Thomas Philippon’s The Great Reversal is an important book in which he warns of the decline of competition in many U.S. product markets. Not having studied antitrust issues for some time, I will leave criticism of his general thesis—that competition has indeed declined and that this is owing to lax antitrust enforcement—to my Cato colleagues writing about that issue. (You can start here.)
But I do want to take issue with Philippon’s argument that finance is one of the industries in which competition has declined. Philippon devotes a whole chapter of the book to the financial services industry, giving it the punchy title, “Why Are Bankers Paid So Much?,” while Martin Wolf, in his Financial Times review of the book, called finance a “rent-extraction machine.”
There are two serious problems with this view of U.S. finance. The first is that two central pieces of evidence Philippon cites in support of his general thesis, high concentration and rising profit rates, are simply missing in the financial sector. Another popular symptom of market power, namely slowing entry, is present in banking but directly attributable to regulators’ actions following the financial crisis.
The second problem is that Philippon’s argument on finance relies heavily on a 2015 paper reporting that the costs of financial intermediation had been largely unchanged for 130 years. From this, Philippon concludes that “improvements in information technologies do not appear to have led to a significant decrease in the unit cost of intermediation.” But because the 2015 paper does not break costs by category, it is not clear whether the mix of costs has also remained stable over the years, or whether (as I suspect) there has in fact been a decline in “market operating” costs, thanks to technology and economies of scale, that has been offset by rising regulatory (including lobbying) costs.
Philippon’s diagnosis of uncompetitive markets may therefore be accurate for U.S. finance, but for reasons to do with government intervention rather than the absence of competitive enforcement by government.
Concentration in Banking Has Increased but Remains Low
The United States has historically had many more banks, relative to its population, than other developed countries. As recently as 1984, we had nearly 15,000 banks and an additional 3,500 savings institutions. Several thousand credit unions also competed with these.
The primary reason for the large number of U.S. banks was states’ policy of restricting branching, which federal regulators allowed for decades. Most banks couldn’t have branches in their own state, let alone other states. Even in the early 1980s, no state allowed interstate banking and only a few tolerated intrastate branching. Things only began to change in the 1980s and 1990s, as states gradually liberalized their branching regulations, signing compacts with other states to reciprocally allow each other’s banks to enter their markets. Finally, in 1994 Congress passed the Riegle-Neal Act, allowing interstate banking throughout the country.
Branching has since ushered in an enduring cycle of consolidation: There are now fewer than 5,000 commercial banks and only a few hundred savings institutions. Small banks, in particular, have dwindled. The number of banks with (inflation-adjusted) assets under $100 million has fallen by 82 percent since 1992, while the number of banks under $1 billion has dropped by nearly half. Consolidation, however, has not resulted in a decline in bank presence or proximity. In fact, the number of bank offices (including branches), at 82,731, is near its all-time high and much higher than the 57,227 bank offices of 1984. The number of offices per capita is also higher than at any time before the passage of Riegle-Neal.
Thanks to longstanding branching restrictions, concentration in American banking has historically been exceptionally low. But that was hardly a good thing. Rather, most U.S. banks have been inefficiently small for decades. Until recently, most of them could not take advantage of economies of scale, could not adequately diversify their loans and product offerings, and therefore could not help being vulnerable to changes in their local economies. Branching restrictions impeded competition and gave banks monopoly power. So effective were these competitive restrictions at making bankers’ lives easier, that it seems bankers had lower mortality rates than other company executives of similar age! Of course, competitive restrictions made everyone else worse off.
What all this means for Philippon’s argument is that, while U.S. banking has indeed become more concentrated over the last 25 years, it is still by no means a concentrated industry. The Department of Justice, in its merger guidelines, uses the standard Herfindahl-Hirschman Index (HHI) to assess market concentration. Markets with an HHI above 2,500 are “highly concentrated,” those with an HHI between 1,500 and 2,500 are “moderately concentrated,” while those below 1,500 are “unconcentrated.” Table 1 shows bank deposit HHIs for the ten most-populated metropolitan statistical areas (MSAs) in the country as of 2019. There are values for 1994, 2018, and 2019. Out of these ten markets, only two (Houston and Philadelphia) are moderately concentrated by the DoJ standard, while another (Boston) met that criterion in 2018 but not 2019. Not a single market is highly concentrated, with most MSA HHIs well below the threshold for moderately concentrated.
