist Emmanuel Todd (2011) in his recent analysis of the development of family systems in EurAsia in the very long run, came to a similar conclusion, based on the spatial distribution of family systems. He pointed to the geographical concentration of 2

M. Dyble et.al. (2015) Sex equality can explain the unique social structure of hunter-gatherer bands, Science 15 May 2015, 796-798, analyse sex equality among hunter gatherers as an adaptive strategy to maximize cooperation, and see this as a ‘shift from hierarchical male philopatry typical of chimpanzees and bonobos’

patriarchal family systems in the heartlands of the continent, and the existence of more female friendly family systems in its margins – in Western Europe, Sri Lanka, Japan, Mongolia and South-East Asia. This links to the findings of Goody (1989, 1996) who points to certain similarities between Asian and Western-European family systems, but in our view these similarities are a phenomenon of the periphery of the EurAsian continent. The hypothesis that Todd formulated was that early states developed patrilineal hierarchies, fundamental changing the original balance of power between men and women which predominated in the nuclear family of the earlier hunter-gatherers. The patrilineal, community family organization type leant itself well to empires based on conquest; as a result early state formation (following the Neolithic Revolution) resulted in family systems that constrained female agency. Only in the margins of EurAsia, at great distance from the centers of the Neolithic Revolution (which spread only slowly), did female friendly family systems survive. The idea that the original form of family organization is female-friendly and has been replaced by a more male-oriented version with the rise of private property harks back to the work of Lewis H. Morgan and Friedrich Engels. In what follows we refer to this as the Engels hypothesis to distinguish it from a different Todd hypothesis, elaborated on in other work which links family-level practices how power is dealt with at a societal level. A similar spatial structure in contemporary value systems was analysed by Olson and Paik (2013, 2015), who found a strong negative link between ‘years since transition to agriculture and contemporary levels of income’, a link which they attribute to the long-term impact of hierarchical values and structures arising in ancient societies, which gave rise to autocratic states. In short, the historical and archeological literature and the anthropological evidence suggests that there is a link between the Neolithic Revolution, the subsequent process of ancient state formation, and the emergence of family systems that suppressed female agency. We test this hypothesis in two steps: first we use anthropological data to reconstruct the ‘female friendlyness’ of Eurasian family systems. In a related paper ‘Towards an ethnographic understanding of the European Marriage Pattern’ Sarah Carmichael and Jan Luiten van Zanden (2015) have used ethnographic information (Murdock’s database and Todd’s studies) to classify the societies of Eurasia on various marriage and family-related institutions, such as monogamy, consensus, female inheritance, exogamy and neo-locality). All these institutions have a ‘girlfriendly’ version: monogamy is from this perspective to be preferred to polygamy; consensus to arranged marriage; female inheritance to systems without them; exogamy to endogamy (which restricts the choice of marriage partners to kinmembers), and neo-locality to patri-locality. An easy and transparent way to classify societies is to let them score on all five dimensions; societies which are monogamous score one point here, and societies with polygamy do not score a point. Female inheritance, exogamy, matrilocality and consensus all score similar points (see for full details Carmichael and Van Zanden (2015)). The scoring is presented in the table below:

Table 1: Scoring for the ‘female-friendly index’ Variable Lowest Score Intermediate Scores

Highest Score

Domestic Organisation Cousin Marriage Monogamy Marital residence

Extended – 0

Inheritance

Patrilineal – 0

Endogamy – 03 Polygamy – 0 Patrilocal and Virilocal – 0

Stem – 0.5

Avunvulocal – 0.25 Ambilocal – 0.5 Neolocal – 0.75 Children daughters less – 0.5

Nuclear – 1 Exogamy – 1 Monogamy – 1 Matrilocal – 1

Children equally – 1 Other matrilineal - 1

The data used for this is derived from a combination of three sources. First and foremost we use Murdock’s Ethnographic Atlas. The atlas was featured in Ethnology from 1962 to 1980. In 1967 the data was compiled into a book. It contains data on 1267 societies for a wide range of characteristics. In each case the data is meant to represent the earliest possible observation for each society made by ethnographers (some economists therefore refer to the data as preindustrial). These were then translated to country-level indicators by Jutta Bolt, using the Atlas Narodov Mira (Bolt 2012). The dataset used here is largely as classified by Murdock, with a number of corrections made on the basis of comparing his categorisations to those of Emmanuel Todd. This is only relevant for the domestic organisation and cousin marriage variables. Using this index, the ‘female-friendlyness’ of family systems in Eurasia can be established: the more points a country scores on the range between 0 and 5, the more its institutions can be said to favour female agency. It is a bit arbitrary to weight all institutions in the same way, but it is highly transparent. Alternative ways of processing this information, via the estimation of a ‘latent variable’ ‘constraints on female agency’ (see Carmichael and Rijpma 2015) give almost identical results, but require more data, as a result of which less observations are available (the R2 between our female friendliness index and the ‘constraints on female agency’ estimated by Carmichael and Rijpma is .78). The results of the FFindex are presented in map 1, which shows that Europe to the west of the Hajnal line is clearly very ‘female-friendly’, but so is South-East Asia (in Carmichael and Van Zanden (2015) we present qualitative information confirming this pattern). When looking more closely, we find that in the other margins of the EurAsian landmass – in south India (Kerala is a famous case), Sri Lanka, Japan, and Mongolia, marriage systems also allow for female agency. The map below presents this visually at the country level for Eurasia. Figure 1: Female-friendliness Index mapped for Eurasia

3

Assigning a score to the extended family variable and the endogamy is complicated as in some cases living in extended, endogamous families can be beneficial to women as it keeps their natal kin close-by and can provide them with a support mechanism in times of need. An argument could therefore be made for assigning a half point for the combination of the two however for simplicities sake this has not yet been implemented here (moreover it has only a marginal effect on the Eurasian distribution).

