Kin Availability and Fertility in a Historical Nuclear Family - - PDF document

Kin Availability and Fertility in a Historical Nuclear Family Society: Sweden in 1900 Martin Dribe and Björn Eriksson Centre for Economic Demography Department of Economic History Lund University Lund, Sweden Martin.Dribe@ekh.lu.se Bjorn.Eriksson@ekh.lu.se October 18, 2017 Abstract Research in anthropology and evolutionary demography has for a long time stressed the importance of kin for fertility. Also in historical demography the kin-fertility link has been studied but mainly in contexts with more complex family systems, such as China or Japan, while the issue has received much less attention in the West. One reason is the lack of good data to study proximity of kin in societies where coresidence with kin was relatively

• unimportant. The aim of this paper is to study the impact of kin on fertility in a nuclear family

society at the beginning of the fertility transition. We use linked full-count census data for Sweden 1880-1900 in which mothers and fathers are linked to their parental household, and where families in both generations are followed in subsequent censuses. This enables us to identify where grandmothers and grandfathers lived at the time of childbearing of their

• children. We study net fertility of women aged 20-44 in 1900 using child woman ratios, and

analyze the association between net fertility and the proximity of kin. We also compare fertility behavior of siblings living in close proximity to their parents to the behavior of siblings living farther away from their parents. Paper for the session “Household, kinship and population dynamics in historical populations” at the 28th International Population Conference (IUSSP), Cape Town, South Africa, October 29 - November 4, 2017. This paper is part of the research program “The Rise and Fall of an Industrial City. Landskrona Population Study”, funded by Riksbankens jubileumsfond.

1

• 1. Introduction

In anthropological and evolutionary demographic research the potential importance of kin, and especially grandmothers, for reproduction has often been stressed (see, e.g., Sear and Coall 2011). The main argument in this literature is that grandmothers, and to a lesser extent also grandfathers, can help maximize the number of surviving grandchildren both through increased fertility of their daughters and improved survival of their grandchildren (e.g. Beise and Voland 2002; Sear et al. 2000, 2003; Lahdenperä et al. 2004; Tymicki 2004). In this way grandmothers can play an important role for reproductive fitness also beyond their own childbearing age, and this has been seen as an important reason for the long post-reproductive life span among humans (Williams 1957; Voland et al. 2005; Hawkes et al. 1998). In historical demography, especially dealing with complex-family societies, the importance of kin for fertility outcomes has also been highlighted (e.g. Wang, Campbell and Lee 2010; Tsuya and Kurosu 2010), while there has been much less focus on these aspects of fertility in Western societies with nuclear family systems (see, however, Hawkes and Smith 2009; Jennings et al. 2012). Instead, most of the research on Western historical fertility and fertility transitions have focused on socioeconomic and cultural variables, and decision making in the context of the nuclear family (e.g. Alter, Neven and Oris 2010; Breschi et al. 2010; Bengtsson and Dribe 2006; 2010a, 2010b; Guinnane 2011; Van Bavel 2004, Szreter 1996). At least partly, this neglect of kin as a determinant of fertility in nuclear-family contexts can be explained by the simple fact that there was not as much coresidence with kin in these societies as there was in societies with different kinds of extended-family households. Coresidence is, however, not necessary for kin influence on fertility behavior. Having grandmothers or other kin alive and living nearby may be just as important as having them in the household, especially if the mechanism is helping with child rearing, but also in terms of socialization and transmission of attitudes and behavior. It has also been shown there are strong patterns of intergenerational transmission of fertility behavior in Western contexts, but especially during

• r after the fertility transition (e.g. Bras et al. 2013; Murphy 1999, Jennings et al. 2012; Booth

and Kee 2009; Reher et al. 2008; Kolk 2014). The aim of this paper is to study the importance of geographic proximity of kin for net fertility in a nuclear-family society (Sweden) in the early phases of the fertility transition. We focus on the role of maternal and paternal grandmothers and grandfathers for net fertility (surviving children under age five), and explicitly study the geographic proximity of the

• grandparents. In the analysis we also exploit sibling comparisons as a way to control for

2 shared family environments which is crucial when trying to assess the mechanisms. In the literature, transmission of attitudes and values regarding childbearing has often been put forward as leading explanations behind intergenerational transmission of fertility, while ideas about cooperative breeding is more related to actual assistance in rearing children to

• adulthood. Comparing sisters growing up in the same family and who presumably were

socialized in a similar way, but who differed in terms of proximity to their parents and parents-in law makes it possible control for the shared value background when studying the impact of the grandmother and grandfather on fertility. In addition, we also distinguish the impact of maternal grandmothers and grandfathers from that of paternal grandmothers and grandfathers. Our results show a clear a positive association between geographic proximity of maternal grandparents and net fertility, but no association for paternal grandparents. They also show a clear negative association between coresidence and net fertility for both maternal and paternal grandparents, as well as between widowhood of grandparents and net fertility.

• 2. Background

It has often been pointed out that women do not raise their children all by themselves but that individuals both within and outside the immediate nuclear family help out in different ways. In addition to fathers who help raise their children through provision of food and other necessities, the crucial role of (helpful) grandmothers for both fertility and child survival has been stressed in anthropology and evolutionary demography (Hawkes et al. 1998, 2000; Hawkes and Coxworth 2013; Voland et al. 2005). The rather long post-menopausal life span in humans may seem as a paradox from an evolutionary perspective, but if we take into account that post-menopausal grandmothers can help their daughters and daughters-in-law to reproduce they will contribute to the multiplication of their genes (biological fitness). In biology this reproductive strategy, where kin help each other to maximize reproductive success, is called cooperative breeding, and has been found important in a number of different species, including humans (see, e.g., Mace and Sear 2005; Mathews and Sear 2013; Sear and Coall 2011). Turning specifically to fertility, there can be different effects of kin on the timing and pace of childbearing. Based on previous research and evolutionary theory, Mathews and Sear (2013) distinguish between kin assistance and kin priming as main mechanisms relating kin to

