ONLY pop music and pornography embraced globalisation more keenly than banks did. Since the 1990s three kinds of international firm have emerged. Investment banks such as Goldman Sachs deal in securities and cater to the rich from a handful of financial hubs such as Hong Kong and Singapore.

A few banks, such as Spain’s Santander, have “gone native”, establishing a deep retail-banking presence in multiple countries. But the most popular approach is the “global network bank”: a jack of all trades, lending to and shifting money for multinationals in scores of countries, and in some places acting like a universal bank doing everything from bond-trading to car loans. The names of the biggest half-dozen such firms adorn skyscrapers all over the world.

This model of the global bank had a reasonable crisis in 2008-09: only Citigroup required a full-scale bail out. Yet it is now in deep trouble. In recent weeks Jamie Dimon, the boss of JPMorgan Chase, has been forced to field questions about breaking up his bank. Stuart Gulliver, the head of HSBC, has abandoned the financial targets that he set upon taking the job in 2011. Citigroup is awaiting the results of its annual exam from the Federal Reserve. If it fails, calls for a mercy killing will be deafening (see next story). Deutsche Bank is likely to shrink further. Standard Chartered, which operates in Asia, Africa and the Middle East, is parting company with its longstanding boss, Peter Sands.

Domestic lenders that global banks have long sneered at are doing far better. In Britain, Lloyds has recovered smartly over the past two years. In America the most highly rated banks—based on their share price relative to their book value—are Wells Fargo and a host of midsized firms.

The panic about global banks reflects their weak recent results: in aggregate the five firms mentioned above reported a return on equity of just 6% last year. Only JPMorgan Chase did passably well (see chart). Investors worry these figures betray a deeper strategic problem.

There is a growing fear that the costs of global reach—in terms of regulation and complexity—exceed the potential benefits.

It all seemed far rosier 20 years ago. Back then banks saw that globalisation would lead to an explosion in trade and capital flows. A handful of firms sought to capture that growth. Most had inherited skeleton global networks of some kind. European lenders such as BNP Paribas and Deutsche Bank had been active abroad for over a century. HSBC and Standard Chartered were bankers to the British empire. Citigroup embarked on a big international expansion a century ago; Chase, now part of JPMorgan Chase, opened many foreign branches in the 1960s and 1970s.

As they expanded in the 1990s and 2000s all of these firms concentrated on multinationals, which required things like trade finance, currency trading and cash management. But all expanded beyond these activities to varying degrees and in different directions; today they typically account for only a quarter of sales. Deutsche and StanChart bulked up in investment banking. BNP built up retail operations in America. At the most extreme end of the spectrum Citi and HSBC tried to do everything for everyone everywhere, through lots of acquisitions.

They sold derivatives in Delhi and originated subprime debt in Detroit.

This model is in trouble for three reasons. First, these giant firms proved hard to manage.

Their subsidiaries struggled to build common IT systems, let alone establish a common culture.

Synergies have been elusive and global banks’ cost-to-income ratios, bloated by the costs of being in lots of countries, have rarely been better than those of local banks. As a result these firms have all too often been tempted to make a fast buck. Citi made a kamikaze excursion into mortgage-backed bonds in 2005-08. StanChart made loans to indebted Asian tycoons.

Second, competition proved to be fiercer than expected. The banking bubble in the 2000s led second-rate firms such as Barclays, Société Générale, ABN Amro and Royal Bank of Scotland (RBS) to expand globally, eroding margins. In 2007 RBS bought ABN in a bid to rival the big network banks.

It promptly went bust, proving that two dogs do not make a tiger. The global giants also lost market share in Asia to so-called “super-regional” banks, such as ANZ of Australia and DBS of Singapore. Big local banks in emerging markets, such as ICBC in China, Itaú in Brazil and ICICI in India, also began to build out cross-border operations.

If mismanagement and fierce competition were problems before the crisis, the regulatory backlash after it has been brutal. American officials have begun to enforce strict rules on money-laundering, tax evasion and sanctions, meaning that global banks must know their customers, and their customers’ customers, if they want to maintain access to America’s financial system—which is essential given that the dollar is the world’s reserve currency. Huge fines have been imposed on StanChart, BNP and HSBC, among others, for breaking these rules.

Bank supervisors, meanwhile, have imposed higher capital standards on global banks. Most face both the international “Basel 3” regime and a hotch-potch of local and regional regimes. A rule of thumb is that big global banks will need buffers of equity (or “core tier one capital”) equivalent to 12-13% of their risk-adjusted assets, compared to about 10% for domestic firms.

National regulators increasingly demand that global banks ring-fence their local operations, limiting their ability to shift capital around the world. The cost of operating the systems that keep regulators happy is huge. HSBC’s compliance costs rose to $2.4 billion in 2014, 50% higher than the year before. JPMorgan is spending $3 billion more on controls than it did in 2011.

One measure of these firms’ viability is their “best case” return on equity (ie, assuming that the huge, supposedly, one-off legal fines and restructuring costs incurred over the past half-decade suddenly stop, but the new capital standards are fully implemented). On this basis most global banks barely achieve 10% (see prior chart). The overall figures hide lots of rot. After three decades of trying to diversify from its base in Asia, HSBC still makes the bulk of its money there; the other two-thirds of its business underperforms badly for the most part. JPMorgan Chase’s profits are more evenly spread, but about two-thirds of its businesses fail to cross the 10% hurdle. The same is true of StanChart.

From the outside—and perhaps from the inside, too—Citi’s reporting system is too crude to make any sensible judgment. Deutsche looks better than most, but many of its rivals question its capital calculations.

Another test of viability is to compare the benefits of being global with the costs. In February JPMorgan Chase said that the revenue uplift and cost savings it gets from its scale boost profits by $6 billion-7 billion a year. There is a plausible case that the extra capital it must carry, and the regulatory costs its complexity incurs, offset a big chunk of that benefit. (If they dared to reveal them, the figures for other banks would probably look far worse.) Scale does not seem to mean cheaper funding. It is no cheaper to buy a credit-default swap, which pays out if a borrower goes bust and so is a reasonable proxy for borrowing costs, on the debt of JPMorgan Chase or Citi than it is on that of mid-sized American banks. All are regarded by debt investors as government-backed.

The financial arguments for global banks no longer appear convincing. Yet unscrambling these firms would be hellish. And in any case both managers and investors see two possible rays of light. One is gradually rising interest rates in America: JPMorgan Chase reckons these might add a fifth to its profits by 2017. The other is declining competition, which would allow them to raise their prices. The withdrawal of second-tier banks should help—on February 26th RBS said it would shrink its commercial and investment-banking operations down to 13 countries from over 50 at the peak of its pomp in 2008.

Yet there are always new competitors to push down margins. Japanese banks are on a cross-border lending bender for the first time since the 1980s. China’s banks are steadily expanding.

The Western network banks were right to assume that globalisation would lead to a big increase in the amounts of money sloshing around the world. They have yet to work out how to prosper from it.