Gfriendly

(-0.00475,0.528] (0.528,1.06] (1.06,1.58] (1.58,2.11] (2.11,2.64] (2.64,3.17] (3.17,3.69] (3.69,4.22] (4.22,4.75]

Shifting the focus to Europe we find, with the exception of Romania and Greece, the pattern is remarkably similar to the Hajnal line, with Poland, the Czech Republic and Austria displaying an intermediary pattern and the UK and the Netherlands, along with Italy and Spain attaining the highest scores. A somewhat similar ‘patriarchy index’ was constructed by Siegfried Gruber and Mikołaj Szoltyzek (2015); although the focus and methodology are rather different, constructed as it is of large micro-datasets concerning demographic behavior, measuring 14 different dimensions, such as ‘familial behavior, including nuptiality and age at marriage, living arrangements, postmarital residence, power relations within domestic groups, the position of the aged, and the sex of the offspring’. Their results, which are however available for only 12 countries and can therefore not be used for the regressions shown below, demonstrate the same West-East gradient in patriarchy as was found in our reconstruction of the ‘female friendly’ index. Their results, averaged at the country level, correlate strongly, but obviously negatively, with our index (R2 = .53). Another check can be conducted using the correlation with current day measures of gender inequality. For this we use the Historical Gender Equality Index developed by Selin Dilli, Sarah Carmichael and Auke Rijpma (2015). This measure captures gender differences in life expectancy, labour force participation, infant mortality, educational attainment, marriage ages, and political participation. Again, our index is highly correlated with contemporary measures of gender inequality, although the correlation is far from perfect (R2=.33); Sweden, for example, is currently world leader in gender equality, but did not score very well on the female-friendly index.

The map above of the spatial distribution of the gender dimension of family systems in Eurasia seems to confirm our hypothesis. However in order to test this more rigorously we estimated three variables reflecting the spread of the Neolithic revolution and the revolution in state formation which followed afterwards. . First, we established the distance of (the capital cities of) all countries to the three centers of the Neolithic Revolution in the Middle East, the Indus Valley and the Yellow River valley in China. On the basis of secondary literature, we selected three cities: Mosul in Iraq, Harappa in Pakistan and Xian in China to geographically represent the transition to agriculture in these three regions. We assumed that the further away a country was from the nearest center the more female friendly the family system would be. This is clearly demonstrated by the evidence for a group of the 47 countries for which we have these data (Figure 3). The second test looks at the underlying mechanism: the process of ancient state formation following the Neolithic Revolution. These states first emerged in Mesopotamia, followed by Egypt, Northern India and Northern China, and then gradually spread to adjacent areas. The ‘World History Atlas and Timelines since 3000 BC’ by GeaCron presents maps per century of the changing boundaries of these ancient states (see the example for 500 BC). We reconstructed for each contemporary country if an ancient state existed on its territory between 3000 and 1 BC, and on that basis constructed an ‘ancient state index’ using the same method as the ‘state antiquity index’ by Putterman and Bockstette (3.1 version), which covers the 1-1950 AD period. All countries are scored per century (and before 1500 BC per half millennium) on the existence of a state, and these scores are added using a discount rate of 10% per century. States with old roots, such as Iraq, Egypt, India/Pakistan, and China, score (nearly) the maximum, whereas regions such as Scandinavia, South-East Asia and Japan, where states emerged or spread to after 1 AD, score zero. The result is presented below (figure 3). It is clear that all very ancient states have rather female-unfriendly family systems, and that the family systems with high levels of female agency are generally found in societies which did not have a state before 1 AD. Moreover, as might be expected, the correlation between the ancient state index and distance to the centers of the Neolithic revolution is high and significant (R2=. 46), confirming that state formation spread from the identified three centers (which is also obvious from the maps published by GeaCron).

7 6 5 4 3 2 1 y = -0,0033x + 3,2321 R² = 0,3717

0 0

100

200

300

400

500

600

700

800

900

1000

Figure 2. FFIndex (vertical axis) and Ancient state index (horizontal axis)

As the ancient state index is constructed in the same way as the state antiquity index by Putterman and Bockstette, we can link our data to theirs and construct a continuous series of ‘state antiquity’ running from 1 AD to the present. Does the link between the ff-index and state formation only relate to ‘ancient states’ (say before the rather arbitrary date of 1 AD), or does it persist over time? The extended state antiquity index for 1300 (again a rather arbitrary date) correlates much less closely with the ff-index (r2=.17), the 1800 version even less (r2=.09). If the state antiquity index is calculated without the ancient states from before 1 AD, then the correlation completely disappears. This suggests that the ‘ancient states’ originating before 1 AD are the ones negatively affected female agency, and that, for example, the spread of states in Europe following the Roman conquest, did not have similar long-term consequences. Next we turn to a similar plot of the FFindex against distance to the nearest center of the Neolithic Revolution. Here we see the expected relationship; the further away a country is from the nearest center of the Neolithic Revolution the more scope there is for it to have female friendly family organisational features. Given the different manners in which the two variables are constructed the R2 of 0.44 is high. We can observe that those societies closest to the centers of the Neolithic Revolution are the only ones to score a 0 on the female friendliness index. It is only as one moves away from these centers that higher scores can be observed.

