One of the highlights of my week is the suspense that follows after I hit send on an issue of Net Interest. It’s usually after 5pm in London, so I grab a beer and wait. After a while, feedback begins to trickle in. There are now over 26,000 subscribers to Net Interest and among them there’s invariably a more qualified expert on almost anything I write. It’s great to hear from you! Best of all, though, is when something I’ve written sparks a new connection.
A couple of weeks ago, for example, I told the origin story of prime brokerage in Prime Time in Crypto. The business traces its roots back to 1971 when “a clerk at a firm called Furman Selz…realised there was a business here, and began cultivating hedge fund clients.” A few days later, that clerk reached out to me via LinkedIn. His name is Lou Ricciardelli. “I read your article. Crypto and PB. I am the person you discussed in the article… It was great to read your article and I believe crypto is the next big play in PB. Thank you.”
Something similar happened in the summer after I published my piece, The Long Slow Short, about the demise of Western Union. While researching the piece, I kept coming across references to a paper on the history of Western Union, but I couldn’t get hold of it. I mentioned it in my notes and a reader immediately got in touch: “I saw you mentioned a thesis written by my friend, Joshua Lachter. I’d be happy to introduce you if you’d like to get the paper.”
These two events illustrate the power of networks. Neither is really a coincidence – each of the pieces was read by over 30,000 people and each of those people probably knows around 1,000 people. So that’s 30 million people covered. No surprise that Lou Ricciardelli and Joshua Lachter are among them out of all the people I’ve referenced over the months.
Networks are a fascinating mental model and once you start looking, they crop up everywhere. I’m interested in them because they are especially prevalent in finance. Credit is built on trust which can be validated via networks, and markets are built on information which is transmitted via networks.
To see how deeply networks permeate finance, let’s look at five case studies. At first glance, the Medici of fifteenth century Florence may have little in common with peer-to-peer lenders or decentralised finance; the Rothschilds may have little in common with credit bureaus like Experian. Yet underpinning the success of all of them is one thing: networks.
The Medici Power Network
In its prime, the Medici Bank of Florence was the largest and most important bank in Europe. It was founded in 1397 by Giovanni di Bicci de’ Medici, who took over the Roman branch of a bank owned by his cousin and relocated it to Florence. Under Giovanni’s stewardship, the bank did well and under his son, Cosimo, it did even better.
Between them, Giovanni and Cosimo invented many banking practices still in use today:
They opened up branches in a number of European cities including Milan, London, Geneva, Lyon and Bruges. Rather than run them directly from Florence, they set them up as ring-fenced entities under a central holding company structure in order to mitigate risk. This has provided the blueprint for banking structures up to the present day.
They also pioneered the use of letters of credit – guarantees of creditworthiness that allow exporters to sell goods abroad without moving cash. These were a particularly profitable enterprise for the Medici. Lending money at interest was banned by the supervisory authority (the Catholic Church) but letters of credit allowed interest to be disguised as FX translation. (Regulatory arbitrage has a rich history.)
Perhaps most famously, the Medici popularised the use of double-entry bookkeeping. By recording both credits and debits, they were able to keep accurate records of their financial condition.
All of these innovations helped propel the Medici to the pinnacle of European finance. However, what cemented it for them was the network they created around their bank.
The Medici’s network consisted of two sets of connections – business dealings and marriages to other elite families. These connections enabled collaboration in an environment where economic contracts were difficult to enforce and where competition could be fierce.
Most elite families had some sort of network around them, but the Medici’s was unique. As well as simply having more connections than other families – almost double the number of marriage and business connections than either of their key competitors – they acted as a central connector to other key families. Sketching out the links between elite families in 15th century Florence shows that while the Acciaiuoli, Ginori, Pazzi and Tornabuoni families are not directly connected, they do connect via the Medici family. The Medici form a key lynchpin in the network that bonds all the main families. In contrast, their main competitors married prolifically among each other – no one family assumed the role of central connector.
The Medici position in the network made them an obvious intermediary for business dealings where trust is essential. Just over half of the shortest paths between other pairs of families in the marriage network pass through the Medici, compared with one in ten for their main competitors. It also gave them political clout, of which Machiavelli was in awe.