Table 1. Herfindahl-Hirschman Index for the Top Five Banks by Local Deposits in the Largest U.S. MSAs
|New York-Newark-Jersey City, NY-NJ-PA||322.40||1173.31||1175.46|
|Los Angeles-Long Beach-Anaheim, CA||602.22||828.81||789.16|
|Dallas-Fort Worth-Arlington, TX||908.11||1401.80||1402.37|
|Houston-The Woodlands-Sugar Land, TX||644.41||2010.78*||2177.85*|
|Miami-Fort Lauderdale-West Palm Beach, FL||263.17||711.57||N/A|
|Atlanta-Sandy Springs-Roswell, GA||927.66||1437.40||N/A|
Source: FDIC, Summary of Deposits, Q2 2019. An asterisk indicates the market is “moderately concentrated,” according to the Department of Justice’s merger guidelines.
Even the European Commission, to whose approach Philippon is more sympathetic, says in its own merger guidelines that it “is unlikely to identify horizontal competition concerns in a market with a post-merger HHI below 1000 [and] in a merger with a post-merger HHI between 1000 and 2000 and a delta [change in HHI following the merger] below 250, or a merger with a post-merger HHI above 2000 and a delta below 150, except where special circumstances . . . are present.” The Europeans, whom Philippon holds up as a model of competition policy, tolerate much higher concentration in banking and have a policy framework that would not be concerned with the changes America has experienced since the mid-1990s.
Nationwide, the U.S. banking market has also seen a significant consolidation of deposits, with the top three bank holding companies jointly accounting for 31 percent of customer deposits in 2019, compared to around 8.5 percent in 1994 (Table 2). But those figures still point to a very unconcentrated market by the DoJ’s merger guidelines.
Table 2. Nationwide Deposit Market Share of Top 10 Bank Holding Companies
|Bank Name||Market Share||Bank Name||Market Share||Bank Name||Market Share|
|Bank of America||10.65%||Bank of America||9.16%||Bankamerica||3.74%|
|JPMorgan Chase||10.23%||JPMorgan Chase||7.17%||Nationsbank||2.70%|
|Wells Fargo||10.15%||Wells Fargo||4.79%||Chemical Bank||2.22%|
|Capital One||2.55%||SunTrust||1.83%||First Union||1.63%|
|TD Bank||2.11%||U.S. Bank||1.82%||First Intrst||1.49%|
|PNC||2.10%||Royal Bank of Scotland||1.67%||Keycorp||1.40%|
|Charles Schwab||1.63%||BB&T||1.21%||Wells Fargo||1.29%|
|Morgan Stanley||1.40%||National City||1.18%||Chase Manhattan||1.17%|
Source: FDIC, Summary of Deposits, Q2 2019.
Profit Rates in U.S. Banking
Philippon also points to the share of corporate profits in GDP as evidence of growing firm market power. He writes that:
over fifty years the profit share remain[ed] stable . . . around 6 or 7 percent, from the end of World War II to the end of the twentieth century. Over the past two decades, however, . . . the after-tax profit share has increased to around 10 percent. (p. 54)
But here again, as in the case of market concentration, the banking sector does not mirror the rise in profit rates apparent in other parts of the U.S. economy. Banks’ return on assets (ROA) and return on equity (ROE), two commonly used measures of their profitability, have remained remarkably stable over the last three decades (Table 3). Between 1992 and 2006, the average ROA of FDIC-insured banks and thrifts was 1.18 percent. From 2011 until the third quarter of 2019, it averaged 1.07 percent. For ROE, the respective figures are 13.54 percent and 8.8 percent. Banks of all sizes have returned less on shareholders’ capital since the end of the financial crisis than before it. Given the proliferation of new regulations, including stiffer capital requirements, after 2010, that should not come as a surprise. Post-crisis regulations may also have reduced bank risk, thereby lowering banks’ required return, but that relationship itself would suggest competitive conditions prevail in banking.