6

5

4 y = 1,3826ln(x) - 8,1666 R² = 0,4464

3

2

1

0 100

1000

10000

-1

Figure 3 FFIndex (vertical axis) and Log of distance to nearest center of Neolithic Revolution The third test of the Neolithic roots of the EurAsian map of family systems is testing for the link between ‘years since agriculture’ (a variable developed by Putterman 2006 to estimate the number of years since a country adopted agriculture) and the FFindex. This link is again negative (which means that early adoption of NR lead to lower FFindex), but it is much less strong (R2 = .17), which perhaps indicates that the connection between the NR and the FFindex runs primarily via state formation and less via the spread of agriculture as such (see Hensen et.al. 2015 for a similar discussion).

We have now established that the EurAsian map of female agency based on ethnographic data from Murdock for the period 1850-1950, is highly correlated with three different variables which all reflect the beginnings and spread of the Neolithic revolution and the process of state formation between about 8000 BC and 1 AD. This is a remarkable result: it appears that the underlying values and norms are highly persistent over time, and were able to remain more or less intact (spatially) during the past 2 millennia. However, similar persistence have been observed in related literature: Alesina, Guiliani and Nunn (2014) and Hansen, Jensen and Skovsgaard (2015) found similar ‘deep roots’ of gender roles, and Olsson and Paik (2013, 2015) found a comparable link between hierarchical value systems emerging after the NR and current beliefs. The fact that these spatial patterns by a large margin predate the emergence of the ‘modern economic growth’ that we are interested in, means that we can consider this pattern to be exogenous to the growth after 1500. This, we think, simplifies the regressions we are going to present

. Explaining the reversal of fortune

We have so far established why and how a distinctive spatial pattern of institutions concerning the family and gender relations emerged in EurAsia, as a byproduct of the Neolithic revolution and the ensuing state formation. Our main argument in this section is that this spatial variation in gender related institutions had important consequences for the pattern of economic growth after say 1500 AD, and helps to explain the reversal of fortune that is a dominant feature of growth between 1500 and 2000. In the current section we test this hypothesis empirically using simple Linear Regression analyses (OLS). As a first step, we show that the relation between female friendly institutions and the distance to the nearest centers of the Neolithic Revolution as set out in Figure 3 is robust to the inclusion of control variables. Therefore the log of the distance to the nearest center of the Neolithic Revolution of country i (NRi) is regressed on the female friendliness index of country i (Fi) in 1500, Fi =  + 1 Zit + NRi  + i ,

(1)

where Zit is a vector of control variables, and it is the error term. Previous studies (e.g. Hansen et al 2015) have shown that agriculture is conducive to the development of female agency. The first set of control variables that enter Zit therefore capture the geographical characteristics of the countries in our sample which could be relevant for the development of agriculture: firstly absolute latitude (measured as distance from the equator) and absolute longitude (measured as distance from the prime meridian). From Nunn and Puga (2012) we added percentage fertile soil (defined as soil which does not have extreme constraints for growing rainfed crops in terms of soil fertility, depth, chemical and drainage properties, or moisture storage capacity), percentage desert (defined as sandy desert, dunes, rocky or lava flows), percentage tropical climate (based on Köppen-Geiger tropical climates) and ruggedness (Terrain Ruggedness Index as devised by Riley, DeGloria, and Elliot (1999) to quantify topographic heterogeneity in wildlife habitats). We also control for the percentage of a country that is within 100 kilometers of an ice-free coast (again from Nunn and Puga, 2012). The second set of control variables that enter Zit are related to economic development, ethnic fractionalization and the quality of institutions. To control for the level of economic development we include the rate of urbanization in 1500 in the regressions below (Bosker et al 2012). Economic growth may have positively enhanced the position of women (see de Moor and van Zanden 2010). Secondly, to control for any effect stemming from ethnic fractionalization, which may have negatively impacted gender relations, we include ELF1 in

the regressions. The ELF1 variable captures the differences in language and ethnicity of the population living in a country. Finally, we control for the quality of institutions by including a set of dummy variables for the legal origin of the countries included in our sample (La Porta et al 2000). Table 2 shows that the results are robust to the inclusion of these controls. Column (1) is the simple bivariate regression as plotted in Figure 3. This shows a strong positive correlation between the distance to the nearest centre of the Neolithic Revolution and Female Friendliness. Column (2) puts the sample equal to that of the urbanisation variable as for this variable there are slightly fewer observations available and shows that the relationship remains. Column (3) controls for the level of economic development in 1500. As expected, the coefficient of urbanisation is positive and significant at the 5% level. This indicates that there is a correlation between female friendly institutions and economic development. Although the coefficient on ELF1 enters with the correct sign in Column (4) it is never found to be statistically significant. Column (5) adds the legal origins variables, none of which are significant and the relationship between distance to the nearest centre of the Neolithic Revolution and the FFI remains. Overall, from the results presented in Table 2, we can conclude that the relationship between female agency and distance to the nearest centre of the Neolithic Revolution is robust and strong. The estimate in Column (5) of Table 2 indicates that a one-percentage point increase in distance from the old centres of the Neolithic Revolution increased the female-friendliness index by 0.86 points. Table 2 OLS regression FFindex and controls (1) (2) (3) (4) (5) ffindex ffindex ffindex ffindex ffindex -----------------------------------------------------------------------------------------lndistancenr 1.358*** 1.364*** 1.046*** 1.011*** 0.859*** (6.63) (6.33) (4.14) (3.80) (3.27) urb