The Medici Bank eventually collapsed as it outgrew its network. Managers were hired to run the branches; they became embroiled in fraud and made a series of bad loans to secular rulers. In 1467, the King of England owed the Medici Bank in London the sum of £10,500. In those days, it was less risky to lend inside a network than to sovereigns. 1
The Rothschilds’ Information Network
Like the Medici, the Rothschilds made a number of contributions to the evolution of finance. They were active in markets ranging from international government debt to commercial bills, commodities, bullion and insurance. Yet also like the Medici, their success owed less to their financial innovation and more to the network they cultivated.
Between 1810 and 1836, the five sons of Mayer Amschel Rothschild fanned out from their home in Frankfurt to establish banking businesses across five major European cities. It didn’t take long for their combined capital to exceed that of their nearest rivals, the Barings.
Rather than marrying into other families, the Rothschild family stayed tight. It wasn’t unusual in those days for cousins to marry each other, but the rate at which it happened among the Rothschilds was especially high. Of twenty-one marriages involving descendents of Mayer Amschel Rothschild between 1824 and 1877, fifteen were between his direct descendents. This did nothing to expand the family network, but it strengthened the bonds within the family. Trust among the principals running the five banking businesses across Europe was assured.
Looking outward, the Rothschilds compensated by building up an equally important network – a network to carry information. In the early nineteenth century, postal services were slow; it took up to a week for mail to reach Frankfurt from London. So the Rothschilds built their own communications service, using private couriers to shuttle information and agents to charter boats across the English Channel. In 1815, Nathan Rothschild was famously the first man in London to hear the news of Napoleon's defeat at Waterloo. His source was a Rothschild courier who brought the news from Brussels via Dunkirk and Deal. Rothschild reportedly found out a full thirty-six hours before Wellington’s official despatch was delivered to the Cabinet.
As well as providing them with market intelligence, this network gave the Rothschild family access to the European elite. They offered excess capacity on their network to others, including Queen Victoria after 1840. And they were able to provide a unique news service. Over time, the network grew as the Rothschilds added further outposts such as Madrid and St Petersburg and later New York and Mexico.
The network provided the Rothschilds with information that bolstered their success in markets – their edge on the Waterloo news allowed them to make some short-term bond gains. The social network it fostered enabled success in banking more broadly. 2
Credit Bureau Network Effects
On December 18 and 19, 1912, the Bank and Currency Committee of the House of Representatives in Washington DC held an inquiry to investigate “money trust” on Wall Street. Giving testimony was John Pierpont Morgan, in the twilight years of his life (he would die three months later). The attorney for the committee asked him some questions:
Attorney: Is not commercial credit based primarily upon money or property?
JP Morgan: No, sir; the first thing is character
Attorney: Before money or property?
JP Morgan: Before money or anything else. Money cannot buy it.
For many years, underwriting character was a difficult process to scale. It was fine in small communities, or inside tight networks but across larger populations it was more difficult. Over time, a solution began to emerge. In 1826, an organisation was formed in Manchester, England: “The Society of Guardians for the Protection of Trade against Swindlers and Sharpers”. Its members included innkeepers, drapers and other merchants who shared information on people who had failed to pay their debts. The information was based on gossip and wasn’t always reliable; in 1857 the Manchester Guardian Society, as it became known, was forced to appoint a “data accuracy officer”.
Similar groups were established in other towns across England, the US and elsewhere. By the time JP Morgan made his comments to Congress, there were over fifty such credit agencies in the US. Some of them innovated. One – the Merchants’ Credit Association in Dallas – began to list good as well as bad credit risks which it sold to subscribing members in an annual credit reference directory.
As these organisations grew and professionalised, they realised the benefits that could accrue from joining forces. The more merchants a bureau has data from, the better the overall credit picture it is able to formulate. This makes it more attractive for additional merchants to join, who contribute more data, leading to a virtuous circle. The Manchester Guardian Society and the Merchants’ Credit Association were both eventually subsumed into Experian, one of three major credit bureaus active in Europe and the US today.
In his new book, The Cold Start Problem, Andrew Chen, a partner at Andreessen Horowitz, highlights the credit bureaus as a beneficiary of data network effects – the ability to better capture value from a customer as a network gets larger. This economic network effect provides a strong defence against new entrants and affords its bearer considerable pricing power. Experian and its peers enjoy healthy 35%-40%+ EBITDA margins on their mature US business.
The Medici and the Rothschilds were both constrained by how much they could scale their networks; each new node was hardly low cost (dowries were expensive and so were information corridors). Not so Experian. The communications era heralded the way for larger networks if their sponsors could get it right.
