This 11-minute podcast was released Nov. 28, 2018.
St. Louis Fed economist Paulina Restrepo-Echavarria (above, right) talks about her research on “search and matching,” particularly as it relates to the marriage market. “We’re finding that people search too little. People should be exerting much more effort to search for their optimal match for there to be more efficiency in the marriage market,” she says. She notes that studying the marriage market is important because it has implications for income inequality and taxation.
Kristie Engemann: Welcome to the Timely Topics series from the St. Louis Fed. I’m Kristie Engemann, your host for this podcast. With me is Paulina Restrepo-Echavarria, an economist in the Research division. She’ll be talking to us about her work on search and matching, particularly as it relates to the marriage market. Welcome, Paulina. Thanks for being here.
Paulina Restrepo-Echavarria: Thank you for having me.
Engemann: Can you first explain what “search and matching” means?
Restrepo-Echavarria: Search and matching is the process of searching for something and finding it and forming a match with that. That can be a partner, so you can be searching for someone to go on a date with. You can make some phone calls or ask your friends if they know someone, and that’s the searching process. And then the matching process is when you say, “Okay, yeah, I like this person. I’m going to ask them out on a date.” And you ask them out, and if they say yes, a match actually happens.
It’s the same thing when you want to buy something. If, let’s say, that you’re looking for a particular perfume and you start searching for all the different ones that you may like, and then when you actually go to the store and find the one that you like, that’s the matching part after the searching process.
Engemann: Can you talk a little bit about your recent research that you’ve done in the search and matching area?
Restrepo-Echavarria: My research is theoretical mostly, but in the literature, there are two types of models. There’s one model where the way you meet someone is by pure chance. Everything is random. You basically sit at your house and just wait for men to knock on your door. You open the door. You look at the guy. You kind of figure out, “Well, I don’t like him so much.” You close the door again. You wait for the next one. At some point, you accept the guy and ask him to come into your house. You decide that based on a tradeoff between waiting a longer amount of time and the characteristics of the guy.
On the other hand, you have the literature where we assume that we know everything about someone and you know where to find them. So, you know who’s your ideal guy. You know exactly where he lives, and you can just go to his house, knock on his door, and he’s going to be super happy to have you there. I think this is completely unrealistic.
What I work on is theoretical models that operate in the middle ground. Basically, you know what type of person you like, but you don’t know where to find them, so you have to search for them, and you have to pay search costs in order to find them more accurately. And, hopefully, when you’re able to find them, they look back at you. Because the other thing is, it’s not obvious that there’s going to always be reciprocation. So, in my model, there’s two-sided heterogeneity, meaning that both sides are searching, and there has to be mutual interest for a match to be formed after the search process.
Engemann: How do you define success in these search and matching models? For instance, in the marriage market, is it just finding a match? So, finding someone to marry, or does it depend on the length of the marriage?
Restrepo-Echavarria: Probably different people will say different things, but if you ask me, I think that success would be an efficient match, meaning that is the best match that you can have because, sure enough, you can find different matches, but they’re going to be inefficient, meaning you’re going to end up with a person that you don’t like so much. For me, it needs to be efficient, so it’s the best possible combination of matches among population.
Engemann: What kinds of characteristics did you look at for this? Did you look at education levels and income levels?
Restrepo-Echavarria: From a theoretical perspective, our model allows for multidimensional types, meaning that each particular individual can have all the characteristics that you may want to define their type. We’re not constraining that part, theoretically. When we take the model to the data to actually see what we can find out about the real world, it’s a different thing because for computational reasons we need to take a stand on what a type is, and we did it in terms of income, education and age.
Engemann: What can you say about what are the optimal matches?
Restrepo-Echavarria: In that sense, we can’t really say that one match is better than another because it depends on your preference for types. What we can say is about the search process, and we know that we’re finding that people search too little. People should be exerting much more effort to search for their optimal match for there to be more efficiency in the marriage market.
Engemann: Do your results have any implications beyond who marries whom? For instance, you mentioned that it could affect income inequality during your Women in Economics podcast.
Restrepo-Echavarria: Yeah, that’s why we care at the end, right? This matters a lot because, ideally, if we want to reduce income inequality, we would like to be able to have people of higher income marrying people of lower income, or higher education with lower education because that can incentivize someone to actually study more and increase their education. This matters a lot for the income composition of the population. So, this is why we care, actually, about this topic.
Engemann: In some countries or cultures, your family finds your spouse instead of you. Does your model have anything to say about that? Are these matches kind of similar to or are they different from the matches that you’re finding in your research?