0.0393** (2.23)

0.0404** (2.40)

0.0449* (1.98)

Latitude

0.0436** (2.47)

0.0426** (2.26)

0.0329 (1.30)

Longitude % Fertile soil % Desert % Tropical climate % Within 100 km. of ice-free~t Ruggedness (Terrai..) elf1 Legal origin indicator: Comm~l

0.00465 (1.15) 0.0118* (1.99) -0.000891 (-0.05) 0.0227** (2.62) 0.00275 (0.65) 0.219* (1.92)

0.00439 (1.02) 0.0109 (1.67) -0.00388 (-0.17) 0.0233** (2.51) 0.00222 (0.52) 0.221* (1.97) -0.517 (-0.35)

-0.000731 (-0.14) 0.00704 (0.79) -0.00124 (-0.05) 0.0241** (2.26) 0.00642 (0.98) 0.242* (1.75) -0.637 (-0.37) -0.102

(-0.11) Legal origin indicator: Fren~c

-0.349 (-0.45)

Legal origin indicator: Soci~s

0.419 (0.51)

Legal origin indicator: Germ~c

0.302 (0.38)

Legal origin indicator: Scan~a

0 (.)

Constant

-8.087*** -8.099*** -9.110*** -8.691*** -7.028*** (-5.37) (-5.09) (-5.87) (-4.33) (-3.23) -----------------------------------------------------------------------------------------r2 0.401 0.398 0.628 0.629 0.645 r2a N 55 50 49 49 49 -----------------------------------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

We now turn to the second part of the analysis where we relate the female friendly-index to estimates of per capita GDP to find out if it helps to explain the reversal of fortune between 1500 and 2000. We regress the female friendliness index on per capita GDP of each country in our sample at different points in time: 1500, 1800, 1870, 1910, 1950 and 2000. Ln Yit =  + 1 Zit + Fi  + it ,

(2)

where Ln Yit denotes the log of per capita GDP in country i in century t, Fi is the female friendliness index of country i, Zit is a vector of control variables which we introduce below, and it is the error term. First we show the results for the individual years, starting with 1500 and ending with 2000. Here we use the same set of controls as in first set of regressions exploring the relation between distance to the Neolithic Revolution and female agency. Geography, ethnic fractionalization and the quality of institutions have impacted economic outcomes, as well as female-friendliness. In addition, we introduce a control variable for oil (averageoil) in 2000 to control for the possibility that oil producing countries experienced a different trajectory of economic development after the Second World War. It may also be correlated with our female friendliness index, as oil producing countries are on average characterized by less female friendly institutions. Finally, we include our distance to the Neolithic Revolution measure, which is, as shown above, highly correlated with our female friendliness index. It may also be directly related to per capita GDP: the countries relatively far away from the old centers were also the ones that have relatively high levels of economic development (see Putterman 2006).

Table 3: OLS results GPD 1500 -----------------------------------------------------------------(1) (2) (3) lngdp lngdp lngdp -----------------------------------------------------------------ffindex 0.0986*** 0.168** 0.171** (4.66) (2.56) (2.63)

Latitude

-0.00620 (-0.78)

-0.00692 (-0.78)

Longitude

0.000978 (0.37)

0.00125 (0.43)

% Fertile soil

-0.00102 (-0.38)

-0.000709 (-0.25)

% Desert

-0.00365 (-0.30)

0.000912 (0.05)

% Tropical climate

-0.00102 (-0.30)

-0.00104 (-0.28)

% Within 100 km. of ice-free~t

-0.00265 (-1.18)

-0.00282 (-1.22)

Ruggedness (Terrai..)

-0.0771 (-1.23)

-0.0778 (-1.23)

lndistancenr

-0.0215 (-0.24)

0.0252 (0.15)

elf1

0.358 (0.40)

Constant

6.468*** 6.977*** 6.596*** (137.46) (10.64) (5.12) -----------------------------------------------------------------r2 0.322 0.481 0.485 r2a N 25 25 25 -----------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 4: OLS Results GDP 1800 -----------------------------------------------------------------(1) (2) (3) lngdp lngdp lngdp -----------------------------------------------------------------ffindex 0.139*** 0.108** 0.113** (3.88) (2.60) (2.71) Latitude Longitude % Fertile soil

0.0119** (2.33)

0.0122** (2.29)

-0.00450** (-2.16)

-0.00435** (-2.07)

0.00183 (1.16)

0.00211 (1.34)

-0.00622 (-0.72)

-0.00387 (-0.37)

% Tropical climate

0.00304 (1.00)

0.00273 (0.88)

% Within 100 km. of ice-free~t

0.00171 (1.02)

0.00176 (1.01)

-0.00695 (-0.13)

-0.00846 (-0.15)

% Desert

Ruggedness (Terrai..) lndistancenr

-0.168** (-2.21)

elf1 Constant

-0.157* (-1.98) 0.192 (0.43)

