What Sort of a Business is Investment Banking?

Plus: Tinkoff, Revolut, Coinbase

Happy Friday and welcome to another issue of Net Interest, where I distil 25 years’ experience of investing in financial companies into a weekly email. If you’re reading this but haven’t yet signed up, please join nearly 16,000 others by clicking on the link. And to my regulars: thanks for spreading the word.

What Sort of a Business is Investment Banking?

Last weekend, Credit Suisse closed its books on the quarter and announced a big loss. It’s a peculiar sort of loss because it stems entirely from a single client – Archegos Capital Management. According to the bank, its dealings with Archegos “negate the very strong performance that had otherwise been achieved by our investment banking business.” In other words, performance would have been really good, befitting of the bull market, were it not for that one pesky client. 

Which leads to a question: what sort of a business is this? Credit Suisse has over 1.6 million individual clients, over 100,000 corporate clients and tens of thousands of institutional clients. Yet a single client can blow through the profits made from all the others. 

It’s not the first time it’s happened. Over the years, investment banks have suffered huge losses – if not at the hands of a single client then at the hands of a single employee. We highlighted some rogue trading cases here a few months ago: Jérôme Kerviel cost his bank $6.9 billion; Nick Leeson cost his bank its solvency. 

Although the frequency of rogue trading incidents has diminished in the recent past, that hasn’t stopped investment banks from piling up some stinking losses. In the living memory of most market participants, Credit Suisse alone has:

  • Written down $2.85 billion of asset-backed bonds after they had been mispriced by traders. (February 2008)

  • Topped up $633 million of write-downs with another $346 million because “even internally the scale of those positions was a surprise for a number of people” and the CEO “was not aware of the existence of those positions on that scale.” (March 2016)

  • And now, a $4.7 billion loss on Archegos.

Welcome to the business of investment banking.

What do investment banks do?

Defined broadly, investment banks do a host of stuff for corporate and institutional clients. They advise on mergers and acquisitions and underwrite securities offerings; they facilitate trading of debt and equity instruments; they structure derivatives to allow clients to hedge or to take on risk.

The biggest of the bunch is Goldman Sachs. At a presentation at Harvard a few years ago, Goldman’s former CFO explained his business: “[A] client has a risk they don’t want or wants a risk they don’t have… we make it happen for them.”

More specifically, investment banks like Goldman intermediate a diverse set of risks:

  • They take risk, like when they lead an IPO. Last week, Goldman oversaw the IPO of Deliveroo in London. The stock fell 30% on the first day of trading and Goldman was forced to go into the market to support it, buying up £75 million worth of stock. Advancing margin loans to firms like Archegos is another example of taking risk, although usually it is mitigated by concentration limits and lending less than the value of the collateral. 

  • They match risk, by transferring it between parties. As a producer of oil, the Government of Mexico may want to protect itself against a drop in oil prices. Airlines, by contrast, may want to protect themselves against a rise in prices. Investment banks provide the search functionality for either side to find each other, but also transform the specific risk that each side holds (from Maya crude in the case of the Mexican government to jet fuel in the case of the airlines). 

  • They source risk, by seeking investment opportunities for clients. Credit Suisse did this badly when it packaged Greensill loans into an investment fund for clients. (Granted, this occurred in its asset management division rather than its investment banking division, but still.) Other times it may package structured notes or baskets of stocks for clients, enabling them to express a multitude of investment views. 

In short, investment banks traffic in risk. The entry-level investment bank recruit’s handbook is Liar’s Poker. In it, Michael Lewis writes:

Risk, I had learned, was a commodity in itself. Risk could be canned and sold like tomatoes. Different investors place different prices on risk. If you are able, as it were, to buy risk from one investor cheaply and sell it to another investor dearly, you can make money without taking any risk yourself. And this is what we did.

Buying and selling risk can be a profitable business, although it’s not as profitable as it used to be. Coalition Greenwich is an analytics company that tracks global investment bank revenues. The people there reckon that last year, investment banks earned nearly $200 billion of revenue, the most in over a decade. Sales and trading made up almost $150 billion of that, the rest being advisory and underwriting fees.

