The Resolution of a Systemically Important Insurance Company during the Great Depression

Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, D.C. The Resolution of a Systemically Important Insurance Company during the Great Depression Jonathan D. Rose 2016-005 Please cite this paper as: Rose, Jonathan D. (2016). “The Resolution of a Systemically Important Insurance Company during the Great Depression,” Finance and Economics Discussion Series 2016-005. Washington: Board of Governors of the Federal Reserve System, http://dx.doi.org/10.17016/FEDS.2016.005. NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

The Resolution of a Systemically Important Insurance Company during the Great Depression Jonathan D. Rose* December 7, 2015 Abstract This paper explores the economic issues related to systemically important insurance companies, using an example from the Great Depression, the National Surety Company. National Surety was a large and diverse insurance company that experienced a major crisis in 1933 due to losses from its guarantees of mortgage-backed securities. A liquidity crisis ensued, as policyholders staged a massive run on the company, demanding the return of their unearned premiums. The New York State Insurance Commissioner stepped in with a reorganization plan that split the company in two, out of fear that a disorderly liquidation would have systemic consequences given the sheer number of the company’s counterparties, scattered all across the United States. A key dynamic of the crisis was that policy holders at an insurance company have a dual role as holders of liabilities and as providers of income. * Federal Reserve Board of Governors. Email: jonathan.d.rose@frb.gov. The views presented in this paper are solely those of the author and do not necessarily represent the views of anyone else associated with the Federal Reserve System. 1

1. Introduction The crisis at the insurance company AIG during 2008 shined a spotlight on the role of insurance companies in the economy, and the potential harm that could occur if a large insurance company were to fail in a disorderly manner. The episode raised many questions about how to regulate such companies and resolve one if it were to experience a crisis. With this experience as motivation, the 2010 Dodd-Frank Act empowered the federal government to regulate systemically important non-bank institutions, including insurance companies. This in turn has further heightened the demand for research on these institutions. Yet, while scholars have studied the banking history of the US in great detail, much less has been written about the history of the insurance industry. This paper uses an historical example from the 1930s, the National Surety Company, to explore the economic issues related to systemically important insurance companies. National Surety’s experience bears a remarkable resemblance to AIG’s seventy-five years later: both companies operated large and diverse insurance businesses which were generally profitable, but both experienced major crises when their guarantees of mortgage backed securities (MBS) soured following a decline in the real estate market and revelations of badly underwritten mortgages. Both AIG and National Surety were spared from disorderly resolution by government intervention. The paper has three main contributions. First, National Surety’s history shows how an insurance company could be considered systemically important, using the modern nomenclature. When National Surety became unable to meet its obligations in April 1933, voices in government and finance spared few adjectives in describing their potential adverse consequences of National Surety’s failure. Various commentators expressed fears that it would be “catastrophic,” “tragic,” “calamitous,” “disastrous,” and the “cause of immeasurable suffering.” The key factor was the sheer size of the National Surety Company, with policies written for vast numbers of policyholders. In addition, many of its counterparties were fiduciaries and governments, societal actors which are often given special protection as a matter of public policy. Lastly, its connections to other insurance companies through reinsurance agreements, and its status as the largest surety company in the world, raised fears about its failure’s particular effect on the insurance sector. 2

As a second contribution, the paper shows how a liquidity crisis could and did develop at an insurance company. Banks are often used as a benchmark for thinking about liquidity crises. Though insurance companies are different than banks and do not owe demand deposits, they do have obligations to repay paid-in but unearned premiums when requested by policyholders. Such repayments often come at a penalty or with a delay for the policyholders, but nevertheless they represent a source of potential cash outflows for insurance companies. National Surety’s case demonstrates a run developing through this mechanism in practice. In the company’s last several weeks, an avalanche of policy cancellations led to a large cash outflow. This was the proximate cause ultimately forcing the firm into resolution. The longer-term roots of the crisis at National Surety were grounded largely in one line of its insurance business, the guaranteeing of mortgage backed securities—one part of a wave of such securitization that took place in the 1920s. When those securities soured in unexpectedly large numbers in the late 1920s and early 1930s, the losses threatened National Surety’s capital buffer and devoured its liquidity reserves. During 1932 and 1933, National Surety then faced two additional large shocks to its solvency and its liquidity: the collapse of the corporate bond market, in which National had invested most of its assets, and the banking holiday in March 1933, which interrupted the company’s normal flow of funds. The egress of policy holders and the accompanying return of unearned premiums was the final shock. The third contribution of this paper is to highlight a few key dynamics that shape any effort to arrange for the orderly resolution of a systemically important insurance company. Of primary importance is the dual position of policy holders as sources of income and as holders of liabilities in the form of unearned premiums. As a result, losing the confidence of those policy holders might not just cause outflows of cash as those policy holders demand the return of unearned premiums, but also reduce cash inflows as they cease paying premiums after having cancelled their policies. Negotiating with policy holders of an insurance company therefore carries an added danger compared to, for example, negotiating with depositors at a bank, as the latter may create a cash outflow if they lose confidence in the bank, but would not directly affect cash inflows. Indeed, the main strategic goal of government officials was to preserve the good will of policy holders and therefore the going concern value of the company for the sake of creditors. 3

With the goal of preserving National Surety’s going concern value, the resolution strategy provides an interesting model for resolving a systemically important insurance company in an orderly fashion. Since National Surety could not meet its obligations, the New York insurance commissioner believed its only available choices by law were to liquidate or rehabilitate, but the rehabilitation options were limited, as the company was deemed too big to reinsure, and fresh capital could not be arranged. Instead, the rehabilitation involved splitting the company in two: liability for existing losses was left in the old company which was liquidated slowly, while liability for future losses was placed in a new company. In addition, the new company bore liability on future losses only for a subset of the old company’s business, mainly excluding the MBS guarantees. A last component of the rehabilitation strategy was the clear communication of the plan at the time of the company’s takeover, with the goal of preserving the good will of policy holders of the new company. All of this contrasted with the less successful rehabilitation of the Globe and Rutgers Fire Insurance Company, which had been taken over by insurance regulators a month prior to National Surety’s takeover. The Globe and Rutgers case serves as an interesting contrast by demonstrating the pitfalls of attempting to negotiate with an insurance company’s policy holders, delaying communication, and not preserving the firm’s going concern value. This paper is largely based on a valuable cache of documents recovered from the New York State Supreme Court at its Manhattan location (referred to throughout the paper as the National Surety Papers). These court filings contain a wealth of information about the historical business practices of the company, as the New York Insurance Commissioner was required to justify the takeover of National Surety to the court. Supplemented by information from the archives of the Reconstruction Finance Corporation, insurance industry periodicals, and other sources, this paper pulls together a fairly complete picture of how National Surety was perceived by government officials, insurance industry peers, and the general public. 2. Background National Surety was, naturally enough, mainly in the business of surety, one form of insurance. As a general matter, surety encompasses any guarantee that an undertaking will be fulfilled. That broad definition encompasses many different types of guarantees, simple forms of which have been practiced for millennia. Indeed, scholars of history have found descriptions of 4

such guarantees as far back in antiquity as circa 2750 BC. In that instance, a farmer leased out his fields in return for a share of the proceeds, and a third party guaranteed the performance of the lessee (Morgan 1927, p. 153). With so deep a history, surety is sometimes described as the oldest form of insurance (although it is common to see that designation given to marine insurance instead). Today, surety contracts are generally grouped into two types. Agreements in which a surety company guarantees the completion of a contract are known as contract surety agreements. A common example of a contract surety agreement is a guarantee of a construction project. In such an agreement, a construction firm pays premiums to a surety company which will, if the firm fails to complete its project, pay out the amount of the bond to the party that commissioned the construction. The second type of surety, involving the bonding of individuals or firms who are required to follow certain laws, is known as commercial surety. This type of insurance might cover those with fiduciary responsibilities such as trustees, administrators, or executors, or also cover public officials, or people involved in court proceedings. National Surety was active in both types of surety, as well as in other forms of insurance. As an example of its contract surety business, it underwrote a construction bond that guaranteed the performance of the companies that constructed the Hoover Dam in 1931.2 The company wrote a large number of blanket and schedule bonds, covering financial institutions and public entities against criminal acts by their employees. It also conducted large amounts of insurance for bank deposits, plate glass, burglary, as well as some insurance related to forgery, fraud, crime, and merchant’s protection bonds.3 During the Depression, some of National Surety’s businesses performed well, but others required large cash outflows. The volume of contract surety business diminished after 1929, as such activity naturally ebbs and flows with the level of macroeconomic activity that generates such contracts. Though such business declined and therefore generated less premium revenue, it did not produce outsized losses. The lines that did perform especially badly during the 2 “Hoover Dam Bonds Written,” New York Times, March 8, 1931, p. 44. Since the Hoover Dam was a very large project, National Surety formed and headed a coalition of surety companies that collectively guaranteed the bonds. 3 The description of National Surety’s business lines is gathered from Best’s Insurance Reports: Casualty and Miscelanneous, editions from 1920s and early 1930s up to 1932; page 6 of Memorandum by Edward McLouglin, Attorney for George S. Van Schaick, Superintendent of Insurance of the State of New York, in folder 1 of the National Surety Papers; and Papers of the Reconstruction Finance Corporation, Loan Application #1 of the National Surety Company. 5

