Economists, journalists and European welfare states

by Henry on September 14, 2007

This “Dean Baker piece”:http://www.prospect.org/csnc/blogs/beat_the_press_archive?month=09&year=2007&base_name=nyt_libels_germany_on_unemploy from a few days ago on how the _New York Times_ misrepresents the German welfare state got some well deserved attention. While I wholeheartedly agree with Baker’s basic point, I think that he perhaps lets the economics profession off the hook a little too easily.

[The NYT]then tells readers that “Some economists worry that the relatively modest labor market changes have fallen far short of what the German economy needs to assure its long-term competitiveness, and that the government might be well advised to use this time of prosperity to tackle the tough issues. Instead, the long-awaited recovery has led to relief and perhaps even a little complacency.” While this no doubt true, it’s also true that many economists don’t see Germany as facing intractable economic problems. These economists point out that Germany is actually running a substantial current account surplus, which means that it is lending money to the rest of the world.

While Dean is right to say that some economists differ on these issues, I think it would be fair to say that the general wisdom among punditizing economists is that European welfare states are a Very Bad Thing. It would also be fair, however, to say that this general wisdom doesn’t have much basis in the available data. “Dani Rodrik”:http://rodrik.typepad.com/dani_rodriks_weblog/2007/09/are-labor-marke.html links to a “new paper”:http://www.newschool.edu/milano/docs/howell,%20Are%20Labor%20Markets.pdf (available via mildly annoying registration process available in an ungated version – thank you Steve Laniel) that studies the comparative economic literature and finds that the empirical evidence for a positive relationship between labour market rigidities _a la mode europeenne_ and high unemployment is very weak indeed. James Heckman, in his “response”:http://www.bepress.com/cas/vol2/iss1/art5/ to the paper, concedes the point, but argues that:

In the absence of better data, and better measurement frameworks, prior beliefs will continue to dominate how one interprets the evidence. This is not as much about dogmatism or conspiracy as it is about good science. In the absence of empirical evidence, logically consistent stories that accord with intuition have great appeal. At both an intuitive level and at the level of formal economic theory, incentives matter. If a person is paid not to work, the person will likely not work. If the costs of hiring a worker rise, fewer workers are likely to be hired.

As Rodrik notes, this isn’t a very satisfactory response. $4.50 and a logically consistent story that accords with intuition will get you a venti latte at Starbucks. More generally, plausible stories are cheap, plentiful, and often mutually contradictory in the social sciences – the proof of the pudding is in the empirical testing. That there isn’t forthright empirical evidence of a relationship between unemployment benefits and rates of unemployment suggests that the simple Economics 101 story about incentives doesn’t have the broad explanatory power that many economists suggest it has. I suspect that the _NYT_ journalists whom Baker criticizes are parrotting what they believe (with some justification) to be the common wisdom of economists on this topic; if I’m right, this suggests that the real problem lies more with the economists than with the journalists.

{ 1 trackback }

Welfare States and Bumblebees at Jacob Christensen
09.16.07 at 11:19 pm

{ 30 comments }

1

Barry 09.14.07 at 6:14 pm

Joke:

Q: Have you ever heard the one about the honest Chicago econ professor?

A: Neither has anybody else.

2

Kieran Healy 09.14.07 at 6:18 pm

That’s the true Chicago voice speaking there, by the way: my theory is the null hypothesis, there is no data that can decisively reject the null, therefore my theory wins on a priori grounds. In formal terms it bears more than a passing similarity to the way Marxists used to assess the evidence (or lack thereof) for their own prior beliefs.

3

Barry 09.14.07 at 6:30 pm

I’m actually surprised that Megan McArdle hasn’t shown up by now. My money is on her arrival in less than 10 more comments.

4

hidari 09.14.07 at 6:41 pm

‘In the absence of better data, and better measurement frameworks, prior beliefs will continue to dominate how one interprets the evidence. This is not as much about dogmatism or conspiracy as it is about good science. In the absence of empirical evidence, logically consistent stories that accord with intuition have great appeal.’

God it’s just as well that dopy fool Copernicus didn’t know about this. After all the logically consistent Ptolemaic system was born in the ‘absence of better data’ and the idea that the Sun orbits the Earth certainly accords with intuition.

Those last two sentences in the paragraph quoted above sum up everything that is wrong with neo-classical economics, and go far to confirming my impression that the whole field is simply ‘unfalsifiable’ in Popper’s sense and is, therefore, simply pseudo-science.

