This is aimed at the presumed business model of transactional banking that Richard offered us on ‘Robin Hood Part Deux’, in this comment defending Robin Hood taxes. He hopes to prove that the transactional costs might be borne by overpaid bankers, who are in any case a cost on the industry.
Richard says: “profit is a residual after costs … those costs clearly include rents purloined by bankers … a significant part of the incidence of a currency transaction tax on foreign exchange, and probably one on derivatives etc as well, will fall on bankers and not just banks”
This seems quite clear. Consider a simple model:
Profit = (Price to Consumer) X (Quantity of product) minus BankerCost minus Tax.
Tax goes up. What happens? Richard is claiming that when Tax goes up, instead of the bank either boosting Price to Consumer, or sitting there and taking a profit hit, he might turn around and go after BankerCost. And, depending on the power relations within the firm, and the competitive situation that the Bank is in, it might happen that way.
Unfortunately, the problem as described above is certainly inaccurate. First, while I acknowledge that the market in Bankers is highly rigged and subject to all sorts of rent-seeking (yes, TUC, I agree), the market in FX is incredibly competitive. This means that the ‘price’ offered to the end users is already driven right down. The spread in the interbank market on FX is 1.3101/1.3103, or something like that. On this spread, the profit is perhaps 0.5 pips per transaction.* Any tax is almost bound to go up on the spread – profits are in fact tiny per transaction.
Now, the really bright ones reading might notice that the spread I have quoted is larger than the transaction cost. So could it not be absorbed? No. Here is why. Imagine you are a market maker. You can make a ‘spread’ of 2 points on, say, EURUSD. This means your price is, say, 1.3102 at 1.3104. If someone wanted to buy off you, they pay 1.3104, and if they sell, 1.3102. Sounds nice, huh? You’re buying low and selling high. If at every moment you were always likely to get a buyer and a seller, then you would make 1 pip per trade – and can pay over to Robin Hood and his Merry Men.
However, you are confronted with a massive number of speculators and other market makers. The market makers all have slightly different prices. Some may be 1.3103/1.3105, for example. Price-aware speculators would buy with you, and sell with the other guys, so that on average you only trade when your price is slightly high or low to average. The speculators also react to news, and you react to the speculators. So, the way news comes out is: something bullish happens, someone buys, you lift your price, hoping to get a seller so that you reduce your exposure. As a result, a buy followed by a sell may in fact only make you 1 pip – or half a pip per trade. This is roughly where I think the FX market is. Profits on the $3.1 trillion of daily turnover at banks are probably about 0.5 pips, I guess.
[If a trade comes in very large size, you may quote a larger spread – say, 1.3098 at 1.3106. If the trader buys a huge amount at 1.3106, you have the comfort – you hope – of being able to ring round and get lots of offers at 1.3103/04/05 etc before the market gets a whiff of what is going on, moves up and wipes out your profit]
If you don’t believe me, try it out. Next time you are watching a game of 2020 cricket, make a market to your many friends in the number of runs scored (here is how it works) and allow yourself a spread of 5 runs. Let people buy or sell a pound a run with you, and you will soon realise how much less your revenue is than your spread. (We used to do this by forcing trainees to make a market in the size of the restaurant bill during a boozy night out – but that contains all sorts of other lessons about rigged markets, and making youngsters cry ….)
So, returning to Richard’s model, what really happens if a transaction cost goes up? Well, the transactional cost is passed on in terms of higher spreads. In this case, you go to making a 3 pip instead of 2 pip spread, as does everyone else. This causes volumes to fall, as speculators find it more difficult to make a profit. With lower volumes, some market makers drop out too – everyone has different cost curves, and the one who considered it marginally worth doing thinks it is no longer.
Because the banker’s basic pay is a fixed cost with regards to transactions, one thing that would NOT help improve the marginal profitability of doing FX would be to cut his salary. If you go from making 0.006% of volume to 0.001%, then what you do is just close the desk. There are other uses of your capital and management time. Instead of 20 market makers making a 2 pip spread, you get 10 market makers making 3 or 4 pips. And when a large end user comes in, instead of getting a quote of 1.3100 at 1.3108, he gets 1.3080 at 1.3125. On $1billion, that is, I think, $2m extra in costs on the way in.
Would bankers get paid less? Probably. There would be lower volumes. The remaining players would normally make a slightly high profit per trade from a more oligopolistic, murkier and less efficient market. They would send Christmas cards to the TUC for sure . . . After all, look at what has happened to profits this year. Liquidity is worse, the remaining players benefiting from wider spreads**. And the end users would face higher transactional costs. According to JP Morgan, what has been suggested so far might be huge.
*the difference between 1.3101 and 1.31015
**I have NO idea what Richard is on about when claiming that everything has been just fine this year. To work out whether things got worse, you need to look at average spreads, as for example the LSE did when studying the short selling ban. This sort of reasoning is what I need:
- Average spreads: stocks subject to the short-selling ban experienced a subsequent increase in spreads that was 150 per cent greater than the increase in spreads in the control sample. During the 30 trading days prior to the introduction of the ban, the average spread had been steady for both groups, but increased by 140 per cent from 15 basis points (bps) to 36 bps for those stocks which were no longer available for short-selling, compared with a rise of only 56 per cent to 20 bps for the control sample.
- Market depth: as measured by calculating the volume required to move the bid and ask price in each stock by one per cent, market depth declined more markedly for the stocks subject to the ban, decreasing by approximately 59 per cent, compared with only 43 per cent for the control group