Depression, creating large losses for National Surety, were mortgage-backed security (MBS) guarantees, depository bonds and workers’ compensation claims. Depository bonds, which guaranteed the safety of bank deposits, especially for municipal depositors, performed quite badly given the number of bank failures in the early 1930s.4 However, the most damaging line of business for National Surety was the guaranteeing of MBS. More than anything else, this business directly caused the liquidity and solvency crisis that forced National Surety’s resolution in 1933. National Surety’s MBS guarantees Among the company’s various lines of business, this was a relatively young one, having been initiated only in 1923, after the New York State Attorney General issued an opinion in late 1922 that surety companies possessed the power under existing law to guarantee MBS.5 National Surety’s business of guaranteeing MBS boomed from 1923 to 1928. It was part of and interacted with a surge in the demand for mortgage credit, a wave of MBS issuance (one of several such waves in US history), and the national construction boom, all of which occurred in the mid 1920s. The MBS guarantees also represents one of many financial innovations that date to the 1920s, especially in the market for retail and institutional investments. National Surety’s role as guarantor appears to have reduced the perceived credit risk on these MBS, both directly through the guarantee and indirectly through the information monitoring that investors believed National Surety to have conducted. For example, the advertisement reprinted in Figure 1 shows the way in which National Surety’s endorsement was used to sell MBS. In terms of the magnitude of the business, National Surety in total guaranteed 80 different MBS series issued by about 30 mortgage companies. Total guarantees peaked at about $73 million in December 1928, representing about 0.4 percent of all residential and commercial mortgage loans in the country, and equaling about 2.5 percent of the total notional value of guarantees made by National Surety in all of its businesses. The MBS contained about 11,000 4 Depository bonds are discussed in The Insurance Field, January 5, 1933, p. 10 and December 28, 1933, p. 10. 5 The history of National Surety’s involvement in this line of business can be found in the “Preliminary report of investigation into the affairs of National Surety Company,” Folder 29, National Surety Papers. A full economic history of this type of MBS is beyond the scope of this paper. For contemporary descriptions of the surety company’s MBS guarantees, see McKenna (1927), Halliburton (1939), and Kniskern (1926). Kniskern was the vice president in charge of this business at National Surety. 6

Figure 1: Advertisement for MBS guaranteed by National Surety Source: New York Times, July 15, 1925, p. 10. 7

mortgages, typically single-family properties with some mixed use properties as well, and a small number of apartment and other small commercial buildings. The properties were dispersed across 38 states, largely reflecting the geographic locations of the different mortgage companies that issued the MBS. By 1933 when National Surety was taken over, there were more than 20,000 investors holding MBS bearing the company’s guarantee, reportedly somewhat below the peak from the late 1920s.6 In October 1928, National Surety ceased writing guarantees on MBS. The decision reflected mounting problems arising in this line of business. The company had already put a halt to new guarantees on MBS with mortgages based in Florida, which was perhaps the state with the most extreme housing boom and bust during the 1920s, and which was the location of National Surety’s first losses on its MBS guarantees. Across all of the MBS that National surety guaranteed, delinquencies on the underlying mortgages first appeared in 1925, and subsequently increased, in line with national trends. By the late 1920s, several of the mortgage companies with which it dealt were experiencing financial difficulties. One of these mortgage companies was responsible for about one-fourth of National Surety’s total guarantees, and its financial weakness was particularly troubling.7 Nationally, mortgage defaults began rising in the second half of the 1920s, prior to the beginning of the recession in August 1929, and then worsened considerably as the Depression progressed. Fundamentally, as house prices fell by 30-40 percent and unemployment became widespread, mortgage borrowers had difficulty making their payments and trouble refinancing their loans, a “double trigger” that economic theory suggests is a recipe for default. As a result. even though National Surety officials ceased writing new MBS guarantees no more than a year or two after experiencing their first losses on that line of business and almost a year before the 6 The exact number of mortgage companies with which National Surety did business varies by source, most likely because a few of them had very small numbers of bonds guaranteed by National Surety which rolled off in the early 1930s and therefore were not counted in later tabulations of mortgage companies with outstanding bonds. The “Preliminary report,” gives 31 as the number (page 5, folder 29), while the National Bondholders claim counts 24 (folder 66), and the Fifth Report of the liquidator counts 26 (Box 922134, no folder). The “Preliminary Report” gives the figures on total guarantees (page 44, folder 29), and the “Report of James A Martin, Esq., Upon the Fairness and Propriety of Reorganization Expenses Passed Upon and Determined by the Reorganization Managers and al matters pertinent thereto” gives the number of MBS, states with mortgaged properties, and the number of investors (pages 5-7, folder 65). Data on the total value of real estate loans in the country is from Fisher (1951) page 64. 7 See page 16 of the “Preliminary Report” for information on National Surety’s first losses in Florida, and page 43 for a discussion of the losses incurred by the mortgage companies. 8

1929 recession began, the scale of the losses on guarantees that had already been made proved to be large enough to push the company into a crisis by 1933. Ex post, it is clear that this line of business suffered from severe underwriting and informational asymmetry problems. The main issues were the quality of the underlying mortgages, and the financial condition of the mortgage companies. National Surety made efforts to evaluate the mortgages and required a 60 percent loan-to-value ratio, but many of the mortgages were evidently poorly conceived, and the mortgage companies that originated them clearly had an incentive to push through as large a volume as they could. National Surety also did not seek much control over the total leverage or risk taking of the mortgage companies whose bonds it guaranteed, even though such factors would affect the risk taken by National Surety in guaranteeing their bonds. The mortgage companies depended heavily on the ability to finance their mortgages by issuing bonds, and once that bond market dried up, they ceased funding new mortgages. Once the first losses appeared, these companies went bankrupt en masse quite quickly, as their capital positions had never been very large. National Surety made many efforts to prevent or mitigate losses on its guarantees. First, as the mortgage companies that issued the MBS began to fail, it reached agreements with them to assume control of their operations as a way to control and conserve the value of the underlying collateral. Second, it initiated a program to buy the MBS it had guaranteed, at large discounts, thus avoiding the need to realize losses (though of course those losses were borne by the bondholders who sold the bonds at a loss). The program commenced in early 1930, but was halted later that year when funds became scarce. Third, National attempted to extend the duration of MBS that were maturing, with the agreement of bondholders. National began securing such extensions in 1931 and continued in 1932. This program was also not particularly successful, as only about $200,000 of bonds were extended. Finally, with the failure of the extension program, National embarked on one last method of postponing the need to pay maturing securities. It created a new subsidiary, which offered to exchange its bonds with the outstanding bonds guaranteed by National Surety. This program, initiated in 1932, quickly led to the exchange of about $8.5 million in bonds. The incentive for bondholders to exchange was a premium of 1.5 to 2.5 percentage points of the principal. However, the program was badly 9