Peer-to-Peer Lending: Failing to Reach Tipping Point
In 2005, Zopa was founded as the world’s first peer-to-peer lending company. In the sixteen years since, it has intermediated £6 billion of loans. But then this week, it announced it’s shutting down its peer-to-peer business to focus fully on its bank.
In the years following the financial crisis, peer-to-peer lending (or marketplace lending) was seen as the future. The thinking was that a technology-powered online marketplace could be a more efficient mechanism to allocate capital between borrowers and investors than the traditional banking system. Consumers and small businesses could borrow through a peer-to-peer marketplace at a lower cost of credit and investors could earn attractive returns from an asset class historically closed to individual investors.
Platforms like Zopa in the UK and Lending Club in the US touted the network effects embedded in the model. Similar to the credit bureaus, increased participation leads to more data which can be used to improve the effectiveness of credit scoring models, enhancing the performance record of the marketplace and increasing trust. Interest rates are a proxy for trust and so, as Lending Club highlighted in its S-1 prospectus:
“As trust increases, we believe investors will continue to demonstrate a willingness to accept lower risk premiums that will allow us to offer lower interest rates and attract additional high-quality borrowers. We believe that these network effects reinforce our market leadership position.”
The Medici, the Rothschilds and the credit bureaus each created new financial products and used various networks to cement their position. Peer-to-peer lending is different – the product is the network. The loans themselves are fairly standard; rather, the innovation is that they are funded via a network.
It’s a shame Andrew Chen doesn’t tackle peer-to-peer lending in his book, because it’s a great case study of how network effects can fail to reach a tipping point. Growth had begun to slow even before Covid hit the market.
The industry suffered many problems en route. It tackled the “cold start problem” of igniting a two-sided marketplace by bringing institutional investors onto its supply side (the hard side). But they had too much leverage. As Chen writes, “And thus the paradox—as a network scales, its hard side will professionalize. Quality and consistency will probably increase, and the most sophisticated players will be able to do it at scale. On the other side of the paradox, this dynamic eventually misaligns incentives—drivers, sellers, and creators might protest.” In the case of peer-to-peer lenders, the result was lower credit standards. 3
Additionally, with no skin-in-the-game, the reputation of a marketplace is critical. But this was spoiled by a few bad actors. Shutting down his peer-to-peer business, the CEO of Zopa said, “We've had some platforms that were potentially not well run fail. As a result, we’ve seen customer sentiment for the industry suffer and a lack of trust that has come through. We haven’t seen that necessarily in our customers but we’ve definitely seen that in our ability to attract new customers. We have seen increased costs associated as a result.”
Anti-network effects around trust overwhelmed any positive effects inherent in the model.
The Future of Finance
All of the networks discussed so far are ultimately quite centralised, with one participant capturing a disproportionate share of the value the network creates. The Medici and Rothschild networks were clearly centred around those families, and the credit bureaus the same. Peer-to-peer lending was meant to share more of the economics with end users, but the marketplace platforms were nevertheless in it to make money.
An alternative is available through decentralised finance (DeFi). DeFi allows participants to conduct financial transactions directly without an intermediary, via so-called smart contracts. Trust is conveyed over a network but, with rules embedded in code, it scales a lot further than the social networks of old. We discussed some of the protocols in My Adventures in CryptoLand and Reinventing the Financial System.
Yet some argue that full decentralisation in DeFi is illusory. All networks need validators. In decentralised networks, those validators are dispersed through the network rather than at its centre, but they need compensation to incentivise them to participate without committing fraud. Blockchains based on proof-of-stake allow validators to stake more of their wealth to give them a higher chance of checking the next block of transactions and receiving compensation. Since the associated operational costs are mostly fixed, this system can lead to concentration. Many blockchains also allocate a substantial part of their initial coins to insiders, exacerbating concentration issues.
Some degree of centralisation seems inevitable for DeFi to take off. This week, Coinbase rolled out a DeFi application to its customers (outside the US) enabling them to earn yield on Dai stablecoin holdings. None of this detracts from the underlying network though. And one thing’s for sure: finance is all about networks.
Always has been.
As well as Andrew Chen’s book, The Cold Start Problem, I would recommend The Human Network by Matthew O. Jackson, The Square and the Tower by Niall Ferguson and, although more technical, Network Science by Albert-László Barabási.