5

Stuart 09.14.07 at 6:49 pm

Surely it can seen (to support the empirical data) that if the way the market behaves is reasonably predictable, then markets adjust. So it doesn’t really matter much if unemployment benefits are relatively generous, because if enough people are unemployed then that will cause people to discount what they consider to be ‘enough’ wages to take the job. The parts of the economy that produce value from labour at close to the level of unemployment benefits are never really exploited by entrepreneurs anyway, as that level of profit isn’t good enough because there are plenty of other better risks to make with their capital and the labour employed.

Unemployment benefits should only ever cause employment problems at the point that everyone is getting nearly everything they could possibly want, and hence there will be businesses trying to chase down the very slimmest profit margins. I’m not sure such a situation is ever likely to exist, because people always want more than they have and there seems to be no real cap on the amount of goods people desire: look at things like cars, phones, TVs, computers, etc – early in their history it was assumed that there was a cap of 1 per household, now lots of people have more than 1 per person is becoming more common as productivity increases and they become relatively cheaper to buy.

6

SamChevre 09.14.07 at 7:01 pm

I think I’m with Heckman.

As I read it, here’s what’s happening:

If you use set A of control factors, unemployment is closely correlated with unemployment benefits.

If you instead use set B of control factors, unemployment is not very correlated with unemployment benefits.

There is a plausible explanation for why you’d use set A and a plausible explanation for why you’d use set B.

In that case, whether you think the set A numbers or the set B numbers is better depends on ideology; you aren’t in the range where empirical evidence is available.

7

Will 09.14.07 at 7:43 pm

“That there isn’t forthright empirical evidence of a relationship between unemployment benefits and rates of unemployment suggests that the simple Economics 101 story about incentives doesn’t have the broad explanatory power that many economists suggest it has.”

This isn’t true (if you mean people don’t respond to incentives). Instead we should read HBGS as a critique of macro empirical work, in general. The economy can be seen a complex network of interacting incentives and HBGS is just saying we need to be more careful about how we untangle that web.

Or as Heckman says… we need better data.

Also, the null is “bad” institutions don’t cause unemployment. HBGS suggest we haven’t rejected the null. Specifically, we haven’t shown that those “bad” institutions are in fact good.

That said, what *should* be the consensus view in economics on this issue?

8

notsneaky 09.14.07 at 8:02 pm

Well, unsurprisingly perhaps, I’m also with Heckman because “However, they do not offer a constructive empirical alternative to existing practices in the literature”. So they weakened the case for one particular explanation. But then the question is still there; why is European unemployment so high? Until I see a compelling alternative story I’ll stick with the fragile empirical studies and a logically consistent story. Though of course one shouldn’t push fragile explanations in op ed pieces or whatever.

Also Dani’s point that there is more than one plausible logically consistent story is made specifically in reference to the ease of firing workers, not things like unemployment benefits which as HBGS note:

“The most compelling finding of the cross-country regression literature is the generally significant and robust effect of the standard measure of unemployment benefit generosity,…”.

And just so I won’t get accused of mis quoting here’s the rest of the sentence:
“…but there are reasons to doubt both the economic importance of this relationship and the direction of causation.”

9

Henry 09.14.07 at 8:20 pm

will – “explanatory power” here refers, as is normal in the social science, to “ability to explain data.” That the simple story doesn’t explain the data suggests that this story isn’t as compelling or powerful as many economists say it is (and certainly not as compelling as OECD recommendations etc would suggest). I’m not sure where you are getting the idea that I am claiming ‘people don’t respond to incentives’ – the point that Rodrik makes is that there are many stories involving economic incentives that are plausible (see further on this broad issue Donald/Deirdre McCloskey’s _The Rhetoric of Economics_ ). It may well be that better data will show a stronger relationship – but at the moment we don’t have data that does this (and there is just a whiff of a Brechtian “then, we must elect a new data-set” about Heckman’s reply). In the absence of same, plausible stories about incentives are just that – plausible stories. They have more credibility than obviously implausible stories, but that’s it.

notsneaky – my particular ire here is aimed exactly at the op-eds/OECD reports etc which do seek to build grand claims on what seem to be very weak foundations. I can understand why people want to stick to theoretical foundations w/o a superior theoretical synthesis in sight – but I get highly annoyed when those foundations become crystallized as representing an empirically well-grounded professional consensus that should guide policy and public debate when they are nothing of the sort.

10

Soullite 09.14.07 at 8:38 pm

“In the absence of empirical evidence, logically consistent stories that accord with intuition have great appeal.” Sounds a whole lot like “Making shit up because it sounds good to me”. Is there an actual difference, or is this a case of attempting to baffle with bullshit?

11

notsneaky 09.14.07 at 9:35 pm

Henry – yeah I understand.

