Wednesday, February 8, 2012

Marriage is an investment

And in an investment you compare the costs to the stream of future returns.

Tyler Cowen has been struggling with definitions lately. First we had "Keynesianism", and now he's calling David Frum and Paul Krugman "economic determinists" for suggesting that maybe the working conditions of the working class have some impact on their social lives. I have no idea why that fairly banal point is "determinism", but let's move on.

He cites Bryan Caplan who points out that marriage increases wages. Being single is expensive, so it doesn't make sense that hard economic times would cause a "decline in social values" in certain classes of workers.

Allow me to blow your minds folks: causality can work in two directions at the same time, and in complex systems like human society, it often does.

Back to the initial point - marriage is an investment. You are making a decision about a whole stream of future benefits - psychic, social, and also of course, monetary. Reduced job prospects reduce the attractiveness of that investment. Now - is Caplan right? Of course he's right. Yes, it's quite well established that marriage has beneficial impacts on income and that relation is causal. I don't see how this refutes Frum and Krugman.

Let's say the present discounted value stream of your earnings and your potential spouse's earnings are both $1,000. And let's say the marriage premium would increase both of those figures by $100. So now you're looking at a combined discounted income of $2,200. I'm not sure how family bargaining models usually work - let's say it's split 50/50 to make this simple. So yes, $1,000 < $1,100. Voila! Krugman and Frum disproved, right?


Marriage is an investment. It's hard. It's costly. I have a fantastic wife and we haven't had any major relationship crises yet, but it's still hard. Being single would be much easier. Plus, if you have a less steady relationship the cost of divorce is very high, and that's a real risk. Not to mention that point when the biological imperative finally lands me with a few kids, which I will have a much harder time ignoring if I'm married than if I were single and sleeping around. There are lots of costs associated with marriage! For me, the costs are well worth the benefits (including the kids... I'm just illustrating the point above - I am excited to have them eventually).

Bryan and Tyler are essentially saying "Duh Paul - $1,000 < $1,100, so what are you citing William Julius Wilson for?". The point is, if we monetize the costs of marriage as being $1,500 and if in an alternative macroeconomic history the working class could have earned $1,4000 on their own with the same $100 marriage premium, you would get more marriage.


  1. The effect of marriage on income is causal?

    1. I don't know the literature well, but people I trust seem to agree there's a genuine wage premium.

      If nothing else, we can just grant this for the sake of argument. I find it plausible (division of labor in the household had got to make people more productive). It still doesn't disprove the Krugman point.

  2. Or as we Austrians would say...a malinvestment!

    Ba dum dish!

  3. Nope, I disagree. I don't think your analysis of costs makes sense on several points:

    1) You mention the marriage effect on income only. Perhaps this is an oversight, but a far larger dollar amount is the savings on expenses from living together, which Bryan mentioned. $1,000 is an order of magnitude too small an estimation of the effect. (Certainly people can obtain this by cohabitating or having roommates, but the coordination problems are harder.)

    2) Kids. As demonstrated, the poor aren't having kids at a lower rate, they're just not getting married and there are a lot more single moms. So your assumption that kids are part of the cost of marriage seems odd. Yes, they are easier for the men to ignore, and unemployed and poor men have less to lose from child support, but shouldn't that have the opposite effect on poor women, making them more likely to insist on marriage? I don't see how you address this point without positing some kind of social or cultural shift, as Murray does.

    3) Monetary costs of marriage. The monetary cost of marriage is largely in divorce, from disrupting the current expense-saving arrangement and from losing wealth. I would think it obvious that the middle class and wealthy have far more to lose monetarily in a divorce than the poor. Anyone who brings more wealth to a marriage has more to lose monetarily. There's a reason why we think that prenuptial agreements are associated with the wealthy. Marriage reduces costs and raises income *right now*, with the risks of divorce being in the far future; that also should favor the poor being married.

    4) Psychological. You allude to psychological costs of marriage, and the stress of being in a relationship. But surely the monetary value of psychological well-being is higher for the middle class and rich. The wealthier people get, the more they're willing to spend on therapists. Psychological health is a luxury good. There are plenty of examples throughout history of people tolerating bad marriages for financial reasons. Therefore, if the poor are getting behind financially, one should expect them to want to suffer bad marriages more for the financial savings.

    As evidence, statistics indicate that the rate of divorce went *down* in the recent recession, as it generally does. People avoid divorce in recessions because divorce is expensive.

    1. Add in more savings from living together. I completely agree (although a lot of singles have roommates or co-habitors... I'm not sure how big this really is).

      It's not a counter-argument to Krugman.

      #2 doesn't disprove Krugman's point either. On this: "but shouldn't that have the opposite effect on poor women, making them more likely to insist on marriage?", it takes two to tango. You think they don't try? Plus, if poor men are a risky bet for poor women they might feel better off without a husband. I'm not sure how robust the marriage premium estimates are to different quintiles. re: "I don't see how you address this point without positing some kind of social or cultural shift, as Murray does." - I'm not entirely opposed to a social shift. I just don't know if (1.) it's a cause or effect, or (2.) why it changes Krugman's point.
      #3 - re: "I would think it obvious that the middle class and wealthy have far more to lose monetarily in a divorce than the poor." - in monetary terms, but there are other costs to divorce besides that. Wealthier couples have lower divorce rates, though (right?) so the risk is different.