Prior to 2020, operating margins in the industry had been coming down, but last year they jumped to 44%. Such high margins require a lot of capital to generate. The sales and trading business in particular is quite capital intensive. Risk has to sit somewhere and in many cases it can hang around for many years. Deutsche Bank has a specific portfolio of interest rate derivatives, for example, whose average life is eight years. Unlike a simple agency broking model, full-scale risk intermediation requires banks to carry a large balance sheet. Consequently, the returns on equity are a bit more pedestrian. Prior to last year, industry return on equity was typically sub-10%; last year it jumped to 13%.

Beneath these aggregate numbers, though, there’s a lot of ups and downs. Last year, Goldman Sachs did $21 billion of revenue in its global markets businesses but some days were good and some were bad. According to disclosures, Goldman lost money on 24 days over the course of the year. On two of those, it dropped more than $75 million on a single day. Yet the firm also scored some massive home runs, making over $100 million per day on no fewer than 50 individual trading days.

Goldman’s skew towards a high number of really profitable days was especially pronounced in 2020. The distribution of daily trading revenue is illustrated in the chart below. Last year’s distribution (the blue line) looks closest in shape to 2011, when the number of “home runs” was last as high, but in that year there were many more loss-making days offsetting the gains. 

As well as the daily ups and downs, which can be a function of market opportunity, there is also the ebb and flow of market share. In 2020, Goldman pulled in around 13% of the total fixed income revenues generated by the top nine global investment banks. Market share tends to be quite sticky over the medium term, since there’s an optimum number of firms clients are happy to deal with – they want more than five banks but they don’t need more than fifteen. Over the past ten years, the biggest loser of market share has been Deutsche Bank, whose share of fixed income trading dropped from around 13% in 2013 to around 9% last year. Importantly, though, Deutsche had to work really hard to lose that share, cutting back its presence in the market significantly.

A Competitive Market

Competition in the industry plays out through a competition for talent – which means the wage bill in the industry can be very high. One of the banks eager to make it into the top bracket of firms before the financial crisis was Barclays. Philip Augar tells the story in his book, The Bank that Lived a Little. He quotes the global head of HR: “We are the highest payer on the street. The competitors all say we are driving up pay rates… No other bank has a scheme like our long-term plan.” Between 2002 and 2009, Barclays’ long-term incentive plan paid on average £170 million each year to 60 people on top of their salary and bonus. Nice work if you can get it!

In those days, employees would typically get a 50% cut of the revenues. This led to perverse incentives where traders would seek to maximise revenue without regard for risk in order to expand the compensation pool. The financial crisis put paid to that and compensation rates have since come down. Those big balance sheets are the gift of shareholders and since the crisis they have demanded a greater share of the economics for supporting them. Last year, Goldman paid a record low 30% in compensation to employees.

Nevertheless, pay still remains relatively high in the industry. Deutsche Bank discloses annual compensation of each of the 2,300 “material risk takers” it employs. Last year, 684 of them took home more than €1 million of total pay and one of them took home more than €10 million.

There’s another way competition plays out, which we became witness to in the Archegos saga. Back in Liar’s Poker, Michael Lewis quotes a leading bond salesman at his firm: “The trading floor is a jungle.” It’s an accurate observation of the industry. We now know one of the reasons why Credit Suisse’s loss on Archegos was so big is that for other firms involved it was so small. By unloading Archegos positions earlier, other firms were able to minimise their losses while at the same time making them worse for Credit Suisse. There’s a zero-sum aspect to the game. 

The situation reminds me of something a divisional head at Morgan Stanley once told me. Reflecting on his competitive strategy, he quoted General George S. Patton: “No bastard ever won a war by dying for his country. He won it by making the other poor dumb bastard die for his country.”

In the Archegos case, because the payoff was non-linear and situations like it come up relatively infrequently, there is little incentive for firms to cooperate. Over the very long term, of course, that strategy can go awry. In 1998, fourteen of the largest investment banks in the world agreed to post $3.65 billion to take over all the assets of failing hedge fund LTCM. Only Bear Stearns declined to participate. Ten years later, the same banks refused to bail out Bear Stearns when it ran into its own troubles. 

