VoxEU: A rationale for the Tobin tax

14 February 2017

This column argues that a rationale for a Tobin tax exists even in competitive and informationally efficient markets when traders have private information and they condition on prices. In this situation traders overreact to private information, and a transactions tax may offset this externality.

In favour of a Tobin tax

In a new paper, I argue that a rationale for a Tobin tax exists, even if markets are competitive and informationally efficient, and traders do not suffer from behavioural biases and extract information from prices in the rational expectations tradition (Vives 2016). The key conditions for this result are that traders have private information, and conditioning on prices – two standard features of financial markets.

Imagine a competitive financial market where there are only two frictions: a transaction cost or trader risk aversion, and private information. Each trader, perhaps out of own research or personal reading of public information, receives a private signal about the fundamental value of an asset. Investors condition on market prices when formulating their trade (more precisely, they submit demand schedules to the market mechanism) and markets clear.

In this context, we may suspect that traders will rely too much on public information. The reason is that traders do not consider that their reaction to private information affects how informative public statistics (prices) and general welfare would be. In other words, traders do not internalise an information externality. This type of externality will make agents insufficiently responsive to their private information (Vives 1997, Amador and Weill 2012) and, in the limit, to disregard it (Banerjee 1992, Bikhchandani et al. 1992). As an example, Morris and Shin (2005) point to the paradox that, by publishing aggregate statistics, a central bank makes them less reliable because it induces agents in the economy to rely less on their private signals.

The previous reasoning, however, disregards one important fact. When traders condition on prices there will be a pecuniary externality in the use of private information. This is that, when traders react to their private information, they do not consider that they are influencing the price, which in turn influences the actions of other traders who are also conditioning on the price. This pecuniary externality may counteract the learning from the price externality, and lead agents to put too much weight on private information. In Vives (2016) I show that, in normal circumstances, where the demand schedules of the traders are downward-sloping, the pecuniary externality is stronger than the learning externality, and so traders overreact to private information. 'Overreaction' is understood with respect to a (second-best) welfare benchmark in which traders internalise collective welfare but respect the decentralised information structure of the economy - that is, a structure in which each trader can act only on his or her information.

The somewhat surprising possibility of prices that are ‘too informative’ may arise because even though more informative prices are good for aggregate efficiency, in a second-best world, this comes at the cost of increasing the dispersion of trades[TIM1] . Indeed, for prices to be more informative, traders have to respond more to their private signals, and this magnifies the noise contained in the signals. The problem for welfare is that when prices are very sensitive to a surprise in fundamental[TIM2] s, this induces traders to trade too much when fundamentals are low and too little when they are high – this[TIM3]  is the outcome of overreaction to private information.

The market inefficiency can be corrected by inducing traders to moderate their response to their private signals, and by making the price less sensitive to surprises in fundamentals. This is where a Tobin-style tax may play a role. If properly designed and calibrated, a transaction tax makes informed traders internalise the pecuniary externality in the use of private information. The end result is a price that contains less information, and possibly even a deeper market. A potential problem is that the regulator typically cannot distinguish between informed and uninformed trade. Fortunately, the tax can be calibrated to apply to all traders to induce an efficient result.

Alternative rationales for financial transaction taxes have been provided by Dow and Rahi (2000), who found conditions in which a transaction tax would be Pareto-improving, even if the tax revenue were wasted. Subrahmanyam (1998) also considered a transaction tax, but found that such a tax would reduce market liquidity. More recently, Davila (2016) found that a positive transaction tax would be optimal if trade were driven by heterogeneous beliefs across investors.

So, are we home and dry? No, because as Tobin pointed out, there are other potential costs associated with an FTT, not least the implementation costs. Regarding the latter, the decision over the range of securities affected is crucial (in principle, the tax has to be levied on shares, debt, and derivative instruments in both organised and OTC markets, though with some exceptions). At the same time, there is the question of how to avoid trade diversion towards jurisdictions with more favourable tax treatment, and what effect the tax may have on the Capital Markets Union. More generally, the impact on the cost of capital of issuers, market liquidity, and volatility have to be examined.

Full article on VoxEU


© VoxEU.org