Friday, April 30, 2010

The Swedish labor market performed remarkably well during the crisis

In 2006 there was a change in government in Sweden. The new center-right government of Fredrik Reinfeldt carried out dramatic reforms of the labor market, including cuts in unemployment benefits, tightening the rules for sick-leave and cutting taxes for work.

Because Sweden is an export driven country, and because Sweden's industrial mix is historically sensitive to business cycles, Sweden was hit by the crisis much harder than most other countries. It is therefore interesting to compare how Swedish employment fared in the crisis. Did the reforms have any effect of dampening the crisis?

The result are a resounding yes.

My method is quite primitive, and not useful for long term analysis. I compare how much GDP changed compared to how much employment changed. Since my focus is Sweden, I initially look at the absolute employment change and change in real GDP between the 4th quarter of 2006 (when labor market reforms in Sweden started) and the 4th quarter of 2009 (the latest available year). All figures are from the OECD.

Since Sweden and most other advanced OECD countries have low rate population growth, that is not a major bias (although it means the true figures for the U.S are even worse than indicated). The big problem is that employment is a component of GDP growth. What is crucial for this analysis is however that this underestimates how well Sweden performed. We use GDP growth to get at the value of the "crisis shock". If one country had supply side labor market reforms during this period, and if this worked, this dampens the decline in growth. The true shock was likely even higher for Sweden than the decline in GDP suggests.

In terms of GDP decline, only 3 countries were hit more by the crisis. Yet, in employment terms, Sweden did better than 15 OECD countries. The decline in GDP in Sweden was twice the average of the Euro-countries. Despite this, employment increased 4% for Sweden during this period, with virtually zero growth of the EU-countries!

There is a strong link between the performance of GDP and employment. Different countries were hit by the crisis with different severity. If we compare how employment was to be predicted to perform based on GDP growth with how employment actually performed, we get an index of how well the labor market did in the crisis. This measures the distance between the countries and the regression line for all countries (look at the red line for Sweden and the U.S).

By this index, Sweden did second best out of 23 OECD countries, after Germany.

Considering how hard Sweden was hit by the crisis, if Sweden had the same employment performance of the average of 23 OECD countries, employment should have fallen by 2.0%. Instead, employment increased by 2.2%.

The decline in GDP for Sweden, Denmark and Finland (who did not undertake similar reforms) are very similar. Yet, while Denmark and Finland witnessed a decline of employment by 4.3% and 1.3% respectively, Sweden had a growth in employment of 2.2%.

Since the core period of the crisis was the second half of 2008 until now, I do the same analysis with starting period the second quarter of 2008 and the last quarter of 2009 (all my data are seasonally adjusted).

Again Sweden does far better than average, being the 4th best country in handling the crisis in employment terms. Since many of the labor market reforms precede the second half of 2008, this underestimates how well Sweden performed.

Note in both cases how poor the United States, the epicenter of the crisis, performed in labor market terms. One possibility is that the employment wedge for the U.S increased in this period, through for example higher minimum wage, expectations of higher future taxes and extending unemployment insurance.

Something that this method does not capture is the depths of the crisis. Notice that the U.S, Spain, Iceland and Ireland perform the worst between the crash and 4th quarter of 2009 (the latest available period). These countries were hit by a banking and housing market shock, whereas most of the effect on Sweden and other countries was decline in export.

Perhaps firms let workers go in countries where the effect was deep and expected to be prolonged, whereas they kept on to them in countries where the crisis was equally severe but expected to last shorter.

Wednesday, April 21, 2010

Does tax visibility matter?

Writing further on the design of the tax system, Casey Mulligan shows evidence indicating that the visibility of taxes does not matter.

Now, Professor Mulligan is the smartest guy I have ever met, so when he says something we should probably listen. However in this case I don't believe the matter is settled.

First, notice the way the nominal employee-employer share of payroll taxes are divided in a large percentage of countries (such as the U.S) is 50-50. You pay half, the employer pays half. Seems fair doesn't it? Economically, this division is silly, since the burden of the tax is completely unrelated to who pays it in. The only thing that matters is incidence.

Just the fact that about half of all advanced countries have chosen the 50-50 illusion tells us tax visibility matters, trying to trick workers.

More importantly, this odd convention restricts the variation in the explanatory variable, so we don't get to see what would happen if lots of countries had created the illusion that the employer "pays" everything.

Second, the argument about visibility is about the total tax burden, and this is looking at only one tax. It is quite possible that voters in countries with low visibility of the payroll tax feel they are taxed less, opening the way for other taxes.

As invisible taxes I have included all employer taxes, the VAT and sales taxes for 28 OECD countries. I compare this with the total tax burden as a share of GDP. Both variables are for 2007, the latest year with full data (and before the crisis).

The relationship is positive, with p value 9.3%.

Here is the same graph with only 23 advanced OECD countries. The p value is 4.5%.

