The second call

A share is usually described as a claim on profits with a floor under the losses: the most you can lose is what you put in. For most of the nineteenth century a bank share meant something else. It carried a second call, an obligation to pay more if the bank failed. Two versions ran side by side, an unlimited one in Scotland and a capped one in the United States, and both were dismantled for opposite stated reasons, neither of which survives inspection. What was being settled was not how much liability a shareholder should carry but whose, and the answer finally arrived at is a weaker version of one already on the table in 1879.

In October 1878 the City of Glasgow Bank failed with losses more than six times its capital. Its 1,819 shareholders, on most accounts, held under unlimited liability: they could be called for any multiple of what they had subscribed.¹² They were called. Every depositor and creditor was paid in full, and only because the liability was unlimited; when the liquidation ended, 254 shareholders were still solvent.¹³ The instrument did what it was designed to do, and within a year Parliament had acted, though not by abolishing it. The Companies Act 1879 added a halfway house, reserve liability, under which the uncalled part of a share could be called only in a failure and a defined multiple of it held in reserve. The second call was capped rather than removed, and pure limited liability in the modern sense became general only decades later.¹

The reason given was that the failure had exposed how far bank ownership had spread to people of modest means, ruined overnight. Graeme Acheson and John Turner went to the ownership records, the trading data and the share prices, and found no support for it, or for the companion claim that shareholders dumped their shares as the crisis broke. Most owners were relatively wealthy, which is exactly why depositors could be paid in full.⁴ The newspapers had written a story about widows, and Parliament legislated on the story.

The United States ran the capped version. Under the National Bank Act, modelled on New York's statutes, a failed national bank's shareholder faced an assessment by the receiver of up to the par value of the stock. The usual name, double liability, is imprecise: the ceiling was par even for shares bought at a premium.⁵⁶ The regime ran from 1865 to 1937 and generated nearly fifty Supreme Court decisions.⁵ Receivers collected about 95 million dollars, reducing losses depositors would otherwise have borne.⁶ Reformers nonetheless retired it on three grounds: that it had failed to protect creditors, that it did not maintain public confidence, and that deposit insurance was preferable.⁵ The Banking Act of 1933 permitted shares free of the liability, and the 1935 Act let banks drop it altogether.⁶

Jonathan Macey and Geoffrey Miller read that abolition as political rather than economically optimal, driven by the hardship of assessing shareholders already in difficulty in the Depression.⁵ Set the two episodes beside each other and the pattern is one. In Scotland the instrument was given up because it worked and the working hurt. In America it was abolished because enforcement hurt at the worst moment and a public guarantee was available to replace it. Neither abolition turned on a finding that liability beyond the subscription failed to absorb losses. Both turned on who got hurt when it did.

That is a different question, and somebody asked it at the time. Charles Goodhart and Natacha Postel-Vinay argue that the Glasgow shareholders were outsiders, much as most bank shareholders are today, with neither the information nor the control that banking decisions require, so limiting their liability made sense on moral-hazard grounds and on grounds of justice. The executive directors had both, and limiting theirs did not. An amendment retaining extended liability for bank directors was put forward and refused outright, on the argument that a heavy liability would deter capable men from taking the job; the authors hold that reason overstated at the time and since.⁷ A tiered answer was available in 1879 and it lost.

Why it keeps losing is more useful than the regret. In both regimes the liability was written into the share, and a share is fungible: its price depends on the identity of its holder being irrelevant. Attach a heavier liability to a director's shares and you have a second class of security that cannot trade at the first's price, which is inconvenient when directors are paid in it. Tiered liability therefore has to attach to the office rather than the holding, and that is the form in which it has come back. Britain's Senior Managers and Certification Regime assigns named responsibilities to named individuals; it is being streamlined through 2026, with the certification element to be lifted out of statute and rebuilt in regulators' rules.⁸ American proposals since the bank failures of 2023 reach for pay rather than capital. The RECOUP Act cleared the Senate Banking Committee by twenty-one votes to two in June 2023 without becoming law; its successor, the Failed Bank Executives Clawback Act, was reintroduced on 11 March 2026 and stands on the record only as introduced.⁹