• fertility. Kin can assist potential mothers in different ways facilitating reproduction. In poorer,

3 high-fertility contexts, this can be related to better nutrition and health, which could serve to improve fecundability. Help from kin could also promote early weaning and hence a faster return of fecundability after birth. In developed countries with lower fertility and higher rates

• f female labor force participation, kin can help by providing childcare as well as giving

financial assistance, access to housing, etc. Kin priming is when relatives encourage potential mothers (and fathers) to have children through the formation of pro-natalist attitudes and influence via direct

• communication. Through social interaction and conversation mothers can affect the fertility

decisions of their daughters and daughters-in-law both through direct persuasion and through a more indirect encouragement of life-course transitions conducive to childbearing (cohabitation, union formation, housing investments, etc). Both kin assistance and kin priming would have the same effect of promoting fertility, and in an empirical analysis using contemporary British survey data, Mathews and Sear (2013) found some evidence for both mechanism being at work. Sear and Coall (2011) review the empirical evidence for a grandmother effect on fertility in different contexts, both before and after the fertility transition. Most studies seem to find a relationship between grandparents and fertility, but both the direction of the association, and which grandparent that is most important for fertility, differ widely across

• studies. Despite all the heterogeneity, Sear and Coall conclude that in pre-transitional

societies paternal grandmothers in particular seem to promote fertility, while maternal grandmothers if anything help to reduce fertility (for a different result see Jennings et al. 2012 for historic Utah). Maternal grandmothers, on the other hand, seem to be especially important in promoting child survival, which does not seem to be the case for paternal grandmothers (Mace and Sear 2005; Voland and Beise 2002). Mace and Sear relate these differences to the greater uncertainty about paternity than about maternity as well as to more serious consequences of maternal death for the maternal kin than for the paternal kin. Both factors make paternal kin more interested in high fertility to maximize their inclusive fitness, while the maternal kin need to trade higher fertility for maternal depletion, which lead them to encourage a bit lower fertility and instead invest more in their offspring. The impact of kin for fertility is highly dependent on the family system; whether nuclear or extended, and whether patrilineal or matrilineal (see, e.g., Sorenson Jamison et al. 2002). In a nuclear-family society such as Sweden, where each generation is expected to live in a separate household, intergenerational coresidence may not be advantageous, and may not be a good indicator of kin assistance from grandmothers to mothers and grandchildren.

4 Instead, it may be an indication of frailty and bad health in the grandparental generation, and thus of caregiving to parents by married daughters or daughters-in-law, which could serve to reduce fertility. Cramped housing could also have implied a lack of privacy in mutigenerational households, which may have reduced coital frequency and hence fertility (Hacker and Roberts 2017). In such contexts geographic proximity, but not coresidence, could be important for both kin assistance and kin priming. Having a mother or mother-in-law living nearby could promote fertility through both priming and direct assistance with childcare, while living in the same household would discourage fertility. In many cases it is difficult to identify kin outside the household in demographic sources, especially for historical contexts where we cannot rely on survey data, and where migration makes direct observation of kin difficult without strong stayer bias. In a recent paper using US census data, Hacker and Roberts (2017) use an innovative method to infer kin in nearby households using surnames and information about the ordering of households in the

• census. They study net fertility (children under five in the household) in the 1880 US census

and find that having “potential mothers-in-law” in a nearby household increased net fertility by about two percent. Coresidence with own mothers (maternal grandmothers) lowered net fertility, while coresidence with mothers-in-law weakly promoted net fertility. The main drawback with their analysis is that it is impossible to directly identify maternal and paternal grandparents in the cross-sectional census to measure exactly how far away they resided. The main contribution of our analysis is to provide a more precise measure

• f geographic proximity of the grandparents using linked census data. We are able to link a

large number of mothers to their own parents and parents-in-law and measure how far away they lived, providing a very precise measure of geographic proximity.

• 3. Context: Sweden in 1900

Around 1900 Sweden witnessed the real breakthrough of industrialization and the emergence

• f industrial society (see Schön, 2010). Earlier in the nineteenth century, a number of

important changes happened that paved the way for the industrial economy. Increased demand for agricultural products, both domestically and from abroad, led to massive investments in agriculture and to profound institutional change. From mid-century, a very dynamic development took place around the railroad investments and innovations in steel making. At the same time, both economic and political institutions were transformed in a more liberal direction, including the introduction of a modern parliamentary system (although universal

5 suffrage for both men and women was not fully implemented until 1921); the deregulation of production, trade, and the labor market; and a strengthening of property rights. The growth of both domestic and external demand led to increasing growth in industrial investments and, ultimately, to a transition to modern economic growth (Jörberg, 1961: 8-28). Iron and timber were the leading industrial sectors, and export of oats and, later, butter was also of great importance. Annual rates of growth in GDP/capita increased from 0.4 percent in the first half of the nineteenth century to 2.1 percent between 1890 and 1930 (Schön, 2010:13). Despite an ongoing process of urbanization Sweden retained its rural

• character. A majority of the population was still employed in agriculture at the turn of the

twentieth century, and it was not until the 1950s that more than half of the population lived in towns (Statistics Sweden, 1969: Table 14). Beginning around 1880 Sweden also entered its fertility transition, about 100 years after the start of the mortality decline (Dribe 2009), and in 1901/05 total fertility (TFR) was 3.9; a decline from 4.5 in 1871/75. In the following 30 years total fertility declined to 1.8 (Statistics Sweden 1999, Table 3.3). Fertility decline started in the upper classes and spread gradually to the lower classes, with the rural classes of farmers and agricultural workers lagging behind (Dribe and Scalone 2014).