designed, as most of the bonds exchanged had maturities in the latter half of the 1930s, whereas the maturities that posed the greatest problem were the more immediate ones.8 3. The crisis National Surety experienced a severe crisis in March and April 1933. Ultimately, once it was determined that the company could not meet its obligations as they came due, the insurance department in New York seized the institution on April 30. The crisis resulted from the combination of four shocks: a longer-term outflow of cash to pay out losses on insurance contracts; the illiquidity of many investments due to the state of the bond market; the bank holiday; and finally a loss of confidence culminating in a run on the firm by policyholders seeking return of unearned premiums. Losses on insurance contracts The losses on guaranteed mortgage securities were the main driving force leading to National Surety’s crisis resolution on April 29, 1933. According to one source, from November 1928 to March 1932, National Surety had a net outflow of about $6.3 million due to this mortgage business.9 A second source states that by a year later in April 1933, the net cash outflow totaled $15 million.10 By the end of April, National was simply unable to meet its obligations.11 8 This paragraph is based on information from the “Preliminary Report,” pages 69-95. The Preliminary Report also gives a few possible explanations for the failure of the bond exchange program. The central problem with the exchanges is that they were conducted by a set of bankers acting as broker intermediaries, and whose incentives were not well aligned with the company’s. Some National Surety officials asserted there was an oral agreement that the bankers would target the early maturities. This was never written into the contract, though some correspondence with the bankers appears to acknowledge that targeting. On the other hand, the bankers clearly had an incentive to exchange a large volume in order to increase their commission, and no particular incentive to work for the exchange of early maturities. The bankers themselves cited some difficulties in procuring early maturing bonds. One excuse was that a large group of these bonds were held by Maine savings banks, who also held later maturities, and demanded the exchange of all their bonds, not just the latter portion. A second excuse was that the new bonds were not yet approved for exchange in all states. 9 National Surety used $11.25 million from Nov 1928 to Mar 1932 for this liquidation, and recovered only $4.92 million. See Papers of the Reconstruction Finance Corporation. 10 In this period, the company paid out about $30 million on mortgage guarantees, and received in salvage only about $15 million. National Underwriter, May 4, 1933, p. 21. 11 The insurance department concluded that “With the large drain on the cash resources of the Company, due to meeting matured mortgage guarantee obligations, the company would have been in a short while not able to meet its current obligations.” Deposition of Herbert C. Clark, an examiner in the NY State Insurance Department, National Surety Papers, folder 1. 10

To put these numbers in context, we must have a sense of National Surety’s buffer in the form of loss reserves and capital. This turns out to be a difficult, because National Surety—like many companies in the 1920s—owned and loaned money to a web of subsidiaries, whose net value to National Surety must be known in order to calculate the true capital position. Nevertheless, the overall size of the company’s balance sheet was about $48 million in 1928 before losses began to accumulate and before most of the subsidiaries were created. At that time, the capital and surplus were listed on the books as totaling $27.2 million together. By year-end 1932, capital and surplus had been reduced to $9 million.12 To meet the demands for cash, National Surety first tapped, in mid 1930, a group of three New York banks. Each of these banks shared a director with the National Surety, smoothing the path to the loans.13 The loans slowly increased in size, and as the bankers demanded additional assurances, the security behind the loans was augmented several times. Eventually, on April 1, 1932, the loans came due and the three banks demanded repayment. By that time, the total amount outstanding was about $7.9 million. The demand for repayment is likely a sign of a deterioration in the perceived creditworthiness of National Surety, and perhaps also reflects that the banks themselves were under pressure to liquidate their assets. To meet the demand for repayment, National Surety turned to the Reconstruction Finance Corporation (RFC). The RFC was established in early 1932, and National had begun borrowing small amounts fairly soon after, in May 1932, while negotiating a larger loan that would allow National to pay off its bank loans. That larger loan ultimately was consummated at the end of June 1932. National continued to receive further funds from the RFC, and as shown in Table 1, the last advance came on January 18, 1933. With that advance, the credit line that National had negotiated with the RFC was maxed out at about $12 million. By February, the RFC’s advances to National Surety were past due, with interest accruing and conversations about repayment growing more urgent.14 Without any additional sources of new cash, National was unable to pay out obligations that came due and was taken over at the end of April.15 12 To the outrage of investors, once the financial plight of the company was understood, National Surety had continued to pay 10 percent dividends ($1.5 million per year) up to 1930 and did not eliminate dividends entirely until 1932. 13 Preliminary report, p. 62. 14 Deposition of Stewart S. Hathaway, manager of the New York loan agency of the RFC, p. 2, National Surety Papers, folder 1. 15 Jones (1951, page 154) puts the amount of the loan at $15,000,000, but this appears to be incorrect. 11

Table 1: RFC loans to National Surety Size of draw Date of loan draw (dollars) Loan 1 May 2, 1932 722,000 May 19, 1932 28,000 June 1, 1932 500,000 June 6, 1932 712,676 August 17, 1932 11,325 1,974,001 Loan 2 June 30, 1932 8,291,000 July 21, 1932 589,000 November 10, 1932 100,000 November 12, 1932 400,000 December 6, 1932 500,000 January 18, 1933 400,000 10,280,000 Total 12,254,001 Amount repaid 253,301 Net amount due (April 30, 1933) 12,000,700 Source: Deposition of Stewart S. Hathaway, page 2, National Surety Papers, Folder 1. National Surety officials complained bitterly about the RFC’s unwillingness to extend further credit, and asserted that it was the cause of their inability to meet their obligations in April. RFC officials, in turn, simply noted that the credit line was a function of available collateral, and that National could not provide any further bonds or other investments suitable as collateral to increase the credit line. In particular, the RFC refused to assign any value to National Surety’s stakes in its subsidiaries, which appeared to be worthless but which National Surety argued were not. At the time, the RFC also had not been given the authority to make 12

preferred stock investments, so legally they had little room to maneuver without making an exception to the collateral standards they applied to all of their loans. Before National was taken over at the end of April, one last plan was floated with the hopes of saving the company. The plan was discussed publicly beginning around April 4, but it did not materialize in time to prevent its resolution. This plan would have involved the RFC making loans to the mortgage companies whose MBS National Surety had guaranteed. These companies in turn would have offered MBS holders the following deal: they would immediately receive cash for a portion of the face value of their bonds, and would receive a new bond for the remainder of the claim, which would bear a longer maturity and a lower interest rate than the previous bond, and would also be guaranteed by National Surety. The goal was to relieve immediate cash outflow pressure and to restructure the bonds, while also requiring the mortgage companies to lower the interest rates on their mortgages as a form of relief to mortgage borrowers. Collapse of the corporate bond market The second factor contributing to National Surety’s crisis was the collapse in the value of corporate bonds that began in late 1931 and continued through the first part of 1933. National Surety invested the large bulk of its assets in the stock and bond markets, and bonds by themselves accounted for about half of its assets. With its bond portfolio having fallen in value by about one-third, National Surety was reluctant to meet cash needs by liquidating securities, which would involve realizing losses. This fall in bond value was one of the rationales for borrowing from the RFC, in order to avoid liquidating bonds at values many expected or hoped would not persist. The RFC’s evaluation of National Surety’s loan application commented that the company’s capital position would be reduced by about $10 million if the bonds were marked to market. Of course, this impairment of bond values affected a wide variety of financial institutions in this period. The bank holiday The bank holiday that took place in mid-March 1933 delivered a third shock to National Surety’s cash flow position. In New York, the holiday began on Saturday, March 4, and the first reopenings were allowed to occur on Monday March 13. National Surety’s own bank deposits 13

