Opponents of the value-added tax complain that it is not sufficiently visible to voters, and is thereby an obstacle to responsible public spending. But in fact government spending is no lower in countries with more visible taxes.
Last week I explained how reforming the income tax system — perhaps by replacing it with a more efficient value-added tax (VAT) — might fuel further growth in government spending because less efficient taxes create more political pressure against big government.
Some economists agree with my conclusion, but for a different reason. Because the VAT is collected at each stage of production and, unlike our state and local sales taxes, is not shown to consumers separately from purchase prices, it has been said that VATs are not visible enough to voters to create much electoral resistance to raising their rates.
John Kass of The Chicago Tribune even suggested perhaps the ultimate in tax visibility — that income taxes be collected only once per year: the day before election day.
The pattern of social security taxes across countries provides a test of these theories, because countries differ in how they collect public pension payroll taxes. Some countries, like the Netherlands, take most of the tax out of employee paychecks, while others, like France, levy most of the payroll tax on employers. The United States and several other countries take an equal amount from employee paychecks and from employers.
Payroll taxes are no less efficient if they are taken from employers rather than employees, but the taxes on employees are clearly more visible to employees. Indeed, economists themselves sometimes forget that their employer is liable for payroll taxes on their behalf!
Interestingly, democracies collect the payroll tax more visibly (that is, a greater share from employee paychecks) than nondemocracies do. But countries that put more of the tax on employees do not manage to spend less on public pensions (Social Security, as we call it in the United States).
The scatter diagram below shows public pension spending as a percentage of gross domestic product, versus the fraction of payroll taxes levied on employees rather than employers. Countries such as the United States with equal payments by employers and employees are shown up the middle of the chart: they all have an employee share of 0.5 but each has its own different propensity to spend on public pensions.
If more visible payroll taxes helped restrain payroll taxation, then we should see countries that take relatively more of the payroll tax from employees rather than employers — those in the right part of the chart — spending less of their G.D.P. on the public pensions for which these taxes are earmarked.
Instead, the correlation between the two variables is essentially zero (neither economically nor statistically significant). This result holds up when adjusting for a host of other variables that determine public pension spending, and excluding the Netherlands (an apparent outlier in its payroll tax collection).
I am not especially surprised that tax visibility is empirically unrelated to the amount of taxation and government spending, because the impressions of voters who see the more visible taxes are by no means the only determinant of government spending. Special interests matter too.
And even governments that are not held accountable by elections manage to restrain their taxation as much (or as little) as democracies do.
In the case of the payroll tax, one of the important interest groups would be the employers themselves, who are of course quite aware of payroll taxes levied upon them. Employers may even resist such taxes more if they thought those taxes were invisible to, and thereby unappreciated by, their employees.
It is wise to consider how transforming our tax system might affect the propensity of government to spend. But making taxes more visible to voters would be all show, and deliver no results.