Saturday, August 11, 2012

Sweden cuts the consumption tax for restaurants


According to basic models of taxation all economic activity should be taxed at the same rate (ignoring externalities and fairness consideration). This is called tax-neutrality. 

If you are taxing the sale of Pepsi at 10% and the sale of Coke at 20% you create distortions by artificially making people drink more Pepsi. Moreover since the economic damage from tax rates is not linear but quadratic in the tax rate, each dollar collected from the tax on Coke at the margin does more damage as the Pepsi tax. There are therefore efficiency gains from harmonizing the two rate rates.  

This model is however an oversimplification. The neutrality principle implicitly assumes that all economic activities are equally tax-sensitive. Ramsey's principle of optimal taxation dictates that we should tax activities proportional to tax sensitiveness. Different economic activities differ in price sensitiveness. For instance taxing land has a very small effect of supply whereas taxing foreign capital can substantially reduce supply. 

For consumption taxes the overriding principle has been to equalize taxes across goods. This is sometimes claimed to be optimal taxation. But I disagree. In Gary Beckers model of household production the ability of households to create substitutes is different for different product categories. Households cannot themselves build high-tech goods such as electronics or vehicles regardless of how high the sales tax is, but can easily produce close substitutes for many services, such as transportation or eating out. Because of this taxes on services can reduce labor supply more than taxes on goods.

Another issue is how labor-intensive each good is. In a welfare state the social cost of unemployment for low-skill workers is higher than the social cost of unemployment for high-skill workers. Therefore it may makes sense (“second-best”) to tax goods and services produced by low-skill workers. 
 
Based on these considerations the Swedish center-right government decided to lower the value added tax on restaurant services from 25 percent to 12 percent. The decision was made in November 2011. 

Konjunkturinstitutet has evaluated the effect of the tax cut on prices. They compare restaurant prices to trend and to other Nordic countries and conclude that so far 30% of the tax cut has passed on to lower prices. The graph shows the price change in the quarter the tax cut was enacted.


I graph food prices and the prices for restaurant services from 2008. The price of restaurant services falls when the tax is cut and remains constant while inflation in food prices continues. Swedish restaurant prices appears to be 2-3 percent lower than they would have been without the tax cut. 

The critique from the left was that this tax cut would not affect sales and employment and would simply be a waste of money. The left-leaning Aftonbladet editorial page writes:

“The theory behind the tax cut is that when prices fall more people will go to restaurants with higher need for hiring. But reality is not so simple. There are more things than the price level which decide if people go to a restaurant, such as habits and the availability of time”

However sales did turn out to substantially increase. Statistics Sweden reports that restaurant sales were 4.9 percent higher in June 2012 than a year before. This should be compared to retail sales growth during the same period of only 0.9 percent. Despite the economic crisis the growth rate in restaurant sales this quarter was the second highest in a decade. Only 60 percent of the restaurant sector was affected by the lower sales tax, which makes these figures even more impressive. 

Though the reform has has some success, it’s interesting that prices have not fallen by the full 10 percent of prices the value added tax cut represents. It may be that some restaurants engaged in tax evasion, so that the taxes paid previously in practice were less than 25 percent to begin with. 

Moreover there is a tendency in policy evaluation to be impatient. Most reforms only take full effect with a long lag as firms and people adjust their behavior. For instance the channel through which tax increases effect hours worked may be slow, generational changes in norms regulating work. In that case you may not notice the full effect of a tax increase for years or even decades, even in cases where the long-run effect is strong. The tax literature focuses on individual-level effects in the short or medium run, ignoring societal effects taking place over the long run.

If the restaurant industry is characterized by something between oligopoly and monopolistic competition a tax cut like this may require new entry for full effect on prices. The part of the sales tax cut which has not passed through in terms of lower prices raises the profitability of the industry. Over time this should put downward pressure on prices. I expect the price of restaurant services to rise less than inflation for some more time.


11 comments:

  1. What models say all economic activity should be taxed at the same rate?

    Even ignoring externalities and fairness, the rates of substitution determine the dead weight loss of taxation (those Harberger's triangles). Optimal taxation will only require all goods to have identical rates if all goods have identical substitution rates.

    ReplyDelete
  2. @OneEyedMan:

    First, Diamond-Mirrlees (1971) showed that differential taxation on intermediate goods is bad.

    For final goods, if you accept some assumptions (among which, that utility is separable in leisure time), you get the Atkinson-Stiglitz (1976) theorem stating that an income tax is normally sufficient for the objectives of a welfarist goverment, because such income tax would be equivalent to a neutral consumption tax.

    I do not dare to say that these textbook prescriptions should be applied to concrete policy. But I am most fearful of letting politicians play with tax rates. I would rather stick with flat-rate taxation (even though I agree in principle with Tino, that distortions are caused in the real world, due to some goods being substitutes with different elasticities within household production functions), and only allow Pigovian taxes as a violation of such neutrality.

    A side note: in my country of origin (Italy), bad politicians use tax rates as election money. Rest assured, the outcome of such frequent changes in tax rates and rebates is a mess and brings additional administrative costs, and all possible uncertainty for the economic agents.
    Maybe the Swedish political system is mature enough to supporta Ramsey-like taxation with a lot of different tax rates. But I would not export this approach to other countries so light-heartedly.

    ReplyDelete
  3. The public choice point is true; if you open up differential taxation in a poorly functioning political system it may open the floodgates of hell with special interests ruining the tax code.

    However Mulligan and Becker have a paper arguing that this may lower taxes, if each interests group focuses on keeping its own rate low.

    http://ideas.repec.org/a/ucp/jlawec/y2003v46i2p293-340.html

    ReplyDelete
  4. How long until restaurants offer free cell service for frequent diners?

    ReplyDelete
  5. I dead consider joyous when I acquire articles pertinent to my process and my issue.
    instant cash out

    ReplyDelete
  6. This comment has been removed by the author.

    ReplyDelete
  7. What will happen to Sweden and other nations if America's financial situation fails? Many nations economic condition is dependent upon the dollar.

    Will we be forced to submit to World control?

    http://www.americasrolerevelation.com

    ReplyDelete
  8. "In a welfare state the social cost of unemployment for low-skill workers is higher than the social cost of unemployment for high-skill workers."

    You seem to be using "social cost" as a synonym for "cost to the (welfare) state", which is not how economists usually use it.

    ReplyDelete
  9. Part of the "loss" given from the fact that prices didn't fall as much as the tax cut could possibly be explained through the difference in tax on take-away food and eating at the restaurant. Take-away is at the same level as before, eating at the restaurant was taxed higher before the tax cut.

    The "loss" could therefore exist because people order take-away and/or because restaurants registered a bigger part of their orders as take-away before the tax cut.

    ReplyDelete
  10. I really appreciate this wonderful post that you have provided for us. I assure this would be beneficial for most of the people.

    Service Stars

    ReplyDelete
  11. Tino, excellent stuff, but are you sure this paragraph is correct?

    "In a welfare state the social cost of unemployment for low-skill workers is higher than the social cost of unemployment for high-skill workers. Therefore it may makes sense (“second-best”) to tax goods and services produced by low-skill workers. "

    ReplyDelete

Google Analytics Alternative