Corporate Governance and Risk Management at Unprotected Banks: National Banks in the 1890s, Study notes of Corporate Finance

Download Corporate Governance and Risk Management at Unprotected Banks: National Banks in the 1890s and more Study notes Corporate Finance in PDF only on Docsity! Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, D.C. Corporate Governance and Risk Management at Unprotected Banks: National Banks in the 1890s Charles W. Calomiris and Mark A. Carlson 2014-08 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. Corporate Governance and Risk Management at Unprotected Banks: National Banks in the 1890s Charles W. Calomiris and Mark Carlson* Abstract Managers’ incentives may conflict with those of shareholders or creditors, particularly at leveraged, opaque banks. Bankers may abuse their control rights to give themselves excessive salaries, favored access to credit, or to take excessive risks that benefit themselves at the expense of depositors. Banks must design contracting and governance structures that sufficiently resolve agency problems so that they can attract funding from outside shareholders and depositors. We examine banks from the 1890s, a period when there were no distortions from deposit insurance or government interventions to assist banks. We use national banks’ Examination Reports to link differences in managerial ownership to different corporate governance policies, risk, and methods of risk management. Formal corporate governance is lower when manager ownership shares are higher. Managerial rent seeking via salaries and insider lending is greater when managerial ownership is higher, and lower when formal governance controls are employed. Banks with higher managerial ownership target lower default risk. Higher managerial ownership and less-formal governance are associated with a greater reliance on cash rather than capital as a means of limiting risk, which we show is consistent both with higher adverse-selection costs of raising outside equity and with greater moral-hazard with respect to risk shifting. Keywords: Manager ownership, corporate governance, rent seeking, risk preferences, bank failures, risk shifting, adverse selection JEL Codes: G21, G32, N21 * Calomiris: Columbia Business School, NBER and IMF. Carlson: Board of Governors of the Federal Reserve. We thank Stijn Claessens, Luc Laeven, and seminar participants at the Board of Governors, at the Federal Reserve Banks of Atlanta, Cleveland, and New York, at Columbia University, at Tilburg University, at Dartmouth’s Tuck School, and at Erasmus University for comments. Ianni Drivas provided valuable research assistance. The views expressed in this paper are solely those of the authors and do not necessarily reflect those of the Federal Reserve Board, the International Monetary Fund, or their staffs. 3 Moral-hazard issues can be mitigated through various measures, including short-term debt contracting; a first-come, first-served rule for bank liquidation; and actions by bankers that credibly signal good risk management, including the maintenance of a minimum amount of cash assets (Calomiris and Kahn 1991, Calomiris, Heider and Hoerova 2013). If bank debt holders are protected by deposit insurance or other guarantees, however, moral hazard can be exacerbated because bank debt holders lose their incentive to monitor and control banks’ risk taking (Calomiris, Heider and Hoerova 2013). Gorton and Rosen (1995) argue that, when faced with a declining industry, managers may boost profits to hide poor prospects from shareholders. These agency issues have received additional attention after the recent financial crisis. Many of studies, in addition to those cited above, have debated the extent to which corporate governance and manager incentive schemes influenced how banks fared during the crisis (Acharya et al. 2009; Berger, Imbierowicz, and Rauch 2012; Ellul and Yeramilli 2010; and Fahlenbrach, Prilmeier and Stulz 2012; Senior Supervisors Group 2008; Mehran, Morrison and Shapiro 2011). Although the nature of conflicts of interest between bankers and their funding sources differ between outside equity and debt, there is also considerable overlap in the usefulness of corporate governance tools for addressing many aspects of conflicts of interest that are common to both types of outside funding sources. For example, the presence of outside directors, or the “bonding” of management, should mitigate the risk of defalcation, which benefits both outside stockholders and debtholders. Corporate governance policies of banks should arise endogenously, in part to reduce the costs related to the two sets of conflicts of interest in risk taking – that is, the conflict between shareholders and debtholders, and the conflict between managers and shareholders. Understanding how governance policies respond to such conflicts, and what effects ownership structure and governance policies have on risk taking, is highly challenging in the current regulatory environment, where policies such as deposit insurance, too-big-to-fail (TBTF) bailouts, 4 and legal restrictions on controlling ownership interests in banks, which remove the disciplinary incentives of debtholders and limit the ability of equity holders to concentrate ownership (on the effects of TBTF, for example, see Acharya, et. al 2009).3 To improve our understanding of how ownership structure affects corporate governance, and how ownership structure and corporate governance affect banks’ risk management, we examine the links among ownership, governance, and risk management during a period prior to the establishment of a regulatory safety net for banks. During the National Banking Era (1863-1914), government protection was absent, and the latitude for voluntary governance decisions by banks was great. We observe large cross-sectional differences in the ownership structure of national banks, as well as great variation in their choices for organizing corporate governance. Banks also structured their portfolios very differently, and displayed important differences in their management of risk – indicated by balance sheet differences and the patterns of bank failure during