The coefficient tells us that if tax invisibility increases by 10 percentage points, taxes as a share of GDP go up by 3.2 percentage points, which I consider economically significant.

Note that it is possible that the VAT and employer contributions rise with higher total taxes because they are more efficient, not because they are hidden. Still, I find the relationship suggestive.

Even trained economists do not take all taxes into account. It is not uncommon for studies of the elasticity of taxable income to ignore consumption taxes. In cases where income taxes have been replaced by a higher VAT this leads to a downward bias for the responsiveness to taxes (the changes in reported income we observe are due to a lower decrease in the tax wedge than the study assumes).

Based on years of talking taxes with Swedes, I have a very strong theoretical prior that tax visibility is important. A very common argument is "we pay 30% taxes, and get back so much, so why should we complain about taxes?".

Of course it is possible that we have reverse causality, with Swedes having high tax rates and low attention to taxes due to ideology. They like the welfare state and don't want to know how much they are taxed. Lastly even if the voters do not know the level of taxes it may be enough for them to know the proposed change, and vote accordingly, to get a efficient outcome.

Nevertheless, since there are no economic benefits to hidden taxes, and since hidden taxes MAY lead to systematic bias, the conclusion remains that we should make tax system as transparent as possible. There is no justification for having an "employer" part of the payroll tax.

Tuesday, April 20, 2010

Immigrant crime in Sweden

In many social outcomes, such as crime Sweden outperforms the US. An important question is whether this is due to Swedish welfare state policies, or due to demographic and cultural factors. Would the United States achieve the same outcomes with Swedish policies, or different outcomes due to American demographics and cultural factors?

In order to answer this immigrant outcomes is an useful indicator. Let us look at crime.

Sweden refuses to report the incarceration rate of immigrants. They do however report the incarceration rate of foreign citizens (people who don't have Swedish passports, not dual citizens). Immigrants who have gotten Swedish citizenship (56% of the foreign born 2008) will be counted as Swedes.

This measure will both underestimate and overestimate crime. It will overestimate crime if immigrants who have not yet become citizens are more criminal than other immigrants. It underestimates crime because the higher crime rate of immigrants that have become citizens are baked into the Swedish figures (also sometimes serious crime leads to deportation).

In the latest available year foreign nationals constituted 6% of the Swedish population and 29% of the prison population. Foreigners are thus 6.2 times as likely to be incarcerated than Swedes, or 520% higher incarceration rate.

The over-representation is even higher for middle eastern citizens, who are 6.6 as likely as Swedish citizens to be in prison.
Lastly African citizens are 10.9 times as likely as Swedish citizens to be in prison.

Swedish citizens have an incarceration rate of 45 per 100.000.

African citizens in Sweden in contrast have an incarceration rate of 490 per 100.000.

These people live under the Swedish welfare system. If the welfare system was the reason Swedes commit so little crime, it should have affected the immigrants similarly. Clearly it has not. My conclusion is that demography is as important or even more important than policy in explaining Swedish crime rates.

Saturday, April 10, 2010

The VAT and the growth of government

Currently the United States is the only country in the OECD that does not have a Value Added Tax. Some in the American left believe that the US should introduce a federal Value Added Tax. The pro-market side argues that a VAT leads to an expansion of government.

Much of the growth of the welfare state in Europe was financed with the VAT. Between 1970-2007 the VAT increased from 2.8% of GDP to 7.5% of GDP in western Europe (EU15). The VAT is currently 19% of tax revenue in the EU15.

Not surprisingly, there is a very strong correlation between the total tax burden and the size of the VAT.

Of course we cannot know what would have happened if Europe had not introduces a VAT, maybe they would have expanded government in a different way. Still there is a strong association between the growth in the VAT and the growth in the size of government. 41% of the increase in tax revenue for the EU15 between 1970-2007 was due to the increase in the VAT.

This graph plots the increase in VAT with the increase in total taxes, countries where the VAT revenue expanded the most tended to be those that had faster growth in the size of the state.

Finally here is the list of OECD countries by total tax revenue and VAT tax revenue.

From an revenue efficiency perspective, the VAT is considered an efficient tax, because it does not tax savings and because tax evasion is low.

However it is a misunderstanding that consumption taxes do not tax work. The reason people work is to earn income and consume. If the price of consumption goes up due to taxes but the price of leisure is constant, this is a tax on work and a distortion in the choice between work and leisure.

Still from a narrow distortion point of view it is as good or better than the income tax.

But efficiency should be interpreted more broadly than pure revenue effects. Democratic efficiency matters as well, and here the VAT does poorly.

The biggest problem with the VAT is that it is hidden taxation. The income tax is easy to understand (the government takes, say, 20% of your income). Consumption taxes in contrast are "baked in" the price of goods and easily forgotten or ignored.

Anyone who opposes further expansion of the U.S government should fight hard against the VAT.