The difference between a clawback and a call is the ceiling. A clawback is bounded by what the executive has been paid, which beside the losses of a large bank is small; a call on personal wealth is bounded only by the wealth. That is the attraction, and where the objections start. The first is empirical and awkward. Richard Grossman found banks in states with extended liability failing less often and holding higher capital and liquidity ratios, the evidence weaker for the 1920s; he and Macey and Miller also found extended liability associated with higher measured leverage, creditors being reassured they would be made whole.¹⁰¹¹ If so, the instrument partly substituted for equity instead of adding to it, and bought reassurance rather than safety. The second objection is that unbounded exposure genuinely deters, the Victorian argument Goodhart and Postel-Vinay call overstated and cannot show to be empty. The third is that modern ownership is intermediated: to call a pension fund is to call its beneficiaries, the diffusion the Victorians wrongly feared in 1878 and would be real now.¹²

The question under all of this has three answers. A bank's losses beyond its capital can be charged to shareholders beyond their stake, to depositors, or to the public. The twentieth century chose the public and charged premiums to the second, and the choice is now so familiar the discarded alternative looks crude. It was not crude. It was aimed at the wrong people, and the aim was correctable in 1879 by an amendment that was refused. What returned in its place has the right target and a ceiling, and the ceiling is no novelty: reserve liability carried one from 1879. What matters is what it is a fraction of. Shareholder capital of every kind covered about seventy per cent of British bank deposits in 1885 and fifteen by 1921.¹

Notes

1Charles Goodhart and Natacha Postel-Vinay, "The case for tiered liability: evidence from the City of Glasgow Bank failure," Financial History Review (2026), pp. 1-26, for the losses exceeding six times the bank's capital, for the 1,819 shareholders holding under unlimited liability and able to be called for an unlimited multiple of subscribed capital, for depositors and creditors being repaid in full, for only 254 shareholders remaining solvent after liquidation, for the 1879 Companies Act giving banks a choice between unlimited liability, limited liability and reserve liability rather than imposing limited liability, for reserve liability being a halfway house under which the uncalled portion could be drawn on only in a failure and a defined multiple of it kept in reserve, for the Act laying the groundwork for the widespread adoption of pure limited liability only decades later, and for shareholder capital of every kind covering around 70 per cent of aggregate deposits in 1885 against 15 per cent by 1921, which the authors take from Turner (2014), p. 128. Read in the LSE Research Online copy. Publisher page and the LSE Research Online copy

2For the failure of the City of Glasgow Bank in October 1878 and its aftermath, including the prosecution of its directors. Bank of England Quarterly Bulletin, "Desperate adventures and men of straw: the failure of City of Glasgow Bank" (2015) and a museum account of the failure

3Sources disagree on the number of shareholders. Most studies give 1,819; a widely used reference account gives about 1,200. The figure of 254 remaining solvent is common to both, which means the proportion ruined differs between accounts even though the count of survivors does not, and nothing in the argument here turns on which total is correct. Reference account giving the lower figure and the study giving 1,819

4Graeme G. Acheson and John D. Turner, "The death blow to unlimited liability in Victorian Britain: the City of Glasgow failure," Explorations in Economic History (2008), for the use of ownership records, trading data and stock prices to test the beliefs that ownership of large joint-stock banks had been diffused to investors of very modest means and that the failure caused bank shareholders to dump their shares, and for the authors' verdict that the evidence offers no support for these widely held beliefs. The related judgement that most shareholders were relatively wealthy, which is what allowed depositors to be repaid in full, is reported in later work citing this paper. Not opened. Publisher page and the University of Strathclyde record

5Jonathan R. Macey and Geoffrey P. Miller, "Double Liability of Bank Shareholders: History and Implications," Wake Forest Law Review 27 (1992), pp. 31-62, for the description of the regime from the Civil War to the Great Depression, for the receiver's power to assess shareholders up to the par value of their stock, for the observation that the label is imprecise because assessments were capped at par even where shares had been bought at a premium, for the nearly fifty decisions of the United States Supreme Court and hundreds more in the state and lower federal courts, for the three grounds on which reformers rejected the system, and for the authors' reading of abolition as a political decision that was not necessarily economically optimal. Not opened. Yale EliScholar copy and a second repository copy