the panics of this era, especially the severe Panic of 1893. Cross-sectional differences in ownership, governance choices, portfolios and risks, under a common and relatively laissez-faire regulatory environment, makes national banks’ experiences in the 1890s an ideal laboratory for examining how manager ownership and board oversight are related to rent seeking, portfolio choice, and failure risk, in an environment free of many of the regulatory distortions that affect those decisions today. Another advantage of focusing on cross-sectional variation among national banks is that their business models were quite similar (in contrast to today’s banking system, in which small banks focus on lending and deposit taking, while global universal banks undertake a much wider range of activities for a quite different customer base). Corporate governance in the historical U.S. banking context has been the subject of numerous prior studies. One of the most important themes of that literature, which is not present in other contexts, has been the connection between stock ownership and lending. In today’s banks, 3 The so-called separation of banking and commerce places special constraints on who is permitted to exercise a controlling interest in a bank. 5 there are strict limits on loans to officers and directors, and it is considered inappropriate to provide better terms to loans offered to officers, directors or other large stockholders. Historically, in the “unit” banking system of the United States, where banks were local, single-office enterprises, banks acted as “loan clubs” for insiders, who were often large shareholders with significant formal or informal control rights. Generally, the empirical literature has taken a benign view of insider lending, arguing that it facilitated value creation and risk management because insiders had strong incentives to screen and monitor one another (Lamoreaux and Glaisek 1991; Lamoreaux 1994; Meissner 2005; Haber and Maurer 2007; Pearson and Taylor 2012; Hansmann and Parglender 2012; Freeman, Pearson and Taylor 2012). Bodenhorn (2013) finds that bank value increases with the number of individual blockholders, but declines with the number of institutional blockholders – that is, blockholders who are not part of the loan club. According to this evidence, loans clubs increased the value of bank stock because insiders valued preferential access to lending that was attached to their blockholding status.4 National banks, which were first chartered during the Civil War, operated alongside state- chartered banks. Although national banks were chartered under the same charter rules throughout the United States, like state-chartered banks they operated as single office (unit) banks. As we will show, national banks, like the state-chartered banks studied by Lamoreaux, Bodenhorn and others, engaged in large amounts of insider lending. Thus, it is important to take into account the effects of ownership and governance rules on this aspect of bank behavior. The data we use come primarily from national banks’ Examination Reports, a source which, to our knowledge, has been little used, and never used for quantitative analysis of the questions we address here. These Reports provide very detailed pictures of the banks and the bank examiners’ 4 Interestingly, 19th century corporate chartering rules often employed voting rights rules that reduced the voting power of large shareholders, largely to reduce concentration of control over corporations. Although these departures from one share-one vote rules were common for many firms, they were less common for banks (Hilt 2008). This may have reflected the desirability of encouraging insider blockholding, as well as the relative absence of the political consequences of control over a bank once banks became chartered freely (roughly around the second quarter of the 19th century). 8 governance affected bank portfolio structure, performance and failure probabilities during the Panic of 1893. Building the analysis out in this way provides a rich perspective on the connections among ownership, governance, rent seeking, and risk choices. Our results on risk taking indicate that managers who own a greater proportion of the bank’s stock take less risk according to any measure of risk we employ. For example, with respect to forward-looking measures of risk, managers with large equity stakes in their banks are less likely to rely upon high-cost “borrowed funds” and are also less likely to be involved in real estate lending. Both activities were perceived by contemporaries as riskier and such perceptions are generally borne out in the Panic. Ex post measures of risk – the proportion of troubled loans, the estimated probability of bank failure, or the forecasted losses anticipated by the bank’s examiner – paint a similar picture; greater management stakes are associated with lower default risk. We view these results as consistent with the idea that managers with a larger share of their wealth invested in the bank were more risk–averse in their risk management practices. Banks with lower managerial stakes, and consequently with more formal governance policies, tended to undertake greater levels of risk. That finding is consistent with outside directors, who represent the interests of all equity holders, as preferring a slightly higher level of risk. The preference for lower risk appears to have been beneficial during the Panic of 1893 as we find that banks with higher manager ownership were less likely to fail. This affect is due largely to how these banks structured their balance sheets as when additional balance sheet controls are included, the direct effect of manager ownership concentration is reduced. Finally, we investigate how banks differed in the financial structures they chose to manage risk. Banks seeking to reduce the risk of default on their debts that is traceable to risks of default on their loans can use two alternative risk management tools in combination: a higher cash-to-asset ratio (on the asset side of the balance sheet), or a higher equity-to-asset ratio (on the liability side of the balance sheet). We find that banks with higher