Monday, July 11, 2011

Economic Welfare

I saw an interesting talk yesterday by Pete Klenow (Stanford) on this paper by Klenow and Jones. The basic idea is to use standard economic theory to come up with measures of economic welfare by country. The paper is quite nice, as the measure allows us to account for how each of the four factors considered contribute to the welfare measure. In general, welfare is increasing in consumption, increasing in leisure, increasing in life expectancy, and decreasing in a measure of inequality. As Klenow showed, while the US and Australia currently have about the same level of per-capita real GDP, the Australians are about 16% (this is the number I remember) better off, in units of consumption. The Australians in the audience certainly seemed to like this result. Australians are doing better as they live longer, enjoy more leisure, and have less income inequality than do Americans.

Comparisons between Western Europe and the US often focus on per-capita GDP, and some researchers, including Ed Prescott, attribute much of the income gap to taxation. But of course the Western Europeans are buying something with their taxes. They are on average healthier than Americans, they live longer, they do not work as hard, and they are better-insured against low-income states. In the year 2000, while Western European per-capita real GDP was 71% of what it was in the US, the Jones/Klenow Western European welfare measure was 87% of the US measure.

Now, a question came up in Klenow's talk about unemployment. Someone was upset that the welfare measure essentially treated not-in-the-labor-force and unemployment as identical states, i.e. leisure. I think Klenow viewed this as the cleanest way to think about this, but do we think there is something to be gained from treating unemployment differently? Certainly many people want us to think about the currently high rate of unemployment in the US as a tragedy, but do they have in mind that we should somehow independently take account of unemployment in a measure of aggregate welfare? Maybe all these people are thinking is that there is a set of policies that could make unemployment lower, and that this would simultaneously make average consumption higher and reduce inequality, thus giving us a welfare improvement by the Jones/Klenow measure?

What does theory tell us about the cost of being unemployed? In some search models, it is costly to look for a job, in terms of effort and the opportunity cost of time. Unemployed people are willing to bear these search costs as they perceive that their potential welfare from being employed is sufficiently high relative to their welfare from being unemployed, given the probability of obtaining work. Thus, costs of search contribute negatively to economic welfare, independent of the factors that go into the Jones/Klenow welfare measure. Of course the measurement here is problematic - the unobservability of search effort is in fact the key problem in designing unemployment insurance systems.

Do we want to think of the unemployment state as somehow more painful than not-in-the-labor-force, i.e. does the average person obtain fewer utils from an hour spent unemployed as opposed to an hour spent not-in-the-labor-force, everything else held constant? An unemployed person may face higher uncertainty, which makes him/her worse off, but is a day spent playing golf less satisfying if I am unemployed than if I am not actively searching for work? But maybe the unemployed person can't afford to play golf, i.e. there are complements to leisure time that we need to take into account.

Sometimes excessive focus on the levels of employment and unemployment as welfare measures can get us in trouble. For example, some people looked at the report of the Council of Economic Advisors on the effects of the stimulus package, which told us that $666 billion was spent to obtain 2.4 million jobs, put the first number in the numerator and the second in the denominator, and determined that each job cost $280,000. Seems like a bad deal. Of course, we know that this is a poor way to evaluate the stimulus program, as we need to take account of the effects on incomes, both now and in the future, among other things. I would be more worried about that 2.4 million number and where it came from.

However, you can't fault the people who made that $280,000 calculation for making it, as people like Paul Krugman are encouraging them to do it. The labor market is important, but if you focus on it too much it can come back to bite you.


  1. My estimate of welfare differences for US and Western Europe and the Jones/Klenow estimates are the same. I only include market consumption and non market productive time (commonly referred to as leisure in aggregate economics) in my calculations.

  2. My understanding is that survey evidence suggests that the unemployed are much more unhappy than people on vacation. For example:

  3. I saw the Jones-Klenow paper a few months ago and I was impressed. It's the only paper I've seen that actually tries to tackle the issue of constructing a metric for economic welfare in a serious way. The clean simplicity of the framework also allows for more variables to be added on and considered with more data.

    On the issue of unemployment versus not in the labor force - I think there does need to be a distinction. I don't think you can just call this leisure. Aside from consumption uncertainty and search costs that you brought up, a long line of psychological and medical research suggests a lot of negative consequences of unemployment that would seemingly cause negative utility. For example, unemployment has been found to be associated with depression; suicide; worsening health; worsening well-being; worsening children's well-being (which matters to an altruistic parent); a decline in an individuals self-respect etc. These harms aren't associated with simple voluntary / out of the labor force. I guess one could also say why you are out of the labor force is important. If you are out of the labor force because you are suffering from a debilitating disease or handicap, I'm not sure how much leisure you are really experiencing. While I believe incorporating these issues of the causes and consequences of involuntary versus voluntary unemployment and out-of-the-labor force are important for a measure of welfare, I understand why Jones-Klenow might not tackle this. I don't think we simply understand the welfare consequences of this well enough, let alone how these interact with the varying degrees of unemployment benefits across nations.

    Really though, it's hard to ride Jones and Klenow too hard on this. Thinking about the issues you mentioned and those I mentioned above in a serious way is a difficult undertaking and arguing unreasonable for a line of literature that hasn't been developed much. I'm actually surprised how little literature there is on the design of welfare metrics! That said, it's hard to argue the Jones-Klenow's metric isn't the best out there. As a measure of welfare, its usefulness far exceeds GDP per capita figures or the Human Development Index.

  4. Ed,

    I assume, since you don't include dispersion and life expectancy like Jones/Klenow, that you are using a different functional form or different parameters.

    One-eyed man,

    This is a happiness study, where you construct a happiness measure based how people answer survey questions. It seems though, especially given Ted's comments, that what we are actually interested in here is mental and physical health, and we can get measures of that, for example from suicide rates and medical statistics. Of course you get into issues of what is reported and treated relative to the population.

  5. I agree that generally speaking survey questions about well being aren't useful unless they are in a panel. But this phenomenon of unemployment being a misery is quite robust to both the measures you specified:

    "Conclusions: Being unemployed was associated with a twofold to threefold increased relative risk of death by suicide, compared with being employed. About half of this association might be attributable to confounding by mental illness." Source:

    "Those who were unemployed but not ill at census time showed a 37% excess mortality over the following 10 years"

  6. Jones and Klenow lump together leisure and home production. I assume this is due to data limitations. Presumably welfare comparisons are quite sensitive to, for a given level of 'leisure + home production,' less time being spent in the leisure component and more time being engaged in home production, and vice versa. I wouldn't be surprised to learn, e.g., that the typical Western European vacation/holiday involved more home production relative to its US counterpart.

    Also, inclusion of inequality in the Jones and Klenow welfare metric seems rather ad-hoc. Wouldn't cross-country data on poverty would be far more helpful, given the goal of the paper?

  7. Yes, home production is important. There is a lot of work on this. Presumably it's hard to measure though. On the Jones/Klenow inequality measure, I think you could argue that this captures what we think of as poverty.