Also, my impression is that in the “standard theory” an increase in unemployment benefits would result primarily in a decrease in labor force participation, not an increase in unemployment (though this might go up due to longer job search – but it would be mitigated by general equilibrium effects) so it shouldn’t be that surprising that with better data the link between unemp benefits and unemp rate is weak or non existent. At least that’s how I understand the standard story but it’s not my area. I’m just wonderin’ what these crazy applied labor people are doing.

12

stostosto 09.14.07 at 10:28 pm

Okay, so it’s the economists’ fault and not the journalists’. As Clearly shown in this paper written by four.. erm… economists.

13

leederick 09.14.07 at 11:56 pm

I think the economists are okay. You can say that cross country differences in unemployment are caused by incentive structures. And you can dodge the empirical evidence bullet because of bad data and bad measurement frameworks.

The problem’s the journalists applying this to reality. You can’t do the above and write op-eds saying that’s why Germany is so crap. The fact is the moment you start saying the Econ 101 theory explains why *particular* countries are having problems – as opposed to just that it applies in general – you’ve just gone and committed yourself a position you can’t get out of by saying the data’s unreliable. If it is then what basis have you got making claims about particular countries?

There’s a real gulf between the theoretical economist’s position which says there is a relationship between labour market rigidities and high unemployment, but isn’t specific about how this maps to countries, and the journalist’s, which is specific and is much more vulnerable.

14

Ragout 09.15.07 at 4:53 am

While it’s true that many economists think that European welfare states lead to high unemployment, the NY Times is making a much dumber claim. The Times says that “Some economists worry that the relatively modest labor market changes have fallen far short of what the German economy needs to assure its long-term competitiveness.”

“Competitiveness” is a fuzzy concept used mainly by journalists, not economists. To the extent competitiveness means anything, it means productivity growth, and there’s no strong theoretical (or empirical) reason to think that labor-market protections affect productivity one way or the other.

15

gr 09.15.07 at 4:54 am

“But then the question is still there; why is European unemployment so high?” One of the points Baker makes in the post linked to above is that it isn’t, once you take into account the differences in the methods for counting unemployment.

16

F. Blair 09.15.07 at 5:54 am

Well, but it is still high, even accounting for the part-time issue. It’s about 50% higher in Germany than in the U.S. Baker suggests we should factor out East Germany, but if you factor out Mississippi and America’s inner cities, the U.S. unemployment would be much lower, as would income growth, inequality, etc. Does that mean these things aren’t problems in the U.S.?

17

notsneaky 09.15.07 at 7:25 am

“One of the points Baker makes in the post linked to above is that it isn’t, once you take into account the differences in the methods for counting unemployment.”

Yes that’s one possibility but 1) it just doesn’t jive with intuition and casual observation, 2) once you control for the difference in counting, the difference in rates could go other way, particularly since in US unemployment benefits are more closely tied to actively looking for a job (ie. being unemployed as opposed to not in labor force) and 3) the authors basically just speculate that this could be the case with little to back it up. At least as far as I’ve gotten into the paper (about a half way)

18

Tim Worstall 09.15.07 at 10:43 am

I thought we were now supposed to distinguish between the three/four different types of “European labour market rigidities” now? Nordic (high unemployment benefits, low job protections), Rhineland (high for both), Mediterranean (high job protections, low benefits) and the sorta fourth, the UK or Anglo Saxon model (not really “European” in this sense, low for both).
Given both the differences between them and the different results are we actually getting anywhere by referring to them all as “European” and treating them the same?

19

alexandre delaigue 09.15.07 at 10:54 am

“my particular ire here is aimed exactly at the op-eds/OECD reports etc which do seek to build grand claims on what seem to be very weak foundations.”

Grand claims like “welfare state is a very bad thing” is not what you get in oecd reports. The last oecd employment outlook :

http://www.oecd.org/document/12/0,3343,en_2649_37457_38792716_1_1_1_37457,00.html

explains that offshoring increases workers’vulnerability, that governments must “do more”, finance welfare with income or VA taxes rather than payroll taxes, and praises high-taxed denmark. OK, it is a bit Econ 101, you may agree with it or not, but it is mostly plausible policies based on plausible stories and facts, not ideologically-based grand claims about the necessity to dismantle the Very Bad European Welfare State.
From a european point of view, this kind of discussion is a bit strange. Unemployment is not a very hot subject in US academic economics, because it is not really a problem in the country; it has been replaced by income inequality and poverty (it seems that as soon as an economic problem is nt an american problem, it is not an economic problem any more). I think it has created a general ignorance (there are, of course, exceptions) about the subject, so that if you ask an american economist about unemployment in Europe, he will answer quickly “ho, probably the cost of the welfare state” an answer which means in fact “I have no idea”.
But if you go out of american economics and look at papers and people who really study the subject, you find a big amount of studies, showing that (surprise!) the subject is hard, but incentive-based economics works quite well.