managerial ownership concentration relied 9 more on cash assets, and less on equity, to control bank default risk. This finding is consistent with the view that high managerial ownership, and informal governance, make bank managers’ behavior less observable and less controllable. It is harder to observe and control risk management at banks without formal governance structures. Similarly, it is harder for outsiders to observe the level of risk in such banks. Under these circumstances, banks will suffer from greater asset substitution risk, and greater adverse-selection problems (if they were to attempt to raise additional sources of outside equity). Greater asset substitution risk will tend to lead banks to rely more on cash as a means of signaling good risk management practices (Calomiris, Heider and Hoerova 2013). Greater adverse-selection problems raise the cost of equity finance, and thus also lead to a greater weight on cash in controlling default risk. Given the paucity of equity offerings in our sample, we think the asset substitution channel is more likely to be the important one. The paper is organized as follows. Section 2 provides an illustrative model of ownership, governance, and asymmetric information. Section 3 discusses the data sources and the sample. That section also explains the construction of our corporate governance measures and the variables we use as indicators of rent seeking and risk preferences. The baseline analysis is contained in Section 4 while a variety of robustness checks are presented in Section 5. Section 6 concludes. Section 2. A Model of Endogenous Asymmetric Information We begin with the simplest possible model of corporate governance choice, where the assets of the bank consist entirely of loans and the financing of the bank consists only of stock. We relax these assumptions subsequently, and show that the central implications of the model – that is, that higher managerial wealth tends to reduce the reliance on formal corporate governance – also hold when we allow for deposit financing and the holding of cash assets. A banker is endowed with wealth (E) and lending opportunities (a given number of profitable potential loans that he might undertake). Each loan is normalized to be of identical, 10 unitary size. The number and amount of loans made, X , is between 0 and Xmax. For simplicity, we assume that the bank holds only loans and is financed entirely by equity provided by the banker and outside investors (there is no bank debt). The manager’s equity share of the bank, m, is therefore E/X. When we add deposit liabilities and cash assets to the model – as in Calomiris, Heider and Hoerova (2013) – the main conclusions of the model are the same, but additional conclusions follow with respect to the role of cash in incentivizing good risk management. Interestingly, in this framework, cash plays an important role in incentivizing good risk management whether or not outsider financing is in the form of debt or equity. This warrants emphasis: unlike the discussion of Jensen and Meckling (1976), the problem of risk shifting in this model is a conflict between the insider/manager and all outside funding sources, not just debt holders.5 In the simplified model, bank managers face incentives to increase risk in value- destroying ways (so-called “asset substitution” or “risk shifting”) even though debt finance is absent. Minority shareholders, like creditors, have an interest in ensuring proper risk management by bank managers, which can either be achieved through higher managerial stakes in the bank or formal corporate governance. Outside equity is provided by a single outside investor. The outside investor and the banker are risk-neutral and have identical reservation returns of R, which represents the gross return they could earn on an alternative to lending. The loan opportunities of the banker are worth pursuing, 5 In the model presented here, the outside equity investor either becomes an insider by being invited to participate in governance, or remains uninvolved in governance knowing that the banker will invest in risk management due to a sufficiently high level of m (the banker’s proportion of ownership). As we discuss further below, and as Calomiris, Heider and Hoerova (2013) show, in this model, the optimal contract for investors who remain outsiders (and therefore are not able to control risk management) would be senior deposits in a bank with cash reserves as well as loans. The key differences in assumptions between that model and the simplified one presented here are the availability of a single large outside investor (assumed here) and the possibility of establishing oversight of risk management by that outside investor. Calomiris, Heider and Hoerova (2013) assume that outside investors are fragmented. Their solution to incentive- compatible risk management entails the use of deposits and cash reserve holdings. In a small bank, with a single large outside investor, and the possibility of direct monitoring of the banker by that outside investor, depositor withdrawal threats and idle cash holdings are not necessary to achieve efficient risk management. The Calomiris, Heider and Hoerova (2013) model captures the role of deposits and reserves in bank corporate governance, while the model presented here highlights the relationship between the banker and a large outside equity investor, which is not considered by Calomiris, Heider and Hoerova (2013). 