Network News

X My Profile
View More Activity

Research Desk responds: How do stimulus size and economic growth compare internationally?

By Dylan Matthews

Some guy named Ezra Klein asks:

See if you can pull together some data comparing the size of a country's stimulus with downturn (measured by gdp growth).

Brookings has tallied up the size of G-20 countries' stimulus packages, as a percentage of GDP. I compared this with the IMF's data on economic growth in these countries from 2009 to 2010, when the stimulus packages took effect; here's how the scatterplot ended up:


The overall trend is positive, but not strongly so. Twenty countries is a rather small sample size, and the countries themselves are different enough -- the U.S., China, Argentina and Saudi Arabia are all counted -- that different effects are to be expected. To see how the trend works out more specifically among developed countries, I took out the data for Australia, Canada, France, Germany, Italy, Japan, Spain, the U.K. and the U.S., and redid the graph:


Again, it is important not to read too much into this, especially with a sample size as small as nine. But with the exception of Spain, which lost economic ground, the correlation is solidly positive here. Here's the raw data, if you want to see for yourself:


By Ezra Klein  |  June 29, 2010; 1:42 PM ET
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: A progressive budget calculator
Next: A price on carbon, but not much of one?


I'm assuming that what they spent the stimulus on has some bearing. Infrastructure versus entitlement spending are generally very different things for the economy.

And does the stimulative effect continue after government stimulus has stopped?

Posted by: Kevin_Willis | June 29, 2010 1:53 PM | Report abuse

A la KW, there's been a lot of talk on this blog about the efficiency of stimulus. Is there any way to assign an efficiency factor to each country's stimulus? It would be interesting to see if it moved the correlation much.

Not easy though as you'd have to look at each country's stimulus, break it into components, assign each component an efficiency rating (from chart you did before), and then get the blended efficiency rating per country. Got a bunch of spare time?

Posted by: BHeffernan1 | June 29, 2010 2:23 PM | Report abuse

One of you needs to learn how to run a simple linear regression and report an r2 value.

Posted by: adamwri | June 29, 2010 2:41 PM | Report abuse

Two questions.
Is it reasonable to adjust the size of the federal stimulus, somehow, to take account of the decrease in state spending (negative stimulus)?
If so, how big was our "net" stimulus, and how does that effect the relationship between stimulus spending and GDP?

Posted by: bill_who | June 29, 2010 2:54 PM | Report abuse

Endogeneity, endogeneity, endogeneity. Countries with greater economic issues enact larger stimulus plans.

Posted by: jeffwacker | June 29, 2010 2:54 PM | Report abuse

There was a bunch of stuff in the stimulus that had a positive multiplier effect like food stamps and other programs that were mostly giveaways with little stimulus value like the housing giveaways. Unless you break the various national stimulus packages down this is a VERY coarse model. Shorter me: half the stimulus was BS put in there to stroke various members of Congress and not stimulate the economy.

Posted by: jamusco | June 29, 2010 2:57 PM | Report abuse

A few points.

I recreated the first chart, and got a linear equation of -0.0266x + 3.7459. In otherwords, the model predicts growth of 3.75% in 2009-2010 and -0.026% for each 1% of stimulus of GDP. Maybe I fat fingered part of the data, but at best this is a flat trend, not a positive one. The r squared was 0.0006. Take out China (which didn't come close to shrinking) and you've got a decent negative correlation, albeit weak with the R squared still at 0.04.

The chart showing the developed countries in Dylan's post does show a positive correlation, of 0.3108 + 1.2789, with an r squared of 0.3475. However, there are biases built into this data. For example, Italy had a small stimulus package and only grew 0.843% - but Italy was never going to be one of the fast growers anyway. The U.S. had a large stimulus package, but benefits from population growth of about 0.9% vs. perhaps 0.1%-0.2% of the other developed countries on average.

If you run a chart of stimulus vs. change in GDP growth for all countries (e.g. Australia would be 2.651%, from 0.85% in 2008-2009 to 3.501% 2009-2010), you get this equation: -0.4038 + 6.2272. In otherwords, the change in GDP growth was 6.23% without stimulus, and fell by 0.4% for every 1% of GDP of stimulus.

Doing this just for the developed countries, and you still see a positive correlation - the equation is 0.1991 + 4.7979, meaning that GDP growth would have improved by 4.8% without stimulus and an extra 0.2% with each 1% of GDP of stimulus. However, the r squared remains just 0.04, and half of the stimulus boost derives from the fact that Australia had a small stimulus but had a mild recession, and so they didn't have a lot of room to 'recover'. In fact, using change in growth you had six countries (US, UK, Canada, Japan, Italy, Germany) see growth improve by about 6%, regardless of whether the stimulus package was 0.3% of 5.9%.

Looking at just the anglo countries, and you see the UK with just a 1.5% stimulus saw a 6.3% improvement in GDP growth, Canada with 2.8% in stimulus saw a 5.8% improvement in GDP growth, and the U.S. had a 5.9% stimulus for a 5.5% improvement in GDP growth. The equation is -0.1477 + 6.36, and the r squared is 0.8399. The correlation is strong and negative, but there are two few data points to take this example seriously.

