Three essays on the microeconomic analysis of long-acting reversible contraception

UNCG Author/Contributor (non-UNCG co-authors, if there are any, appear on document)
Lorissa Charis Pagan (Creator)
The University of North Carolina at Greensboro (UNCG )
Web Site:
Christopher Swann

Abstract: In this dissertation I use a variety of methods to study the effect of and choice to use long-acting reversible contraceptives (LARCs) using data from the National Survey of Family Growth. Effective contraception may increase women’s welfare by allowing better control of the timing of fertility. However, many contraceptives rely on consistent use to achieve low failure rates, potentially leaving benefits of fertility control forgone due to user error. As LARCs do not rely on contraceptive adherence, they may be welfare improving through eliminating the gap between the “perfect use” and “typical use” failure rates. How much LARCs can reduce the risk of unintended pregnancies depends on the degree to which LARC users would have used other methods consistently and correctly, how sensitive the other methods would be to inconsistent use, and how long LARCs are used. In my first essay, I implement survival analysis techniques to analyze which reversible methods women used before transitioning to a long-acting method and the duration of LARC use. Consistent with the literature, I find that LARC use is associated with high continuation rates. I also find that contraceptive spells of methods that are more sensitive to inconsistent use are not at a greater risk of ending due to switching into LARC use, which may dampen the effect of increased LARC use on pregnancies among contracepting women. Evaluating the additional effectiveness of using a LARC compared to other methods on pregnancy is difficult as women may select into LARC use due to their risk of pregnancy. In my second essay, I use an exogenous change in provider recommendations to get around the selection issue and evaluate the causal effect of LARC use on pregnancies and births using an instrumental variables approach. First, I show that the release of the recommendation had a differential effect for younger mothers of one child compared to older mothers of one child. Then using this exogenous variation, I find that LARC use decreases the probability of pregnancies in the current year, in the following year, and births in the following year compared to other methods, at least among young mothers who were affected by the recommendation. Using a correlated random coefficients model, I find evidence that women who choose LARCS would have been more likely to experience pregnancy in the following year. Finally, LARCs can only be welfare improving if women choose them over other alternatives. As with any product, each contraceptive can be thought of as a bundle of its different characteristics such as its maximum duration of use, its effectiveness, if it is hormonal, and if the method requires the insertion of a device. In my third essay, I use discrete choice models to estimate how attributes of contraceptives affect method choice. I find that the LARC-specific attribute of being a physical device may discourage some women from choosing a LARC method.

Additional Information

Language: English
Date: 2020
Contraceptives, IUD, LARC, Long-Acting Reversible Contraception
Contraceptives $v Statistics
Contraception $v Statistics

Email this document to