13 the bank at Xmax) depends on how much the banker receives under each of those alternatives. Recall that, if the outside investor is included in corporate governance, he will split the rents with the banker, and therefore, both the banker and the outside investor/director will each earn an identical “salary” of S = (Y – R)/2. If the following condition is satisfied, the banker will earn more by choosing to include the outside investor in governance and operate the bank at Xmax: (5) ER + Xmax(Y – R)/2 > ER + X*(Y – R). So long as Xmax > 2X*, this condition is satisfied. Note that, in any comparative static calculation, Y affects the governance decision only indirectly through the positive effect of Y on X* (i.e., dX*/dY >0, implying that, ceteris paribus, higher Y makes it less likely that outside investor’s will be invited to participate in governance). The above model has clear implications for corporate governance decisions and their consequences. Depending on the size of rents per loan, and the number of loans available to the banker, he will decide whether to run the bank with no outside oversight or to include the outside investor in oversight. If the outside investor is included in oversight, then “asymmetric information” and “asset substitution risk” will be eliminated, and the banker’s salary will be lower, as he is forced to share rent with the outside investor. Section 2.1. Adding Cash Assets and Deposits To the Model The model can be extended to allow bankers to choose to hold cash in a credible and observable form. If cash assets are added to the model without also allowing for senior deposit claims, cash holdings serve no purpose. To see why, consider the effect on equation (1) of bank cash holdings, C. Because cash is riskless, the banker will receive, in addition to the payoffs described in equation (1), an amount mC irrespective of whether the banker undertakes risk management. Thus, cash has no effect on the banker’s risk management effort. 14 As Calomiris, Heider and Hoerova (2013) point out, however, outsider financing via equity is not generally the optimal contract under these circumstances. By giving outsiders a senior claim on the cash flows of the bank, the banker ensures that when risk is not managed properly, and when low payoffs occur, outsiders will receive all of the cash, not just (1-m)C. Thus when outsider financing is partly in the form of deposits, and bankers are able to hold cash, bankers are able to commit to proper risk management by holding a sufficient amount of cash assets. Deposits and cash affect risk management because, unlike outside equity financing, deposit financing does not dilute the upside of the banker’s profit, and unlike outside equity holders, deposits receive all of the bank’s cash assets when the banker chooses not to invest in risk management and the bad outcome occurs (with probability 1-p). A fully realistic model (which would have to be much more complicated that either the framework presented here, or that of Calomiris, Heider and Hoerova 2013) could allow for both deposits and outside equity sources of funding. For example, with a more continuous distribution of bank earnings outcomes, one could derive two forms of outside financing by assuming that depositors are relatively risk-averse small investors who desire fixed claims, but that a single, large investor, who is less risk-averse, is willing to provide outside equity financing. In that setup, we conjecture that the same two basic results derived above and in Calomiris, Heider and Hoerova (2013), respectively, will hold: (1) bankers that limit the amount of their risky lending relative to their own equity interest in the bank, will be able to attract both depositors and an outside equity investor without establishing formal corporate governance protections, and (2) bankers that choose not to establish formal corporate governance protections can expand the amount of risky lending in which they can engage, and thereby increase the rents they derive from banking, by raising much of their outside financing in the form of senior debt and holding sufficient cash assets. In other words, this framework implies that bankers that choose not to engage in formal corporate 15 governance will tend to rely less on equity in their financing, and will hold a larger fraction of their assets in cash. In our empirical results, we will test, and confirm, these predictions. Section 3. Data We gather a variety of information on individual banks using the Call Reports and the Examination Reports. In this section, we describe the data sources and the definitions of the variables used in this study. Section 3.1 The Sample Our sample contains 206 banking institutions, which consists of all the national banks located in 37 cities. As national banks (i.e., those chartered by the federal government), these institutions were subject to the same set of rules and regulations regardless of where they were located. Beneficially for our purposes, all the banks were unit, or single office, banks which makes it easier to control for differences in local economic conditions. National banks were required to provide information to the Comptroller of the Currency, their primary regulator, several times a year. One method was through the Call Report, which contains information on the banks’ balance sheets and was filed about five times a year. The second method of providing information about themselves consisted of Examination Reports filed by examiners who visited each bank once or twice a year. To be included in our sample, the banks needed to have provided information for the September 1892 Call Report and to have had at least one Examination Report completed prior to May 1893 (the onset of the Panic). Those are the Reports that provide the information used for the analysis.6 6 Two banks file the September 1892 call report but close prior to May 1893. For these institutions, we use the examination report nearest closure, so long as it was filed at least [four] months prior to closure. 