Tuesday, October 04, 2011

Americans Don't Really Hate Government That Much

The economic theory behind public good provision is largely unsatisfactory. Bergstrom, Blume and Varian (1986) remains the benchmark paper in the literature and under pretty general conditions it finds a unique Nash Equilibrium that taxes funding public goods will completely crowd out private provision.

An implication of this is that higher taxes should not, all else equal, be associated with greater public goods provision. This is not supported by the data. So what's happening?

In 2008 American government expenditure over GDP was 0.37, markedly below the EU average of 0.47. With the Tea Party being the talk of the town on this side of the Atlantic, I wondered how much of this was attributable to Americans simply trusting Congress less. Does the old cliche of Americans really distrusting government hold true? Not really.
Although the United States is below the (unweighted) EU average for trust/confidence in its national parliament, the members of Congress still retain more support than their counterparts in Belgium, Germany, France, the UK, Greece, Ireland and Portugal.

Could it be that a small fraction of Americans are particularly averse to government intervention? Possibly, but this tabulation below seems to suggest Americans are more centrist than they let on.
I'm not sure what's going on here, but I have a few ideas. It could be that a small fraction of government-averse American voters are pivotal in elections; or that differences in wording of the questions between US and EU surveys leads to data incomparability; American levels of mistrust would be much higher if G/GDP reached European levels; or that trust in government has more to do with the latest scandal than the efficacy of its expenditure.

At any rate, contrary to popular perception, it's interesting to see that support for government in Boston does not seem all that different to Berlin.


Liam Delaney said...

Good post Enda. Given that you are here, I will throw out our old favourite about whether Americans are interpreting the trust question differently. But that's just to be annoying!

The other thing I would say is that variables such as trust in institutions vary markedly in Europe. Trust in countries such as Greece and most Eastern European countries is really low. Ireland and UK tend to be above mid-table. The Chelseas/ManUs of this division are always the Scandinavian countries.

I think when making those types of comparisons, it is useful to think of the high-trust countries in Europe separately. I think the idea that there is some historical multiple equilibrium whereby high trust countries tend towards more comprehensive health and welfare systems which in turn reinforce the initial trust is a reasonable attempt at explanation.

Liam Delaney said...

The other thing about your table is where Ireland currently are. Hardly surprising given the times but this has potentially very far-reaching consequences if they can't restore trust.

Enda Hargaden said...

I accept that the minor tweaks in the question ("Confidence" vs "Trust") could be very important, and also that the time spans (US=2000-2008; EU=2008) could matter. If you have data suggestions I'd like to hear them.

I also agree that the data lend themselves to notions of multiple equilibria but, as you say yourself, Greece is a fine counterexample suggesting high G/GDP doesn't seem to be explained by a trust-dependent path alone. I have yet to come across a paper that models these cross-country differences in a multiple equilibria setting without resorting to increasing returns to scale over ranges in some countries but not others, or just assuming people like giving money away. These approaches are both fair enough but it's hardly surprising that if you assume some people like giving money away, you find that they give money away.

Liam Delaney said...

Some of the literature on trust and multiple equilibrium is reviewed here by Fehr


Greece is an interesting example alright. I think in that case, it relates to willingness to pay tax and is a similar story.

As reviewed by Fehr above, and as pointed out by you, estimating multiple equilbrium models like this using standard econometrics is very tricky. But not as tautological an excercise as your comment implies. Again, Fehr is a good case in point.

Regarding the anchoring and question-wording issue, Gary King has been working a lot on these types of issues but I haven't seen a paper specifically on trust


Liam Delaney said...

Another thing that might be interesting Enda is how trust in the national parliament and preferences for government size are related. I dont know this relationship at all. In my head, I can see Democrats trusting government less but wanting it to be bigger.

Liam Delaney said...

In terms of data Enda, the World Values survey is one place to look. The Gallup World Poll is another but is too expensive for the vast majority of places. If you could live with sampling differences pooling the European Social Survey with the General Social Survey in the US would be the best strategy I could think of.

Liam Delaney said...

Just saw from your table that pooling GSS and ESS was what you had done!! - there is a lot more in both files and also they are both done pretty frequently. If you were taking it on as a bigger project, a lot of data to work with and is mostly clean as a whistle.

Liam Delaney said...

sorry for bombarding with comment Enda but this is really interesting area. my first empirical papers were simple estimates of attitudes to public good provision. One issue that came up a lot is the median voter theorem. in those types of models, even if there is full crowding out, you can still get preferences for public spending to the extent that the median voter wants redistributive tax and spend policies.