Table 3. Average Return on Assets (ROA) and Return on Equity (ROE), by Asset Size
|Asset Size Group||ROA||ROE||ROA||ROE|
|Assets > $250 Billion||1.06||14.21||1.03||9.90|
|Assets $10 Billion – $250 Billion||1.21||14.27||1.14||9.22|
|Assets $1 Billion – $10 Billion||1.27||13.84||1.10||9.30|
|Assets $100 Million – $1 Billion||1.14||11.90||0.98||8.79|
|Assets < $100 Million||1.02||9.38||0.80||6.36|
|All Insured Institutions||1.18||13.54||1.07||9.48|
Source: FDIC, Quarterly Banking Profile, Q3 2019.
Other measures of bank earnings also contradict Philippon’s claim of growing market power. Listed banks’ average price-to-book-value ratio, a proxy for the market’s confidence in a firm’s future profitability, is 1.1—a figure well below the market-wide average of 2.66. For the largest U.S. retail banks by assets, which presumably have been best able to adapt to the post-crisis environment thanks to their scale and too-big-to-fail status, price-to-book measures are much lower than before 2008, with the sole exception of JPMorgan Chase (Table 4).
Table 4. Ratio of Bank Stock Prices to Book Values, 2006 and 2019
|Bank Name||Price-to-Book Ratio, 2006||Price-to-Book Ratio, 2019|
Philippon himself allows that value-added (profits plus wages) in the financial services sector as a share of financial assets has stayed constant, rather than increased, for 130 years. According to his own calculation, the financial services industry’s share of income in intermediated assets declined from around 1.8 percent to around 1.4 percent between the mid-1990s and the mid-2010s, and is now where it was in the mid-1970s. (Philippon’s original paper only goes up to 2012, but in his book he extends the series to 2016. The unit cost of intermediation stayed at late-1970s levels long after the recovery from the financial crisis began.)
Table 5. Philippon’s Estimate of the Unit Cost of Financial Intermediation, 1886-2012
Source: Thomas Philippon, “Has the U.S. Finance Industry Become Less Efficient? On the Theory and Measurement of Financial Intermediation,” American Economic Review Vol. 105 No. 4 (April 2015), 1412. Author’s edit in red.
Bank profitability has been a headache for bank shareholders and managers for some years, but it should not (for now) worry antitrust regulators.
Bank Entry Has Declined, but Not Because of Market Power
Philippon also worries that low rates of entry, and high rates of incumbency, are another symptom that the market power of U.S. firms has increased, to the detriment of consumer welfare. And there has indeed been little entry into banking since the financial crisis. However, the causes of this appear to be quite different from those Philippon posits—namely, low interest rates and much tighter chartering rules by financial regulators.
Low interest rates since 2008 have squeezed bank profits by narrowing the margin that banks earn from lending funds borrowed from depositors. The net interest margin is especially significant for small and new banks because, unlike larger banks, they have few other sources of revenue. Researchers at the Richmond Fed attribute 75 percent of the decline in new (de novo) bank charters in the years after the crisis to low interest rates.
The FDIC also made it costlier to start and operate a new bank between 2009 and 2016. During that time, de novo banks were subject to higher capital requirements and more frequent examinations for their first seven years of operation, rather than the three years previously provided for. (The FDIC rescinded the seven-year rule in 2016.) Higher capital requirements, other things equal, mean a lower return to shareholders per dollar of loans, while more frequent examinations slow down business and raise compliance costs. While this extra caution may have been justified, it also impeded entry into banking.
Why Is the Unit Cost of Financial Intermediation Stable?
Philippon’s 2015 AER article finding that the unit cost of financial intermediation has remained stable for more than a century has received a great deal of scholarly and media attention. The finding puzzles those who would assume information technology, globalization, and increasing competition put downward pressure on the prices financial intermediaries can charge. It also contradicts the anecdotal evidence from retail asset managers, whose fees have dropped steadily with the advent of online trading platforms and passively managed funds.
Philippon’s methods are painstaking. His adjustments for changes in the risk associated with financial intermediation, as startups and lower-income households gained access to capital, are particularly impressive (Table 6). They suggest a moderate decline in costs since the 1990s but no long-term trend of decline.
Table 6. Quality Adjustments Lower the Unit Cost of Intermediation in Recent Years
Source: Philippon, “Has U.S. Finance Become Less Efficient?,” 1433.