6.258***

7.197***

7.057***

(93.94) (13.86) (11.22) -----------------------------------------------------------------r2 0.283 0.655 0.657 r2a N 42 42 42 -----------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 5: OLS Results, GDP 1870 -----------------------------------------------------------------(1) (2) (3) lngdp lngdp lngdp -----------------------------------------------------------------ffindex 0.179*** 0.102** 0.108** (3.92) (2.20) (2.34) Latitude Longitude

0.0210*** (3.06)

0.0212*** (3.02)

-0.00711*** -0.00699*** (-3.39) (-3.22)

% Fertile soil

0.00132 (0.65)

0.00155 (0.73)

% Desert

0.00160 (0.14)

0.00423 (0.34)

% Tropical climate

0.00643 (1.62)

0.00608 (1.48)

% Within 100 km. of ice-free~t

0.00135 (0.83)

0.00146 (0.84)

0.0160 (0.21)

0.0149 (0.19)

-0.0814 (-0.97)

-0.0700 (-0.84)

Ruggedness (Terrai..) lndistancenr elf1

0.218 (0.46)

Constant

6.420*** 6.511*** 6.361*** (80.10) (11.33) (10.13) -----------------------------------------------------------------r2 0.257 0.753 0.754 r2a N 47 46 46 -----------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 6: OLS Results GDP 1910 -----------------------------------------------------------------------------(1) (2) (3) (4) lngdp lngdp lngdp lngdp -----------------------------------------------------------------------------ffindex 0.203*** 0.104* 0.0959* 0.108** (3.86) (1.94) (1.88) (2.24) Latitude Longitude

0.0287*** (3.16) -0.00712** (-2.50)

0.0312*** (3.27)

0.0322*** (3.30)

-0.00887*** -0.00880*** (-3.23) (-3.15)

% Fertile soil

0.00313 (1.29)

0.00138 (0.53)

0.00175 (0.63)

% Desert

0.00618 (0.41)

0.00301 (0.22)

0.00953 (0.60)

% Tropical climate

0.00844

0.0115**

0.0112**

(1.63) % Within 100 km. of ice-free~t Ruggedness (Terrai..)

0.000128 (0.07)

(2.16) -0.0000705 (-0.04)

(2.10) 0.000296 (0.15)

0.0373 (0.38)

-0.0102 (-0.15)

-0.0149 (-0.21)

-0.0456 (-0.42)

-0.0996 (-0.94)

-0.0792 (-0.81)

Legal origin indicator: Comm~l

-0.568** (-2.15)

-0.630** (-2.25)

Legal origin indicator: Fren~c

-0.519* (-1.97)

-0.563* (-2.04)

Legal origin indicator: Soci~s

-0.601** (-2.40)

-0.642** (-2.41)

lndistancenr

Legal origin indicator: Germ~c Legal origin indicator: Scan~a elf1

0 (.) -0.718** (-2.11)

0 (.) -0.788** (-2.18) 0.554 (1.31)

Constant

6.778*** 6.259*** 7.288*** 7.010*** (68.32) (8.37) (9.84) (8.79) -----------------------------------------------------------------------------r2 0.245 0.752 0.812 0.818 r2a N 47 46 46 46 -----------------------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

Tables 3 to 6 summarize the results for 1500-1910. The results indicate that there is a consistent and strong connection between the female friendliness index, our measure for the level of female agency, and the log of per capita GDP, our measure for economic development. Important to note is that the relationship between the variables becomes stronger over time. This indicates that growth is fastest in the countries with relatively high levels of female agency. Adding the control variables to the regressions reduces the coefficient to about 0.1 (with the exception of 1500 in Table 3, where control variables increase the size of the coefficient), but the pattern remains intact. What is also interesting is that latitude and longitude are not significantly correlated with GDP in 1500, but that from 1800 onwards they show the patterns known from other studies: the further away a country is from the equator, the higher the GDP will be, and the greater the distance to the Greenwichmeridian, the lower GDP is. In 1500 the geographic differences in GDP per capita are limited, but they appear to emerge between 1500 and 1800. In 1950 and 2000 the coefficient of the univariate regression with the female friendliness index remains significant and sizable, but when adding the set of control variables it result in a collapse of this pattern. In particular the geographic variables seem to dominate the picture. The tentative explanation for this may be that from the early centers of the Industrial Revolution in North-Western Europe and Japan that were characterized by high levels of female agency, modern economic growth has in the course of the 20th century spread to other

parts of the world (Russia and China for example) that were also located in temperate climate zones, but did not share the same institutions as the original centers of growth. Another explanation for this is related to the size of our sample. We have a limited number of observations to work with (ca. 25 to 55), and therefore, when adding a large set of controls, we may lose too many degrees of freedom, making it harder to interpret the regression results. Therefore we will now turn to the regression analysis where we collapse all the information into one single model to verify the robustness of our results. Table 7: OLS Results GDP 1950 -----------------------------------------------------------------------------(1) (2) (3) (4) lngdp lngdp lngdp lngdp -----------------------------------------------------------------------------ffindex 0.192*** -0.0123 0.00994 0.0230 (2.74) (-0.15) (0.12) (0.33) Latitude

0.0383*** (4.85)

0.0372*** (3.27)

0.0429*** (4.25)

Longitude

-0.00892*** -0.00842** (-2.88) (-2.55)

% Fertile soil

0.000800 (0.30)

0.00177 (0.58)

0.00269 (0.83)

% Desert

0.00231 (0.19)

0.000115 (0.01)

0.00863 (0.60)

% Tropical climate

0.00855* (1.70)

0.00920* (1.74)

0.00960* (1.99)

% Within 100 km. of ice-free~t

0.00379* (1.78)

0.00103 (0.45)

0.00227 (0.94)

Ruggedness (Terrai..)

