Turnover tax, transaction cost and stock trading volume revisited: investigation of the Japanese case

Abstract

Global financial turmoil in recent years has resulted in renewed interest in Stock Transaction Taxes (STT). Given that the existing literature on the quantitative impact of STT on turnover is limited and/or outdated, this article reinvestigates the issue, taking up the Japanese market reforms during the 1990s as an example. The analysis using an ordinary, fixed parameter model and Bayesian, variable parameter model finds that STT and, more generally, increased transaction cost significantly reduced trading volume. It is also found that the elasticity of turnover somewhat increased as the reforms were implemented. Finally, it is found that even in 2003 the elasticity was considerably smaller in the Japanese market than in European markets many years before.

Keywords:

• turnover tax
• taxation on financial transactions
• transaction cost
• stock trading volume
• Bayesian analysis
• Japanese market

JEL Classification:

• C11
• G12
• G18
• H24
• H25

Acknowledgements

We thank the Tokyo Stock Exchange and the Japan Securities Research Institute for providing essential data to conduct this study. We also thank an anonymous referee and the editor for useful comments on the earlier version of this article. All remaining errors are ours.

Notes

¹ They include Angela Merkel, Nicolas Sarközy and Gordon Brown.

² Global Financial Stability Report: Meeting New Challenges to Stability and Building a Safer System.

³ For instance, many economists from all over the United States cosigned ‘An Open Letter from Economists in Support of Financial Transaction Taxes’, in December 2009, which was made public through the Center for Economic Policy Research.

⁴ Song and Zhang (2005) bridge both views by a theoretical model that incorporates two types of offsetting risks that noise traders bring into the market.

⁵ This is the most recent among similar tax changes of major advanced economies; except that the UK reduced the tax burden in March 2008 only partially.

⁶ Exact institutions vary from country to country in regards to the coverage of taxable assets and the taxpayer. The Japanese STT was levied on both stocks and bonds and only the seller had to pay it.

⁷ Until January 2003, however, investors were given the choice of paying capital gains tax just like STT, besides paying it on the stock's increased value in a ‘normal’ way, in order to avoid cumbersome duties of reporting the purchase price. If this option was selected, the capital gains tax was calculated in the guise of a 1.05% sales tax. In this case, it should obviously be considered as an addition to the transaction cost. However, it is difficult to obtain numerical values on the degree to which this payment option of capital gains tax contributed to the transaction cost, because the investor could decide which method to choose on a sale-by-sale basis. Since there were no changes in this selection system during the period of our investigation, this is completely omitted from our analysis of transaction cost.

⁸ The first among those was the United States, which abolished the federal tax in 1965 and the last remaining New York state tax in 1981. In Europe, the UK cut its turnover tax rate to half in 1984 and again in 1986. In the 1990s, France and Denmark followed the same path as the UK. Germany and Sweden completely abolished turnover tax in 1991. See SIA (1994).

⁹ More detailed description of the historical changes of STT can be found in Shouken Dantai Kyougikai (1992), for instance.

¹⁰ See footnote 7.

¹¹ See, for instance, Barclay et al. (1998).

¹² Jackson and O’Donnell (1985) include the bid–ask spread in the transaction cost variable using a fairly rough estimate for their analysis of the UK market, but report that the results did not change much from the case without it.

¹³ We obtained this data from the TSE. Other fees such as account management fees and stock storage fees are omitted because they are quite small and considered negligible.

¹⁴ In the regressions conducted below, we allow only for a one-period lag, because the variable for longer lag would provide insignificant estimated coefficients.

¹⁵ See, for instance, Tauchen and Pitts (1983), Karpoff (1987), Lamoureux and Lastrapes (1994) and Andersen (1996).

¹⁶ Jackson and O’Donnell (1985) used the total inflow of money into insurance companies and pension funds for the proxy of the presence of institutional investors. However, such data is not available for Japan.

¹⁷ As TV and Cost are both found to be trend-stationary, we regressed the residuals on time trend, and obtained an insignificant coefficient.

¹⁸ Alternatively, one could interpret that, being used to a highly regulated market, Japanese investors are not as ‘mature’ in the sense that they react less sensitively to transaction costs in quest of higher returns.

¹⁹ Notable examples include China, Korea, Hong Kong and India.