Must Do Better

Donald MacKenzie

Sometimes, the most important – and perturbing – insights make their way into the world without fanfare. As yet, few have picked up on an analysis by the New York University economist Thomas Philippon of the history of the unit cost of financial intermediation. The unit cost is a measure of the efficiency of the financial system, and Philippon tracks its level in the United States since 1884, the time of pens and paper ledgers, when a ‘computer’ was still a human being, equipped at best with a mechanical calculator. Remarkably, Philippon finds that despite more than a hundred years of technological innovation, the efficiency of US finance hasn’t improved since the 1880s.

What is financial intermediation? Imagine you’ve got savings of £100. You want them to be safe, and to have ready access to them. That means you need what an economist would call ‘liquidity services’. A bank can provide them, and will maybe even pay you a modest rate of interest, although less than you might get if you were prepared to tie up your money in an unsafe investment. Let’s say that when your savings are deposited in the bank they earn 1 per cent a year.

Let’s also assume there is someone else who needs ‘credit services’. She wants to borrow £100. The bank lends her the money, charging her 5 per cent per year, or £5. Altogether, the bank has provided £200 of intermediation services (£100 of liquidity services to you; £100 of credit services to her) at a cost to the consumers of those services of £4. The unit cost of financial intermediation in this case is £4 divided by £200: 0.02, or 2 per cent per year. Taken in isolation, 2 per cent annually may not sound a lot, but costs at that level can have a substantial impact on both the cumulative returns that savers receive and the amounts borrowers pay.

The cost of financial intermediation resembles a tax on the rest of the economy, slowing its growth – the difference from a tax being that, instead of paying for schools, hospitals, economically beneficial infrastructure and so on, much of the cost of intermediation makes up the pay packets of senior staff in banks and other financial businesses. Costly intermediation thus increases inequalities of income and wealth.

It isn’t easy to work out the unit cost of intermediation for an entire financial system in a way that’s consistent through time, but essentially the task involves doing the same three things as in my little example: first, calculate the total amount of intermediation services provided each year; second, work out the total annual cost of those services; finally, divide the result of the second calculation by that of the first. Philippon does the first calculation by adding together the sums of money involved in four categories of financial activity: the total amounts held in bank accounts and similar ‘safe’ deposits; the money lent to firms and the value the market gives their shares; the money lent to households; and the total value of corporate mergers and acquisitions. He does the second calculation (working out the total annual cost of intermediation) by adding up the profits and staff salaries of the entire gamut of financial intermediaries: banks, investment-management companies, insurance companies, private equity firms and so on.

The graph shows how Philippon’s estimate of the unit cost of financial intermediation fluctuates through time. It rises to a first main peak during the financial excesses of the 1920s, falls during the middle decades of the 20th century, then rapidly increases again. The unit cost has come down a little in recent years, but only to roughly its level in the late 1880s. (The lower line in the graph is Philippon’s estimate of the unit cost corrected for the changing aggregate level of the difficulty of the task of intermediation. For example, investing wisely in start-ups involves more screening and monitoring – and is thus intrinsically more expensive – than buying the shares of established corporations with lengthy track records; while making a single big loan to a wealthy household is cheaper per dollar lent than making multiple smaller loans to less well-to-do households. However, adjusting the unit cost of intermediation to take this into account produces an only slightly improved picture of change through time.)

Although research on efficiency (in Philippon’s sense of the term) is in its infancy, its early results suggest that there’s nothing specific to the US in the discovery that the financial system has become no more efficient over an astonishingly lengthy period. His data stretch back only to 1950, but Guillaume Bazot of the Ecole d’économie de Paris finds broadly similar patterns of stagnation or increase in the unit cost of financial intermediation in Europe. In the UK, for example, the unit cost was about 1.3 per cent in 1950. By 2007 (the endpoint of Bazot’s time series), it was around 1.8 per cent.

Why hasn’t the efficiency of finance improved during a period in which its core technology, computing, has advanced so much? A good part of the answer seems likely to be that much of the economic benefit of technical improvements has been captured by finance’s senior employees in the form of higher pay. In separate work with Ariell Reshef, also of the Ecole d’économie de Paris, Philippon has shown that in the middle decades of the 20th century, levels of pay in US finance were broadly similar to those in other industries (taking different levels of education into account). At the end of the century, however, pay in finance accelerated quickly compared to other sectors. By 2005-6, average pay in finance (adjusted for educational levels) was 50 per cent higher than elsewhere, and executives of financial firms took home two and a half times what their counterparts earned in other sectors.

Some areas of finance have cartel-like features, including high barriers to the entry of potential competitors. Where retail banking, say, has seen some recent new entrants, creating a new fully fledged investment bank would be a hugely expensive, dauntingly complicated process, and no one seems to be trying it. Yet it’s too simple to think that the inefficiency of finance and its high levels of pay are just the result of oligopolists’ market power. As Philippon and Reshef show, work in finance has become more complex, and in particular more demanding of mathematical skill. You don’t have to spend long with today’s financial intermediaries to realise that they’re usually clever, hard-working people.

Unfortunately, far too much of that intelligence and energy seems devoted to activities that amount to efforts to influence, outwit or outrun their fellow intermediaries. These add to the cost of intermediation without necessarily improving the effectiveness with which the financial system channels savers’ money to productive uses in the non-financial economy. As the economist John Kay points out, being an intermediary tends to create an inherent bias towards activity, even when it’s wiser and cheaper to do nothing, because you have to give the appearance of ‘earning your keep’. Investment managers, for example, cannot realistically expect systematically to beat the market (in aggregate, they essentially are the market), but large amounts of investors’ money are wasted in usually fruitless efforts – by dedicated, highly paid, skilled managers – to do so, thus contributing to the aggregate cost of financial intermediation.

Viewed through the lens of Philippon’s analysis, it’s possible to see the virtues of the smaller, simpler, safer, cheaper financial system of the 1950s and 1960s, and indeed all four of those attributes ought to be goals of financial policy. If there’s reason for at least a little optimism, it’s that increasingly firms with roots not in finance but in information technology are experimenting with new approaches to lending, payment systems, financial advice and other aspects of financial intermediation. Of course, those firms are in business to make money, and a sector of finance that became dominated by a single firm to the extent that Google, Uber or Airbnb dominate theirs could be just as expensive and inequitable as one dominated by Citigroup, Goldman Sachs or the Royal Bank of Scotland. It is, however, worth keeping a close eye on the rich variety of these new ‘fin tech’ experiments, in the hope of spotting ways of creating a better – and cheaper – financial system.