18 Section 3.3 Ownership and Governance Variables The individuals most responsible for running the bank were its senior managers, in particular the president, vice-president, and cashier (essentially, the chief operating officer of the bank). They played a large role in making loans and arranging the funding of the bank, including whether the bank borrowed from other banks via short-term, higher-interest “borrowed” money. These individuals tended to own shares in the bank and were frequently also on the board of directors (the President of the bank was always on the Board, and the others typically were, too). A key variable in our analysis is the share of the bank’s stock owned by the officers of the bank. We focus, in particular, on the fraction of outstanding bank shares owned by the president, vice- president, and cashier.10 The average portion of shares owned by these three officers, as reported in Table 2, was 25 percent. The histogram in Chart 1 provides a better indication of the distribution of managerial ownership. At most banks in the sample, ownership by the managers is fairly modest; the three top managers owned less than 6 percent of outstanding shares for about 30 percent of the sample. There are also cases of significant ownership concentration; the top three managers owned at least half the outstanding shares in nearly 10 percent of the sample. The behavior of the managers could be constrained by the Board of Directors. Boards ranged in size from 4 members to 23 members. Some Board members owned significant stakes in the bank. Others were prominent businessmen that might provide business to the bank.11 A histogram of ownership by outside directors is shown in Chart 2. The average portion of shares owned by outside (non-officer) directors was 15 percent but it reached as high as 57 percent. Presumably, the larger the portion of shares owned by the outside directors, the more they could 10 We obtain the number of outstanding bank shares by dividing bank capital by 100 (as bank capital was typically carried at book value based on share prices of $100 per share). In a few cases the examiner indicated the number of shares outstanding and these reports confirm that our procedure is correct. In a few other cases the examiner reported that the value of capital had previously been written down and shares revalued. We believe that we have made all the appropriate corrections for these write-downs. 11 For instance, a Mr. Proctor and a Mr. Gamble served on the board of the Citizens National in Cincinnati. 19 influence the behavior of managers. The ownership by all other individuals is shown in chart 3. As can be seen from this chart, individuals who are neither managers nor bank managers own a majority of the shares in about two-thirds of the banks in our sample. There were also other ways that the board could exert control over managers. One way was by maintaining an active independent discount committee containing at least one outside director to review and approve loans proposed by the managers. Such a committee was maintained by 60 percent of banks. Another way of exerting control was by meeting frequently. Boards that met infrequently, such as semi-annually, presumably had little influence on the managers. The board met monthly or more frequently in nearly two-thirds of the banks in our sample. In cases where the managers comprised a significant portion of the board, there was presumably little independent oversight; when outside directors dominated the board, they could presumably exert more control. In our sample, the median portion of the board that consisted of outside directors was 71 percent; we create an indicator variable equal to one when the portion of directors are outsiders is above the median and is zero otherwise. Our measures of the reliance on independent directors, of the existence of a loan review committee, and of the frequency with which it met are similar to other measures used to analyze corporate risk management in modern financial institutions, such as the “active board risk committee” of Ellul and Yerramilli (2010). Another way of influencing bank management was requiring bank managers to post surety bonds. These bonds would offer the directors (or receiver) a way of recovering funds in the event the manager committed some specified act, typically some type of fraud that caused losses to the bank. Bonds could be personal or provided through a surety bond agency (which often required that the person being insured post some type of collateral).12 Surety bonds were most often required for the cashier, who oversaw the books and for whom the possibility of fraud was therefore highest (nearly 60 percent of cashiers posted bonds). Other managers also were required 12 For more information on surety bonds see Lunt (1922). 20 to post such bonds (the President posted a bond in 33 percent of our sample and the vice-president did so in 12 percent of the sample). In Chart 4, we illustrate the relationship between manager ownership and one of the indicators of corporate oversight: the fraction of the Board consisting of outside directors. The negative relationship between these two measures indicates that more manager ownership tends to be associated with less oversight. Moreover, not only are each of the measures of Board oversight negatively correlated with manager ownership, Table 3, but they are all positively correlated with each other. Although we investigated the impact of each of these measures of Board control on managerial behavior, it is useful for our purposes to create an index that aggregates the different measures into a single corporate governance index. We do so by summing the five indicator variables.13 Examiners seem to have understood that banks could achieve good management of risk with or without active oversight of management by the Board. Below are excerpts from the Examination Reports of two banks, one with the minimum corporate oversight score of 0 and the other with the maximum score of 5. In neither case did the examiner have concerns about the management of the bank or the soundness of the bank, even though the examiner was aware of the clear differences in the oversight being exercised by the Board. Oversight score of 0 - Comment on the Board: Frequent meetings are not held by the directors of this bank and records only show that formal meetings are held to declare dividends. No mention being made of their having examined or approved loans and discounts at such times, and there is no report of discount and examining committee having acted. The management is apparently with Mr. Gates, the president of the bank. 13 We also tried aggregating the five indicators by taking the first principle component, similar to Ellul and Yerramilli (2010). All the five indicators had positive and roughly equal weights. Thus, the first principle component was not so different than the simple average so we stick with the average for simplicity. 