• 4. Data and method

Most historical studies of kin availability and fertility use data from regional or local samples, sometimes plagued by small numbers and strong selection following migration. In this paper we use full-count census data for Sweden, covering the entire population 1880 and

• 1900. The Swedish historical full count censuses are a unique source for studying kin and
• fertility. Unlike other contemporary censuses women appear with their maiden names even

when married. This peculiarity makes it possible to link women between the censuses to nearly the same extent as men. It is thus possible to follow women from first observation as children to marriage and family formation Individuals have been linked between the censuses using a probabilistic linking method which results in high linkage rates, a representative sample, and few false positive links (see Eriksson 2015). In total we are able to link close to 71 percent of all individuals between the two censuses.1 The linkage rate varies somewhat for men and women (73 percent for men and 68 percent for women) but not dramatically so. The resulting linked sample is a

1 71 percent is the backward linkage rate, i.e. the share of all individuals appearing in the 1900 census that were

born in or before 1880 which we were able to link to the 1880 census.

6 rich data source which allows for precise identification a kinship and variation in proximity to kin over time. Our study population consists of all married women aged 20-44 in 1900. Parents and parents-in-law are identified by observing them together with the woman or man in 1880 and then again 20 years later in 1900. These intergenerational links makes it possible to identify and locate kin residing not only in the same household (which is possible with most census data) but also residing elsewhere in the country. We use coordinates of parish geographic mid points to calculate the actual distance to kin. The identification of parents also allows us to identify sisters who can be followed between 1880 and 1900. To be defined as sisters we require that the considered women’s biological mother and father are identical in 1880. This restriction leaves us with a sister sample consisting of 30,070 women belonging to 13,599 groups of sisters that range in size from 2 to 7. We measure net fertility by the number of own children under five in the household (child-woman ratios). This measure of net fertility has been used extensively in past research and forms the basis of the own-children method to estimate standard aggregate fertility rates (Cho et al 1986; Breschi et al. 2003). For the Swedish censuses used here it has also been demonstrated that the unadjusted child woman ratios reflect socioeconomic differentials in total marital fertility quite closely (Dribe and Scalone 2014; Scalone and Dribe 2017). In addition, we have information for all men (husbands, brothers-in-law, fathers and fathers-in- law) on social class (occupation), as well as other household and family related conditions for all individuals, such as place of residence (at the parish level). We analyze the association between the geographic proximity of grandparents and net fertility using linear regression (OLS). The dependent variable is the number of own children 0-4, and the main explanatory variable is a categorical variable measuring the proximity of maternal and paternal grandmothers/fathers. It is divided into the following categories: same household, same parish, different parish within 10 kilometers, 10-49 kilometers and more than 49 kilometers. In addition, we have a category for grandparents that could not be linked between the censuses, which either means that they had died or could not be linked for other reasons. In separate regressions we assess the role of unobserved geographic heterogeneity using parish-level fixed effects and shared family characteristics using sister fixed effects. We control for socioeconomic status of the husband, age of woman, age difference between spouses, presence of children over 4 (as an indirect control for marital duration), woman’s employment status, and widowhood of the grandparent under

7

• consideration. Socioeconomic status is measured by HISCLASS, an internationally

comparable historical class scheme (Van Leeuven and Maas 2011) which in turn is based on the coding scheme HISCO (Van Leeuven et al. 2002).

• 5. Results

The descriptive statistics, presented in table 1, compare the analytical samples with all women in the selected age group in the 1900 census. Overall the full analytical sample is similar to the whole census in terms of net fertility (1.13 compared to 1.11), children over 4 in the household (65 percent and 64 percent), age of woman (33.9 years and 34.2 years), and age difference between spouses (a difference of 1 percent or less in all categories). The class distributions also look quite similar. There is some overrepresentation of farmers in the analytical sample (33 percent compared to 29 percent in the census), but apart from that proportions in different classes are similar. The higher proportions of farmers in the sample is related to their lower migration propensities which increases the linkage rate. The sister sample is very similar to the analytical sample in terms of class distribution, but deviates more for fertility and spousal age difference. In the analysis below, we compare results of identical models in the two samples to assess differences between them in the importance of grandparents. Table 1 here A high proportion of grandparents have not been linked between the censuses (57-77 percent). Some are not linked because they died between 1880 and 1900, others because there were more than one possible match in the census given the variables we link on. In the sister sample we condition upon the maternal grandparents being linked since there is no between- sister variation in the grandparent-not-linked category. Our sister sample is somewhat younger than our analytical sample, a result of the fact that it is more probable to observe a woman with her sister in 1880 if she is younger rather than older. This explains much of the difference between the two analytical samples. If we only look at the women with linked parents/parents-in-law, 6 percent coresided with their own mothers (2.7 percent of the whole sample), and 44 percent lived in the same parish, 37 percent had their mothers in a different parish but within 50 kilometers, while 13 percent had them farther away. Looking instead at mothers-in-law (paternal grandmothers in

8 table 1), the figures are quite similar. 13 percent of the linked women coreside, 45 percent live in the same parish, 29 percent have their mothers-in-law in a different parish within 50 kilometers and 12 percent farther away. Overall, these numbers show the nuclear character of the Swedish family system of the time. Intergenerational coresidence was rather uncommon also in cases when it was possible (i.e. when both generations were alive), but a majority of women lived in rather close proximity to both their mothers and mothers-in-law. Still there is a sufficiently high proportion of women whose kin lived farther away for the analysis to be meaningful. Table 2 shows the results for separate models for maternal and paternal grandmothers and grandfathers. The models include all control variables. Column 1 is the model for the full sample without parish fixed effects, column 2 includes the parish fixed