0.0503 (0.54)

-0.0215 (-0.31)

-0.0367 (-0.49)

lndistancenr

0.116 (0.94)

0.0959 (0.75)

0.132 (1.16)

Legal origin indicator: Comm~l

-0.121 (-0.35)

-0.0807 (-0.25)

Legal origin indicator: Fren~c

-0.0868 (-0.26)

0.0631 (0.21)

Legal origin indicator: Soci~s

-0.411 (-1.37)

-0.279 (-0.96)

Legal origin indicator: Germ~c

0.371 (0.96)

0.612 (1.51)

Legal origin indicator: Scan~a

0 (.)

0 (.)

elf1 Constant

-0.00827** (-2.67)

1.340** (2.39)

6.992*** 5.214*** 5.621*** 4.714*** (47.58) (5.16) (4.29) (4.06) -----------------------------------------------------------------------------r2 0.122 0.720 0.767 0.793 r2a N 54 53 53 53 -----------------------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 8: OLS Results GDP 2000 -----------------------------------------------------------------------------------------(1) (2) (3) (4) (5) lngdp lngdp lngdp lngdp lngdp -----------------------------------------------------------------------------------------ffindex 0.320*** 0.0746 0.110 0.121 0.121 (3.71) (0.59) (0.88) (1.07) (1.05) Latitude Longitude

0.0339*** (3.12)

0.0256* (1.74)

0.0303** (2.20)

0.0299** (2.10)

-0.00471 (-1.08)

-0.00595 (-1.57)

-0.00583 (-1.59)

-0.00599 (-1.57)

% Fertile soil

0.00145 (0.37)

0.00243 (0.59)

0.00318 (0.75)

0.00391 (0.86)

% Desert

0.00668 (0.41)

0.00312 (0.20)

0.0101 (0.55)

0.00384 (0.18)

0.000133 (0.02)

0.0000126 (0.00)

0.000344 (0.05)

0.000719 (0.10)

0.00678** (2.15)

0.00780** (2.36)

0.00775** (2.33)

% Tropical climate % Within 100 km. of ice-free~t

0.00998*** (3.45)

Ruggedness (Terrai..)

0.0445 (0.39)

-0.0788 (-0.93)

-0.0914 (-1.03)

-0.0755 (-0.86)

lndistancenr

0.259 (1.27)

0.181 (0.99)

0.211 (1.26)

0.203 (1.10)

Legal origin indicator: Comm~l

-0.254 (-0.58)

-0.221 (-0.54)

-0.236 (-0.55)

Legal origin indicator: Fren~c

-0.407 (-0.99)

-0.284 (-0.75)

-0.333 (-0.80)

Legal origin indicator: Soci~s

-0.644* (-1.84)

-0.535 (-1.62)

-0.558 (-1.62)

Legal origin indicator: Germ~c

0.721 (1.55)

Legal origin indicator: Scan~a

0 (.)

elf1

0.918* (1.93)

0.868* (1.77)

0 (.)

0 (.)

1.098 (1.16)

0.815 (0.73)

averageoil

0.00920 (0.52)

Constant

7.918*** 4.986*** 6.426*** 5.682*** 5.748*** (39.14) (3.27) (3.77) (3.67) (3.31) -----------------------------------------------------------------------------------------r2 0.209 0.647 0.742 0.753 0.755 r2a N 54 53 53 53 53 -----------------------------------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.0

Since the female friendliness index is time-invariant, as are many of our control variables, we estimate the relationship using a simple OLS model. All panel regressions include a full set of century dummies, and report the robust standard errors. Table 9: OLS Results, unbalanced panel, GDP. ----------------------------------------------------------------------------(1) (2) (3) (4) lngdp lngdp lngdp lngdp ------------------------------------------------------------------------------

ffindex

0.196*** (7.78)

0.0749** (2.32)

0.0814** (2.54)

0.0878*** (2.78)

year==

1500.0000

0 (.)

-0.0629 (-0.51)

-0.0953 (-0.73)

-0.0881 (-0.66)

year==

1800.0000

-0.120 (-1.43)

0 (.)

0 (.)

0 (.)

year==

1870.0000

0.134 (1.45)

0.250*** (3.45)

0.263*** (3.54)

0.260*** (3.52)

year==

1910.0000

0.548*** (5.40)

0.661*** (8.57)

0.674*** (8.72)

0.671*** (8.72)

year==

1950.0000

0.738*** (5.94)

0.904*** (10.10)

0.923*** (10.33)

0.914*** (10.32)

year==

2000.0000

1.950*** (13.32)

2.116*** (17.72)

2.136*** (18.21)

2.126*** (18.19)

0.0252*** (5.93)

0.0268*** (4.89)

0.0283*** (5.16)

Latitude Longitude

-0.00577*** -0.00602*** -0.00596*** (-3.84) (-3.58) (-3.57)

% Fertile soil

0.00235 (1.64)

0.00173 (1.10)

0.00202 (1.27)

% Desert

0.00318 (0.46)

0.000670 (0.10)

0.00462 (0.59)

% Tropical climate

0.00423* (1.67)

0.00557** (1.98)

0.00545* (1.94)

% Within 100 km. of ice-free~t

0.00315*** (3.02)

0.00181 (1.49)

0.00221* (1.76)

0.0145 (0.35)

-0.0416 (-1.16)

-0.0466 (-1.27)

-0.0175 (-0.30)

-0.0372 (-0.58)

-0.0192 (-0.31)

Legal origin indicator: Comm~l

0.105 (0.66)

0.121 (0.76)

Legal origin indicator: Fren~c

0.110 (0.68)

0.155 (0.96)

Legal origin indicator: Soci~s

-0.111 (-0.69)

-0.0647 (-0.40)

Ruggedness (Terrai..) lndistancenr

Legal origin indicator: Germ~c

0.501*** (2.62)

0.581*** (3.02)

Legal origin indicator: Scan~a

0 (.)