6Roger Tufts and Graham Tufts, "Shareholder Double Liability and Depositor Losses in Failed National Banks," Office of the Comptroller of the Currency, Moments in History (November 2021), for double liability running from 1865 to 1937, for New York's laws being the model for the provision in the National Bank Act, for receiverships collecting 95 million dollars in assessments and thereby decreasing losses that would otherwise have been borne by depositors, and for the 1933 Banking Act permitting the issue of shares carrying no double liability and the 1935 Act allowing banks to drop the provision. The OCC's own catalogue entry dates the study 1865 to 1935 while its text runs to 1937, a discrepancy the sources do not resolve. Not opened. OCC landing page and the study itself

7Goodhart and Postel-Vinay, as above, for the argument that the bank's shareholders could be considered outsiders like most of today's bank shareholders, that the executive directors were presumably those with both fuller information and the necessary control powers to make banking decisions, that limiting liability in the same way for both groups made sense from a moral hazard point of view and from the point of view of justice only for most shareholders and not for executives, that Joseph McKenna's amendment, which would have kept directors on unlimited liability until a year after they left office, was rejected at once by the Chancellor of the Exchequer on the objection that it would discourage men who would make desirable directors from incurring the responsibility, and that the reasons for rejecting tiered liability for insiders were overstated both then and subsequently. Read in the LSE Research Online copy. Publisher page and the LSE Research Online copy

8For the Senior Managers and Certification Regime remaining in force and being reformed in phases during 2026: the Financial Conduct Authority and Prudential Regulation Authority confirmed the first phase in April 2026, with most changes effective 24 April 2026, further changes on 10 July 2026 and the non-financial misconduct provisions on 1 September 2026, the regulators stating that the core principle of individual accountability is unchanged; and HM Treasury confirmed its intention to remove the Certification Regime from the Financial Services and Markets Act 2000 and empower the regulators to make replacement rules. Bank of England announcement and a law firm's account of the first phase

9For the Recovering Executive Compensation from Unaccountable Practices Act being voted out of the Senate Banking Committee by 21 votes to 2 in June 2023 and not becoming law, and for the Failed Bank Executives Clawback Act, S. 4050, being reintroduced by a bipartisan group of senators on 11 March 2026, with the Government Publishing Office record showing it as introduced and no later action. The bill would require the Federal Deposit Insurance Corporation to claw back compensation received in the three years before a bank's failure. I could not confirm any committee action on S. 4050 after introduction. Report of the 2023 committee vote, the bill record and a report of the 2026 reintroduction

10Richard S. Grossman, "Double Liability and Bank Risk Taking," Journal of Money, Credit and Banking 33, no. 2 (2001), pp. 143-159, for the findings that banks in states with contingent liability had lower failure rates and higher capital and liquidity ratios than banks under limited liability, and that the evidence is weaker for the 1920s. No reachable copy of the paper itself could be found, so the findings are taken from a summary that reports them; the figures are therefore second-hand. A summary reporting Grossman's findings

11For the finding, attributed to Grossman and to Macey and Miller, that double liability was associated with higher measured bank leverage, explained by double liability reassuring creditors that they would be made whole in the event of failure, and for the judgement that the wider literature provides mixed evidence on whether double liability constrained risk taking. Office of Financial Research working paper 18-06, "Reducing Moral Hazard at the Expense of Market Discipline" and a summary of the mixed evidence

12Felipe Aldunate and Dirk Jenter, "Shareholder Liability and Bank Failure," for the finding that double liability reduced bank failures and that it was quickly repealed after 1933; and a study of bankers' liability and risk taking in New England between 1867 and 1880, for the wider evidence on extended liability and bank behaviour. Neither was opened. Conference listing for Aldunate and Jenter, the New England working paper and a column summarising the latter