      I just don't understand the thrust of your comment. You claim you disagree with me, and you provide lots more details that I generally think make sense. But you've nothing to challenge my point, which is that Krugman's argument make sense and Caplan's argument is in no way a refutation of Krugman's point. So aside from adding detail to the basic Caplan point that I already said I agree with, what exactly are you diagreeing with here????

  4. Throughout history people have entered into and stayed in bad marriages for financial reasons. I think that you'd need alternate explanations. Isn't it much more plausible to say that AFDC and other welfare support, combined with increasing female opportunities in general, made women much less dependent on marriage for financial support, and allowed them to avoid bad marriages? That isn't necessarily a bad thing, is it?

    1. No - social safety nets and female labor force participation are hugely important for this.

      Again - this doesn't really seem to challenge Krugman's point. It does provide more detail.

      You seem to be misunderstanding me - I'm not claiming Tyler/Bryan are making bad arguments. I'm just claiming that their good arguments don't constitute a refutation of Krugman's good argument.

      Don't read me as rejecting the wage premium point or any point related to that. I haven't rejected that, and a quick review of my post will prove I haven't.

    2. The Bryan/Tyler points are fine as far as they go.

      A big reason for me highlighting that they are not a refutation of Krugman's excellent points is the condescension of Tyler's post. As good as the points he adds are, it really didn't merit that kind of trashing of Krugman, Frum, etc. It really wasn't the smack down he was acting like it was - not by a long shot. Krugman's point still stands.

  5. I'm in a hurry, I'll expand on this later...

    * Before 1960 marriage rates were higher even though people were poorer. How come income is so important for marriage now when it wasn't in, say, the 19th century.

    * The same is true if you compare many poorer countries with the US.

    * The real income data is dubious, the "stagnating real income" from the 70s to today comes from making various assumptions to derive real income. This is now well known.

    I think L. Roger Devlin gave the most important reasons for declining marriage in his review of Murray in the Occidental Review.

  6. I think you are actually agreeing more with Cowen and less with Krugman. Cowen quotes Caplan, who looks at it from an economic perspective, but Cowen's real point is that it's not all just about economics -- marriage is a sociological topic and requires a broader, sociological explanation.

    Current, do you know of any good papers/books that describe how real income has been derived? And then, how these derivations might be wrong?

  7. Krugman's data on real wages is taken from the EPI, which is a left leaning think tank - see their webpage for example . Their blog is full of the fashionable moaning about "the 1%" and their briefing papers are full of opposition to right-to-work laws. Of course none of that makes them wrong.

    Now, I don't know exactly where the EPI got their data from. I expect it was from a modified version of the BLS data series. Krugman's graph doesn't say exactly what it's plotting, is it median wages? Is it wages of some band of people such as "the 99%" or bottom 25%? It looks like the BLS "non-supervisory workers" series.

    The main problem with the BLS data is that it doesn't cover non-pecuniary income. Or, at least, it covers it in a strange and inconsistent way. It doesn't cover employer provided healthcare or all employer pension contributions. Data from the bureau of economic analysis, on the other hand, does include these things but it doesn't include paid vacations.

    A few years ago the St.Louis fed added back in non-pecuniary income to derive the proportion of GDP paid as wages. They found that once this was done that proportion hadn't changed:

    The CBO did something similar, they tried to add back in non-pecuniary income. They found that inequality has increased greatly, but real incomes have increased too. Specifically:
    "CBO finds that, between 1979 and 2007, income grew by:
    275 percent for the top 1 percent of households,
    65 percent for the next 19 percent,
    Just under 40 percent for the next 60 percent, and
    18 percent for the bottom 20 percent."
    So, even the US bottom 20% are richer than they were thirty years ago. This CBO data compares households though, not individual incomes. Now, Daniel is quite right that falls in individual incomes could cause people to avoid forming households. However, the starting point for investigating that should be individual real income. If that has fallen then it's sensible to see it driving a fall in household formation leading to more and smaller households and consequently changing household data.

    The Minneapolis fed looked at the problem of non-pecuniary benefits and changes in household size in two reports in 2007 and 2008. Terry Fitzgerald showed the median incomes have risen substantially if the problems I mention above are removed. See:
    (I think Fitzgerald's second article is the best of the ones I mention here, if you read only one, read that one).

    I find it odd that Keynesians like the BLS data and similar data. If you look at it carefully it implies that real GDP data must be wrong. All incomes become household incomes at some point. So, if real GDP growth was 108% over 1970-2008 and median household income growth was 16% then income growth for the rich must have been enormous. Most research of this kind shows that it's very big, but not enough to make total income growth equal to real GDP growth. Tino Sanandaji looked at Pickety & Saez's numbers. They find that real taxable income of all households rose by only 29% from 1970-2008, in that case how did GDP double?

    There is also the thorny problem of the accuracy of CPI figures. As several people have pointed out quality improvements are not taken into account well. Mass manufacturing and distributions improvements naturally favour cheaper products, this also isn't taken into account in CPI. These issues affect the entire history of CPI measurements though going back well before 1970 . The Minneapolis fed data also shows that picking different CPI series can have a huge impact on the final answer.


All anonymous comments will be deleted. Consistent pseudonyms are fine.