I was reading an article in EE Times (an electronic engineering trade site) about future development of very high speed Ethernet networks. At the end they had this chart:
Notice that the financial sector has the highest growth rate for networking bandwidth, at 95%. Why would those guys need so much bandwidth? Compared to video, the amount of data for each financial transaction is small. They must be doing a very large number, and the number they plan to do is increasing fast.
So I looked up that source for that chart, which is a report from the IEEE (Institute of Electrical and Electronics Engineers). It included this chart (Figure 14):
The vast majority of financial industry network traffic is for options. I assume that would include all of the fancy financial instruments made so infamous by the financial crisis. The volume of over 4,000,000 messages per second indicates that this traffic is being initiated by computers, not people. Notice also that the rate did not slow down after 2008, instead it accelerated.
The supposed purpose of a financial services sector is to allocate resources to productive uses. The high speed trading and massive shift to "financial engineering" with all sorts of options and derivatives does nothing to allocate any resources to anything useful. Instead, it appears to be intended to skim money off the flow of money in the real economy. Even worse, the automated high speed nature of the trading, combined with the huge volume, threatens instability. The "flash crash" that happened after the hacked AP report is just an small example of what could go wrong. Computer programs now trade in enormous volumes. Anyone who thinks that these programs have no bugs in them has never dealt with computer software. But even if the programs were perfect, the assumptions they are based on could be faulty, just as assumptions about housing prices never going down contributed to the financial crisis in 2008. With the volumes and speed of computer trading, and the complexity of the financial instruments and then algorithms being used, it would be very hard to show that major instability won't occur.
Paul Krugman has always said that banking should be boring. The period from the New Deal to 1980 had more financial regulations and was free of financial crises, while rapid economic growth was taking place during most of that period. The economy should be based on real production of goods and services, not financial engineering.