were of course inaccessible during the holiday. Likewise, the deposits of National Surety’s counterparties were also tied up. As a result, while National’s policy holders continued to file claims, they could not make premium payments. Insurance companies were directed to halt claim payments until the end of the holiday, putting the casualty and surety businesses in a state of “paralysis.”16 A portion of National Surety’s funds remained inaccessible after the holiday because they were deposited in banks that were not licensed to reopen. At the end of April, National Surety still had $1.18 million out of reach in suspended banks While the temporary interruption of funding flows may have been overcome with time, bank closures also led to an additional shock to National Surety’s capital position because of its guarantees of bank deposits. As it became clear that not all banks would reopen, National Surety began to receive sizable claims on its deposit guarantees. These losses appear to have totaled at least $1.9 million, including $1.1 million tied to banks Pennsylvania and the rest to banks in the Midwest.17 By the winter of 1933, National Surety’s stock price had fallen to around five dollars per share, down from a highs of over one hundred fifty dollars per share, established in early 1928 just as National Surety was pulling away from the mortgage business. Clearly the market had learned of National Surety’s business troubles long before the run that commenced in March 1933. Of course, stock prices in general fell by large amounts in the same period, as the Dow Jones index fell by about 80 percent in the same period. 16 Weekly Underwriter, March 11, 1933, p. 515. One exception was made to the cessation of business: to a limited extent, banks were allowed to make payment related to workers’ compensation claims and payrolls. 17 See Exhibit A, table on pages 10-11, National Surety Papers, folder 1. Further details are available from the National Surety loan application to the RFC (in Reconstruction Finance Corporation Papers, report dated 1932-04- 22, page 2), which states that National Surety owed about $1.1 million to the state of Pennsylvania on depository bonds covering more than one bank in that state. As one example, the Wall Street Journal reported that National Surety owed $400k to the state of Pennsylvania on depository bonds covering Franklin Trust Company, based in that state (December 13, 1932, p. 7). Also, the National Underwriter, May 5, 1933, page 21, notes that the key impact of the losses in Pennsylvania.. The National Surety Papers show additional losses on a few more banks located in the Midwest. Also see exhibit A of the National Surety Papers (folder 1, page 11), which states that National Surety held $650,000 of its own cash in closed banks, and $1,250,000 in restricted or limited withdrawal banks. 14

Figure 2: National Surety Stock Price Notes: Prices are indexed to 100 in January 1927. Monthly values are computed using the average of the high and low closing prices in each month. National Surety stock prices are taken from CRSP, adjusted for a split in 1928. The Dow Jones index data are taken from the NBER Macro History Database, series 11009B. The run By April, 1933 it appears that National Surety had widely lost the confidence of its policy holders and other counterparties. The loss of confidence led to an “avalanche of cancellations” during April by its existing policyholders. These cancellations created a large cash outflow problem for the company, as policy holders were owed the return of unearned premiums within 30 days. (Unearned premiums are the portion of premium payments that cover future time periods. Insurance policyholders pay for insurance in advance, so that at time of payment the 15

entire payment is unearned.) At the same time, revenue fell off sharply as policy holders withheld payments to renew their policies or enter into new ones.18 The magnitude of the cancellations appears to be quite large. At the time of its takeover, the company listed $2.25 million in liabilities for bonds or policies where notice of cancellation had been given. This is roughly one quarter the size of total unearned premiums as of December 31, 1932, suggesting the same percentage of policy holders (by dollar volume) had cancelled their policies.19 The size of the run is all the more remarkable because some types of the bonds that National Surety issued were not easily cancellable: as one example, about two percent of National Surety’s business (measured by aggregate bond amounts) was in judicial bonds, which are not cancellable at will by the policyholder since those bonds must remain in force for the life of the judicial proceedings. Another twelve percent of National Surety’s business was for fiduciaries, which cannot cancel their bonds while still in their positions. Indeed, National Surety suggested to the court at the time of its takeover that not all of the policy cancellation requests it had received were valid, though I have not found data to quantify the extent to which any cancellations were nullified.20 In court filings, the New York Superintendent of Insurance asked the judge to think of these demands for return of unearned premiums as being equivalent to a bank run. That analogy was adopted by the trial court and ultimately the appellate court as well. The key similarity is that unearned premiums represent 30-day obligations, and therefore are subject to a run with that lag in the same manner that bank deposits can be subject to a run. Just like a bank would have difficulty redeeming deposits en masse if requested with little notice, an insurance company would likewise have difficulty in returning all unearned premiums if demanded all at once. Contemporaries widely attributed the run on National Surety to rumors that spread about the company’s impending collapse.21 Certainly, the company’s troubles were well known by 18 Exhibit A of the National Surety Papers, folder 1, describes the loss of confidence of counterparties. The avalanche quote is from Eastern Underwriter, May 5, 1933, p. 17. The National Underwriter also described “widespread cancellations” in National Surety’s final days (May 4, 1933, page 21). 19 The size of unearned premiums may vary over the course of a year depending on when policy holders enter into contracts. If, for example, all policy holders enter their contract in December, the unearned premium would be large at year-end and then decline for the rest of the year. 20 Mackall (1929) discusses the ability of policyholders to cancel different types of surety bonds. 21The chairman of National Surety, William Joyce, complained bitterly about “unfounded rumors” regarding his company’s condition that led to the cancellations. See the deposition of William B. Joyce National Surety Papers, folder 1. In a newspaper, he was quoted as equating the rumors about his company with “financial treason” (“Gossip Mongering” in Brooklyn Daily Eagle, April 1, 1933, p. 20). 16

1932, especially with the reduction in its published capital position in that year, and the cessation of dividends.22 The rumors appear to have changed in character, though, in late March 1933. A scanning of the financial press at the time suggests that the rumors that the company would fail became widespread around the very end of March.23 (In turn, this explains the resolution of National Surety at the end of April, when the 30 day period for returning unearned premiums would have begun to expire.) The timing of the rumors at the very end of March suggests a link to the failure of another insurance company, the Globe and Rutgers, which failed on March 24.24 Though National Surety had no direct link to Globe and Rutgers and the two companies had no investments in common, they were linked in investors’ minds because of common misadventures in mortgage securitization. Globe and Rutgers, a fire insurance company, failed largely because of losses on equity stake it had taken in an MBS issuer.25 In fact, these were the two largest insurance companies in New York to be taken over by the state during the Depression, and both owed their collapses to mistakes in the mortgage field. Though the rumors may have been based on limited information at the time, in the end they seem to have been well founded. (I discuss the resolution strategy for Globe and Rutgers more below. Its somewhat disorderly resolution was an important precedent that made salient to regulators the need to plan for a more orderly resolution for National Surety.) National Surety was not the only company to face a run by its policy holders in these years. Globe and Rutgers appears to have failed following a run of its own, as one report stated that “hundreds of assureds cancelled their policies and asked for the return of the unearned premiums.”26 Elsewhere in the insurance field, while large life insurance companies survived the Depression with very few failures, liquidity pressures created significant problems at times, 22 For example, an insurance periodical, the Spectator, noted that losses on MBS guarantees had hit National Surety and its peers “heavily,” and described how “one of the largest surety companies” had been forced to set up a subsidiary to administer and sequester this business, obviously referring to National Surety. 23 For example, see New York Times, March 31, 1933, p. 30; Wall Street Journal, March 27, 1933, p. 2; and Milwaukee Journal, March 28, 1933, p. 8. 24 To be specific, Globe and Rutgers was taken over at the end of the day on Friday March 24, and thus its first impact on the market came on Monday March 27. If National Surety received cancellations immediately that Monday, the 30 day period on those requests would have expired on Thursday April 27. This timing matches up quite well, as National Surety was taken over the following weekend (technically dated Sunday April 30). 25 The MBS issuer in which Globe and Rutgers had invested was the Lawyers Title and Guaranty Company, which was an eligible investment because it was a New York based title insurance company. This was one of many New York based issuers of MBS in the 1920s, but was not one of the companies that National Surety was involved with, which were all outside of the New York Area. For a description of Globe and Rutgers investments, see Eastern Underwriter, March 31, 1933, p. 19. 26 See “Insurance Lines Becoming Normal,” New York Times, April 3, 1933. 17

