In-Vitro Fertilization and Educational Gradients in Fetility: Evidence from Denmark

Anna Barbuscia, Oxford University
Sander Wagner, ENSAE, CREST, Université Paris Saclay

We explore the effect of the emergence of a new technology, in-vitro fertilization, on the relationship between education and fertility. Using the Danish IVF Register, in combination with the population register from 1994 to 2005 we look at all births that took place in the country and show how the educational gradient of in-vitro-fertility changed over that time period, allowing us to gain insights into how the adaption of this new technology evolves across different socio-economic groups. We also look at how that gradient compares to the fertility-education gradient of non-IVF births and how it influenced the overall relationship between fertility and education in the country. Finally we decompose the contribution of self-selection into IVF-treatment, choice of treatment method and differential success rates across educational strata to the relationship between education and IVF births we found.

I. Introduction

Education and Fertility are key aspects of individual lifecourses as well as of the societal metabolism. They are also intricately linked. This link can be regarded directly or through the prism of its interplay with institutions. The insight that educated women tend to have later births and in most places lower overall fertility looks at the direct interplay between the two interactions. Whereas the insight that the educational expansion in industrialized countries has been accompanied by declining fertility rates, looks at a potential interaction of the linkage between these two variables and institutions. We propose adding a new technological and institutional change that is quickly gaining in importance to the analysis: the emergence of in-vitro fertilization (IVF). Denmark has been among the countries with the earliest and most complete embrace of IVF. It was sanctioned as a public health intervention in 1986 and its usage has been stedily growing since. In 2010 about 8% of Danish children were
conceived via IVF.

II. Data
We use the Danish IVF register provided by the Danish National Board of Health for the years 1994 until 2005 in which reporting has been mandatory. This means completeness of IVF registrations is close to a 100 percent. From the IVF register we obtain information on the date of treatment and date of birth when birth took place, on the reason for infertility, the mode of treatment, the number of fertilized eggs transferred back to the womb as well as on wether the treatment resulted in birth, abortion, stillbirth, or no pregnancy. The IVF register contains information on 32,007 women receiving one or several IVF treatments. We merged the IVF register with other registry data to obtain information on educational attainment, the labor market situation, age, number of children and marital status of the women undergoing IVF treatments as well as of their partners. We also use the registry data on the enture Danish population from 1980 to 2005 to compare women undergoing IVF treatment to those that do not.

III. Methods
In the first stage of our analysis we look at the educational gradient of IVF births and non-IVF births for mothers and fathers and look at how this gradient
changed over the course of the 11 years between 1994 and 2005. In regression form we look at the results of a simple logit model

(1) y(birth) = alpha + beta education + epsilon

where education is measures as years of education. We then augment this model, stepwise with different covariates X, such as previous number of children, partner''s education, and income to obtain estimates of the conditional effects of education on IVF fertility

(2) y(birth) = alpha + beta education + X + epsilon

In the next step we use a nested logit model to look at the extent to which education in uences rst selection into IVF treatment, choice of treatment method,
and nally success of treatment, to look at how these dierent stages contribute to the overall education-fertility gradient. Finally we analyse the extent to which the overall fertility-education gradient has been influenced by IVF births in Denmark over time.

IV. Expected Results

We expect to test the following hypotheses.

1) IVF mothers are higher educated and have higher income than non-ivf mothers. IVF fathers are higher educated and have higher income than non-ivf
fathers, but the difference is less strong than that of mothers and mostly explained by assortative mating patterns

2) the main factor explaining the higher educational gradient of IVF mothers is selection into treatment, however treatment success rates also contribute
to the gradient

3) As the technology becomes more widely available and used the difference between IVF births and non-IVF births in terms of maternal education
begins to shrink

4) overall the availability of IVF technology has a noteable impact on the fertility education gradient.

Presented in Session 1159: Fertility