Monday, December 16, 2013

A blow to the Prescott theory of labor



Ed Prescott - he of RBC model fame - is also known for advancing a theory of why Americans work more than Europeans. In this 2004 paper, Prescott attributed the difference to taxes - Europeans are taxed much more than Americans, in order to pay for their generous welfare states. Like many other economists (e.g. Greg Mankiw), Prescott thinks that high taxes make people work a lot less.

When I studied the Prescott paper in grad school, the instructor noted that Canada might be an exception to the high-tax, low-labor pattern. Now, David Andolfatto brings us even more dramatic news of Canadian exceptionalism. Andolfatto (a hard-working Canuck himself) breaks down the Canadian labor force participation rate by age and gender, and shows that young and prime-age Canadians work more than do their American counterparts, and that this has been true since around the year 2000:










The effect is more pronounced among the young, and in fact is reversed among the old; Canadians 55+ work less than do Americans of that age. For more data see Andolfatto's blog.

Overall, Canada's labor force participation rate is 67%, to our 64%, despite the fact that their median age is about 4 years higher than ours. Since Canada's unemployment rate is lower than ours, this definitely means that Canadians are working more than Americans. (American people, when employed, actually work slightly more hours than Canadians, but that difference is not enough to make up the difference in employment rates.)

Now let's think about this fact in terms of Prescott's theory. Canada is a somewhat low-tax country by OECD standards, but has a substantially higher tax burden than the U.S. (Note: As many have pointed out, the link I provided represents total tax burden, not the effective marginal tax rate. Prescott's 2004 paper lists America's tax rate as 40% and Canada's as 44%.) Canada uses these taxes to pay for a large welfare state, of which its nationalized health care is a part.

In addition, the crossover in labor force participation happened right around the time that America passed the enormous Bush tax cuts. Right when Prescott's theory predicts that American industriousness should have pulled away from Canada's, the exact opposite happened.

Also note that the labor gap is by far the widest among young people. Since Canada's tax code (including government benefits) is more progressive than America's, the gap in effective marginal tax rates for young people is probably even bigger than the 4% tax rate gap reported in Prescott's paper. This implies that, according to Prescott, Canadian teens should be working much less than American teens, and yet we see the opposite, by absolutely enormous margins.

(Notice that since America has a much bigger ratio of national debt to GDP than Canada, a very forward-looking model, like the kind Prescott likes to use, would predict that Americans would be working even more harder than Canadians, since as we all know, high national debt implies future tax increases!)

In the genteel terminology of academic econ, this Canadian data presents a "puzzle" for Prescott's theory. But I don't really like that terminology; it implies that someday we will inevitably find a reason why tax differences do, in fact, explain cross-country differences in labor. I think we should elevate the alternative hypothesis - that taxes matter only a little in determining how much a country's people work - to a plane of equal prior plausibility.

Thus, I think this Canadian data should make us signficantly more skeptical about the entire idea that taxes largely determine how much people work in different countries.


Update: Frances Woolley, another industrious Canadian (They're everywhere! Auuughhh!), responds. I think her response really validates my points, since it points out lots of other factors besides effective marginal tax rates that can affect labor supply. In other words, Prescott's analysis was just hopelessly simplistic.

48 comments:

  1. Anonymous6:36 PM

    Is there any easy way to get these participation-rate numbers broken down by rough individual incomes?

    ReplyDelete
  2. Anonymous7:09 PM

    I think one should be careful with the claim that taxes don't matter about work decisions. This may be true when it comes to the breadwinner of a family, but I think that taxes can have a big effect one seconed jobs or seconed earners. For example, in Germany, the tax system works in such a way that it is favourable for a family to have a big earnings differential between the man and the women(it doesn't matter which one of them earns more, but the gap is important). And I think this is one of the major reasons why most of the married women in Germany only work in small part-time jobs (called 450 € jobs) or not at all.