20

Henry 09.15.07 at 2:12 pm

alexandre – as it happens (not that there’s any reason you should know this) I am a transplanted European, and European political economy is my academic specialization. Not unemployment and employment in particular, but I know enough about the debate to be able to say with confidence that you are quite wrong when you say that this discussion is a “bit strange” from an European point of view. And given that one of the ur-texts in the debate is the 1994 OECD jobs study, which was (correctly) perceived as a concerted attack on European labour market institutions, the role of the OECD is not only directly pertinent but overwhelmingly important. And the point of the Howell et al. paper is that the broad literature on this topic is based on crap econometrics (I’ll note for the record that I am not an econometrician myself, but also note that Heckman seems satisfied with this aspect of the paper, except insofar as he believes that it hasn’t gone far enough).

Tim – you are absolutely right on the different models etc – but my target here was less the European welfare state as it actually exists than as it is represented in US debate. It should be noted perhaps in passing that many US lefties who like the ‘European model’ aren’t much more sophisticated than right wing critics in making these important distinctions.

21

mq 09.15.07 at 3:58 pm

Following up on 18: I thought there was some pretty good evidence that high job protections for current employees increase the duration of unemployment (though not necessarily the level), probably by making employers reluctant to hire. I thought there was general consensus on how this contributed to an “insider/outsider” situation in the labor market.

This is different than the question of the relationship between unemployment benefits and level of unemployment.

Am I wrong about this? I haven’t kept up with the recent literature.

22

notsneaky 09.15.07 at 10:03 pm

If it increases the duration than it would probably also increase the level since at any point in time there would be more people in between jobs.

23

Tim Worstall 09.16.07 at 7:56 am

Henry, I end up having the same sorts of arguments with many US types from both sides of the ideological divide. Whether or not you like the Nordic models is more a matter of your a priori views on siuch things as social justice, the level of redistribution etc. The Med models, certainly in this unemployment issue, seem simply not to work, in the sense that they actively create unemployment when they’re supposed to do the opposite.
Unless these differences are teased out (and it’s not just in the US that these distinctions need to be made) then we’ll never get the debate to where it ought to be: which things shouldn’t we be doing because they are counter productive and which things should we indeed be arguing about from our different world views.

24

derrida derider 09.16.07 at 11:51 am

not so, notsneaky. High job protection certainly decreases flows from unemployment to employment, raising average duration of unemployment, but it also decreases flows from employment to unemployment – the theoretic effect on the level of unemployment is ambiguous.

I agree with Tim – to speak of “Europe” in this context is to conflate very different labour market institutions and outcomes. Some countries have lower unemployment than the US, some much higher. And the point of this paper is that those outcomes seem only weakly correlated, at best, with the institutions.

As an aside, when the 1994 OECD Jobs Study came out its studied ignoring of fairly obvious points such as high unemployment leading to high benefits rates (rather than the reverse), along with its adoption of arbitrary lags with no prior theoretical justification to try and make the econometrics fit, led me to dismiss it as an ideological hatchet job at the time.

25

rich 09.16.07 at 1:06 pm

Wait, are there still people who, knowing his recent history, still take Jim Heckman seriously?

26

Matt Weiner 09.16.07 at 9:33 pm

What are you talking about, Rich?

27

notsneaky 09.17.07 at 12:05 am

That’s why I said “probably” – the consensus view is that the second effect is dominated by the first. The pro-employment effect of protection is to keep workers who would otherwise be fired in their jobs. The anti-employment effect is that employers, not knowing in advance the productivity of individual workers, will limit the hiring – but that also means that the inflow of otherwise-would-be-fired workers into the employment pool is limited which would make that effect small.

28

notsneaky 09.17.07 at 1:10 am

…or to put it another way, if the length of job search for the *average* worker goes up (where in computing this average we include the otherwise-would-be-fired workers who’s job search length is zero) then overall unemployment level still has to go up.

29

derrida derider 09.18.07 at 7:03 am

No, the average length of job search for the average worker who is not working goes up. But there are now more otherwise-would-be-fired workers with a search length of zero, so the average search length overall may or may not go up. Do the math.

That said, employment protection shifts the burden of unemployment from the average worker onto those who are “bad risks” from an employer’s POV, in exactly the same way as making it hard for landlords to evict bad tenants ensures that homelessness is confined to the most marginalised (who are likely to be the worst risks from a landlord’s POV). Neither is a good thing if you’re worried about “social exclusion”.

30

derrida derider 09.18.07 at 7:06 am

PS – that shifting of the burden is the political economy reason why both employment protection and landlord-tenant laws are popular with most workers and renters respectively.

Comments on this entry are closed.