My concern is with Philippon’s interpretation of his findings. In particular, he dismisses regulation as a countervailing force keeping costs from declining. “Regulatory barriers to entry,” he says, “have been reduced in banking since the 1970s and yet this is when the unit cost goes up.” In his book, Philippon does not dismiss regulation quite so laconically. But there he relates rising compensation for financial sector workers to a supposed deregulation that began in 1980. He adds: “When an industry is deregulated, wages and prices usually fall. In finance, they seem to rise.” (p. 214)
But there’s much more to regulation in banking than rent-seeking by financial firms, which may have increased in the last few decades, and branching and investment-banking restrictions, which disappeared in the 1990s. Since 1970, regulations on credit intermediation have increased rapidly. According to the Mercatus Center at George Mason University, credit intermediaries are subject to six times as many mandates and restrictions as they were 50 years ago. Rules on securities, derivatives, and fund managers have also mounted since 2000. The Dodd-Frank Act alone added more than 27,000 new regulations to the books. Other proxies for bank regulatory compliance costs, such as non-interest expenses for salaries, legal fees, and other activities have increased significantly since the financial crisis.
The basic problem with Philippon’s findings on the unit cost of financial intermediation is that he does not break costs down by type. Financial intermediaries have operating (personnel and capital) as well as financial (cost of capital) expenses. They also face significant costs of compliance with government regulation. Technological innovation can particularly lower operating costs, though better information processing and tech-enabled competition may also reduce funding costs. But innovation has a harder time lowering regulatory costs, because regulatory requirements tend to be fixed and rigid. Indeed, regulation often acts as a barrier to innovation.
What Philippon Gets Right
Although the evidence suggests that the thesis of The Great Reversal does not hold in the case of banking, Philippon is right to claim that competition in this sector is less than it might be. As Table 2 illustrates, any examination of bank rankings by deposits and assets cannot fail to show a remarkable persistence among the largest institutions. Where other sectors, such as retail, telecommunications, and even the auto industry have seen significant turnover in the last quarter-century, the leading names in banking as of 2019 would ring familiar to anyone living in 2006 and even 1992 (if not 1907!).
Here again, regulation seems to be the chief contributor to this state of affairs. The dominant firms in other sectors tend to have entered the market relatively recently. In banking, entry since 2008 has been minimal. Furthermore, regulation has encouraged consolidation because many of the costs of regulatory compliance are fixed. Finally, and more controversially, the government’s readiness to rescue large financial institutions encourages size and promotes incumbency. Citigroup would not be around today were it not for Uncle Sam.
Philippon is also right to complain about America’s unusual separation of commerce and banking, passed into law by the 1956 Bank Holding Company Act. Not only is there little logic to this legal prohibition, “as if debit cards and savings accounts were magical products that a retail firm could not provide,” (p. 216) but it helps to entrench the largest financial institutions in their dominant positions by preventing similarly sized outside firms from challenging them. When Walmart applied for an industrial loan company (ILC) charter, the one banking charter that nonfinancial companies are allowed to hold, in 2005, resistance from banks was so fierce that Walmart eventually relented two years later. Now that Japanese e-commerce and financial services firm Rakuten has applied for an ILC charter, banks large and small have again responded with vociferous opposition.
America’s strange segregation of commercial and financial companies is also bad for financial inclusion. Contrary to the popular understanding, the reason millions of Americans lack a bank account is not that they are far away from branches, but that they find bank accounts expensive and do not trust banks. But many of these Americans recognize and regularly patronize businesses like Walmart and other retailers. Moreover, some of these businesses have specialized precisely in serving people on tight budgets. Allowing commercial companies to provide a range of basic financial services could therefore help to reduce America’s unbanked problem.
We Need a Great Reversal in U.S. Banking
Philippon’s book has rightly gained a lot of attention since its publication. He raises important questions about trends in corporate concentration and market power, and the government’s role in them, in recent decades. But U.S. banking has different problems. Regulatory accumulation has made financial intermediation costlier. America’s longstanding fragmentation due to restrictions on mergers and branching means many banks are inefficiently small. The post-crisis environment of low interest rates and increased compliance costs have made banks significantly less profitable. Regulators are not helping to increase competition when they block entry. The best legislators can do to promote efficiency and innovation in banking is tear down the wall between commerce and finance.
These changes will constitute not just a great, but a long-overdue reversal in the direction of better financial markets in America.