But striving for a Nash equilibrium in a round of Prisoner’s Dilemma is not the sole reason some firms end up doing worse than others. Credit Suisse is in the crosshairs now, but it’s part of a broader phenomenon whereby European banks just aren’t that good at investment banking compared with American banks. They can be run by Americans, staffed with Americans, they can even have grown by acquiring American firms, but they’re not that good. Understanding why gets to the heart of what it takes to be good at investment banking. 

European Investment Banks

The first thing that differentiates European investment banks from US ones is that they weren’t founded as investment banks. We’ve discussed the origins of both Goldman Sachs and Deutsche Bank here before. Goldman was founded in 1869 as Marcus Goldman, Banker and Broker, with an initial focus on the commercial paper market. Deutsche Bank was founded one year later to provide long-term loans to German industry. It wasn’t until the 1990s that Deutsche Bank threw itself into investment banking, in response to the structurally low level of profitability it faced in its domestic banking market. In 1990, it acquired London based Morgan Grenfell and, after spending heavily in an attempt to build a global presence organically, went on to buy Bankers Trust in 1999. 

Credit Suisse has a longer legacy in investment banking and may even be seen as the creator of the first truly global investment bank. Credit Suisse entered the market via joint venture, first with White, Weld and Co. (later acquired by Merrill Lynch) in 1962 and then, in 1978, with First Boston. At the time, First Boston was one of the leading firms on Wall Street; Credit Suisse took a 25% stake and they each contributed capital to a joint venture. The deal was struck just before a change in the law made it an impossible structure to replicate. The Glass-Steagall Act had prevented US commercial banks from entering investment banking since 1933, but the stipulation was only extended to foreign banks in 1978. Consequently, Credit Suisse had a twenty year headstart on European banks buying into the US investment banking market. 

Through the 1980s, the relationship worked very well. Credit Suisse looked after the Swiss market; First Boston, the American and Australian markets; and the joint venture – CSFB – was responsible for Europe and the rest of the world. Across the three groups, the franchise became a top three player in M&A and established leading positions in equity and debt underwriting, with market shares of 11-15% and 9-15% respectively. 

However, as markets increasingly globalised, the three groups started treading on each others’ toes. In 1988, they were merged into a single firm headquartered in New York, in which Credit Suisse took a 45% stake (employees held 25% and institutional investors 30%). One year later, disaster struck. CS First Boston had become a leading player in the junk bond market. In 1988 it was ranked #2, behind Drexel Burnham Lambert. When the market collapsed, CS First Boston was left holding $1.1 billion of paper it couldn’t shift. With the firm’s future in doubt, Credit Suisse was forced to bail it out, taking majority control and slashing its headcount and balance sheet. 

CS First Boston never really recovered. Although it was run autonomously, it didn’t get the resources – either capital or staff budget – that other firms did. Staff defected in droves. Having been a top 3 player in M&A and underwriting in the 1980s, the firm slipped to top 5 in the 1990s. It had its license revoked in Japan following misconduct there and suffered huge losses in Russia. To restore its position, Credit Suisse acquired parts of Barclays’ investment banking business in 1997 and then Donaldson, Lufkin & Jenrette in 2000. Yet the DLJ acquisition turned out to be one of the most expensive in investment banking history; sixteen years later its goodwill was finally written off. 

One possible explanation for the sustained poor performance of European investment banks is that they are always playing catch-up. In an effort to close the gap with the market leaders, they take short cuts, taking on excessive risk either via leverage, concentration or duration. (On duration, it is notable that Credit Suisse still has an ‘asset resolution unit’ consisting of $14 billion of assets it’s been trying to get rid of for years; Deutsche Bank has a ‘capital release unit’ of $240 billion.) 

Another explanation is that they rely too much on models, over-intellectualising the risk management process. One Twitter thread describes how Credit Suisse’s Archegos exposure may have bypassed its risk models completely, even though the underlying risk was plain. We’ve discussed risk management here before, in Wimbledon and the Art of Risk Management, and the answers aren’t always in the models. Way back in 2012, Goldman’s CFO said, “While metrics and quantitative measures are an important part of risk management the judgment and experience of our people that overlay these models is a key component.” 

The fact is, for investment banks, risk management is their business. If they take risk, match risk and source risk, they can’t outsource the management of that to a chief risk officer; it’s the job of the frontline staff. How that all hangs together – how the incentives of staff are reconciled with the health of the firm, particularly in an environment where individual compensation can be very high – comes down to the culture of the firm. And culture takes a long time to build, longer than most participants in fast-moving markets have the energy to invest. 