especially during the bank holiday when policy holders requested loans against their policies or sought to surrender their policies for their cash value. In response, the New York state insurance commissioner used emergency powers to ban such loans and surrenderings on life insurance policies.27 That ban was kept in for several months, into September 1933. The insurance commissioner never placed similar protections for other types of insurance companies, which is why policy holders were able to stage a run on National Surety and Globe and Rutgers. 4. Systemic importance In the 1930s, the phrases “too big to fail” and “systemically important” had not yet been coined. Nevertheless, those writing about National Surety in 1933 clearly expressed fears about the wide-reaching damage that would result from the potential disorderly liquidation of the company, what would today be called systemic consequences. For example, in the insurance industry press, leading weekly insurance magazines issued strong statements describing potential systemic consequences. In an editorial, the National Underwriter wrote of the “tragic consequences that inevitably would follow the complete failure of an institution of the prominence of the National Surety.”28 The Eastern Underwriter’s editorial stated “It would have been a calamity if the National Surety Co., with all of its good will in the world of finance, business and industry, had disappeared.”29 In the political sphere, Jesse Jones, the head of the Reconstruction Finance Corporation, wrote in his memoir that “somber tragedies were averted” by the successful rehabilitation program, in which the RFC participated.30 Similarly, the deputy of the insurance commissioner, who was in charge of the National Surety case, stated in a deposition that “...if a situation were created calling for the cancellation and termination of these policies, the business, commercial, industrial, financial and judicial interest of the country would have suffered immeasurably… [The superintendent] felt it incumbent upon him as a matter of public policy to do all that he could to prevent such a catastrophe from arising.”31 27 See “Insurance concerns tighten loan rules,” New York Times, March 9, 1933, p. 6. 28 National Underwriter May 4, 1933, p. 21. 29 See Eastern Underwriter, May 5, 1933, p. 16. The National Underwriter and the Eastern Underwriter were two of the four leading insurance newspapers at the time. The other two were the Spectator and the Weekly Underwriter. The Spectator wrote an editorial about how the National Surety’s resolution was handled better than the Globe and Rutger’s, but did not opine on the possible effects of a collapse. The Weekly Underwriter never wrote an editorial on the matter. 30 Jones (1951), p. 153. 31 Deposition of Samuel R. Feller, page 2 National Surety Papers, Folder 1. He continues, on page 8, to say that “The attendant disturbances will be material and disastrous.” 18

This section considers the mechanisms underlying these alarming statements, examining the different ways that National Surety was connected to the rest of the economy. Sources of systemic importance32 One source of National Surety’s systemic importance was its sheer size. Indeed, National Surety was widely described as the largest surety company in the world, including when it described itself in its advertisements. The insurance commissioner of New York, in justifying his act to rehabilitate National Surety to the court, warned that National was connected to a large and varied number of counterparties: So huge in amount and so varied in character are the bonds and policies of this company that unless a sound and practicable plan of reorganization is effected the consequences will constitute a nationwide calamity, vitally affecting banks, insurance companies, and the entire industrial and commercial life of the country.... The general effect upon the life and business of the country by such a disaster will be appreciated by those outside as well as inside the business of bonding and insurance. With the multitude of business and financial transactions which are based primarily upon the soundness of the Company issuing the bond or policy, the results are unfortunately so tragic and far reaching that further comment or analysis is unnecessary. Quantitatively, there is more than one way to measure the size of an insurance company. A typical yardstick is premium income, as it is a direct indicator of the volume of business being conducted, and because such data are generally readily available. That measurement confirms National Surety as the largest surety company in the United States during the 1920s and early 1930s. In 1928, for example, the company accounted for about 13 percent of the volume of surety business as measured by premium income, for example. That said, the surety sector appears to have been fairly concentrated among five companies, National Surety and another New York company, American Surety, and three Maryland-based companies: Fidelity and Deposit, Maryland Casualty, and United States Fidelity & Guaranty. Together, these five 32 This section rests heavily on material found in National Surety Papers, Folder 1, Exhibit A, pages. 4-6. 19

companies controlled just over 50 percent of all surety premiums written nationally. As shown in Table 2, two of those companies did much more non-surety business than National Surety, primarily in the fields of automobile liability and workers’ compensation. Table 2: Premiums written during 1928 by the insurance companies with the largest surety, fidelity, and credit businesses Premiums Surety, Fidelity, and Workers Note: Total Company Credit comp. Auto Other Total Assets National Surety 13.9 4.4 18.4 48.3 American Surety 8.8 1.1 9.9 23.3 Maryland Casualty 4.8 8.5 9.6 5.1 28.0 46.7 Fidelity & Deposit 10.7 1.7 12.4 28.7 11.1 10.9 11.3 8.8 42.1 68.0 United States Fidelity & Guaranty Note: Figures are in millions of dollars. Source: New York Insurance Commissioner, Annual Report for the year 1928, p. xl. Another yardstick might be the total dollar size of contingent liabilities, which court filings reveal totaled roughly $3 billion at National Surety at the time of its failure. As one measure of scale, this equaled about 5½ percent of GDP at the time. (Though GDP was quite low in 1933, the business volume across the surety industry had fallen significantly as well, so on net it is not clear if this figure is higher because of the contraction in GDP.) Unfortunately, such data are not readily available for other insurance companies in that era for purposes of comparison. An insurance company’s size could also be measured by its geographic reach. Here, National Surety appears to have been particularly national in scope. While many other surety companies were more regional in focus, the insurance commissioner stated that “The National Surety had outstanding bonds… affecting money, interests, and responsibilities in practically every city and town in the country.”33 It also had operations in several foreign countries, with the largest businesses in France and Germany. Along with its size, the identity of its particular counterparties were also a source of concern, along with the sheer number of those counterparties. Federal, state, and local 33 Fuller deposition, page 3, National Surety Papers, folder 1. 20

governments were counterparties on about one-third of National Surety’s contingent liabilities (roughly equally split between federal and state/local). These entities took out contract surety bonds on a variety of contractors, such as construction companies on infrastructure projects, as well as a wide array of commercial surety bonds on government officials, such as postmasters, who were required to follow certain laws as part of their jobs. National Surety also provided deposit insurance to governmental entities with funds in commercial banks, as such deposits are usually required to carry some form of security. In some cases, banks post collateral against such deposits. In others, banks took out bonds, issued by companies like National Surety, as an alternative form of security for the deposits. These bonds were estimated to total $40 million, $15 million of which covered deposits at New York banks. The cancellation of those bonds would have required the banks to repay the deposits if they could not post collateral or arrange for new surety bonds, both of which would have been difficult given the depressed state of the bond market in 1933 and, again, the weakness among other surety companies. Outside of government, a large portion of National Surety’s other counterparties on commercial surety policies were administrators, executors, guardians, and other fiduciaries. Traditionally, such actors have been afforded special protections by the government, while at the same time being subject to fairly strict regulation. For example, over 70,000 fiduciaries had bonds issued by National Surety, with total contingent liability of about $350 million. About 17,000 people took out judicial bonds totaling $56 million. Such counterparties were required by law to obtain insurance or bonding policies covering their activities, and therefore would be forced to find new policies if National had failed. Such a search would likely have been acutely difficult in the financial environment during spring 1933. A third channel of systemic importance was through reinsurance contracts. Government officials feared that National Surety’s failure could drag down its reinsurance counterparties, concluding this type of contagion was “extremely dangerous.” (Exhibit A, part E, page 6) Likewise, the insurance press noted relief among officials at other insurance companies that there would be little contagion through reinsurance contracts.34 The reinsurance contracts were provided both from and to National Surety: 10 other surety and casualty companies ceded approximately $64 million in reinsurance to it, and in turn National Surety ceded approximately 34 The National Underwriter, May 5, 1933, page 21, observed that “Relief is expressed that reorganization of the National Surety will not cause serious injury to other companies under reinsurance agreements”. 21