The primary take away if you look at a scatterplot which looks like a shotgun blast is that there is no obvious relationship.

Posted by: justin84 | June 29, 2010 3:11 PM | Report abuse

On endogeniety, if you run a scatterplot of the severity of the 2008-2009 downturn vs. size of stimulus package, there is no real correlation (actually, it suggests that stimulus packages were larger in countries with smaller 2008-2009 downturns, although the r squared here is 0.04). Countries like Saudi Arabia and China enacted large stimulus programs despite no severe recession, whereas places like Italy and Turkey didn't bother with them despite a severe recession (note that Turkey rocketed out of the recession and Italy has gone nowhere).

Posted by: justin84 | June 29, 2010 3:22 PM | Report abuse

The homebuilders would probably disagree with the opinion of the "little stimulus value like the housing giveaways." They booked their best quarters in years.

Whenever anyone mentions the effectiveness of debt spending on public infrastructure as a form of stimulus, take a look at Japan and their economic growth over the past 20 years. They've had their share of bridges to nowhere.

The infrastructure spend that the US could most benefit from in the present and in the future would be rebuilding a more robust and more accessible high-speed Internet infrastructure. For better or worse, the US has ceded this responsibility to private interests subject to short term quarterly outlooks.

Posted by: tuber | June 29, 2010 3:30 PM | Report abuse

your data is bogus...for China you take one year stimulus data and for US you are taking 2 years stimulus data...China's stimulus was close to 600 billion dollar whereas US stimulus 800 Billion China's economy is one fourth of the US so howz that percentage justified

Posted by: amritpalsidhu | June 29, 2010 3:52 PM | Report abuse

You get a better regression if you use GDP per capita. I took your GDP data and subtracted the 2008 growth rates given by Google. Adjusted R^2 popped up to ~53%, change in GDP(%) = 0.575 + 0.326 * stimulus as % of GDP

Posted by: mathewbinkley | June 29, 2010 4:02 PM | Report abuse

Just want to say what a great idea I think the "Research Desk" feature is and what great work the "Desk" is doing in responding to the inquiries.

And, Ezra, I read your renderings daily, as I used to read Dave Weigel's. I'm waiting for Weigel to emerge on some other platform

Thanks for the good work.

Posted by: ChrisBrown11 | June 29, 2010 4:20 PM | Report abuse

@ChrisBrown1: Dave Weigel now has a gig with MSNBC and will be making appearances on Countdown with Keith Olbermann.

Weigel also had a piece yesterday on Big Journalism:

So, there ya go.

Posted by: Kevin_Willis | June 29, 2010 4:39 PM | Report abuse

As long as we're playing with numbers, the Brookings chart shows which countries concentrated their stimulus in the form of tax cuts, as opposed to new spending. Russia, Brazil, and Indonesia were majority or all tax cuts, and they all got big positive GDP effects from a very small stimulus. Maybe that's the lesson to take away from this data.

Posted by: tomtildrum | June 29, 2010 5:23 PM | Report abuse

(*# you scott brown!!!

with an acid tipped $#@!!!

LOL. I guess you can't write that here.

Posted by: DROSE01 | June 29, 2010 5:44 PM | Report abuse

Tough comparison to make based solely on GDP growth and stimulus. Obviously there are other factors like currency value and trade balances which can also generate growth in the absence of purely domestic spending. Monetary policy plays into the balance as well.

Given the austerity measures at the state level -- tax increases and cut-backs; it's pretty clear that without the boost of federal spending, things would have been much much worse in 2009-2Q2010. It's too bad that the GOP seems more interested in sabatoging the recovery than supporting American workers at this stage. It might make for smart near-term politics, but the consequences for the economy could be pretty nasty.

Posted by: JPRS | June 29, 2010 6:17 PM | Report abuse

Request for the research desk:

A lot of claims get made about how high prices paid for medical care in the US are subsidizing medical R&D for the planet. We know that the per capita costs in the US are higher on average.

What I haven't seen is good statistics on what fraction of revenues and profits for pharmaceutical companies and medical supply/technology comes from the US versus abroad.

Can you check it out? It doesn't seem obvious to me than revenues and profits in R&D intensive industries necessarily follow from higher per-capita expenditures in all industries.

Posted by: zosima | June 29, 2010 10:19 PM | Report abuse

As Krugman puts it "Spend now, pay later." Can sound like blasphemy in the ears of majority of the people (w/ good reasons). But we simple have no choice.

Stimulus money in our city and State (Reno,NV) has helped tremendously, esp. in the public works sector (roads, etc) so even though we're still at 12% unemployment our local economy is doing much better (see
Than if we didn't have the stimulus. I don't agree with it completely but I've seen the positive effects of it.

Posted by: RenoRealEstate | June 29, 2010 11:07 PM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company