Folie 1

1
The economic consequences of a Tobin Tax An
experimental analysis
Michael Hanke Jürgen Huber Michael
Kirchler Matthias Sutter, University of
Innsbruck University of Göteborg ESA World
Meeting Rome 2007
2
Outline

• Motivation
• Experimental design and procedure
• Effects of the introduction of a Tobin Tax on
• Trading volume
• Market shares the trivial issues
• Tax revenues
• Volatility
• Market efficiency the disputed issues
• Individual trading patterns
• Conclusion

3
Motivation

• James Tobin proposed a transaction tax on foreign
  exchange markets in the 1970ies.
• Alleged effects
• reduces short-term speculation
• reduces volatility (stabilizes markets)
• tax revenues (often downplayed as side-effects)
  
• Hard facts missing the introduction of a Tobin
  Tax is still discussed.
• Run an experiment to clarify the disputed
  issues.

4
The issues

• The trivial issues (Haq et al. 1996 Weaver et
  al. 2003)
• Market volume is expected to decrease since
  those who do not trade for bona fide commercial
  reasons might be driven from the market.
• The market shares of taxed markets in relation
  to tax havens should decrease due to tax
  avoidance.
• Tax revenues should accrue, but due to tax
  avoidance they should be smaller than naïve
  estimates would predict.
• The disputed issues
• Volatility Up or down (Aliber et al. 2003 or Hau
  2006 vs. Westerhoff 2003 or Ehrenstein et al.
  2005).
• Market efficiency Up or down (Ehrenstein 2002 or
  Westerhoff 2003 vs. Kupiec 1995 or Subrahmanyam
  1998).
• Individual trading patterns Which trader types
  are affected? (Bloomfield et al. 2006
  examination of tax havens missing)

5
Experimental design

• Two markets (LEFT and RIGHT).
• Two currencies (GULDEN and TALER).
• Both currencies are tradable on both markets
  (Gulden as home currency).
• Continuous double auction markets (100 seconds
  per period).
• Limit and market orders without restrictions.
• Short selling is prohibited.
• Switching between markets costless.
• 20 traders. Initial endowment of 8,000 GULDEN and
  200 TALER (worth 40 GULDEN each).
• 18 periods.
• No interest is earned on any currency
• Fundamental value of Taler revealed at the
  beginning of each period.
• Fundamental value follows random walk.

6
Designing the treatments

• Periods 1-6 no tax on either market.
• Periods 7-12a two-way transactions tax of 0.5
  is introduced
• on one market ? other market is tax haven
• on both markets ? encompassing Tobin Tax.
• Periods 13-18
• if previously introduced on both markets, the tax
  is removed in one market
• if previously introduced on one market, it is
• either removed there, or
• introduced in the other market as well.

7
Balanced treatments
Periods 1-6 Periods 1-6 Periods 7-12 Periods 7-12 Periods 13-18 Periods 13-18
Treatment LEFT RIGHT LEFT RIGHT LEFT RIGHT
0L0 - - Tax - - -
0R0 - - - Tax - -
0L2 - - Tax - Tax Tax
0R2 - - - Tax Tax Tax
02L - - Tax Tax Tax -
02R - - Tax Tax - Tax
8
Trading screen
9
Results

• Some intuitive illustrations
• Econometric estimations

10
Trading volume

• In general, very active trading overall (one
  transaction every two seconds).
• Unilateral introduction of Tobin tax on one
  market leads to a huge drop in trading volume.
  Most of this drop is due to a shift of trading to
  the untaxed market (the tax haven).
• An encompassing introduction of the Tobin Tax on
  both markets reduces trading volume by about 25.

11
Trading volume
12
Market share

• In the first 6 periods of each market, conditions
  on both markets are completely identical.
• However, market shares are strikingly different
• 68 of volume on LEFT,
• 32 of volume on RIGHT.
• This endogenously evolved pattern creates a
  perfect opportunity to study how a Tobin Tax
  interacts with the relative size of the taxed
  market.

13
Market shares in 010-marketsMarket share of
LEFT market
RIGHT taxed
Tax removed
LEFT taxed

• 0L0 Huge drop of LEFT after period 6. Loss of
  market share not fully regained after period 12.
• 0R0 Large drop of RIGHT first, but losses
  fully regained after period 12.

14
Market shares in 012-marketsMarket share of
LEFT market

• 0L2 Huge drop of LEFT after period 6. Loss of
  market share not fully regained after period 12.
• 0R2 Hardly any influence on RIGHT after period
  6.

15
Market shares in 021-marketsMarket share of
LEFT market

• 02L and 02R Taxed market after period 12 has
  less than 10 of market share.

16
Tax revenues
Trade volume before the introduction of the tax
is a very bad predictor for the tax base after
the introduction of the tax.
17
How we measurethe disputed issues

• Volatility Absolute returns in .
• Market efficiency Mean Absolute Error (MAE) of
  prices and fundamental values (provided to
  traders).
• Speculative trading Relative frequency of
  switching from buyer to seller position and vice
  versa within a given period (fundamental value
  does not change within period).

18
Econometric estimations
Tb dummy if both markets taxed Tt dummy if
market taxed, but other one untaxed Tu dummy if
market untaxed, but other one taxed
19
Conclusion

• If introduced unilaterally, the Tobin Tax causes
  a dramatic shift of trading volume to the untaxed
  market.
• The taxed market loses a large share in market
  volume, in particular if the large market is
  taxed.
• Tax revenues are smaller than naïve estimates
  would predict.
• Volatility is reduced if both markets are taxed
  or in the untaxed market, if the other is taxed.
• Market efficiency increases, expect in the taxed
  market when the other one is untaxed.
• Speculation is reduced in taxed markets, but it
  shows up again in untaxed markets.

20

• Thank you very much for your attention!

21

• Appendix (on request)

22
Main results on disputed issues

• Volatility
• If only one market taxed, volatility does not
  decrease in the taxed market, but rather in the
  untaxed one (due to a large shift of trading
  volume to the untaxed one).
• If both markets are taxed, volatility decreases
  (rather sharply).
• Market efficiency
• If only one market taxed, market efficiency
  increases in the untaxed market (due to
  speculators shifting to this market), but stays
  the same in the taxed market.
• If both markets are taxed, market efficiency
  increases.

23
Main results on disputed issues

• Speculative trading
• If only one market taxed, speculation is sharply
  reduced in the taxed market (due to massive
  trading shifts), but is increased in the untaxed
  market (speculators show up again).
• If both markets are taxed, speculation is
  significantly reduced, but not by very much.

24
Sample transaction plot of a 010 market where a
tax is introduced on the LEFT market in periods
7-12
25
Number of tradesLarge LEFT market taxed in
periods 7-12

• If taxed, the large LEFT market loses a lot of
  trades. The previous level is not regained if the
  tax is removed or the other market taxed as well.
  Same results apply to market shares.

26
Number of trades Small RIGHT market taxed in
periods 7-12

• If taxed, the small RIGHT market loses some
  trades. The previous level is fully regained if
  the tax is removed or the other market is taxed
  as well. Same results apply to market shares.

27
Number of trades Tax removed on one market in
periods 13-18

• Large shift of trading activity if an
  encompassing Tobin Tax is removed in one market
  only.