$70 million in reinsurance with the same companies. These companies would have been liable to National Surety’s customers if it had been liquidated. In addition, if National Surety’s reinsurance to them were voided, they might have been forced to raise capital or find new reinsurance partners. Since the reinsurance appears to have covered workmen’s compensation insurance in particular, a line of insurance that was performing very poorly in the early 1930s and delivering large losses to insurance companies, there was a real potential for sizable losses and inability to find new reinsurance partners. Finally, government officials also feared an indirect effect on the surety business as a whole. Given National’s position at the top of the surety business, they thought its failure might have undermined the confidence of other surety companies’ counterparties. Court documents suggest that this is one reason the RFC agreed to the rehabilitation plan.35 “It would have hurt the prestige of American suretyship world-wide because the National’s operation extended over a wide territory, including several European countries.”36 Assessing National Surety’s systemic importance In summary, the case for National Surety’s systemic importance relied on its connections to large numbers of policy holders, the particular nature of those policy holders as government entities or those with trustee or administrator responsibilities that are often afforded special government protection, and its connection to other insurance companies directly through reinsurance contracts and indirectly given the possibility of contagion. One way to assess these claims is to ask whether anyone disagreed. Company officials supported all of these arguments, but they had a strong incentive to do so, because if the company had been liquidated they would have been out of work and the stock they owned would be worthless. Officials from the state Insurance Commissioner’s office appeared to have agreed with these arguments completely. The Reconstruction Finance Corporation might have had an incentive to disagree, since the reorganization increased the riskiness of the collateral behind its loan (see section 5), but RFC officials supported the plan, specifically citing the benefits not just 35 In his deposition, Hathaway, the manager of the New York loan agency of the RFC, stated that The RFC executive committee decided to go along with Van Schaick’s plan because “such action would prove of benefit not only to the National Surety Co. and its creditors but also to the surety bond business as a whole.” 36 Eastern Underwriter, editorial, May 5, 1933, page 16. 22

to National Surety but also to “the surety bond business as a whole.”37 Finally, while some of National Surety’s creditors appealed the reorganization plan, they did not do so on the grounds that National Surety’s failure would not have been disruptive, but rather that the Commissioner did not have the legal powers to conduct the reorganization as he did. One source of ambiguity in the Insurance Commissioner’s arguments relates to the matter of whether business could have been easily transferred from National Surety to other insurance companies. If such transfers were relatively speedy and low in cost, then National Surety’s failure might not have greatly affected its policyholders, as long as they had no outstanding claims. However, officials in the New York insurance department saw obstacles to such transfers of business. Issuing an insurance policy requires a certain amount of underwriting and accounting work. That work, along with other factors, could lead to long delays, during which “the business, commercial, industrial, financial, and judicial interests of the country would have suffered immeasurably.”38 However, this testimony is a bit at odds with efforts the insurance department made to prevent other companies from raiding business immediately after National Surety was placed in rehabilitation, mostly in the form of moral suasion.39 Potentially, there could have been some heterogeneity in the substitutability of different types of insurance policies, so that competitors might have been able to steal the most substitutable policies but nevertheless other policy holders would have been left without much ability to find new insurance coverage, at least quickly. Overall, though, the transferability argument may be somewhat weaker than officials stated. At the very least, it is clear that the NY State legislature granted the Insurance Commissioner emergency powers for rehabilitation specifically because it feared the very sort of fallout that appeared imminent when National Surety experienced its crisis. Indeed, the Yale Law Review (1934) described the purpose of that legislation in this way: “The broad scope of the powers conferred on the Superintendent indicate that this act leaves him free to adopt any reasonable course of action that seems appropriate to attain the desired end of financial stability” (page 1148). The Superintendent of Insurance made the same observation to the trial court.40 37 Deposition of Stewart S. Hathaway, page 4, National Surety Papers, folder 1. 38 National Surety Papers, folder 1, deposition of Samuel R. Feller, first deputy superintendent of the New York State Insurance Department, p. 3. 39 This is discussed in all the major insurance periodicals following National Surety’s takeover. 40 Deposition of Samuel R. Feller, page 8, National Surety Papers, Folder 1. 23

5. Strategy for rehabilitation and resolution This section describes how National Surety was reorganized, with an emphasis on the principles that guided government officials in designing the reorganization. Reorganization mechanics The basic reorganization actions involved the splitting of National Surety into two companies. One company continued to conduct the surety business of the old company, but only took over the subset of the old company’s business that was expected to be profitable. Mainly, this excluded mortgage guarantees and depository bonds on closed banks, which were responsible for the great bulk of the old company’s losses.41 This “new” company’s liabilities were also limited to future losses on the old company’s business, and only on the part of the old company it took over. In other words, the new company admitted no responsibility for any claims that had already been filed, either for losses or for return premiums. The name for this new company was National Surety Corporation – a deliberate piece of wordsmithing meant to invoke continuity with the old National Surety Company. The second company created by the reorganization was a liquidating corporation. This corporation was responsible for the existing claims on the old company, such as the repayment of unearned premiums from cancelled policyholders or loss claims on insurance policies that were already filed. It was also responsible all future claims on the lines of business not taken over by the new company. Table 2 provides a guide to how the assets and liabilities of the old company were split between the new company and the liquidating corporation. On the asset side, the most liquid assets went to the new company, including cash, and the more readily salable stocks, bonds, mortgages, and real estate. One item on the asset side of the new company, premiums due on the business taken over by the new company, was of course intangible and would have been worth nothing in liquidation without the reorganization. That asset nevertheless could only be placed 41 The other liabilities not taken over were fiduciary bonds covering estate administration, but only temporarily until an accounting of the estate as of May 1, 1933 were taken. In addition, liabilities on policies covered by reinsurance agreements accepted or ceded by the old company were assumed by the new company only after it had negotiated an agreement with the reinsurance partners. 24

in the new company since it was the one that would continue operating. The less liquid assets went to the liquidating corporation, including the less liquid stocks, bonds, and mortgages, as well as cash trapped in closed banks, debts due from reinsurance partners, and a large debt owed to the company by its subsidiaries.42 These assets were to be slowly sold off and the proceeds given to the liquidating company’s liability holders. The new company had a sizable capital buffer created through a transaction with the Reconstruction Finance Corporation. Indeed, the RFC played a key role in this reorganization. As discussed earlier, the RFC owned a claim to a large chunk of National Surety’s best assets, as security to the sizable loan National Surety had accumulated to the RFC prior the reorganization. Without the release of those assets by the RFC, the reorganization could not have moved those assets into the new company. The RFC agreed to release a portion of those assets from the collateral pool, but only in return for the addition of a new asset to the pool. That new asset was the entire capital stock of the new company, which was held by the liquidating corporation but which the RFC had the first claim to. In the process, the stockholders of the old company were wiped out. As noted above, financial stability was part of the motivation for RFC officials to allow this transaction. At the same time, though, RFC collateral rules were quite strict. The RFC put real value on the capital stock of the new company, one sign of confidence that the new company would prosper. As part of the reorganization process, company officials tabulated data showing the profitability of its business lines, from 1923 to 1932, broken down into the business lines that would be discontinued (and placed into the liquidating corporation) and those that would be continued. These data, shown in Table 4, were used to support the idea that the business placed into the new company would be profitable. Those business lines earned a 4 percent profit from 1923 to 1932, while the discontinued business lines created a 24 percent loss. 42 Constructing a consolidated balance sheet is not entirely impossible but would almost certainly be meaningless given the misleading accounting used at the subsidiaries. 25

Table 3: Reorganization of National Surety New company Assets Liabilities Stocks and bonds 4.9 Capital + Surplus 4.0 Premiums due 4.0 Reserves 7.1 Cash 1.2 Past due premiums 0.7 Mortgage loans 1.3 Real estate 0.4 Other 0.0 11.9 11.9 Contingent Liabilities Future claims on continuing lines of business Liquidating company Assets Liabilities Capital stock of new corp 4.0 Capital + Surplus 9.0 Stocks and bonds 15.0 Reserves 11.0 Loans to subsidiary 9.9 Borrowed money (including RFC loan) 12.4 Mortgage loans 1.3 Accrued Commissions 0.4 Money in closed banks 1.1 Due from reinsurance 0.2 Other 1.3 32.8 32.8 Contingent Liabilities Past claims on all lines of business Future claims on discontinued lines of business All unearned premium claims Notes: Figures are in millions of dollars. 26

