New NBER Paper Analyzes Sales Tax on Services: A Wonkish Review

June 13, 2014

A recent National Bureau of Economic Research working paper assessing optimal sales tax policies has found that exempting some services (like health and education) from a hypothetical national sales tax would amount to only a 3 percent revenue loss. They also find that charging different rates on different categories of goods rather than a flat sales tax rate can improve society-wide “utility” (how well off people feel) by 30 percent for any given level of revenues.

These findings are provocative. Sales tax base expansion to services is widely regarded as being a major revenue boost for states, and charging different rates on different goods is also widely regarded as ill-advised policy except in perhaps a few special circumstances.

However, a look at the paper itself reveals that its claims, while totally reasonable within the paper’s context, need some bracketing.

First, the paper assesses eight categories of consumption:

  1. Food
  2. Transportation
  3. Entertainment
  4. Owned Dwellings
  5. Rented Dwellings
  6. Alcohol and Tobacco
  7. Other (taxed)
  8. Other (not taxed)

It’s important to keep these categories in mind while reviewing the paper’s findings. How you break down the categories for analysis has a powerful impact on how to interpret the findings.

Would a shift to different rates on different tax bases be a good thing?

Probably not.

The authors identify a 30 percent loss in society-wide utility if policymakers adopt ideal differentiated tax rates on the eight categories of goods rather than a flat tax. However, while that number may sound large, the paper actually regards the 30 percent utility loss from differential tax rates as modest. That 30 percent utility loss assumes perfectly efficient policymakers who set tax policies perfectly equal to relative demand elasticities: in other words, policymakers who aren’t influenced by voter preferences. In reality, policymakers set taxes based on what people are willing to bear, which means they often tax highly elastic goods rather than inelastic goods.

Moreover, this paper lumps together almost every good currently covered by the sales tax in “Other (taxed).” This paper provides no support for taxing, for example, medicines at a different rate than furniture. Really what this paper assesses is excise taxes, sales taxes on food, and services taxes. And the paper implicitly argues that grocery taxes and taxes on services should be higher than other goods. Indeed, the authors argue that currently taxed goods should probably be taxed at a lower rate than almost anything else.

Finally, the paper notes that, if policymakers aren’t perfectly efficient (they obviously aren’t), utility losses from a flat sales tax are likely to be lower than 30 percent. The authors also suggest that administrative burdens and complexity due to differential tax rates could outweigh the utility gains from differential tax rates. This would be even more likely for state tax codes: if having eight different tax rates nationally could counterbalance a 30 percent utility loss, imagine what having 8 different rates in 50 different states could do.

Thus, the paper makes an interesting theoretical argument for differential tax rates, but practical concerns about how sales taxes are actually structured, political viability, and administrative costs mean that a flat sales tax is, as the authors themselves say, an entirely reasonable tax reform option.

Does exempting services from taxation only amount to a 3 percent revenue loss?

No. That is an egregious misreading of the paper’s finding.

The authors compare two systems: a flat sales tax on all goods, and a flat sales tax on all goods except “Other (untaxed)” which is mostly services like education, insurance, and healthcare. They find that, if policymakers have set a perfectly efficient flat tax rate in both cases, there’s a 3 percent revenue loss.

But that’s not a circumstance that faces legislators. First of all, the paper measures 3 percent from a baseline that includes all taxes on property transfers and mortgage payments, sales taxes on rent, sales taxes on all entertainment, all taxes on alcohol, tobacco, and transportation, and all general sales taxes. This isn’t 3 percent of sales tax revenues, it’s 3 percent of a theoretically perfect sales tax regime.

Furthermore, the authors erroneously synonymize “Other (untaxed)” with services. In reality, entertainment and rented dwellings are also large portions of what makes up “services.” Some services, like utilities, are also bundled into the other taxed category and the transportation category. Recall as well the authors earlier suggested that the optimal tax rate for services was higher than for other goods.

Finally, the authors suppose that, if policymakers exempt some services, they would raise the general rate slightly to compensate for revenues lost.

So what does this 3 percent revenue loss really mean? It means if policymakers massively broaden the tax base to include the majority of services, and if they raise the tax rate at the same time, the relative revenue losses from exempting some services may not “overly impair” budgets. More to the point, if policymakers expanded their sales tax bases to most services tomorrow, new revenues raised from the general sales tax would be orders of magnitude more vast than 3 percent.

This paper makes a very interesting, but extraordinarily nuanced claim, and is a valuable contribution to tax policy research. But it doesn’t supply numbers that accurately reflect current debates about state general sales taxes: because that isn’t what it was trying to do.

When we’re to the point in the sales tax base discussion where we have a perfectly flat, zero-exemptions tax for everything except just a few categories of goods, we can have the debate this paper outlines. But we’re not there yet.

Read more on sales taxation and services here.

Read more on general sales taxes here.

Follow Lyman on Twitter.



Related Articles