Thursday, June 30, 2011

Is Government the Answer, or the Question?

In honour of Canada's 144th birthday, which I'm sure you are celebrating tomorrow, today we'll look at some Canadian data. I'm not just drawing attention to Canadians to make them feel good. Indeed, some Canadians are so shy that they would rather be ignored. I'm thinking this exercise might actually shed some light on some recent issues. My title is inspired by a prolific Canadian who is endowed with a surplus of integrity, Neil Young, who once wrote (and sang): "love's the answer, love's the question." Sounds very sixties, and it seems somehow profound, but I can't tell you why.

In this previous post, I looked at some trends in US GDP, which basically replicated part of what Lucas was discussing in these slides. Basically, if you look at departures of US real GDP from a linear trend, the current observation is substantially below trend. Indeed, the data we have seen is consistent with either of two views. In one view there was a level adjustment in US real GDP that put us on a lower growth path. In the other view we will return to the trend growth path we were on previously.

Now, the standard Keynesian narrative we hear - and it's hard not to hear it as these people are blasting it at us every hour of the day - is that aggregate demand is too low, i.e. deficient, and the government is the answer. The US federal government can, and should, purchase more goods and services and/or cut taxes, and this will return us to trend. Lucas's slides suggest another explanation. He considers the possibility that there has been a level adjustment in the real GDP growth path, so that we are now on a trajectory that looks more like that of Western Europe than the US in days of yore. Why would this happen? Lucas argues that we may now have too much government rather than too little; various incursions or expected incursions of the federal government have essentially made us less productive, or potentially less productive in the future.

What do we make of this? Well, rather than comparing ourselves with Western Europe, let's go the back yard, as it were, and see what the Canadians (our socialists to the north) have been up to. The first chart shows what has happened to real GDP in Canada and the US since the beginning of the last recession. The recession in Canada was clearly shorter, and not quite as severe, and Canada is now doing somewhat better than the US, relative to where it was at the beginning of the recession. This is to be expected since, as everyone knows, Canada went through the financial crisis with no significant problems in the banking sector, there was no collapse in housing prices, and residential construction bounced back in the recovery. Still, the Canadian experience is not that different.

Further, the next chart shows a longer time series for the log of real GDP in Canada, with a linear time trend. Over this period, the average rate of growth is essentially identical to what it was in the US, roughly 3.1% per year. What is striking is the next chart, which plots the deviations of the log of real GDP from a linear trend for both the US and Canada. The US and Canada are currently in about the same place, i.e. about 10% below trend; indeed the US is doing a little better. This certainly is not consistent with what Lucas is suggesting. If more government intervention, or the threat of it, caused a negative level adjustment to the real GDP growth path, then there should be a negative level adjustment relative to the Canadians who, if anything, are moving in the direction of less intervention.

So is government the answer? We can certainly look at what governments have been doing recently in the US and Canada. The next chart shows the total government surpluses in the US and Canada as percentages of GDP. Now focus on the time series since 2000, when the US and Canada are in approximately the same fiscal position. The Bush tax cuts contribute to what looks like a drop of about 4% in the surplus in the US relative to GDP. Then, during the recent recession, there is a drop in peak-to-trough surpluses of about 7.3% of GDP in Canada, and about 8.9% of GDP in the US, presumably reflecting a bigger stimulus program in the US. So, whether you think that deficits work as Keynesian aggregate-demand-boosters or that tax cuts work to encourage labor supply, you have a lot of explaining to do. Why does 10% of GDP in stimulation in the US, and 4% in Canada, put both countries in the same place, at 10% below trend?

Now, finally, what is also very interesting is how the relative GDP paths in Canada and the US compare to relative labor market performance in the two countries. The next chart shows unemployment rates. Before the recent recession, the behavior of the unemployment rates in the two countries is typical, i.e. the Canadian unemployment rate is about 2 percentage points higher, due to more generous unemployment insurance and a somewhat different sectoral composition of employment. During the recession, that relationship flips. The US unemployment rate increases by about 5 percentage points during the recession, while in Canada the increase is 2.5 percentage points. Further, the unemployment rate has dropped faster in Canada during the recovery. Indeed, the Canadian unemployment rate is at its 2003 level, while the US unemployment rate is three points higher than in 2003.

