“Is the first zone wide recession in the short history of the eurozone about to be registered?” asks Edward Hugh. I was curious to apply the algorithm for calculating my U.S. recession indicator index to a euro area GDP measure to get an answer.
We don’t have a long enough time series on the European Union to estimate the parameters characterizing expansion and contraction. But I wondered what would happen if you take the parameter values estimated on the basis of 60 years of U.S. real GDP growth rates and just apply them directly to the European data. That procedure is not immune from criticism. European growth rates are lower than the U.S., meaning that the algorithm may be a little too quick to call a recession with the European data, since it is expecting to see stronger growth rates in an expansion. On the other hand, we know the values of these parameters pretty well for U.S. data, and using prior information from another sample in this way can eliminate a lot of the statistical noise that would complicate an effort to start from scratch on the European data. So I was interested to take a look.
|
The top panel in the figure above plots the quarterly Euro-15 real GDP growth rate for each quarter at an annual rate, while the bottom panel reports the recession indicator index. It is interesting that the algorithm would have characterized Europe as being in a recession in 2002 following the U.S. downturn in 2001, if you adopt my rule of making a call when the index exceeds 67%.
I’m also following the procedure I use with U.S. data of waiting to see one extra quarter of data before trying to make a call. So for example, the most recent entry in the bottom panel is for 2008:Q1, when the recession indicator index for Europe is up to 32.8%. For comparison, the 2008:Q1 indicator for the U.S. is 38.4%
If this was a contest between the U.S. and Europe to see who’s going to get there first, it just reached the point where it’s about to get interesting.
Technorati Tags: GDP,
recession probability,
recession indicator index,
macroeconomics,
economics
This is a very interesting formula and while probably not 100% accurate just as you stated is a good model to forecast possibly when people should be paying extra close attention to the details.
Hello James,
Thanks for the hat-tip, and very interesting result.
I have my doubts on theoretical grounds about the direct transferability of material from the US elsewhere, but as a simple “rule of thumb guide” something seems to work.
My main issue, following Chad Jones, would be to ask why the US seems to have had something resembling steady state growth over such a long period of time. This question can be looked at from two directions, namely should we expect other economies to reveal patterns which resemble the US, or should we be asking ourselves just why the US is so different from everyone else.
I tend to go down the latter route, and tend to look to underlying demographic patterns to provide part of the answer. At least this approach seems to be giving Claus and me good results so far.
Which is why I like a lot of “hands on” data, and spend a lot of time peering into charts, trying hard to “read the runes”.
“If this was a contest between the U.S. and Europe to see who’s going to get there first, it just reached the point where it’s about to get interesting.”
And if you look at the latest batch of data in from Germany yesterday, even more so. I think, since Spain is going off a cliff, and Italy is hovering around zero, what happens in Germany really decides the issue.
Obviously Germany’s dependence on Russia and the CEE generally means that the anti has been driven upwards quite considerably in the last week or so.
Anyway, keep up the good work.
Best wishes,
Edward
I have done something similar in the past in order to estimate recession years in other countries besides the US. In order to do that I estimated the US model with standardized regressors (mean 0 and variance 1). In my case it seemed to work quite well for the countries I was looking at (Canada and the UK). Without controlling for the scale and the variance of the different variables the results were not very good. (the model I used was a simple probit model with unemployment variation and gdp growth and allowing for assymmetric effects in both variables).
Just for curiosity I went back to the recession dating model I had estimated previously for the US and used the most available data to forecast the current situation.
My results were that if 2008Q3 is similar to 2008Q2 (similar in terms of gdp growth and unemployment rise), then, the recession probability is below 75%. If unemployment rises at the same rate as in 2008Q2 and GDP is muted (FED expectations), then, the recession probability in 2008Q2 is below 90%. If in 2008Q3 GDP growth is 0 and unemployment rises more than in 2008Q2, then both 2008Q2 and 2008Q3 are recession periods already.
“My main issue, following Chad Jones, would be to ask why the US seems to have had something resembling steady state growth over such a long period of time. This question can be looked at from two directions, namely should we expect other economies to reveal patterns which resemble the US, or should we be asking ourselves just why the US is so different from everyone else.” — Edward
…..
If the US has had steadier and higher growth rates in GDP and productivity since 1981 than the EU or Japan — especially multi-factor productivity (proxied by technological innovation and diffusion — it seems due to a combination of factors:
1) The US alone, with two or three minor exceptions — mainly Sweden and Finland — has been the only advanced industrial country to reap the productivity benefits of the newest Schumpeterian long-wave of radical technologies: the computer chip and big information and communication innovations. As a recent study by Robert Gordon and a British economist showed, essentially only the US has benefitted from the Internet revolution.
(Productivity growth in the EU in the 1980s and 1990s was largely a matter of reducing excessive employment in basic industries and services and reallocating many of the jobless to long-term unemployment. For a few years now, there has been decent job growth in the EU (until the last few months), but — as with the US in the late 1970s until the mid-90’s — at the expense of growth rates in labor productivity. Oh, forgot to mention: much of the EU job-growth is in part-time employment, most probably involuntary.)
..
2) The faster growth of the richer Eurozone countries and Japan until the mid- or late-1980s reflected the advantages of “economic backwardness” — more specifically, the benefits of convergence catch-up growth. As Japan and Germany moved toward the technological frontier with the US in the 1980s, their growth rates slowed considerably. Since then, roughly 1991’s end, both have vied to see which can rack up the worst economic performance in the industrial world since the 1930s . . . though German export industries have impressive restructured and become very efficient the last five years or so.
Convergence catch-up growth is still going on in the EU, but mainly in Spain and Poland and two or three other new member-states, but the catch-up is with the richer EU countries. Essentially, per capita income in the US in PPP terms is about 25-30% higher than in France, Germany, or Britain.
