Trading could be divided in two depending on the rationale behind the transaction.

You can either take prop risks - on behalf of the investment bank- or pursue trading activities just because you are servicing client needs.

Proprietary trading or "prop" for "connoisseurs" is the art of taking positions for the bank in any way, shape or form (equities, bonds, derivatives, FX, commodities, etc). Prop comes with huge risk for obvious reasons and it has been very useful to maximize return on equity (ROE) because it achieves amazing returns with little infrastructure costs. Goldman Sachs has always been the paradigm of prop.

An example of "client trading" could very well be the fixed income business where profits are made out of bid/offer spreads. This means that you need to close the position in order to make money out of the business. When an investment bank is selling bonds, it is generally taking a short position because it does not have all the assets in the balance sheet. If, by the same token, the bank buy bonds back it will have to keep them in the balance sheet until it can offload the risk.

So the fixed income cash business could be considered a trading source of revenues and a risky one!
CitiGroup is one of the most active "client trading" animals in the arena, mostly in the fixed income side.

On the other hand, brokering means buying and selling a position with no other than "operating risk" (e.g.: Lets say that the phone line collapses for that matter). As the reader can imagine margins are low in brokering activities so we are talking big volumes here. Banks with reduced risk appetite are condemned to arbitrage positions of risk producers and risk takers at the expense of limited ROEs.

Finally client activities are customer related fees like selling equities, M&A fees, structuring fees, etc. Everyone is keen on client activities because they do not involve trading risk so they are not volatile but they come with a huge infrastructure cost. Nurturing equity, credit and macro research teams. Client trading means providing with liquidity and therefore increasing balance sheet costs for the sake of keeping clientele happy and, of course, maintaining a well paid sales force centrally - NY and London - and locally in the satellite offices.

M&A fees are noisy but... very little compared to equity or fixed income market ones... and arguably not financially profitable. If you do not take into account indirect business like debt issuance, derivatives or wealth management (for example after a reach family IPOs their company).

The very best example of client fees is the equity cash model where UBS and Morgan Stanley are kings.

There´s always been two investment banking models. The ones that rely on customer revenues and the ones that are backed by trading.

Remember this article was about banking models and... their sustainability.

15 years ago, in the mid-nineties top tier investment banks took a dramatic decision: Goldman Sachs decided to link their revenues to trading and Deutshe Bank opted to pursue client (mainly structuring-wise) revenues.

Goldman Sachs model had an exception: Elephant trades. A nice $100m revenues M&A or structuring trade does not need a huge infrastructure if you have the "rainmaker" (marketer) in place. Goldman achieved 50%-70% ROEs. Excluding rainmakers (star marketers) only traders had a decent office at Goldman´s at the time.

Deutsche Bank also pursued trading but mainly linked to capturing client fees. Deutsche Bank model followers needed a huge infrastructure in place as we discussed before, therefore their ROE was more amongst 15%-20% as an average. But their revenues were much more predictable.

So the decision was there. Lehman, Morgan Stanley, Bear Stearns followed Goldman´s approach and BNP, Societe General, Merrill Lynch, JP Morgan, Credit Suisse and Barclays tracked Deutsche Bank one.

I still remember when 10 years ago Anshu Jain (Head of Deutsche Bank Investment Bank and probably the best paid banker in the city) told us how much he liked Lehman and Goldman Sachs model...

I can understand why.... Its was a sexy model... Being able to move the pound like Soros... I wondered to myself if these traders could have "alpha" forever.

We needed more growth and we thought prop. trading was the easy way to achieve Goldman´s $ 20,000m revenues range...

Market has answered in a crude fashion: Bear Stearns, Lehman have gone out of the game.

What can we learn from this exercise?

I believe prop. trading will always be there but only combined with a huge client platform. And by that I mean either a commercial bank (Bank of America-> Merrill Lynch) or a private bank (UBS, Credit Suisse).

Otherwise it will be very difficult to sail the volatile waves of this new century market ocean.

So... Bankers keep your clientele because that will help you get to the 22th century alive...