26 FebControl II – How to miss an open window, lose $2mm and survive

I once lost $2mm of Goldman Sachs’ money and lived to tell the tale. This is a post about Corporate Actions. In banking this refers to events relating to the bonds or shares of a company. At the centre is Rupert Murdoch, or rather News Corp. For once not the villain in the story and not involved in any phone hacking, even if in the annals of local Goldman history, this became known as the bad news story.
A normal bond issued by a company will pay interest, known as a coupon, and mature sometime in the future. These are mandatory events, and even if the bondholder does not do anything, she will receive her money. Some bonds often have extra features that require the bondholder to do something if they want to take advantage of that feature. These are voluntary events. If you are not in, you can’t win; you have to take the action. One of those with special features is a “Put Bond”. This allows the holder to give back the bond to the issuer for a fixed price during a certain time window. Back in the day, News Corp issued several of these bonds with the put feature there to make the bonds more attractive to investors. At Goldman Sachs, these bonds were traded by the convertible desk, at the time led by Michael Hintze. I got on very well with Mike and he always made fun of me for the “loud” ties I wore when I saw him in the London office. By the sartorial standards of the day, “loud” meant anything that had more than two colours and / or, was asymmetrical. One of the wags in London, John Frankel, once greeted me by putting the phone to my tie and bellowing to the personal on the other end of the line: “Olaf’s here. Another of those loud ties. Can you hear it?”
One afternoon, I had a call from one of the converts traders, asking me to exercise the put feature on a large position of Swiss franc denominated News Corp bonds. For these events, there is an agent, which in this case was Credit Suisse. So, we phoned up, asking to put the bonds, only to be told we had missed the put window. Now we had done favours for CS in the past, but none were coming back our way; they and News Corp refused to budge. The trader has some $2mm in unrealised P&L riding on this. Unrealised because he had the money in theory and not in practice.
This was a real loss to the trading desk and given the sum involved, there was an investigation into what went wrong. The standard set up for managing these voluntary events is one with dual controls. In the front-office the trader will have his eye on the window or bond feature. In the back-office it is standard practice to buy an announcement service from a supplier. Every security has one or more unique identifiers; like a social security number or even a car number plate. In the banking industry there are firms that supply a service that will tell you the user: “on security X, the event Y has been announced”. Now, if you have done a good job making sure that those identifiers are in your systems, then you when the announcement flies past in the night, as one of many, you have the magnet to suck in that information. Turns out, Goldman had received the information and that led to an enormously long print out of all the positions that day. This was part of a very long report. That particular day, yours truly had been covering for one of the team and reviewing the report. I missed this item. It was never printed again. No news is good news. So the back-office part of checks & balances failed. In the front-office, the traders held six different News Corp bonds, each with a put feature. Five were in USD and the sixth in Swiss Francs. The traders had their systems set with the same details about the put window on all the bonds. This is data they set-up and maintained themselves. The bad news was that the window for the Swiss bond was not the same as the one for the one. So Murphy’s Law had triumphed to the tune of $2mm and it was all my fault. I had missed the announcement. No amount of pleading with CS or the issuer helped. Bad news.
Reference or static data has a role to play here too; this is data that describes the products you trade. Like a catalogue with technical details. Almost all the information about a product is known when the product is first set-up. There are a couple of keys to success: getting the unique identifiers on to the product. The typical one is the ISIN. Without this, you will be unable to source external services since you will not be talking the same language. As the techies would have it; no unique key, no service. The second is the “golden source”. Having multiple arms of the firm update things independently is a recipe for disaster. my won view of this is: Reference data is the either the root of all evil or the fountain of all good. And that is an 80:20 game.
$2mm was a big amount of money then and is still a big amount of money. Big enough to warrant the partner in charge of equity  trading, David Silfen, to pay me a personal visit in Zurich. He sat me down and explained how much money $2mm was; so many commissions, so many spreads, so much turnover. Almost like Chinese water torture. But he did not rant, did not lose it and wanted to understand how we dropped the ball. I explained the weaknesses in the process and how I bungled the only chance I had had to spot the event. A great listener. He asked me to draw up a plan to fix things. This was one of those seminal moments when an organisation wants to invest in making things better in the back-office. Normally, investment is on must do stuff and driven top down. In this case, as a wise old manager at Goldman Sachs, Mike Sammons, put it: from bad comes good. We had lost enough money that we were willing to invest in making the tools better. This did not endear me to the Head of European Ops, who thought the process was not that bad, but we did end up fixing things in a way that made life better for the Operations staff.
Lessons Learned: If you have correctly triaged your reports between Control and Reference types, you need to make sure that the control reports are cumulative. So if something is a problem today and it is still there tomorrow, keep showing it. Ideally, these types of activities should be on-line and “in your face”. On-line is not though “E-Mail”; Controls will only work effectively if they are shown in scorecards, that allow you to drill won to the individual item, make notes on it and administer it. The tools for this today are readily available and very good. Reference data is vital too. This must be centrally maintained and distributed.
Epilogue: At the time, the head trader, Mike Hintze, was just important and one of my internal clients. Since then has gone to both fame and fortune as the head of the hedge fund CQS and is active as a philanthropist and donor to the Conservative Party. Thankfully he was charitable to a young, loud-tied Operations manager in his early years

2 Responsesso far.

  1. John Frankel says:

    Those were the days. The ties were loud, is all I can say in my defense.

    I am not sure if GS would react the same today, but this story is typical of the “good old days” when the firm was run by partners with sense and understanding and a realization that any process can be improved. By the way, your ties were only matched, years later, by Tim Craighead’s shirts, but that is another story.

  2. Marcel Estermann says:

    Another one of these “fun” stories…and as always wittingly written !
    Keep going…I really enjoy it !

Leave a Reply

Your email address will not be published.

CAPTCHA, to Submit please solve this: * Time limit is exhausted. Please reload CAPTCHA.