Financial Transactions Taxes: Potential Revenue and Economic Implications

Abstract

Given Donald Trump’s extreme income tax–cut proposal, even useful tax hikes may not be likely. But down the road, there may be a call for higher taxes again to support social policies and manage, in the minds of Republican sponsors of tax cuts, a bulging deficit. This economist argues that a financial transactions tax can both rein in financial excesses and raise significant revenues. Computing the benefits is not easy, and we present his full analysis, which we think is a significant contribution.

Notes

Italy taxes over-the-counter stock transactions at a 0.2 percent rate.

More recently the Tax Policy Center did an analysis of the Sanders plan and explained most of its differences with the PHH analysis (Nunns 2016). This analysis put the revenue from a Sanders-type tax at $52 billion, slightly higher than its earlier calculation. The discussion in this study focuses on the fuller analysis in Burman et al. 2016, but little would be changed by addressing the Nunns analysis.

The tax applies to derivatives on stock, but only if they are exercised. This would exclude the overwhelming majority of derivate transactions.

This calculation uses data from Inland Revenue Service (2010).

The figure in the table for Japan for 1990 contradicts other evidence. The Japan Securities Research Institute (1992) put the revenue raised from the Japanese financial tax in that year (the peak of the stock bubble) at 1.1 percent of GDP.

Pollin et al. (2003) proposed a rate structure that was intended to double current transactions costs, so as not to bias the tax against specific instruments. This, of course, requires knowledge of what current transactions costs are.

The SIFMA data are available at http://www.sifma.org/uploadedFiles/Research/Statistics/StatisticsFiles/CM-US-Equity-SIFMA.xls?n=53461.

http://www.bis.org/statistics/extderiv.htm.

http://www.bis.org/statistics/derstats3y.htm.

The 2013 data are adjusted up by 10 percent for growth between 2013 and 2015, but then reduced by 10 percent under the assumption that U.S. trades account for 90 percent of North American trades, which is the category given by BIS. This figure is similar to the one calculated in Nunns (2016).

Nunns (2016) uses the same higher equity transactions volume as PHH.

This figure is obtained by calculating the response of revenue to changes in the size of the tax.

This figure would not include intermediate inputs, like the use of electricity and other utilities and supplies of paper, but these are almost certainly of little consequence in this sector. On the other side, not all the income in this sector stems from trading. The firms in this sector derive substantial revenue from underwriting new issues and also from management fees not directly associated with trading. The amount of income derived from trading would almost certainly be considerably less than the total income calculated for the industry.

In this respect, it is worth noting that Schulmeister (2008) reports trading costs for S&P 500 index contracts of just 0.0016 percent of the nominal value and for three-month interest contracts in eurodollars of 0.0006 percent (table A-4). This is consistent with the view that the most widely traded assets are generally traded at extremely low cost.

One of the reasons that the response of trading volume was so small is that these taxes have raised relatively little revenue to date. The tax in France only raised 700 million euros in 2015. Adjusting for GDP, this would be the equivalent of roughly $4 billion in the United States (Barbière 2015). It should not be surprising that a tax that has raised little revenue would have little effect on trading volume.

Since trading is an intermediate good (it does not directly provide a benefit to consumers), there is not an issue about its being a luxury good, the relative demand for which increases as income rises. (An exception would be if trading is actually a form of gambling that people do value as an end in itself. If this is the case, then an FTT would take the form of a sin tax, discouraging people from engaging in potentially harmful behavior.).

Burman et al. (2016) note the trend of declining transactions costs. Their calculations assume that transactions costs fall by 50 percent from 2010 to 2035. Matheson (2011) notes that bid-ask spreads on the NYSE have fallen more than 90 percent since the mid-1980s.

A fuller discussion of design issues can be found in Pollin et al. (2003).

Note that this is not a revenue-maximizing tax rate. The revenue-maximizing rate would be 0.8 percent on stocks, assuming an elasticity of −1.25. While it may prove desirable to move to a revenue-maximizing rate, it would probably not be advisable to introduce the tax at such a high rate. If the tax proved not to lead to major market disruptions, then it should be possible to raise the rate after it was implemented. Burman et al. (2016) lowered their revenue estimate by $29 billion to account for lower taxable income as a result of the tax. The extent to which the tax leads to a reduction in taxable income would depend on the extent to which a rise in transactions costs affect the economy, which is the topic of the next section. Most immediately, the tax does not reduce the income of investors, as the tax payments are more than fully offset by reductions in trading costs.

The size of the sector was calculated from BEA (2016) by taking the lines for compensation in the securities and commodities trading industry and also investment funds and trusts (table 6.2D, lines 59 and 61 for 2014, and table 6.2B, lines 55 and 59, for 1970). The calculation attributes income in the financial sector from proprietorships (table 6.12D, line 14 for 2014, and table 6.14D, line 14, for 1970) and corporate profits (table 6.16D, line 12, for 2014 and table 6.16B, line 12, for 1970) in proportion to the narrower trading sector’s share of employee compensation to employee compensation in the financial industry as a whole.

Employment in the securities and commodities trading sector was a bit less than 900,000 in 2014 (National Income and Product Accounts, table 6.4D, line 49).

There is a possibility that some people do view trading as a form of recreation like casino gambling. This discussion assumes that the bulk of trading is not done for recreational purposes.

Burman et al. (2016) include a useful table (table 3) showing the impact of tax rates on the rate of return for different holding periods.

Matheson (2011) has a good summary of the research on the relationship between trading costs and volatility.

These figures are obtained from BEA (2016) by dividing total employee compensation (table 6.2B, lines 55 and 59, and table 6.2D, lines 59 and 61) by the number of full-time equivalent workers in the sector (table 6.5B, line 55, and table 6.5D, line 59.).

Even conservative critics have agreed that this figure is a reasonable estimate of the cost of the proposal (e.g., Pethokoukis 2016).

Additional information

Notes on contributors

Dean Baker

Dean Baker is co-director at the Center for Economic and Policy Research, Washington, DC.