23 management’s belief that earnings will persist. Dividend payment differences may also reflect different growth opportunities; retaining profits raise the amount of equity invested in the bank, which ceteris paribus, lowers the bank’s default risk, and thus increases the capacity of the bank to grow its assets. To analyze dividend payouts, we consider is the ratio of dividend payments relative to shares outstanding if dividends were paid during the past six (banks typically paid dividends semi-annually, in June and December). 17 Section 3.5 Risk We consider several indicators of the risk of the bank. Some of these indicators focus on aspects of bank asset risk – based either on objective criteria about the composition of bank assets (e.g., the ratio of real estate loans), or on examiner expectations (e.g., the ratio of “troubled” loans, or the losses forecast on assets). Another asset-side indicator of risk is the amount of other real estate owned among bank assets. This asset category typically represented properties seized when loans went into default. Finally, we measure risk based on failure outcomes. Our measures of risk that employ examiner opinions use specific categories contained in the Examination Reports. Examiners reported the amount of bad debts or other suspended or overdue paper; the proportion of loans consisting of these items – which we define as “troubled loans” – is a useful metric of loan quality. Examiners also provided estimates of likely losses on assets (not just loans but on securities and other items as well, such as non-income generating assets such as furnishings). The two primary tools of risk management for banks were the equity-to-asset ratio and the cash assets-to-total-assets ratio. Equity, or net worth, is measured as the sum of paid in capital plus cumulative retained earnings held as surplus or undivided profits. There were no equity ratio requirements, although banks were required to maintain minimum amounts of capital and surplus. 17 As an alternative, we also looked at whether the bank paid out dividends during the past six months. The implications from those results are similar. 24 Estimating the demand for cash assets is complicated by the legal minimum requirements of cash relative to deposits. Cash reserve requirements specified a certain level of cash and deposits in reserve city banks relative to deposits and net due to banks. As we show in our regression analysis, however, regulatory constraints on holdings of cash reserves were not binding on banks’ demands for cash assets.18 Section 3.6 Other Controls We also include a number of variables to control for local conditions. At the county level, we gather information on population and the share of county income from agriculture. These variables are from the 1890 census. An important feature of the banking system during the National Banking Era was the system of interbank depositing of reserves. National banks were required to hold cash and interbank deposits against their own deposit liabilities. Banks outside major cities need to hold a 15 percent reserve, three-fifths of which could be held as deposits at banks in larger “Reserve” cities or, “Central Reserve” cities—New York, Chicago, or St. Louis. Banks in Reserve cities needed to hold a 25 percent reserve, half of which could consist of deposits in a “Central Reserve” city. Deposits in New York played a key role in the settling of interregional payments. Many banks held deposits with banks in New York. Moreover, banks in New York provided a substantial amount of interbank loans through rediscounting. To capture the potential importance of proximity to New York in affecting banks’ risks and operations, we include the log of the distance of banks from New York as a control. We also include an indicator for whether the city in which the bank is located is a 18 The Examination Reports are based on non-scheduled (surprise) examinations of national banks. In addition to these examinations, there were regularly scheduled call reports of bank balance sheets in June and December. We hypothesize that banks did not maintain required reserves continually through the year, but that they may have engaged in window dressing to meet their reserve requirements temporarily on the June and December call dates. The penalties available to the Office of the Comptroller of the Currency appear to have been limited to rather extreme measures (suspension of dividends or revocation of charter). Banks that failed to meet their reserve requirements on examination dates likely were told to correct the problem, which they could do by window dressing their balance sheets on the next reporting date (the June or December call). 25 reserve city, to capture the possible effects of differences in interbank relationships and reserve requirements on bank behavior. We do not include bank asset size as a control variable because, as our model shows, it is an endogenous variable. Treating bank size as an exogenous variable, and adding it to the list of controls would not affect any of the conclusions derived below. Section 4. Analysis We are interested in how the different ownership and corporate governance variables affected behavior. As these variables are clearly inter-related, we start by presenting our approach to identifying the linkages among ownership structure, governance choices, rent seeking, and risk management. We then review our findings. As we noted in our review of the literature on manager/stockholder conflicts, it is not clear a priori whether increased shareholdings by managers lead them to take more or less risk. Risk- averse managers that hold a large share of their wealth in the form of bank stock, and whose human capital depends on the fortunes of the bank, should generally prefer less risk than the outside shareholders. In some states of the world, however – if hidden losses are large – managers may prefer to undertake more risk than outside shareholders. With respect to rent extractions, we expect that all managers would prefer to extract greater rents from the bank, and that those with more equity shares will be more successful in doing so, especially if they are not subject to formal