Gordon Brown and Nicholas Sarkozy’s recent comments at the EU Summit have raised again the idea of imposing a financial transaction levy – a broader version of the 1972 idea by James Tobin to discourage speculative behaviour by taxing foreign exchange transactions. Whilst their motivation is clearly driven by the politics of extracting a greater contribution from the financial sector, the economic case for a Tobin tax rests on whether it would, in fact, stabilize financial markets. The traditional economic arguments for and against such a tax rest on the impact of the tax on market liquidity, since, if the tax drastically reduced market liquidity, it could in theory increase volatility rather than reduce it. However, these arguments often miss the most important point. What matters is not whether the tax reduces normal market volatility by some small margin – but whether such a tax would help to prevent manias and panics.
Unfortunately, the Tobin Tax fails to address the problem of spikes, since taxing transactions merely reduces the amplitude of the entire price time series by a small amount. The fundamental problem with the “fat tailed” distributions observed in financial markets lies not with the amplitude of the price series, but with its frequency distribution. As all engineers know, time series can be transformed into the sum of a series of sines and cosines of different frequencies through a Fourier transform. The existence of spikes in a distribution reflects the presences of high frequency components in the price series. Engineers routinely remove undesired high frequency elements through the use of “low-pass filters”. In electronics, the simplest low-pass filter consists of simply putting an inductor (a tightly wound coil of wire) into the circuit. Inductors have the property that their resistance is proportional to the rate of change of the current going through them – when the frequency is changing slowly they let the current through with minimal resistance – but when the frequency is high they block the pass through of current. This is precisely the property needed to remove the spikes in price series. Thus it might be possible to implement an Inductance Tax which would tax transactions based on the rate of change of the aggregate market price. Sales and purchases during normal periods would incur virtually no tax – but sales during crashes would be heavily penalized, as would purchases during booms. By dramatically reducing the returns associated with selling during panics or buying during manias, one could substantially reduce the spikes observed in such time series. (In fact a similar idea has already been proposed by Spahn who has argued for a two-tier Tobin Tax. However, the Inductance Tax has the advantage that it would not require the specification of arbitrary thresholds between the lower and upper tier tax rates).
Although the implementation of an inductance tax would be the first-best solution to the problem removing spikes in price series, it may be difficult to persuade market participants to pay a tax the value of which cannot be known more than a very short period in advance. Luckily there is a simple analogue to the inductance tax which is easy to implement and is familiar to us all – queuing. Forcing market participants who wish to buy or sell an asset to wait in a queue is effectively a tax. The more participants who are attempting to leave (or join) the market at the same time, the longer the queue and therefore the higher the effective tax. Of course deliberately introducing delay into trades which, technologically, could be instantaneous would also not be popular. However, again there is a clear analogy to this in the real world – we are all familiar with the idea that, in the event of fire or emergency we should walk, not run, and form an orderly queue to leave the building. Doing so is likely to result in a better chance that everyone survives the emergency. Exactly the same thing is true in finance – forcing investors to form an orderly queue might ensure that all market participants are a little less “burned”.
A final twist on the above idea takes into account that some market participants might value exiting or entering a market more highly than others. Thus it would be perfectly feasible to implement a queuing system as described above, and then allow market participants to sell their place in the queue. This maintains the principle that rapid aggregate exit/entry is heavily taxed, but allows a market participant that really needs to sell now to purchase the place in the queue of someone who doesn’t mind waiting a little longer. It is also possible, with or without such trades, for people who have joined the queue to drop out. Indeed the whole point of forcing queuing is that people joining the back of the queue during a panic may give up trying to exit and, by doing so, lessen the panic.
Brown and Sarkozy would therefore do better to focus their attention less on extracting more revenue from the financial sector via a uniform Tobin Tax, and more on stabilizing markets through the imposition of an Inductance Tax.
Institute of Development Studies