    ReplyDelete
    Replies
    1. Bill Ellis3:57 PM

      While the lower income partner gets penalized compared to a single earner the higher income partner gets special benefits compared to a single earner. So it evens out for the household.(or a bit better) It seems like this does penalize spouses who want to keep their income independent of each other, and don't want to correct for the government's tax policies.
      Germany is what we would call a community property nation. Legally everything is owned is 50/50 in a marriage.

      Delete
  3. Anonymous7:30 PM

    Well, I think one should be careful with the claim that taxes *do* matter about work decisions.

    ReplyDelete
  4. I think in general fluctuations in unemployment reflect changing preferences in peoples desire to express laziness. So I think Prescott's theory still holds.

    ReplyDelete
    Replies
    1. So you hold with the "great vacation" theory of the late economic unpleasantness? You have an actual theory to support that?

      Delete
  5. That's a bit of a thin piece if evidence to rest your hat on. The evidence the other way is much stronger. The unemployment rate over the cycle in e.u. For example is consistently higher then in the us or other English speaking (lighter taxed) countries. On top of that the transition in average hours worked in continental Europe in the 70's when the taxation burden exploded is highly suggestive of a significant impact of taxes. As is the slowdown and then reversal in catchup growth between continental Europe and the us over the last 30 years. Prescott's hypothesis looks pretty safe to me.

    ReplyDelete
  6. Anonymous10:28 PM

    Are you sure that the tax burden in Canada is higher than the U.S.? I'm guessing that a disproportionate share of U.S. GDP is pumped out on the coasts (e.g., California, New York, etc.) where the state income tax burden is high. You also have property taxes, sales taxes etc. in the U.S. Not saying you're wrong, but I'm a little skeptical. I would also think the key metric is the marginal tax rate, because that is what would incent a worker to work more or less at the margin. Again, I'm not sure of the specifics, but if Canada has some sort of VAT, which I don't think is a very progressive tax, then you might find the U.S. marginal taxes are as high, or higher than the taxes in Canada. Again, I have no specific data, so I could be wrong.

    ReplyDelete
    Replies
    1. Progressive income taxation does not result in higher marginal tax rates for most people; just for the top brackets.

      Delete
  7. Prescott's paper presented a very partial (biased?) analysis of European labor markets. The European countries he selected did have high taxes and low hours relative to the US. But there are plenty of European countries with higher taxes that have higher participation rates and working hours than the US (Northern Europe). So in a larger sample of European countries (+the US and Canada) you would not see that correlation. Also, if you look at the time series you also do not find much supporting evidence for his hypothesis.

    ReplyDelete
    Replies
    1. Northern Europe has longer working hours than the US? Try and find ONE:
      http://stats.oecd.org/Index.aspx?DataSetCode=ANHRS

      Even Canada has lower working hours.
      It's obvious that the regulation of the labor market plays a role, too since Scandinavian flexecurity helps with working hours compared to the continental EUrope, still all those countries have lower working hours. it's not all about taxes. Other things matter, too.

      Yet, Prescott is looking pretty good.

      Delete
    2. Norther Europe has higher participation rates than the US. What matters (to Prescott's argument) is hours to population not hours per week.

      And check the time series of participation rates and hours worked and see if there is any correlation with tax rates. None.

      Delete
    3. You wouldn't expect much correlation one way or the other since it's a small sample and it's a noisy measure.
      Working hours per week are a cleaner measure, since participation can be affected by even more factors than the labor supplied buy those already participating in the labor force. So, filtering of the signal is even harder.
      Either way, the fact that poorer countries (like northern Europe) than the b us have lower working hours than the us is quite peculiar. Even more damning it's the fact of the dramatic changes in working hours in the 70's. Just look at the time series from the source I gave you and tell me what happened then.

      Delete
    4. "Either way, the fact that poorer countries (like northern Europe) than the b us have lower working hours than the us is quite peculiar."

      Leave aside out the argument over whether it's hours per population or hours per week that matters. Isn't US the oddity here, not relatively prosperous N Europeans working less than, say, less prosperous Greeks?