British readers will be familiar with Trigger’s Broom (comedy gold, if you haven’t seen it). You can change the head multiple times, you can change the stick, but it stays the same broom. Credit Suisse has gone through four investment banking heads in the past ten years and its staff has turned over, but the culture remains the same. 

After announcing his Archegos losses, the CEO of Credit Suisse said, “Serious lessons will be learned. Credit Suisse remains a formidable institution with a rich history.” Unfortunately, it’s a rich history of not learning its lessons. 

Full disclosure: I was a managing director at Credit Suisse once and still have a soft spot for the firm. I really do hope they sort this out.

Share Net Interest

More Net Interest


Tinkoff is Russia’s premier digital bank. We wrote about it in depth earlier this year. This week the company held an investor event in which it set a target to hit $1 billion of profit in 2023, up from ~$700 million guided for this year. Its profit forecasts are predicated on two drivers – winning more customers and extracting more revenue from them. 

Right now, the company has 10 million customers and targets 16.5 million by 2023. Last year, active customers held an average of 1.4 “revenue-generating products” each; the target is to increase that to 1.7. The sorts of products the company is talking about include its Tinkoff Black debit card and brokerage accounts in addition to its core credit card offering.

In its presentation, the company gives very detailed data on the lifetime value (LTV) of customers across each of its products. Tinkoff Black customers are the cheapest to acquire ($18) but generate low revenues, so the LTV is only $117. The investment business has the highest return currently – it costs $35 to acquire a brokerage account, but LTV is $1,164. However, that is likely due to high current levels of trading activity. Going forward, it’s layering the revenue streams of these products on top of each other that may provide the upside. 

The stock has been strong recently. Market cap is over two times what was agreed Yandex would pay for the company before that deal was called off. But at ~11 times 2023 earnings, it continues to trade at a discount to other digital banks globally, perhaps due to a Russian discount. 


At its investor event, Tinkoff took a couple of sideswipes at Revolut, one of the largest digital banks in the UK. On one slide it presented its average current account balance alongside the UK neobank average ($850 versus $340) and compared that to average monthly wages in the countries (129% versus 8%).

Then, in answer to a question about international expansion, CEO Oliver Hughes said, “our model that we're thinking about at the moment is not a light model, where you skim off a few hundred thousand customers in 20 different markets. That's not what we’re thinking about.”

A few weeks earlier, Revolut CEO Nik Storonsky was interviewed by Tinkoff founder Oleg Tinkov on his show, Business Secrets (in Russian). According to one translation, Storonsky said that Revolut is now in 35 countries and has 20 million customers. One country it is currently looking to expand in is the US, where it is in the process of applying for a banking license. 

Storonsky outlined his ambition: “The aim of the company is to build a global bank that offers all financial services that people need every day 10x cheaper than the average bank.”

The company has one of the most extensive product suites of any digital bank. Revolut has been very quick to copy products offered by other fintechs to expand that suite further. On the show, Storonsky said that Revolut would be launching a ‘salary advance’ product (similar to what Salary Finance and Wagestream offer in the UK, and Earnd, before it collapsed as part of the Greensill empire). It is also looking at social trading, similar to what eToro offers.

However, the company has been aided by high levels of retail trading activity. “When the Covid pandemic started, everything closed and our revenue dropped by 40% because a lot of our revenue is from payments (interchange), and in two months we made lots of optimisations our revenue increased x2 gross profit increased 10x. People stayed at home, they increased the trading of stocks, it wasn’t worth keeping money in savings accounts. So the apps that had stock trading benefited during that time.”

Retail trading (and crypto trading) has subsidised a lot of fintech investment over the past year. The question is what happens when it stops. Revolut may argue that “people will start travelling and payments revenues will return” but the handover will be a key one to watch. 


Coinbase comes to the market via direct listing next Wednesday, 14 April. I contributed to an analysis of the company as part of Mario Gabriele’s S-1 Club, which you can read here. In a wonderful innovation by Mario, the report was tokenised. Mario’s thread describes the process. A new way to monetise content?