Thus, in spite of similar GDP performance in Canada and the US, behavior in the labor market is remarkably different. You can't reconcile these differences with some story about "aggregate demand deficiencies," which at the best of times is not helpful anyway, being another way to say "output is below trend." I did not calculate productivity numbers for the US and Canada, but the flip side of the labor market and GDP observations has to be reflected in different average labor productivity growth too.

My conclusion is that, to understand why the US recovery is weak, and what this means for the future, we should not be thinking about the government. In my mind, the answers will come from understanding the barriers to mobility in labor across sectors, sectoral productivity, and links between the labor market and financial factors. Government may or may not be the answer, but you have to know what the question is first.


  1. "In my mind, the answers will come from understanding...links between the labor market and financial factors."

    there has already been some IMF research on that topic, and, i think also some fed research. links to follow.

    why do you say bigger government would lead to a LEVEL adjustment? surely it would lead to an ongoing lower rate of growth via lower productivity rather than what is effectively a 'jump' to a lower level of GDP.

  2. "why do you say bigger government would lead to a LEVEL adjustment?"

    I was just following what Lucas seemed to be suggesting, which was related to empirical observations regarding the US vs. Western Europe. Among the set of rich countries, the data make it look like there is convergence in growth rates across these countries, but there are level differences. Lucas seems to be saying that he thinks the level difference between the US and Western Europe is due to bigger government in Western Europe. However, as you say, you could imagine government affecting levels and growth rates. Further, the effects could go either way. I can imagine the government doing things that would increase the growth rate in GDP.

  3. Stephe!
    The U.S. unemployment benefit system seems to be lower in level, but strongly counter-cyclical, increasing benefit duration from 26 weeks to 99, clearly a governmental choice an order of magnitude larger than in previous recessions.
    I dont know what happened in Canada with the benefit system, but would you argue that these type of governmental policies could not potentially contribute to the slow recovery, possibly in line what Lucas might have had in mind?

  4. Average duration of unemployment has gone up from about 17 weeks before the recession to about 40 now. Certainly some of that could be due to extended UI benefits, but it could also be greater friction in the labor market. Even if it's UI benefits, this may not help you in explaining US vs. Canada. I think the Canadian "employment insurance" system (that's actually what it's called) provides for extended benefits in a region of the country where the unemployment rate is cyclically high.

  5. There are two views of the labor market. The conventional is that people really really care about being employed (or about the consumption that comes with this) and explaining cyclical fluctuations in employment is very hard (think Shimer Puzzle). The other is that people are kind of indifferent (think Hagedorn and Manovski's solution).

    I bet if you put large shocks to b in a Shimer type calibration you don't get so much movement in unemployment, but you would get a lot in a Hagedorn and Manovski calibration.

    Has someone done that?


  6. sure a shimer calibration with countercyclical benefits gives you the HM results conditional on reproducing
    the same wage elasticity say .7
    it is also (almost) equivalent to solutions of
    rigid wages Michaillard et al what matters is the
    induced surplus reaction to a productivity shock making
    wages mildly rigid

    too large to explains stephen pointed out increase from 17 to 40
    but it might well explain the difference to past shocks given
    that the additional incease would have to account "only" for say 10 weeks
    but how to identify to stephen matching friction story

  7. Yes, clearly the alternative calibrations are going to matter for the response to changes in UI compensation schedules. For the "matching frictions" story you have to load everything into the matching function in a conventional search model. That does not seem satisfactory. You need heterogeneity across workers and firms and some notion of matching between skills and occupations. In principle you can model it, but is the data rich enough?

  8. "I can imagine the government doing things that would increase the growth rate in GDP."

    No Xmas card for you, comrade Williamson.

    There has been empirical work done on the effect of expanded unemployment benefits. (No cite right offhand.) I think that the estimates from a couple of studies suggest that it currently raises the measured unemployment rate by about 1 percentage point. That is what I recall, anyway.

  9. Chris,

    e.g. You go from anarchy to a system with property rights, contractual law, and a police force. That would be good for growth, right?

  10. I was just joking. That is difficult to convey in text, I guess.

    I agree that property rights are good for growth, generally. Though I have some sympathy for Michele and David's arguments, to the limited extent that I understand them.

  11. I was just thinking about basic things, like enforcement of written contracts and laws against fraud, for example. Whether and how you protect intellectual property - the Boldrin-Levine issue - is another question.