      Delete
  8. Anonymous1:15 AM

    Rogerson (Prescott taught) has a paper that "explains" why Northern European countries have high tax high work hours. Shock horror it depends upon what you do with the tax revenues. Remember Prescott lump sums them back, rogerson impacts the effort margin. What happens to the revenue gets ignored when people cite Prescott.

    ReplyDelete
  9. Martin3:38 AM

    "The effect is more pronounced among the young, and in fact is reversed among the old; Canadians 55+ work less than do Americans of that age."

    So individuals at the peak of their earnings work less in a country with high marginal tax rates than individuals at the peak of their earnings in a country with low marginal tax rates?

    Is that not evidence for Prescott's theory rather than against it? One could read the data as: Canadians who are for some reason work harder than Americans do start to work less once they face higher marginal tax rates than Americans do.

    ReplyDelete
    Replies
    1. Anonymous9:01 AM

      Nope. Means the better social welfare structure allows for earlier retirement.

      Delete
  10. Couldn't we get a much better picture with panel data across countries and time? Why just do cross-sections when adding in time would allow us to see the impact of tax changes within countries? Cross country comparisons are tricky, and only having cross-sectional data makes the evidence inconclusive.

    Also, shouldn't it be after-tax income, and not tax rate, that determines hours worked? (Unless Prescott is making a fairness argument, which I doubt). The U.S. has higher after-tax incomes, of which lower taxes is only one part, with higher before-tax incomes also being relevant.

    - Nate K (not sure why the comment preview shows me as anonymous).

    ReplyDelete
  11. Christian9:09 AM

    Why would one focus on taxes anyway - shouldn't it be wages after taxes that determines my decision whether or not / how much to work, if we follow standard theory?

    And here we find that wages explain labour market participation even less. There are obviously huge differences in wages across countries & across time. If we assume that high taxes reduce the incentive to work by lowering the net wages - well why would people in the nineteenth century have bothered to work at all?

    ReplyDelete
    Replies
    1. Christian9:13 AM

      just noticed that faithandpublicpolicy made the same point. So anyone out there that could enlighten the two of us?

      Delete
    2. Why would one focus on taxes anyway - shouldn't it be wages after taxes that determines my decision whether or not / how much to work, if we follow standard theory?

      In reality? Yes. In Prescott's mind? No.

      Delete
    3. Is this a serious question? No, taxes are not the only thing that affect supply of labor, which doesn't affect the issue one bit. Richer countries work less in general (leisure vs. income and all that jazz), which makes it even more peculiar that poor western Europeans work less than us (richer Americans).

      Delete
    4. Christian11:44 AM

      yes, Krzys, serious question.

      "No, taxes are not the only thing that affect supply of labor" Yepp. where did I say otherwise? My point is that taxes are not even the only thing that affect wages, but any affect that taxes may have on the labour supply has to come from the fact that they reduce wages. So why not see first whether wages have any clear effect on the labour supply at all?

      Btw. when you notice that "Richer countries work less in general" you stipulate that labour supply depends negatively on wages. That goes in the complete opposite direction of Prescott, who argues that taxes (by lowering wages) lower labour supply - and increasing taxes would increase labour supply, thereby ending the current crises. Expansionary austerity at last!

      Delete
    5. Taxes are not the only thing that affects labor supplied and it's not obvious, which effect will be stronger, that's why we look at the evidence available. The evidence suggests that richer countries work less but higher taxation and regulatory burden lowers labor supplied even further.

      Delete
    6. Christian1:04 PM

      Sorry, once more, since my last sentence might have been a bit confusing.
      We both agree that labour supply depends on many things (social norms: e.g. is it 'normal' for women to stay at home and look after the kids).

      The question at hand is whether ceteris paribus labour supply depends positively or negatively on hourly wages after taxes. Your observation that countries with higher wages have a lower labour supply would point to a negative relationship. Prescott argues for a positive relationship: (lower taxes ->) higher wages -> higher labour supply.