….
3) US labor markets have been and remain far more flexible than in Europe or Japan, and so it’s been much easier to reallocate labor and capital away from standardized low-productivity sectors of the economy toward the new ICT sectors and those in retail and wholesale business that have quickly tapped ICT innovations.
Venture capital helped too in creating and diffusing ICT businesses.
…
4) Rapid globalization of the US economy, along with de-regulation since 1979, has further added to greater competitive pressures in the US for business firms to perform effectively. (Thank heavens for Japanese and German cars here! The Japanese are also to be thanked for their quality and price performance in consumer electronics).
….
5) Finally, don’t overlook the role of the Federal Reserve since the early 1980s in helping to stabilize the US economy’s performance, using different guides: the Taylor rule, the monetarist rules (it seemed initially), Greenspan’s optimistic-rule, and Bernanke’s own rescue-mission to salvage credit-institutions and markets . . . an outcome of his own intellectual immersion in the similar and catastrophic performance of those markets in the 1930s.
….
There are no doubt some other factors here, but these seem to be the major causal influences of convergence-reversal for the US. Essentially, since such convergence growth for the other advanced industrial countries ended after 3-4 decades of catch-up after 1945.
Will China continue to converge? Not likely unless huge changes in the rigid controls of the CP and elephantine state bureaucracies are undertaken . . . an unlikely development, with the existing elites involved in an orgy of self-enrichment.
….
Michael Gordon, Aka, the buggy professor
I wonder if we should modify business cycle timing to make variables in per capita terms rather than in absolute terms.
I per capita terms, US real GDP fell in Q4:2007 and Q1:2008; it may fall in Q3:2008 and very likely fall in Q4:2008.
We may have a per capita recession but not an absolute recession.
The usefulness of using per capita data is: (1) for an individual country, if population growth varies over time, we can make more valid intertemporal analysis; (2) comparing business cycles across countries that have different population growth rates, per capita variables are more relevant.
One interesting feature of your graph not mentioned is that perhaps the CEPR business cycle dating committee should have called a European recession in 2002.
The same question about a 2002 recession arose in the results from a Hamilton-style Markov switching model with country-level European data in
Dueker and Sola (2008).
An excellent post. I fear that from the importance of capital goods exports to its economy, Deutschland may be unter alles as the gobal economy turns down.
I wonder if we should modify business cycle timing to make variables in per capita terms rather than in absolute terms.
In general I think that’s useful, and you’re quite right to point out that GDP growth less than population growth isn’t really growth. Of course, the same applies to all those “household” figures that get thrown about; since there’s been a steady long-term trend towards smaller households, measurements based on household income tend to underestimate growth. (Now, to the extent that people have less children than they otherwise would if they were wealthier, looking at household income makes sense. But that’s not anything close to the entire explanation for decreasing household size.)
One problem with this kind of current analysis is that it ignores the fact that the most recent observations tend to be the most revised. As a result, models which fit well historically can be more erratic when used with preliminary estimates.
(For example, see the exchange on this blog on 23 Dec 2007.)
If we look at historical revisions of Eurozone GDP over the past several years, we find that these quarterly growth rates have a standard deviation of about 1.7%, but that their revisions have a standard deviation of about 0.5%.
What does that mean for a regime-classifying model? As a rule of thumb, it means that the recent data are less informative than we think since they come from a distribution with a higher (by about 0.5%) standard deviation than we’ve allowed for. If we think that those revisions have the same distribution in recessions and expansions, I think this will move our recession probability estimates towards their unconditional estimates (e.g. lower them at the moment.)
Agreed, Simon. Do you know where I can find real-time Eurozone data?
Funny you should ask. I’m teaching a course for Euro-system central bankers this week in Vienna. This morning’s lecture included a comparison of real-time data resources around the world.
I know of two real-time data sources for Euro-aggregates. The ECB compiles figures and makes them available to researchers affiliated with the Euro-Area Business Cycle Network. (Economics professors seem to be able to joint pretty freely by asking.) These figures are just taken from past issues of the ECB Monthly Bulletin and the first vintage is January 2001. You can find more information at http://www.eabcn.org/data/rtdb/euro_area_rtdb.htm
The OECD has eurozone aggregates in its realtime database at http://stats.oecd.org/mei/default.asp?rev=1 The construction is very similar, with data simply taken from past issues of the OECD Main Economic Indicators. The first vintage is late 1999.
Britain in recession 2008! Europe, part of the western hemisphere and part of SE Asia in recession 2009! Fully blown world in recession 2010! This seems a fairly confident and predictable outcome considering that our banks, who have caused the worst economic crisis in modern times, now have no real money of their own in reserve. The reason why they cannot lend anymore at the level they could and why the world is heading for the socio-economic turmoil that is clearly upon the horizon now. Many of our banks will undoubtedly go to the wall according to the edicts of capitalism and the market forces dictum. With share losses of western banks, insurance groups and investment institutions of US$2.7 TRILLION over just the last 18 months and bank losses in excess of well over US$1 TRILLION predicted when this is all over, the forecast becomes more-or-less a certainty. Indeed, who would have thought that Europes once largest bank UBS (and one of the largest in the world) would have been worth a mere 25% of what it was only a mere 15 months ago. But adding to this many are now worth less than 60% of what they were. Unfortunately the unseen suffering around the world that these great bank losses will cause to all people should be the main point of our anxiety. For in this respect millions will die through a global recession that had nothing at all to do with them. Let us hope that when we are all over this terrible malaise in a decade from now, that governments around the world will make perfectly sure that our banks can never crucify us ever again. Do we really learn is the question?
Dr David Hill
World Innovation Foundation Charity (WIFC)
Bern, Switzerland