oversight. Outside directors presumably will try to influence the behavior of the managers, particularly if that behavior deviates from what outside stockholders would prefer. We expect that when the managers own a larger equity stake, and outside directors own a smaller stake, that fewer oversight tools will to be employed. We also expect that outside directors will try to reduce any rent seeking by the managers. With respect to risk preferences, outside directors should represent the interests of outside shareholders. If, for example, managers with large equity stakes tend to 28 (relative to assets) increase the expenses of the bank and reduce funds paid out to shareholders as dividends. When non-management shareholders own a greater proportion of stock, they may be able to better limit salaries. The results, shown in Table 5, are consistent with that idea and indicate that when the managers own more shares, they tend to pay themselves higher salaries. Another way of extracting rents from a bank is for the owners to lend to themselves to finance their outside projects. There has been considerable prior academic analysis of this issue, which indicates that insider lending is not always value-destroying or risky (Lamoreaux 1994, Haber 1995). We look at two variables related to insider lending. The first is the amount of loans made to all insiders (board members and management) relative to all loans. The second is the proportion of all insider loans going to managers. We expect that managerial ownership and governance measures will be associated with both these variables. Interestingly, in regressions not reported here, we do not find any evidence that our measures of ownership or Board oversight are associated with insider lending. We do find, however, that ownership and governance structure strongly influence who receives those insider loans (Table 6). At banks where the management owned a greater proportion of the stock, a greater fraction of insider loans went to the management. When there were more corporate governance controls, more of the insider loans were made to the outside directors. With respect to dividends, we find, in Table 7, that when more shares are owned by managers, then dividend payments are higher. While this finding is consistent with the idea that institutions with higher managerial ownership provide greater payouts to owners, it is also consistent with the idea that these institutions are more profitable. Taken together, our results regarding salaries, insider lending, and dividend payments are consistent with the idea that when managers own a greater fraction of the equity shares of the bank, they extract greater rents from the bank through higher salaries and more loans to themselves. They do not, however, limit dividend payments (because this means of rent extraction 29 would harm them as stockholders). Stronger oversight by the Board of Directors tends to be associated with less rent extraction by the managers but somewhat greater extraction by the outsiders on the Board (insider lending became skewed more toward the outsiders on the Board). All parties appeared interested in maintaining strong dividend payments. Section 4.3 Corporate Governance, Balance Sheet Composition, and Risk Taking We begin our analysis of the relationship between risk choices and ownership and governance structure by focusing on measures of risk from the liability side of the balance sheet. With respect to the composition of liabilities, we examine bank reliance on the use of borrowed funds, which previous research has shown is a forecaster of bank distress (Calomiris and Mason 1997, 2003, Carlson 2010). Borrowed funds were more expensive and had to be secured; use of these funds suggests a greater level of risk. As noted earlier, due to data limitations in tracking the exact amounts of borrowed funds, we use a probit specification to test whether our ownership or governance variables are associated with the use of such funds. We find, in Table 8, that banks where managers are more significant owners are less likely to rely on borrowed funds from other banks. To economize on the reporting of results, our subsequent findings for other endogenous variables are summarized in Table 9, which omits the various control variables and focuses on the key coefficients of interest (the relationship among managerial ownership, governance score, and other variables of interest). With respect to measures of risk based on the asset side of the balance sheet, we consider the composition of loans. As noted earlier, real estate loans were generally considered to be riskier and were forbidden by the National Bank Act, but banks could use mortgages to secure debts previously entered into. As shown in Table 9, when management owns more shares in the bank, the bank tends to have fewer mortgages on its books. 30 There are a number of outcome variables that also reflect the risk preference of the banks. The three measures we consider are other real estate owned relative to assets, as well as the examiner’s assessment of problems, measured by the share of troubled loans to total loans and by the estimated losses on assets relative to total assets. We report in Table 9 that greater ownership by management is associated with lower values of all these measures. There is no association between board controls and troubled loans. However, estimated losses do appear to be reduced by increased board oversight measures, a finding that suggests that these governance structures are at least partially effective. On closer inspection of the composition of expected losses, we found that loan losses were not driving this result; rather, losses related to greater expenditures on furnishings are a primary contributor to the greater expected losses of banks with both low managerial ownership and low board oversight. That result is intuitively appealing: excessive expenditures on furnishings are a wasteful, value-destroying use of funds that would not be chosen in a disciplined environment. Table 9 also examines the effects of ownership and governance on bank survival. We find that increased ownership by management is associated with