      So if you believe both, that people in richer countries work less because of higher wages and that higher taxes reduce labour supply, that's self contradicting.

      To sum up: I think the whole concept of a wage dependend labour supply doesn't help us to make sense of the evidence available. I guess the world will just have accept my disbelief ;)

      Delete
    7. no, there is no contradiction. There are two separate issues. Wealthier/richer countries work less, but at the margin higher taxes (and regulatory burden) lower labor supplied further. It's also perfectly consistent with the evidence, especially the sudden transition of the 70's.

      Delete
  12. Wonks Anonymous10:42 AM

    A Canadian economist responds:
    http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/12/so-we-work-harder-in-canada-eh.html

    ReplyDelete
  13. I always presumed that more Americans worked more hours, more years because we got paid substantially less than Europeans -- because of the stark disparity in bargaining power in the respective labor markets; deunionization, etc.

    In any case -- just as there is no free lunch -- you must pay for lunch whether individually or through taxes. Ergo, Europeans and Canadians pay more for taxes but maybe less for medical care (a lot less per head) and education, etc. So, the higher taxes ultimately shouldn't amount to much of a disincentive.

    ReplyDelete
    Replies
    1. Except we get paid more. Whether measured by median or average household income (PPP).

      Delete
    2. My understanding is that we get paid more -- higher per capita GDP -- because more of us work more hours more years. Same technology, starkly weaker labor bargaining power -- how else could we possibly get paid more?

      American median wage unchanged pretty much since 1973 -- minimum wage now $3.50 an hour lower than 1968. Would have to triple our minimum wage to catch up with the doubling of productivity since 1968!

      Delete
    3. We are more productive per hour, too. Yes, we do get paid more no matter the reason.

      Labor bargaining laws and higher minimum wages do not make you richer, as the comparison to Western Europe shows.

      Delete
    4. Wonks Anonymous1:53 PM

      The minimum wage raises measured productivity because it prices the least productive workers out of the market. Productivity stats do not include the unemployed in their denominator.

      Delete
  14. Already the first paragraph shows a serious flaw with those studies. People in Europe spend less time at their job but it does not mean they work less. Many of the services done by outside "contractors" in the US (childcare, landscaping, eating out) are done by most Europeans themselves. A long time ago I read an article where they included any productive activity as "hours worked" and they came to the conclusion that Americans and Germans work the same number of hours per year. Germans do more of it at home s it is not captured in those statistics. Does not mean they just sit at home and enjoy their life.

    ReplyDelete
    Replies
    1. Anonymous10:18 PM

      Funny thing: this is EXACTLY what prescott's paper says.

      Delete
  15. Anonymous12:42 PM

    Inflation could be considered similar to a 'tax' and so look what has happened to the CAD since 2000...more than a 25% appreciation against the dollar! In other words, perhaps Canadians are working more as they are being compensated in a currency that has been increasing in value.

    ReplyDelete
  16. The American labor force participation rate is suppressed by its draconian justice system. People in prison can't work.

    In my subjective experience with self employed professionals an increase in taxes leads to increased individual work effort to maintain life style and a cut in taxes results in a reduction in work effort.

    ReplyDelete
  17. "(Notice that since America has a much bigger ratio of national debt to GDP than Canada, a very forward-looking model, like the kind Prescott likes to use, would predict that Americans would be working even more harder than Canadians, since as we all know, high national debt implies future tax increases!)" To which it may be added, Canadian wealth per capita now exceeds U.S. wealth per capita.

    ReplyDelete
  18. Noah, on the Canadian part of your argument at least, a Canadian blogger claims to have caught in the blogger sin of going a conclusion too far,
    http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/12/so-we-work-harder-in-canada-eh.html

    ReplyDelete
    Replies
    1. I looked very carefully at all the numbers she discusses.