Table 4: Profit and loss by business line, 1923-1932 Business lines Business lines to be All business to be continued discontinued lines combined Earned premiums 138.6 32.0 170.6 Losses incurred 63.4 24.8 88.2 Expenses incurred 69.2 14.8 84.0 Underwriting profit 6.1 -7.6 -1.5 Profit as percent of premiums 4.4% -23.8% -0.9% Notes: Figures are in millions of dollars. The new company survived the reorganization. Ultimately, the capital stock of the new company was sold in 1936 for $10 million, with the proceeds benefiting creditors and paying down the RFC loan.43 Creditors of the old company also benefited from the proceeds of a lawsuit against the directors of the old company, settled for $1.35 million in 1937. Principles of the reorganization A few principles guided this reorganization. One principle was the preservation of the good will of policyholders. In the process, this would also maintain the going concern value of the business. The concern was that if policyholders fled the new company the end result would be a disorderly liquidation that officials were trying to avoid. In addition, the going concern value was embodied in the capital stock of the new company, which was a key asset held by the liquidating corporation to satisfy the claims of the old company’s debtors, and secure the RFC loan. If the company had been liquidated, that going concern value would have been dashed, and the creditors would have had fewer assets available. To this end, the reorganization was designed to put the new company into a very strong operating position, with liquid assets, a sizable capital buffer, and no exposure to unprofitable business lines. 43 Commercial Investment Trust (later known by its abbreviation, CIT) was the buyer. CIT sold National Surety to Fireman’s Fund Insurance Group, based in San Francisco, in 1954. In 1991, Allianz, the Munich-based insurance giant, acquired Fireman’s Fund. 27

The engineering of a capital stock with positive value out of a failing company was touted by the plan’s proponents as creating “assets from ashes.”44 In a way, the creditors of the liquidating corporation capitalized the new firm, not by infusing new cash, but by ceding claim to the liquid assets of the old company in return for ownership of the new company. From an economic point of view, a key question is who the winners and losers were. In evaluating this, the key counterfactual is a disorderly liquidation. In the counterfactual, stockholders would have been wiped out, and all creditors would have had equal claim to the firm’s assets. As events actually unfolded, stockholders were still wiped out, but a distinction was made between creditors placed in the liquidating corporation and those placed in the new company. The creditors placed in the new company fared well. The creditors placed in the liquidating corporation did better than the counterfactual only if the value of the capital stock of the new company, which they had claim to, exceeded the cost to them of surrendering the best quality assets to the new company, as well as the conversion of their claims into long-term obligations with no fixed maturity. In order to preserve the going concern value of the new company, a second and related principle guided the reorganization: negotiating with policy holders of the new company was considered dangerous and was avoided. In particular, it was feared that negotiating with policy holders about altering the extent of their contingent liabilities could lead those policy holders to lose confidence in the firm, cancel their policies, and therefore undercut the firm’s revenue stream. Fundamentally, this is a dynamic that would complicate any attempt to rehabilitate an insurance company: policy holders are both the holders of an insurance company’s (contingent) liabilities, and providers of income through premium payments. An insurance company therefore cannot negotiate with liability holders without affecting their revenue, since policy holders are both. To understand this dynamic, it is instructive to contrast insurance companies with commercial banks. If a bank were to negotiate with its liability holders, even if those negotiations soured and the liability holders demanded the return of their funds, that run would not affect the income stream from the assets of the bank. Though a typical bank does have some counterparties who both hold deposits and borrow money, banks’ liabilities and income streams are much more separated. 44 See, for example, the obituary of William Joyce, the head of National Surety. New York Times, August 6, 1962, p. 25. 28

A third principle was that reinsurance was not a viable alternative. Typically, reinsurance is one option for a troubled insurance company. In this case, though, National Surety was widely viewed as too big for even a group of other companies to reinsure. The Weekly Underwriter stated in an editorial that that “So great were the transactions of the company, ordinary methods of reinsurance would probably have failed to adjust the situation to the best advantage of all concerned.” (May 6th 1933). Specifically, it was reported that the size of the reinsurance reserve was too large, and that companies who had been approached had offered only 40-50 cents on the dollar for that reserve.45 The immediate historical context In designing the rehabilitation program for National Surety, the insurance department sought to avoid a repeat of the loss of good will that followed their takeover of a different insurance company, the Globe and Rutgers Fire Insurance Company, about a month earlier on March 24, 1933. An insurance magazine called this episode “One of the most sensational financial dramas which this town of many dramas had witnessed.”46 When the state took over the Globe and Rutgers, the insurance department announced its intention to rehabilitate the company, much as it later announced its intention to rehabilitate National Surety. But unlike in the case of National Surety, no actual rehabilitation plan was in place at the time of the takeover. Instead, it was announced that one would be worked out soon. The uncertainty led to cancellations by Globe and Rutgers’ policyholders, even though Globe and Rutgers couldn’t even pay out the unearned premiums because it was under court-ordered payment restrictions. Worse, as time went on and government officials had difficulty in quickly announcing a rehabilitation plan, the rush to cancel policies reportedly gathered momentum. This in turn delayed the rehabilitation planning even more, because with the loss of business from cancelled policies, plans for rehabilitation needed to be modified, especially because original estimates for additional capital became too low and new sources of capital needed to be found. At one point, the insurance department decided that so much of the business had been lost that there was not 45 Weekly Underwriter, May 6 1933, p. 889. 46 Eastern Underwriter March 31, 1933, page 19. 29

much of a company left to rehabilitate, and proposed liquidating it. Eventually, though, a rehabilitation plan was enacted.47 Part of the reason for the disorderly nature of the treatment of Globe and Rutgers is that the insurance department spent most of its efforts before taking over the company in seeking reinsurance arrangements for the firm. Ultimately, negotiations reportedly broke down as reinsurance partners were not willing to insure all of Globe and Rutgers liabilities. The partners were willing to reinsure the premium reserve creditors but not the loss creditors. Evidently, they also did not view Globe and Rutgers as a threat to financial stability, or could not overcome collective action problems to form a reinsurance pool.48 The Superintendent of Insurance directly cited this experience in the court papers justifying the rehabilitation strategy for National Surety, stating that the experience in New York has been that taking over an insurance company for rehabilitation leads to a flood of cancellations, leading to cash demands for return of unearned premiums and destruction of the good will. Legal basis for the reorganization The legal basis for this reorganization derived from two pieces of legislation passed in 1932 and in 1933. The first, an amendment to the New York Insurance Laws, added rehabilitation to the powers available to the superintendent in the case of an insolvent insurance company.49 The second was a piece of emergency legislation passed in early March 1933.50 The act was sweeping in its scope, declaring that a “public emergency exists affecting the health, comfort, and safety of the people of the state,” and essentially allowing the superintendent to suspend as necessary any aspect of insurance law during the duration of the emergency, as well as alter rules and regulations of the insurance department. In the case of National Surety, the superintendent was able to suspend certain laws that regulated the way in which new insurance corporations could be formed. For example, no notice was given of the proposed formation of 47 For a discussion of cancellations affecting Globe and Rutgers, see Eastern Underwriter, March 31, 1933, page 21 and Spectator, May 4, 1933, page 8. See also the Feller deposition in the National Surety Papers, folder 1, which notes the insurance department officials’ desire to avoid a repeat of the Globe and Rutgers experience. The National Underwriter (June 8, 1933) quotes Ignatius, a deputy of Van Schaick, the insurance commissioner, who discussed the need to prevent cancellations. 48 Eastern Underwriter, March 31, 1933, page 19 49 Chapter 191 of the laws of 1933 in New York, which amended article XI of the insurance laws. 50 Chapter 30 of the laws of 1933 in New York. 30