• effects. Columns 3-5 are models for the sister sample, with no fixed effects (column 3), parish

fixed effects (column 4) and sister fixed effects (column 5). The final model with sister fixed effects is the most restrictive, constraining identification of kin effects to between-sister variation in proximity of their parents and parents-in-law. Table 2 here Panel A displays the results for maternal grandmothers. Looking first at proximity, the reference category is the maternal grandmother living more than 50 kilometers away from the daughter and her family. In the first model geographic proximity of the maternal grandmother is related to higher net fertility, except in cases of coresidence in the same household, which is negatively related to net fertility. There are no big differences within the 50-kilometer range, implying that grandmothers living in the same parish have about as much influence as grandmothers living 10-49 kilometers away. Women with unlinked mothers have similar fertility as those with mothers living more than 50 kilometers away. It also seems that widowhood is negatively related to daughters’ net fertility, but the coefficients are not statistically significant in the main sample. Adding the parish fixed effects does not change the main pattern, but alters the magnitudes somewhat. The sister sample also look fairly similar to the main sample in terms of the basic pattern of the association between maternal grandmother proximity and daughter net fertility (comparing columns 1 and 3, and 2 and 4 respectively). The main exception is that the coefficient for widowhood is now statistically significant and of a sizeable magnitude. Moreover, adding the sister fixed effects does not change the picture to any greater extent. Maternal grandmothers residing in their own

9 household within a range of 50 kilometers appear to promote the fertility of their daughters, while coresidence generally seems to discourage fertility. In panel B we look at maternal grandfathers, and the pattern is very similar to that of the maternal grandmothers. Proximity of maternal grandfathers promotes fertility as long as it does not involve coresidence. Widowhood is also associated with lower daughter fertility, and the association is stronger than was the case for the maternal grandmothers. As was the case for maternal grandmothers, these results are robust both to the inclusion of parish fixed effects in the full sample, and the inclusion of sister fixed effects in the reduced sample. Having maternal grandfathers living in independent households within a 50-kilometer range promotes net fertility of their daughters in all specifications, while the opposite is true for coresidence. In panel C we turn to paternal grandmothers. Overall there is much less of fertility- promoting effects of geographic proximity for paternal grandmothers than was the case for maternal grandmothers. However, there is the same negative association between coresidence and net fertility as we saw for the maternal grandmothers, while the negative association for widowhood is much stronger in this case than it was for maternal grandmothers. Adding parish fixed effects do not change the picture, but in the sister sample there is no association at all between proximity and net fertility except for the remaining negative coefficient for coresidence. Finally, panel D shows the estimates for paternal grandfathers. Overall, the results are similar to those for paternal grandmothers. Proximity is not closely associated with daughters’ net fertility, but coresidence is negatively associated with net fertility, and the same is true for women with non-linked fathers-in-law. There is also a strong association with widowhood, both in the main sample and in the sister sample, implying that women whose mothers-in-law are dead have lower net fertility. This effect is particularly strong in the model with sister fixed effects. It should be noted that this is a very small group (see table 1). In Table 3 we turn to models including both the maternal and paternal grandmother/grandfather. Panel A shows the results for grandmothers and panel B for

• grandfathers. Looking first at the grandmothers, the results confirm what we saw previously

in the separate models. Including the paternal grandmother does not alter the main pattern for the maternal grandmother that geographic proximity is related to higher net fertility. As before, there are no systematic differences in the impact of proximity within the 50-kilometer

• range. For paternal grandmothers there is no positive association between geographic

proximity and net fertility. For both maternal and paternal grandmothers there is a negative association between coresidence and daughters’ net fertility. There also seems to be negative

10 relationships between widowhood and daughters’ net fertility, but the patterns differ somewhat across different model specifications. Table 3 here Panel B shows the results for grandfathers. Overall the patterns are very similar to the one for grandmothers. For maternal grandfathers there is a clear positive relationship between geographic proximity and daughters’ net fertility. For paternal grandfathers there is no such association. For both grandfathers, however, coresidence is negatively related to net fertility, and the same is the case for widowhood. Overall, the patterns are similar in the different model specifications, which shows that the results are robust to both geographic unobserved heterogeneity and shared characteristics at the family level.

• 6. Discussion

In biological and anthropological models of reproduction kin often play an important part, acting as cooperative breeders facilitating childbearing and promoting child survival. These and other intergenerational transfers are crucial not only to understand decisions about childbearing and childrearing, but also for our understanding of human aging and mortality selection (see Lee 2003). The role of kin for reproduction is mediated by institutions, such as family systems, welfare states, and markets and to fully understand the kin influence studies need to be contextualized. Especially in nuclear-family societies where intergenerational coresidence is both relatively unusual and strongly selected, standard measures of presence of kin in the household will not be valid indicators of the influence of kin. Instead we need information about the geographic location of kin in relation to the family under consideration. In historical contexts we usually rely either on parish-level population registers of smaller areas or cross-sectional censuses without identifiers of kin relationships outside the

• household. Except in rare cases where genealogical data is available to complement the other

sources it is therefore impossible to identify the location of kin outside the household without introducing bias from only looking at the stayer population. In this study we have overcome this problem by linking census data, thereby identifying the place of residence of paternal and maternal grandparents in relation to parents and children. Our results show that geographic proximity of maternal grandmothers and grandfathers appears to have a fertility-promoting effect on their daughters, as long as they do