      Delete
  19. Anonymous2:27 PM

    Prescott may be mixing correlation and causation here. You get high taxes in countries with one set of preferences, and low taxes in countries with other preferences. So the level of taxes and the balance between work and leisure may both correlate with the same traits in the population. For a Norwegian that sound more plausible than that I go skiing because of the marginal tax rate. Maybe there is a link between good skiing and hours worked?

    ReplyDelete
  20. Anonymous8:09 PM

    I love it. Frances Woolley points out that you have misinterpreted the data, and your response is that "her response really validates my points".
    Oh my sides.

    ReplyDelete
    Replies
    1. What??? How so?

      I didn't catch that.

      Delete
    2. Anonymous12:46 AM

      Paragraphs 2 and 3:

      "David - having learnt the hard way what happens to people who give their opinions too freely - is careful not to speculate on what might be behind these trends. Noah Smith has no such reservations. His interpretation of these graphs is that "young and prime-age Canadians work more than do their American counterparts."

      In fact, the graphs show no such thing. Let me just list all of the things wrong with seeing a difference in labour force participation rates and concluding that Canadians "work more." "

      Delete
  21. Anonymous8:18 PM

    Having declined the weasel word 'puzzle', you don't take the next obvious step. Theories are R*E*F*U*T*E*D by data contrary to those that they predict. Say it!

    ReplyDelete
    Replies
    1. Oh sure, it's refuted. But it didn't need this data to be refuted. Lots of other people have shown lots of other data that refutes it!

      Delete
  22. Anonymous12:27 PM

    That you don't understand that the Prescott paper is all that is wrong with econ..
    If you thought for a few seconds, one difference is in safety net; people worry a lot about paying for healthcare, retirement, education etc
    A paper that asks about why people in the US and France work differently, without putting this up front and center in the introduction,
    is an idiotic paper.
    and the substitution of math for intelligence..why bother to argue

    ReplyDelete
  23. Anonymous10:10 PM

    Prescott's paper is about the change in effective marginal tax rates on labor over time in the OECD. They went up in Europe, the stayed relatively constant in US and Canada. Hours worked went down in Europe, went slightly up in the US, stayed roughly where they were in canada. You seem to be saying that prescott's theory fails big time because canadians' participation rates didn't fell during the great recession, while in the states they did. Do you have any evidence to what happened to the effective marginal tax rates in Canada and the US recently? You seem to be implying that the effective marginal tax rate on labor went up in Canada and down in the US, but hours (prescott checks hours worked, you do participation rates, why?) did the opposite of what prescott's theory predicts. Did you calculate effective marginal tax rates for these 2 countries during the great recession? Also, you seem to be slashing on prescott theory using a very special time period (the greatest recession of the last 50 years), while this theory seems to be dealing with long term changes (from the 70s, when europeans were working more hours than americans). Do you think that this situation in participation rates is permanent? Why?
    Finally, Rogerson already re-did prescott's exercise including spending in public goods and the theory works pretty well. Usually people criticize this kind of papers for their assumption on labor elasticities, not because the theory doesn't work well. Indeed, if you look at the paper, the prediction of the model for Canada's hours worked is pretty close to the data...so I guess I don't get this post. What are you blowing?

    ReplyDelete
    Replies
    1. Anonymous1:14 AM

      As I said earlier:

      "Rogerson (Prescott taught) has a paper that "explains" why Northern European countries have high tax high work hours. Shock horror it depends upon what you do with the tax revenues. Remember Prescott lump sums them back, rogerson impacts the effort margin. What happens to the revenue gets ignored when people cite Prescott."

      To cite the Prescott's work to defend the "taxes kills effort" idea ignores the very very important assumption which is what do you do with the revenues. Also, remember Prescott sidesteps Japan by saying "they have different preferences".

      Finally if you look at the work that looks at WHO is working more it tends to be females (and those with kids) that are working more. Hence, household production, child-care availability and a whole host of things matter.

      To simply pin it on taxes is just plain silly.

      Delete