Wednesday, April 7, 2010

The likelyhood of a Republican takeover of the House based on historic patterns

Historically republican voters have been more likely to turn out in midterm elections, so getting (say) 50-50% in polls of registered voters in polls means the Republicans win the election. Also, the out-party outperforms the generic polls, and this year they are the out party.

Here is the historic relationship between poll results and vote outcomes in congressional elections from famous Columbia political scientist Andrew Gelman. It can be used to predict the results in the 2010 election.

Excluding Rasmussen, the republican advantage on the generic ballot is 50.6% to 49.4% according to Real Clear Politics. This means "Generic vote Dems minus 50%" is -0.6%. Let us err on the side of caution and say it is exactly 50-50.

We are about 200 days from the election. The incumbent party is Democrat, the blue line applies.

I have plotted the current data on the relevant graph. If this election follows the average for 7 historic elections, Republicans will win the national vote share by a decent margin.

Is that enough to capture the house? Again, following historical patterns, it should be.

Yet the prediction markets only asses the probability of a Republican takeover at 41-44%.

Either the markets anticipate that (perhaps because the country has changed, or due to 2010-specific reasons) the Republicans under-perform relative to their poll figures and relative to historical patterns.

Or the markets are miss pricing the probability of a shift in power, in which case we should buy at the current prices.

Monday, April 5, 2010

The NYT misleads readers about Tea Party unemployment

In an article last week, the NYT tries to give the impression that a disproportional fraction of Tea Partiers are unemployed and living off the government (obviously in order to de-legitimize the movement and make them appear hypocritical). The article is called "With No Jobs, Plenty of Time for Tea Party".

What is noteworthy is that the article is entirely anecdotal. It starts with some guy who lost his job and contacted the government for help.

"When Tom Grimes lost his job as a financial consultant 15 months ago, he called his congressman, a Democrat, for help getting government health care."

Second, some scholar is cited to support the NYT story, although if you read it carefully he doesn't really give any new facts about the current Tea Party either.

"The Great Depression, too, mobilized many middle-class people who had fallen on hard times. Though, as Michael Kazin, the author of “The Populist Persuasion,” notes, they tended to push for more government involvement. The Tea Party vehemently wants less — though a number of its members acknowledge that they are relying on government programs for help"

There are no statistics cited by the NYT to support their claim that Tea Partiers disproportionately live of the government. This is the problem with articles that do not use statistics, anecdotes can be used to "prove" anything.

Today, Gallup released a poll about the demographics of Tea Party supporters.

It turns out the NYT is dead wrong. Tea Party supporters are substantially less likely to be unemployed than U.S adults in general. Only 6% of Tea Partiers are unemployed.

Now, this poll is about supporters, while the NYT claim was about activists. However this CNN poll indicates that Tea Party activists (7% of the population) are much better off financially than the population at large and better off than the total population of Tea Party supporters (28% of the population). 66% of Tea Party activist make more than 50 k, compared to 42% in the total sample of the adult population in the CNN poll.

From an scientific point of view the NYT article is nonsense, using anecdotes to give their readers the opposite impression of what statistics tell us. This is what the NYT specializes in: well written articles with an intellectual tone that are generally wrong in substance. The NYT is high quality junk.

Interestingly David Frum's homepage cited this NYT article, even though they must have known it contained no evidence of their smear of Tea Partiers. That tells us something about David Frum.

Friday, April 2, 2010

Median earnings higher in U.S than in Europe.

I have previously written about the fact that the U.S has higher average income than European countries. This is not only due to the rich earning more. The U.S also has higher median income than most European countries. Here is median income in PPP adjusted dollars in the "mid-2000s" by the OECD.

I should warn that estimates of median income are much less reliable than GDP per capita.

1 Luxembourg (35.200)
2 United States (27.768)

3 Switzerland (27.228)
4 Norway (27.098)
5 Netherlands (25.876)
6 Canada (25.507)
7 Austria (22.916)
8 Denmark (22.796)
9 UK (22.306)
10 Ireland (21.402)
11 Iceland (21.317)
12 Australia (21.068)
13 Finland (20.915)
14 Korea (20.892)
15 Japan (20.879)
16 Germany (20.586)
17 Belgium (20.388)
18 Sweden (19.895)
19 France (19.047)
20 New Zealand (16.798)
21 Spain (16.456)
22 Italy (16.140)
23 Greece (15.996)
24 Portugal (11.927)
25 Czech Rep. (10.760)
26 Hungary (8.714)
27 Slovak Rep. (7.838)
28 Poland (7.326)
29 Mexico (4.350)

France, which Social Democrat Bruce Bartlett thinks is a pretty positive role model for America, has a median income 31% lower than the U.S. Do you think the median American would voluntarily reduce his or her standard of living by 31% to get "better restaurants; and greater income equality"?

Not only does the educated class do much better in the U.S compared to Europe, the median earner are also better of here.

But why let objective facts stand in the way of leftist theory?
Google Analytics Alternative