11 not actually coreside. Living within 50 kilometers from their mother and father increases net fertility of women, while there is not much difference within this range. Coresidence with parents, on the other hand, reduces net fertility significantly. For paternal grandmothers and grandfathers the picture is quite different. Here there are no associations with geographic proximity, but a clear negative relationship between coresidence and net fertility, just as for maternal grandparents. For all kinds of grandparents, widowhood also reduces net fertility of daughters/daughters-in-law. These associations are present also when taking geographic unobserved heterogeneity and shared family characteristics into account. Thus, even if we did not apply a fully causal empirical design, we are certain that the associations shown are not a simple result of compositional differences between families or across geography. These results clearly indicate that proximity to kin was important for fertility decisions also in Sweden in 1900; a nuclear-family society undergoing the fertility transition. Grandparents seem to have mattered, but there were important differences between maternal and paternal grandparents. Maternal grandparents were more important for net fertility than paternal grandparents. These results do not support the hypothesis that paternal grandparents promote fertility while maternal are more directed on improving child survival, as has been argued in previous research. The context, however, is somewhat different from a classical natural fertility society. Fertility was already quite low and in most cases not close to levels where maternal depletion was a major concern. Instead, assistance in childcare and/or a pro- natalist discourse may have been important. Maternal grandmothers may have been more likely to give assistance, and women may also have been more susceptible to pressure to have children from their own mothers than from their mothers-in-law. It is also possible that living close to one’s parents reduced the likelihood of adopting new attitudes on birth control, which may have delayed the onset of family limitation. That innovation diffusion was an important aspect of the fertility decline has often been stressed (e.g., Cleland and Wilson 1987; Cleland 2001; Lesthaeghe and Vanderheoft 2001), and a tighter relationship to the parental generation might well have reduced the responsiveness to such new attitudes and innovative behavior, which could be one explanation for the kin influence observed. One limitation of this study is that we are not measuring fertility but net fertility, i.e. the number of surviving children under age five. Even though previous research has shown that infant and early child mortality was low in relation to fertility, and that differentials in net fertility in terms of socioeconomic status were similar to real fertility differentials (Scalone and Dribe 2017), it is still possible that some of the grandparent effect works through improved survival rather than higher fertility.

12 As has already been pointed out, intergenerational coresidence can be expected to have signaled that the grandparents were unhealthy or lacked the means to support themselves, which in both cases would have required time and resources of the middle generation (i.e. the mothers). Hence, instead of receiving assistance in caring for their children they had to care for their parents, which contributed to reduce their fertility. It was more difficult to separate the associations for grandmothers and grandfathers. Because the widowhood rate was so low in the linked sample, in most cases both the grandmother and the grandfather was alive. Nonetheless, in cases when they were widowed it was detrimental to fertility, and this was quite uniform for all grandparents. Hence, there is not much support for a more important role for grandmothers than for grandfathers, but the result is consistent with explanations focusing on resources and time availability to provide

• assistance. The authority of grandparents in priming their children, and thereby affecting their

fertility decisions, may also have been greater when they were both alive and functioned as a family themselves. As a whole, our analysis points to the importance of kin for fertility decisions also in nuclear-family contexts undergoing fertility decline. On a general level our results are in accordance with the hypothesis of cooperative breeding, but they are equally consistent with more traditional explanations for the historic fertility decline.

13 References Alter, G., M. Neven, and M. Oris (2010). Economic change and differential fertility in rural eastern Belgium, 1812 to 1875. In N. O. Tsuya, et al. Prudence and Pressure. Reproduction and Human Agency in Europe and Asia, 1700-1900. Cambridge, MA: MIT Press. Beise, J. and E. Voland (2002). Opposite effects of maternal and paternal grandmothers on infant survival in historical Krummhörn. Behavioral Ecology and Sociobiology 52: 435-443 Bengtsson, T. and M. Dribe (2010a). Agency, social class, and fertility in southern Sweden, 1766 to 1865. In N. O. Tsuya, et al. Prudence and Pressure. Reproduction and Human Agency in Europe and Asia, 1700-1900. Cambridge, MA: MIT Press. Bengtsson, T. and M. Dribe (2010b). Economic stress and reproductive responses. In N. O. Tsuya, et al. Prudence and Pressure. Reproduction and Human Agency in Europe and Asia, 1700-1900. Cambridge, MA: MIT Press. Booth, A. L. amd H. J. Kee (2009). Intergenerational transmission of fertility patterns. Oxford Bulletin of Economics and Statistics 71: 183-208 Bras, H., J. Van Bavel, J., and K. Mandemakers. (2013). Unraveling the intergenerational transmission of fertility: Genetic and shared-environment effects during the demographic transition in the Netherlands, 1810–1910. The History of the Family 18: 116–134. Breschi, M., S. Kurosu, and M. Oris (eds.) (2003). The Own-Children Method of Fertility Estimation: Applications in Historical Demography. Udine: Forum. Breschi, M., et al. (2010). Patterns of reproductive behavior in preindustrial Italy: Casalguidi, 1819 to 1859, and Venice, 1850 to 1869. In N. O. Tsuya, et al. Prudence and

• Pressure. Reproduction and Human Agency in Europe and Asia, 1700-1900.

Cambridge, MA: MIT Press. Cho, L. J., R. D. Retherford, and M. K. Choe (1986). The Own-Children Method of Fertility

• Estimation. Honolulu: The University of Hawaii Press.

Cleland, J. (2001). Potatoes and pills: An overview of innovation-diffusion contributions to explanations of fertility decline, in J. B. Casterline (ed), Diffusion Processes and Fertility Transition: Selected Perspectives. Washington, D. C.: National Research Council, pp. 39–65.

14 Cleland, J. R. and C. Wilson (1987). Demand theories of the fertility transition: An iconoclastic view. Population Studies 41: 5-30. Dribe, M. (2009). Demand and supply factors in the fertility transition: A county level analysis of age-specific marital fertility in Sweden 1880-1930. European Review of Economic History 13:65-94. Dribe, M. and F. Scalone (2014). Social class and net fertility before, during and after the demographic transition: A micro-level analysis of Sweden 1880-1970. Demographic Research 30: 429-464. Eriksson, B. (2015). Dynamic Decades: A micro perspective on late nineteenth century Sweden, PhD dissertation. Lund: Media-Tryck. Guinnane, T. W. (2011). The historical fertility transition: A guide for economists. Journal of Economic Literature 49: 589-614. Hacker, J. D. and E. Roberts (2017). The impact of kin availability, parental religiosity, and nativity of fertility differentials in the late nineteenth-century United States. Demographic Research, forthcoming. Hawkes, K. et al. (1998). Grandmothering, menopause, and the evolution of human life

• histories. Proceedings of the National Academy of Sciences 95:1336-1339.