0 (.)

elf1

0.517 (1.61)

Constant

6.245*** 5.538*** 5.667*** 5.331*** (74.99) (11.56) (9.50) (8.88) -----------------------------------------------------------------------------r2 0.584 0.768 0.789 0.791 r2a N 269 265 265 265 -----------------------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

Table 10. OLS results, balanced panel, GDP ------------------------------------------------------------------------------

(1) (2) (3) (4) lngdp lngdp lngdp lngdp -----------------------------------------------------------------------------ffindex 0.292*** 0.145*** 0.161*** 0.161*** (12.41) (3.91) (4.08) (4.02) year==

1500.0000

-1.123*** (-7.58)

-1.123*** (-7.86)

-1.123*** (-7.94)

-1.123*** (-7.91)

year==

1800.0000

-1.061*** (-7.69)

-1.061*** (-8.73)

-1.061*** (-9.12)

-1.061*** (-9.10)

year==

1870.0000

-0.739*** (-5.32)

-0.739*** (-6.70)

-0.739*** (-7.16)

-0.739*** (-7.13)

year==

1910.0000

-0.315** (-2.13)

-0.315*** (-2.70)

-0.315*** (-2.92)

-0.315*** (-2.91)

year==

1950.0000

0 (.)

0 (.)

0 (.)

0 (.)

year==

2000.0000

1.333*** (7.18)

1.333*** (8.66)

1.333*** (8.98)

1.333*** (8.94)

0.0254*** (3.84)

0.0443*** (4.39)

0.0451*** (4.20)

Latitude Longitude % Fertile soil % Desert

-0.00200 (-1.26)

0.00108 (0.48)

0.00109 (0.49)

0.00527** (2.06)

0.00540* (1.93)

0.00535* (1.87)

0.0136 (0.89)

0.0243 (1.59)

0.0237 (1.37)

0.00348 (0.67)

0.00372 (0.69)

% Tropical climate

-0.000705 (-0.17)

% Within 100 km. of ice-free~t

0.000426 (0.36)

-0.00132 (-0.72)

-0.00128 (-0.71)

0.0592 (1.48)

0.0253 (0.58)

0.0260 (0.61)

-0.0170 (-0.18)

-0.0428 (-0.39)

-0.0570 (-0.44)

Legal origin indicator: Comm~l

0 (.)

0 (.)

Legal origin indicator: Fren~c

0.188 (1.43)

0.199 (1.41)

Legal origin indicator: Soci~s

-0.343 (-1.42)

-0.342 (-1.41)

Legal origin indicator: Germ~c

0.115 (0.56)

0.119 (0.58)

Legal origin indicator: Scan~a

-0.170 (-0.86)

-0.170 (-0.85)

Ruggedness (Terrai..) lndistancenr

elf1 Constant

-0.0900 (-0.12)

7.150*** 6.243*** 5.531*** 5.604*** (52.10) (8.92) (6.42) (5.60) -----------------------------------------------------------------------------r2 0.806 0.849 0.859 0.859 r2a N 150 150 150 150 -----------------------------------------------------------------------------t statistics in parentheses * p<0.10, ** p<0.05, *** p<0.01

Both the balanced and the unbalanced panel show a strong connection between FFindex and GDP and the more or less expected signs of the control variables.

Conclusion This paper sheds new light, we think, on the character of long-term economic development in EurAsia. It is possible to distinguish two different phases of development, a first process of ‘ancient’ economic and political development, and a second stage of ‘modern’ economic and institutional growth. During the first stage, which was concentrated in and near the centers of the Neolithic Revolution, development resulted in – co-evolved with – the creation of hierarchical structures, both at the level of the state and at the micro level. Development and inequality went hand in hand. The EurAsian economy as a result was characterized by a large band of (strong) states, high levels of urbanization and relatively intense international trade stretching from the Mediterranean to the Yangze Delta. This created the spatial structure of family systems and gender relations we mapped in this paper (see Figure 1). Growth after 1500 was fundamentally different to growth before 1500: it started and was most intense in the margins of the EurAsian continent, in regions with relatively low levels of patriarchy, in Western Europe and Japan. Growth was not based on the creation of large hierarchical structures subjugating the working population, on ‘extractive institutions’, but on ‘bottom up’ processes of market participation and investment in human capital (by households), which required radically different, inclusive institutions. The margins of EurAsia, where the hierarchization of the previous millennia had not occurred, were much better placed for this second stage of growth. Here we find most intense forms of preindustrial growth (in the North Sea area), followed by the industrial revolution and the ‘Great Divergence’. Our explanation for this ‘reversal of fortune’ is a combination of the gendered North hypothesis and the gendered Becker hypothesis, but at this point we can only demonstrate that female friendly institutions appear to have mattered for long-term economic success in EurAsia after 1500. Which mechanisms translated female agency into growth cannot easily be established at the level of EurAsia as a whole, but we have suggested elsewhere – in case studies of the economic and demographic effects of the European Marriage Pattern for example – which mechanisms may have played a role. Our analysis also helps to explain the persistent regional differences in family systems and gender relations, and argues that their roots are to be sought in the differential impact of the first stage of economic development. Moreover, the persistence of these institutions in the heartland of EurAsia hindered their economic modernization. The reversal of fortune within EurAsia is in the end explained by this interplay of geographic and institutional factors – such as distance to the centers of the NR, the institutions that emerged in the core and in the margins of the Continent. In conclusion differences in the position of women within family systems help to explain the Little Divergence and the Great Divergence.