Restrepo-Echavarria: In the model, it would depend on the amount of information that they have about you. If you can claim that your parents have perfect information about your preferences over types, then there would be no difference. But clearly, that’s probably not the case. These matches are going to be based on preferences on some dimensions, which are probably going to be different from the dimensions that you would focus on if you were to take the decision by yourself. But from the perspective of the model, if you assume that they have all the information about your preferences, then it’s equivalent.
Engemann: You sometimes hear people say that divorce rates in the U.S. are around 50 percent. Could your matching model reduce that sort of number?
Restrepo-Echavarria: Our model doesn’t have divorce, but the fact that it implies that you should search harder for a match, in principle, should result in a longer-term equilibrium because if you find a match that is better suited for you, then, in principle, the marriage should last longer. But our model doesn’t have anything to say about that because it’s a static, one-period model.
Engemann: And what about dating sites like Match.com? Are those examples of matching models in some way?
Restrepo-Echavarria: These are platforms that reduce search costs. When you think about what are the search costs for you, basically, you have this ideal image of the person that you would like to be with. But how do you know where to find them, right? We usually look for them in the same place that we’re studying, so in college, or at your workplace or something like that. But there are potentially many other better matches for you, but you don’t know where they are.
And what Match.com and these platforms do is they reduce the search costs because they basically give you a summary of everyone that is out there and allows you to contact them. This implies a huge reduction in search costs. And if there’s a reduction in search costs, then this means that you’re going to be able to search harder at a lower cost, so it’s going to be more efficient. So, on that dimension, in terms of the searching costs, they help.
Now, in terms whether they are a search model, it depends on how they’re constructed. Different sites are constructed with different purposes. For example, there’s a big difference between Match.com and Chemistry.com. They’re from the same owners, but the difference is that Match.com is much more informal than Chemistry.com. Chemistry.com is for those people who are actually looking to get married, while Match.com is for those people who want to go on a date.
And the main difference is, actually, that the algorithm to make the matches in Chemistry.com was done by Helen Fisher. She’s worked an endless amount of years on these things. She’s an anthropologist, and she has these very deep studies about types of people and how they interact with each other and how they can form longer-term relationships rather than short-term relationships. And she was the one who did the algorithm for Chemistry.com, and that one is based on much deeper questions than, for example, Match.com.
The outcome is going to depend on the website that you use, and they have different algorithms that are based on different types of questions, depending on how serious people are.
Engemann: You talked about buying perfume and other things like that. What other kinds of markets can you apply these models to, and have you done that sort of thing, and what kind of theoretical and empirical results have you found so far?
Restrepo-Echavarria: You can apply it to many, many markets. So, you can apply it to, for example, colleges and applications, so how students get matched with colleges. You can apply it to the organ market. So, how do you decide when to give an organ that is being donated to someone who needs it? You can apply it to goods and services, and you can apply it to the labor market.
And we’re currently working on an application for the labor market, which is fairly different from the marriage market because in the marriage market the theoretical model is a simultaneous model. So, both men and women make decisions at the same time. In the labor market, it’s different because it’s sequential. It’s not simultaneous. Then, you have, first, workers applying for jobs, and then firms take the applications that they receive, and they decide among those. So, we’re now working on an application for the labor market to specifically explain the wage dispersion that we see in the data.
Engemann: In the research that you’ve done, have you found anything that flies in the face of what people actually do? For instance, they do something different from what your model says they should do versus what they actually do?
Restrepo-Echavarria: Yeah, they don’t search enough. They’re doing a very poor job. We are all doing a very poor job searching for our matches.
Engemann: Anything else you want our audience to know?
Restrepo-Echavarria: I think that we care about these things from an economic point of view because it affects income. And a lot of people may think, “Why are these people studying marriage markets or who marries whom or how do you go on a date?” It matters for the economy, and it potentially has a lot of implications also for the government in terms of taxation. Because given that here in the U.S. you can file your taxes married filing jointly or filing separately, and all of these details, it matters a lot if there are big differences between the income of one spouse and the other one or the extreme case where one works and the other one doesn’t work at all.
So, we want to better understand how these matches are formed in order to have a better understanding of how can we improve taxation? How can we reduce income inequality? All of these things, which are what, at the end, matter to us as economists.
Engemann: All right. Well, that was great. Thanks, Paulina.
For more on Paulina’s research on this topic, go to her website. You can find that by starting out at stlouisfed.org, then clicking on “Research and Data,” and then on “Economists.” To listen to more of our podcasts, go to stlouisfed.org/timelytopics.