Hawkes, K. and K. R. Smith. (2009). Brief communication: Evaluating grandmother effects. American Journal of Physical Anthropology 140: 173-176. Jennings, J. A., A. R. Sullivan, and J. D. Hacker (2012). Intergenerational transmission of reproductive behavior during the demographic transition. Journal of Interdisciplinary History 42: 543-569. Jörberg, Lennart, 1961. Growth and Fluctuations of Swedish Industry 1869-1912. Stockholm: Almqvist & Wiksell. Kolk, M. (2014). Understanding transmission of fertility across multiple generations – Socialization or socioeconomics? Research in Social Stratification and Mobility 35: 89-103. Lahdenperä, M., Lummaa, V., Helle, S., Tremblay, M., & Russel, A. F. 2004. Fitness benefits

• f prolonged post-reproductive lifespan in women. Nature 428: 178-181.

Lee, R. D. (2003). Rethinking the evolutionary theory of aging: Transfers, not births, shape senescence in social species. Proceedings of the National Academy of Sciences 100:9637-9642. Lesthaeghe, R. and C. Vanderhoeft (2001). Ready, willing, and able: A conceptualization of transitions to new behavioral forms, in J. B. Casterline (ed), Diffusion Processes and

15 Fertility Transition: Selected Perspectives. Washington, D. C.: National Research Council, pp. 240–264. Mathews, P. and R. Sear (2013). Family and fertility: Kin influence on the Progression to a second birth in the British Household Panel Study. PLoS ONE 8(e56941):1-10. Murphy, M. 1999. Is the relationship between fertility of parents and children really weak? Social Biology 46: 122-145. Reher, D. S., J. A. Ortega, and A. Sanz-Gimeno (2008). Intergenerational transmission of reproductive traits in Spain during the demographic transition. Human Nature 19:23- 43 Scalone, F. and M. Dribe (2017). Testing child-woman ratios and the own-children method on 1900 Swedish census data: Examples of indirect fertility estimates by socioeconomic status in a historical population. Historical Methods 50:16-29 Schön, Lennart, 2010. Sweden’s Road to Modernity. An Economic History. Stockholm: SNS. Sear, R. and D. Coall (2011). How much does family matter? Cooperative breeding and the demographic transition. Population and Development Review 37 (Suppl.): 81-112 Sear, R. R. Mace, and I. A. McGregor. (2000). Maternal grandmothers improve the nutritional status and survival of children in rural Gambia. Proceedings of the Royal Society of London, Series B, Biological Sciences 267: 461–467. Sear, R. R. Mace, and I. A. McGregor (2003). The effects of kin on female fertility in rural

• Gambia. Evolution and Human Behavior 24: 25–42.

Sorenson Jamison, C., L. L. Cornell, P. L. Jamison, and H. Nakazato (2002). Are all grandmothers equal? A review and preliminary test of the “grandmother hypothesis” in Tokugawa Japan. American Journal of Physical Anthropology 119: 67-76. Statistics Sweden, 1969. Historisk statistik för Sverige Del 1. Befolkning 1720-1967. Stockholm: SCB. Statistics Sweden, 1999. Befolkningsutvecklingen under 250 år. Historisk statistik för Sverige. Demografiska rapporter 1999:2. Stockholm: SCB. Szreter, S. (1996). Fertility, Class and Gender in Britain 1860-1940. Cambridge: Cambridge University Press. Tsuya, N. O. and S. Kurosu (2010). Family, household, and reproduction in northeastern Japan, 1716 to 1870. In N. O. Tsuya, et al. Prudence and Pressure. Reproduction and Human Agency in Europe and Asia, 1700-1900. Cambridge, MA: MIT Press.

16 Tymicki, K. (2004). The kin influence on female reproductive behavior. The evidence from the reconstitution of Bejsce parish registers, 18th – 20th centuries, Poland. American Journal of Human Biology 16:508-522 Van Bavel, J. (2004). Deliberate birth spacing before the fertility transition in Europe: Evidence from nineteenth-century Belgium. Population Studies 58: 95-107. Van Leeuwen, M. H. D. and Maas, I. (2011). HISCLASS. A Historical International Social Class Scheme. Leuven: Leuven University Press. Van Leeuwen, M. H. D., Maas, I., and Miles, A. (2002). HISCO. Historical International Standard Classification of Occupations. Leuven: Leuven University Press. Voland, E. and J. Beise (2002). Opposite effects of maternal and paternal grandmothers on infant survival in historical Krummhörn. Behavioral Ecology and Sociobiology 52:435-443. Voland, E. et al. (eds.) (2005). Grandmotherhood. The Evolutionary Significance of the Second Half of Female Life. New Brunswick, NJ and London: Rutgers University Press. Wang F., C. Campbell, and J. Z. Lee (2010). Agency, hierarchies, and reproduction northeastern China, 1789 to 1840. In N. O. Tsuya, et al. Prudence and Pressure. Reproduction and Human Agency in Europe and Asia, 1700-1900. Cambridge, MA: MIT Press. Williams, G. C. 1957. Pleiotropy, natural selection and the evolution of senescence. Evolution 11:398-411.

17 Table 1. Descriptive statistics for the sample of women in 1900 census.

1900 Census Analytical sample Sister sample Mean

• Std. Dev.

Mean Std. Dev. Mean

• Std. Dev.