References (incomplete) Acemoglu, Daron, James A. Robinson, and Dan Woren. Why Nations Fail: The Origins of Power, Prosperity and Poverty. New York: Crown Business, 2012. Acemoglu, D., Johnson, S., & Robinson, J. A. (2002). Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution. Quarterly Journal of Economics, 117(4), 1231-1294. Acemoglu, Daron, Simon Johnson, and James Robinson. “The Rise of Europe: Atlantic Trade, Institutional Change, and Economic Growth.” American economic review 95, no.3 (2005): 546-579. Becker, Gary S. “An Economic Analysis of Fertility.” In Demographic and economic change in developed countries, 209-40. New York: Columbia University Press, 1960. __________.”A Theory of the Allocation of Time.” The economic journal 75 (1965): 493517. Becker, Gary S., and H. Gregg Lewis. “On the Interaction between the Quantity and Quality of Children.” The Journal of Political Economy 81 (1973): S279-S288. Becker, Gary S., and Nigel Tomes. “Child Endowments, and the Quantity and Quality of Children.” The Journal of Political Economy 85, no. 4(2) (1976): S143-S162. Becker, Sascha O., Francesco Cinnirella, and Ludger Woessmann. “The Trade-Off between Fertility and Education: Evidence from before the Demographic Transition.” Journal of Economic Growth 15, no. 3 (2010): 177-204. Carmichael, Sarah., Tine De Moor, and Jan Luiten van Zanden. “Introduction.” History of the Family 16, no. 4 (2011): 309-311. Carmichael, Sarah, and Jan Luiten van Zanden. “Towards an Ethnographic Understanding of the European Marriage Pattern: Global Correlates and Links with Female Status”. Population Histories in Context (forthcoming). Currently online available as CGEH Working Paper no. 67, 2015. http://www.cgeh.nl/sites/default/files/WorkingPapers/cgehwp67_carmichaelvanzande n.pdf Comin, Diego, William Eastery and Erick Gong (2010), Was the Wealth of Nations Determined in 1000BC? American Economic Journal: Macroeconomics 2 (July 2010): 65–97. Desmet, Klaus, Ignacio Ortuño-Ortín, Romain Wacziarg (2012) The Political Economy of Linguistic Cleavages. Journal of Development Economics, 97, 322-336. Diamond, Jared (1997) Guns, Germs and Steel. The Fate of Human Societies. Norton. Siegfried Gruber and Mikołaj Szołtysek (2015) The patriarchy index: a comparative study of power relations across historical Europe. In: The History of the Family, [DOI: 10.1080/1081602X.2014.1001769 Dyble, M., et.al. (2015) Sex equality can explain the unique social structure of huntergatherer bands, Science 15 May 2015, 796-798. Hansen, C.W., P.S. Jensen and C.V. Skovsgaard (2015) Modern gender Roles and Agricultural History: the Neolithic Inheritance. Journal of Economic Growth North, Douglass C. Structure and Change in Economic History. New York: Norton, 1981. North, Douglass C., John Joseph Wallis, and Barry R. Weingast. Violence and Social Orders: A Conceptual Framework for Interpreting Recorded Human History. Cambridge: Cambridge University Press, 2009. Olsson, Ola & Paik, Christopher, 2013. "A Western Reversal since the Neolithic? The longrun impact of early agriculture," Working Papers in Economics 552, University of Gothenburg, Department of Economics.

Olsson, Ola & Paik, Christopher, 2015. "Long-Run Cultural Divergence: Evidence From the Neolithic Revolution," Working Papers in Economics 620, University of Gothenburg, Department of Economics Schultz, Theodore W. “Investment in Human Capital.” The American economic review 51 (1961): 1-17. Teignier, Marc and David Cuberes. “Aggregate Costs of Gender Gaps in the Labor Market: A Quantitative Estimate.” UB Economics Working Paper no. E14/308, 2014. http://ssrn.com/abstract=2405006. Todd, Emmanuel. The Explanation of Ideology: Family Structures and Social Systems. Oxford: Blackwell, 1985. __________. The Causes of Progress: Culture, Authority and Change. Oxford: Blackwell, 1987. ------------------- L'Origine des systèmes familiaux : Tome 1 L'Eurasie, Gallimard, 2011. Trigger, B. G. (2003). Understanding early civilizations: A comparative study. Cambridge University Press Wright, Katherine I. (2007) Women and the Emergence of Urban Society in Mesopotamia. In Sue Hamilton et.al. (eds) Archeology and Women. Ancient and Modern Issues. Left Coast Press, Walnut Creek, CA., pp. 199-246.