the new company.51 New York was not exceptional in passing such legislation; many other states passed similar acts, applying to various types of financial firms including insurance companies. Yale Law Journal (1934) discusses the legal basis for the rehabilitation, as well as the challenges brought by creditors of the old company. 6. Conclusion: National Surety and today’s systemically important insurance companies There have been proposals at various points over US history to establish national regulation and supervision of insurance companies, but such activity remained at the state level until the Dodd-Frank Act in 2010. The Dodd-Frank Act created the Financial Stability Oversight Council (FSOC) and charged it with designating and regulating non-bank systemically important financial institutions, including insurance companies. Since then, the FSOC has designated three insurance companies as systemically important, AIG, Prudential, and MetLife, though the latter has asked a federal court to review its designation. With this new regime of insurance company supervision, policymakers at the national level are devoting greater attention to insurance companies, and in the process are grappling with fundamental questions about those companies. How can an insurance company be systemically important? Can an insurance company have a liquidity crisis? How should a systemically important insurance company be regulated? How should such a company be resolved? In this conclusion, I consider each of these questions in turn, and use the National Surety episode as a source of historical perspective to answer those questions.52 How can an insurance company be systemically important? In designating MetLife, Prudential, and AIG as systemically important, the FSOC delineated three mechanisms by which financial distress at those companies could be transmitted to the economy: exposures of economic actors to the insurance company, the effect of liquidating the insurance company’s assets, and any critical market roles fulfilled by the insurance company. Only the exposures channel was raised in the discussion of National Surety by contemporaries. In terms of asset liquidation, though National Surety was the largest surety company in the world, as a surety 51 The insurance commissioner also suspended the requirement that the Attorney General review the charter of the new company, the requirement that the capital of the new company be paid in cash and that additional cash be deposited with the insurance department, and certain other formalities about the payment of capital. 52 For other recent papers that discuss systemic risk in the insurance sector, see Acharya and Richardson (2014), Cummins and Weiss (2014), Harrington (2009). 31

company its assets were still significantly smaller than other types of financial institutions, such as life insurance companies. For illustration, in 1932, the largest life insurance company was Metropolitan Life, which had assets of $3.8 billion in 1932, about 80 times the size of National Surety’s balance sheet. Fundamentally, the business models of these companies necessitated very different balance sheets. Because life insurance policies function essentially as lifetime savings plans, life insurance companies invest premiums in long-term assets which they accumulate to match the long-term nature of their liabilities. Surety policies are shorter term in nature and therefore such companies do not operate with balance sheets that are nearly as large. The reverse is true for contingent liabilities, though. Life insurance companies’ contingent liabilities are not nearly as large relative to their assets. In 1932, Metropolitan Life’s life insurance policies, for example, had total value of about three times its balance sheet, whereas National Surety’s contingent liabilities were about sixty times the size as its balance sheet. Contemporaries also said little about National Surety’s key role in markets, although there was some discussion about whether National Surety’s customers would be able to find insurance at other companies. Can an insurance company have a liquidity crisis? After MetLife was designed by the FSOC as systemically important, its officials have often discussed this question publicly. They point out that insurance companies run different business models than banks. Since insurance companies’ liabilities are longer in term than banks’ liabilities, they have less of a maturity mismatch. MetLife officials also note that there are disincentives for policyholders to call their policies early, in the form of surrender charges and tax penalties.53 Nevertheless, two aspects of the 1930s history suggests that liquidity crises could arise at insurance companies. First, insurance companies could face short-term obligations if policyholders decide to cancel their policies and demand the return of unearned premiums, as they did to National Surety. The main disincentive in the National Surety episode was a 30-day waiting period, which was evidently not nearly enough to discourage a wide swath of its policyholders from running. Second, in another part of the insurance world, life insurance companies in the 1930s had liquidity problems of their own, leading the New York insurance commissioner, for example, to prohibit life insurance companies from being able to pay out surrender values or make policy loans for six months during 1933. 53 See Wheeler (2012) and Kandarian (2013). 32

How should a systemically important insurance company be regulated? One school of thought focuses on an activities-based mode of regulation, in which regulatory attention is devoted to nontraditional insurance activities at any insurance company, not just systemically important ones. This view, promoted for example by officials from MetLife, is clearly responsive to the experience of AIG, whose problems were concentrated in its financial products division. Likewise, had National Surety not entered into the mortgage security guarantee field, it almost certainly would not have experienced this crisis. Yet, National Surety was not the only surety company that engaged in this business, and such an activities-based mode of regulation would require all of its peers to have been subject to extra scrutiny, regardless of their size. In addition, National Surety’s mortgage business was no secret to regulators. It sought and received direct permission from the New York Attorney General and the state insurance commissioner to engage in this activity. While it was a new activity, evidently, neither the risks of the activity nor the scope of the corporate governance and other incentive problems were apparent. Activitiesbased modes of regulation would rely on supervisors being able to accurately assess the risks of nontraditional businesses. Finally, how should a systemically important insurance company be resolved? The experience of the NY insurance department in the 1930s highlights the pitfalls of negotiating with the policyholders of an insurance company in the same way that debt holders of a bank may be negotiated with. Negotiating with insurance policy holders is particularly problematic because policyholders are also a core source of income through the payment of premiums. If policyholders’ confidence is undermined, they could request the return of unearned premiums, undermining the financial position of the company. In other words, the debt structure and the income stream of insurance companies are linked in a way that they are not at banks. National Surety provides an interesting example of a good insurance company/bad insurance company resolution strategy. The resolution appears to have been relatively successful, even in the absence of new capital injection, though with the full and public-spirited cooperation of National Surety’s single largest creditor, the Reconstruction Finance Corporation. 33

References Acharya, Viral and Matthew Richardson. (2014) “Is the Insurance Industry Systemically Risky?” Chapter 9 of Modernizing Insurance Regulations, John Biggs and Matthew Richardson, eds. Best’s Insurance Reports, Casualty and Miscellaneous. (Various issues) Cummins, J. David and Mary A Weiss. (2014) “Systemic Risk and the U.S. Insurance Sector” in Journal of Risk and Insurance, vol. 81 no. 3, pp. 489-528. Eastern Underwriter. (Various issues) Fisher, Ernest M. (1951) Urban Real Estate Markets: Characteristics and Financing. New York: National Bureau of Economic Research. Foote, Christopher L., Kristopher S. Gerardi, and Paul S. Willen. (2008) “Negative Equity and Foreclosure: Theory and Evidence.” Federal Reserve Bank of Boston Public Policy Discussion Papers 08-3. Halliburton, Robert A. (1939) The Real Estate Bond House. Franklin, IN: Indiana University. Harrington, Scott E. (2009) “The Financial Crisis, Systemic Risk, and the Future of Insurance Regulation” in Journal of Risk and Insurance, vol. 76 no. 4, pp. 785-819. Insurance Field. (Various issues) Kandarian, Steven A. (2013) “Life Insurers as SIFIs: A Case of Mistaken Identity?” Speech given at the US Chamber of Commerce. https://www.metlife.com/assets/cao/pr/Capital- Markets-Summit-Remarks-FINAL.pdf Kniskern, Philip W. (1926) “Methods of Issuing Guaranteed Mortgages.” Annals of Real Estate Practice, Volume IV: Real Estate Finance, Proceedings & Reports of the Mortgage & Finance Division. Chicago: National Association of Real Estate Boards. Jones, Jesse. (1951) Fifty Billion Dollars: My Thirteen Years with the RFC, 1932-1945. New York: Macmillian. Mackall, Luther E. (1929) Principles of Surety Underwriting. Chicago: The Spectator Company. McKenna, Edward D. (1927) “The Guarantee of Mortgage Bonds by Surety Companies.” Annals of the American Academy of Political and Social Science. Vol. 130, pp. 131-136. Moody’s Investor Services. Morgan, Willis D. (1927) “The History and Economics of Suretyship.” Cornell Law Review, vol. 12, no. 2, pp. 153-171. National Convention of Insurance Commissioners. (1933) Proceedings of the Sixty-Fourth Annual Session of the National Convention of Insurance Commissioners. Columbia, SC: State Co Morgan, Willis D. (1927) “History and Economics of Suretyship.” Cornell Law Review, Vol. 12, No. 2, pp. 153-171. National Surety Papers. “In the matter of the application of The People of the State of New York, by George S. Van Schaick, as Superintendent of Insurance of the State of New York, for an order to take possession of the property and rehabilitate the National Surety Company.” Index number 16491/1933. Papers filed in the Supreme Court of New York State, New York County. National Underwriter. (Various issues) New York Times. (Various Issues) Reconstruction Finance Corporation Papers. Paid Loan Files, Box 874, Record Group 234, National Archives, College Park MD. 34

Spectator. (Various issues) Wall Street Journal. (Various issues) Weekly Underwriter. (Various issues) Wheeler, William J. (2012) Testimony before the United States House of Representatives, Financial Institutions and Consumer Credit Committee. The Yale Law Journal. (1934) “Rehabilitation of the National Surety Company.” Vol. 43, No. 7 (May), pp. 1146-1156. 35