Children 0-4 1.11 0.94 1.13 0.93 1.28 0.94 Children >4 years at home 63.7% 65.4% 54.2% SES

• 1. Higher managers

0.4% 0.4% 0.2%

• 2. Higher professionals

1.3% 1.1% 0.8%

• 3. Lower managers

2.7% 2.6% 2.4%

• 4. Lower proffesional, clerical and sales

4.4% 3.9% 3.8%

• 5. Lower clerical and sales

1.3% 0.9% 0.8%

• 6. Foremen

2.4% 2.2% 1.7%

• 7. Medium skilled workers

13.4% 12.5% 11.6%

• 8. Farmers and fishermen

28.9% 33.3% 33.8%

• 9. Lower-skilled workers

12.9% 12.3% 12.7%

• 10. Lower-skilled farm workers

1.8% 2.0% 2.5%

• 11. Unskilled workers

15.6% 14.8% 15.8%

• 12. Unskilled farm workers

7.0% 7.7% 7.7% Missing/not classified 7.8% 6.3% 6.1% Age of woman 33.93 6.41 34.20 6.50 30.91 5.40 Age difference between spouses Wife older 23.7% 22.7% 20.2% Husband 0-2 older 23.5% 23.1% 24.5% Husband 3-6 older 27.5% 28.0% 29.6% Husband >6 older 25.4% 26.1% 25.6% Woman employed 1.0% 0.5% 0.4% Maternal grandmother proximity >50 km 5.6% 12.9% 10-49 km 11.0% 27.4% 0-9 km 4.8% 11.2% Same parish 19.0% 46.6% Same household 2.7% 1.9% Not linked 57.0% Maternal grandmother widowed 0.0% 0.1% Maternal gradfather proximity >50 km 5.7% 13.0% 10-49 km 11.0% 27.2% 0-9 km 4.9% 11.2% Same parish 19.1% 46.7% Same household 2.0% 1.9% Not linked 57.4% Maternal grandfather widowed 0.0% 0.1% Paternal grandmother proximity >50 km 2.8% 3.6% 10-49 km 4.9% 6.5% 0-9 km 2.2% 2.6%

18

Same parish 10.9% 15.0% Same household 3.2% 4.4% Not linked 76.0% 67.9% Paternal grandmother widowed 0.0% 0.0% Paternal grandfather proximity >50 km 2.8% 3.5% 10-49 km 4.9% 6.5% 0-9 km 2.2% 2.7% Same parish 10.8% 14.9% Same household 2.6% 3.7% Not linked 76.7% 68.7% Paternal grandfather widowed 0.0% 0.0% N 450,597 265,254 30,070

19 Table 2. OLS estimates of number of own children under 5 in the household in 1900. Separate models for each grandparent.

Full sample Sister sample 1 2 3 4 5 Panel A: Maternal grandmother (mother) Proximity >50 km ref. ref. ref. ref. ref. 10-49 km 0.032 *** 0.058 *** 0.048 *** 0.074 *** 0.061 ** (0.009) (0.009) (0.018) (0.019) (0.027) 0-9 km 0.015 0.043 *** 0.029 0.059 ** 0.024 (0.011) (0.011) (0.022) (0.024) (0.033) Same parish 0.027 *** 0.028 *** 0.057 *** 0.049 *** 0.054 ** (0.009) (0.009) (0.017) (0.018) (0.026) Same household

• 0.144 ***
• 0.094 ***
• 0.172 ***
• 0.144 ***
• 0.098 *

(0.013) (0.013) (0.040) (0.044) (0.056) Not linked 0.005 0.032 *** (0.008) (0.008) Widow

• 0.032
• 0.030
• 0.409 **
• 0.520 ***

(0.081) (0.081) (0.177) (0.185) F-statistic 632.640 608.384 45.718 42.104 24.046 R2 0.087 0.087 0.056 0.057 0.056 N 265254 265254 30070 30070 30070 Panel B: Maternal grandfather (father) Proximity >50 km ref. ref. ref. ref. ref. 10-49 km 0.030 *** 0.055 *** 0.045 ** 0.072 *** 0.057 ** (0.009) (0.009) (0.018) (0.019) (0.027) 0-9 km 0.022 ** 0.052 *** 0.027 0.058 ** 0.022 (0.011) (0.011) (0.022) (0.024) (0.033) Same parish 0.028 *** 0.025 *** 0.054 *** 0.046 ** 0.051 ** (0.009) (0.009) (0.017) (0.018) (0.026) Same household

• 0.156 ***
• 0.111 ***
• 0.179 ***
• 0.150 ***
• 0.119 **

(0.014) (0.015) (0.040) (0.044) (0.057) Not linked 0.011 0.028 *** (0.008) (0.008) Widower

• 0.125
• 0.086
• 0.546 ***
• 0.674 ***

(0.082) (0.083) (0.170) (0.194) F-statistic 631.891 611.776 45.795 42.161 24.076 R2 0.087 0.087 0.056 0.057 0.056 N 265254 265254 30070 30070 30070

20

Panel C: Paternal grandmother (mother-in-law Proximity >50 km ref. ref. ref. ref. ref. 10-49 km 0.005 0.017

• 0.021
• 0.023
• 0.007

(0.013) (0.013) (0.035) (0.036) (0.046) 0-9 km 0.009 0.022

• 0.012
• 0.006
• 0.012

(0.016) (0.016) (0.043) (0.045) (0.058) Same parish 0.021 * 0.005

• 0.014
• 0.040
• 0.027

(0.012) (0.012) (0.032) (0.033) (0.042) Same household

• 0.073 ***
• 0.070 ***
• 0.112 ***
• 0.126 ***
• 0.108 **

(0.014) (0.015) (0.039) (0.040) (0.051) Not linked

• 0.013
• 0.010
• 0.030
• 0.042
• 0.042

(0.011) (0.011) (0.029) (0.030) (0.039) Widow

• 0.241 **
• 0.250 **

0.277 0.116 0.257 (0.117) (0.118) (0.287) (0.281) (0.403) F-statistic 628.885 608.052 44.107 40.825 22.979 R2 0.087 0.086 0.055 0.056 0.056 N 265254 265254 30070 30070 30070 Panel D: Paternal grandfather(father-in-law) Proximity >50 km ref. ref. ref. ref. ref. 10-49 km

