Economic Uncertainty and Fertility: A Meta-Analysis

Daniele Vignoli, University of Florence
Giammarco Alderotti, Università di Roma - Sapienza
Michela Baccini, University of Florence

The relation between economic uncertainty and fertility has always been an important topic in demographic research, and the Great Recession, spanning 2007 to 2009, made it increasingly crucial over the last years. Although studied since ever, the interplay between economic uncertainty and fertility is far from being clearly understood: uncertainty is usually deemed to be a negative condition, but different fertility reactions are both advocated by sociological theories and supported by micro-level evidence, which is still fragmented and sometimes contradictory. Given this state of the art, we decided to perform a meta-analysis (i.e., a quantitative literature review) in order to synthesize all the evidence coming from the existing literature and to draw general conclusions about the size and the direction of the impact of economic uncertainty on fertility.

Articles were collected systematically, using an electronic database (Scopus) and applying inclusion criteria to decide which ones to include in our sample. In order to avoid ecological correlation fallacy, we considered only micro-level studies, limiting the search to Europe to make the analyses more comparable. So far, 120 articles were collected and 60 of them were correctly coded and included in the meta-sample.

Preliminary results are available and they show that significant differences exist between men and women, between transition to parenthood and to higher parities, and over time. Nevertheless, the hypothesis of homogeneity within groups was rejected by all the tests done. On this basis, we need to be careful when interpreting the preliminary results and we have to go deeper into the analysis, performing a meta-regression in order to study the heterogeneity, which is going to be the next step.


1.    Introductionand objective

Theinterplay between economic uncertainty and fertility has always been part of theresearch agenda of demographers and sociologists, and over the last years ithas become increasingly crucial. The notion of economic uncertainty refers toclarity, or the lack thereof, about future economic activities (Bloom,2014; Moore, 2016) and reflects the likelihood ofexperiencing adverse labour market conditions over the life course. Startingfrom the 80s, the labour market experienced a strong process of deregulationthat renewed growth of labour insecurity (Standing, 1997), generating anunprecedented level of structural uncertainty in contemporary societies (Millsand Blossfeld, 2005), strengthened by the recent Great Recession. Such ascenario affects all workers and it is really likely to have an impact onfamily formation as well (Blossfeld et al., 2006; Esping-Andersen, 1999).

Thenature of the relationship between economic uncertainty and fertility is debated,because both the hypotheses of a negative and positive interplay are supportedby sociological theories and empirical evidence. For example, Ranjan (1999)concluded that increasing uncertainty about the future income could lead peopleto postpone childbearing, while Friedman and colleagues (1994) advocated that womenmight choose the career of mother when chances in labour market are limited.Empirical evidence on this topic is mixed, fragmented and, at times, contradictory:data showed that the relationship between uncertainty and fertility can benegative (Pailhé 2012, Hofman 2013), positive (Perelli-Harris 2006, Kreyenfeld2010), non-significant (deLange et al. 2014).

Webelieve that before conducting another micro-level study, there is a sensibleneed to get a closer understanding of previous research to draw generalconclusions about the size and the direction of the impact of economicuncertainty on fertility in Europe. To this end, we performed a quantitativeliterature review, namely a meta-analysis (e.g., Matysiak and Vignoli 2008) tosynthesise, merge and interpret all the evidence coming from heterogeneousstudies, standardised for country, method applied, control variables, and soforth.

Methods

Theprocedure to perform a meta-analysis can be summarized into three steps:collection of information, quantitative synthesis, analysis of heterogeneity.

Wecollected articles using the largest electronic database, Scopus (www.scopus.com),and then we applied inclusion criteria to assess which articles could enter ourmeta-sample. Conference papers were disregarded to ensure a high-qualitymeta-analysis. Then, we want to avoid ecological correlation fallacy exludingmacro-level studies. Third, we restricted the search to studies in Europeancountries because they display an interesting variation in fertility patternsand labour market, while also sharing certain socio-economic and culturalcharacteristics that minimize heterogeneity. Finally, it is important to decidewhich measures of uncertainty are to include. A growing number of studiespointed a connection between economic uncertainty, in the form of unemployment,temporary contracts and unstable labour market situations, and fertility (Millset al. 2011). Hence, this work will focus on objective measurements of economicuncertainty deriving from the labour market, looking at employment conditionsand type of contract (considering fixed-time contracts, part-time employmentand unemployment as “treatments”, opposed to more stable positions). Nevertheless,studies including other measures of economic uncertainty (income-relatedvariables, perceived uncertainty) were also collected to be used in a moreadvanced stage. To date, 115 articles were collected and 60 of them were alreadycoded and included in the meta-sample.

Preliminaryresults

Althoughthe meta-sample is not completed, preliminary results show interesting general patterns.Table 1 displays the estimations of the average effect-sizes separately bygender, parity and before/after 1995 (the hypothesis is that the effect mighthave changed significantly over time).

Fromthis descriptive analysis, some differences arise. First, we note a positiveand significant association between economic uncertainty and transition toparenthood, but non-significant for higher parities. Second, women’s employmentuncertainty seems to be positively related to fertility, while for men thisassociation is negative. Third, the general positive effect of economicuncertainty on fertility decreases in recent times.

Clearly,these preliminary results must be interpreted with caution because within-groupsheterogeneity is high. For example, the positive result for women could be theconsequence of a mixture of positive effects (volunteer part-time), negativeeffects (precarious employment) and more theoretically ambivalent effects(unemployed women). In this regard, all the tests that we performed rejectedthe hypothesis of homogeneity within groups. Therefore, our next step will bethe analysis of heterogeneity, by performing a meta-regression (forthcoming).

Ina more advanced stage of our work, we would like to improve the results of ourmeta-analysis adding the results from a network meta-analysis (NMA), namely anextension of pairwise meta-analysis that facilitates indirect comparisons ofmultiple competing treatments that have not yet been studied in head-to-headand allows the visualisation of a larger amount of evidence, estimation of therelative effectiveness among all treatments, and rank ordering of the treatments.

Presented in Session 1165: Fertility