• 0.005

0.008

• 0.021
• 0.024
• 0.040

(0.013) (0.013) (0.034) (0.034) (0.047) 0-9 km

• 0.011

0.003

• 0.024
• 0.022
• 0.056

(0.016) (0.016) (0.043) (0.046) (0.059) Same parish 0.008

• 0.009
• 0.023
• 0.045
• 0.068

(0.012) (0.012) (0.031) (0.033) (0.042) Same household

• 0.095 ***
• 0.099 ***
• 0.102 **
• 0.109 **
• 0.117 **

(0.015) (0.016) (0.040) (0.044) (0.053) Not linked

• 0.019 *
• 0.021 **
• 0.033
• 0.046
• 0.069 *

(0.011) (0.011) (0.029) (0.030) (0.039) Widower

• 0.245 *
• 0.184
• 0.503 **
• 0.543 ***
• 0.832 ***

(0.147) (0.143) (0.233) (0.196) (0.219) F-statistic 628.605 607.646 44.042 40.831 23.325 R2 0.087 0.086 0.055 0.056 0.056 N 265254 265254 30070 30070 30070 Parish of birth fixed effects X X Sister fixed effects X

Note: Models control for SES, # children>4, age of woman, age difference between spouses, and woman employed

21 Table 3. OLS estimates of number of own children under 5 in the household in 1900. Paternal and maternal grandparents in the same model.

Full sample Sister sample 1 2 3 4 5 Panel A: Grandmothers (mother and mother-in-law) Maternal grandmother proximity >50 km ref. ref. ref. ref. ref. 10-49 km 0.032 *** 0.058 *** 0.051 *** 0.079 *** 0.064 ** (0.009) (0.010) (0.018) (0.020) (0.027) 0-9 km 0.014 0.043 *** 0.032 0.065 *** 0.030 (0.011) (0.011) (0.022) (0.025) (0.033) Same parish 0.026 *** 0.029 *** 0.062 *** 0.056 *** 0.060 ** (0.009) (0.009) (0.017) (0.018) (0.026) Same household

• 0.148 ***
• 0.100 ***
• 0.176 ***
• 0.151 ***
• 0.108 *

(0.013) (0.013) (0.040) (0.045) (0.056) Not linked 0.005 0.032 *** (0.008) (0.008) Widow

• 0.030
• 0.028
• 0.406 **
• 0.524 ***

(0.081) (0.081) (0.177) (0.188) Paternal grandmother proximity >50 km ref. ref. ref. ref. ref. 10-49 km 0.001 0.010

• 0.037
• 0.045
• 0.023

(0.013) (0.013) (0.035) (0.036) (0.047) 0-9 km 0.008 0.017

• 0.022
• 0.024
• 0.021

(0.016) (0.016) (0.043) (0.045) (0.059) Same parish 0.017 0.000

• 0.032
• 0.059 *
• 0.043

(0.012) (0.012) (0.032) (0.034) (0.043) Same household

• 0.084 ***
• 0.084 ***
• 0.138 ***
• 0.157 ***
• 0.135 ***

(0.015) (0.015) (0.039) (0.041) (0.052) Not linked

• 0.016
• 0.015
• 0.044
• 0.059 *
• 0.055

(0.011) (0.011) (0.030) (0.031) (0.039) Widow

• 0.245 **
• 0.251 **

0.262 0.094 0.244 (0.117) (0.118) (0.286) (0.280) (0.404) F-statistic 563.151 542.295 40.816 37.740 21.435 R2 0.088 0.087 0.057 0.058 0.057 N 265254 265254 30070 30070 30070

22

Panel B: Grandfathers (father and father-in-law) Maternal grandfather proximity >50 km ref. ref. ref. ref. ref. 10-49 km 0.031 *** 0.056 *** 0.049 *** 0.078 *** 0.065 ** (0.009) (0.009) (0.018) (0.020) (0.027) 0-9 km 0.023 ** 0.054 *** 0.032 0.065 *** 0.032 (0.011) (0.011) (0.022) (0.025) (0.034) Same parish 0.029 *** 0.028 *** 0.060 *** 0.054 *** 0.062 ** (0.009) (0.009) (0.017) (0.018) (0.026) Same household

• 0.160 ***
• 0.117 ***
• 0.182 ***
• 0.155 ***
• 0.121 **

(0.014) (0.015) (0.040) (0.044) (0.057) Not linked 0.012 0.030 *** (0.008) (0.008) Widower

• 0.126
• 0.087
• 0.543 ***
• 0.675 ***

(0.082) (0.083) (0.170) (0.195) Paternal grandfather proximity >50 km ref. ref. ref. ref. ref. 10-49 km

• 0.009

0.000

• 0.038
• 0.048
• 0.058

(0.013) (0.013) (0.035) (0.035) (0.047) 0-9 km

• 0.014
• 0.004
• 0.035
• 0.040
• 0.065

(0.016) (0.016) (0.043) (0.046) (0.059) Same parish 0.003

• 0.014
• 0.043
• 0.065 *
• 0.084 **

(0.012) (0.012) (0.032) (0.033) (0.043) Same household

• 0.108 ***
• 0.114 ***
• 0.132 ***
• 0.144 ***
• 0.148 ***

(0.015) (0.016) (0.040) (0.045) (0.054) Not linked

• 0.022 **
• 0.027 **
• 0.049 *
• 0.065 **
• 0.083 **

(0.011) (0.011) (0.029) (0.030) (0.039) Widower

• 0.252 *
• 0.188
• 0.516 **
• 0.557 ***
• 0.849 ***

(0.147) (0.143) (0.233) (0.199) (0.220) F-statistic 562.245 545.304 40.834 37.819 21.780 R2 0.088 0.087 0.057 0.058 0.057 N 265254 265254 30070 30070 30070 Parish of birth fixed effects X X Sister fixed effects X

Note: See Table 2.