a reduced likelihood that the bank closes between October 1892 and December 1893, though the bulk of the closures occur during the Panic of 1893. When managers had a greater ownership stake, they took less risk and were thus less likely to succumb. However, the direct effect of managerial ownership is reduced when additional balance sheet controls are included. This finding suggests that the benefits of managerial ownership on survival operate largely through the balance sheet choices made by the managers. We now turn to the question of how ownership and governance structure are related to greater or lesser reliance on particular tools of risk management. Our analysis of bank loan composition in Table 9 showed that higher managerial stakes and greater oversight were associated with less risky lending, but this is only one of the main influences on bank default risk. In finance theory, the default risk of a bank is mainly determined by three variables: the riskiness of 33 These are particularly apt examples of the sorts of behaviors we identify in the empirical analysis. More generally, in reviewing the anecdotal information, we find that there tended to be more concerns about banks with low management ownership and low governance and few concerns about banks with high ownership and governance. Section 5 Robustness and extensions Here we report a variety of robustness checks and extensions of the baseline analysis. In conducting this additional analysis, we focus on selected indicators of risk. We summarize these extensions as follows: First, we re-run our regressions, allowing stock ownership structure to be endogenous. As an instrument, we use managerial turnover events – moments when one bank president replaced another. Second, we provide an alternative conditional-mean analysis of banks to test the proposition that managerial ownership and formal corporate governance were substitute forms of discipline over managerial behavior. Third, we show that our results are robust to separately considering the components of the governance score used above. Fourth, we show that managerial compensation structure also affects management incentives toward risk; managers with more of their compensation in dividends relative to salary undertook lower risk. Fifth, we show that the association between greater risk and formal corporate governance is accentuated in the presence of a large outside blockholder (analogously to Laeven and Levine 2009). Sixth, and finally, we find that our results were robust to the inclusion of various controls. Section 5.1 Endogenizing Ownership Structure Using Managerial Turnover Here we treat managerial ownership as endogenous, and we instrument either managerial ownership or corporate governance score using events associated with managerial turnover. We 22 Columbia National Bank, Tacoma, WA 34 expect (and find) that a managerial turnover event (such as the death of a bank president), which can be thought of as an exogenous reduction in the level of E in our model, is associated with a reduction in the managerial ownership share of the bank and an increase in corporate governance. To verify that managerial turnover is traceable to exogenous events, we performed web- based searches, and also searched through newspapers available through the various digitized search engines maintained by the Library of Congress, to find information about the changes in bank presidents between 1882 and 1892 for banks in our sample. We used both the bank names and the presidents’ names to obtain information about the reason for managerial turnover. Because the sources covered by these digital databases tend to be biased toward larger cities’ newspapers and national publications, we were not able to find information about many of these management changes. For the 137 relevant turnover events in our sample, we find information explaining the reason for the management change for 37 of the events. For 65 of the events for which information was lacking, we were unable to locate any newspapers for the relevant time period and location. For 35 of the events for which information was lacking, local newspapers for the relevant time period were available, but we were unable to find any story about the changes in bank presidents. Managerial turnover generally was associated with clearly identifiable exogenous events.. In the cases we were able to trace, the causes of turnover included death or severe illness (23 cases), election to public office or other new career opportunity (9 cases), retirement (2 cases), and other apparently exogenous circumstances (one departure in the wake of a cashier embezzlement, one because of business problems unrelated to the bank, and one because the president declined re- election). We also checked to see if there were notable changes in the condition of the banks as indicated by changes in the capital stock around the time the president changed. We found no evidence that changes in capital systematically preceded, followed, or were coincident with turnover. 35 In our regressions, the turnover instrument is measured as the number of times the President of the bank changed between 1882 and 1892. The first-stage regressions – measuring the effect of turnover on managerial ownership and corporate governance score – are reported in Table 12. Clearly, turnover results in reduced managerial ownership and greater use of formal governance. Table 13 presents, for each endogenous variable of interest, the results for the key parameter of interest (the effect of instrumented managerial ownership, or alternatively, instrumented governance score, on the endogenous variable of interest). For purposes of comparison, Table 13 reports the estimated coefficient for the second-stage regression alongside the comparable coefficients reported in Table 9. The results are quite similar, although, not surprisingly, the new IV results are less precisely estimated. Section 5.2 Conditional Mean Tests Another way to test the proposition that managerial ownership and formal corporate governance acted as substitute means for disciplining managerial rent seeking and promoting risk management is to divide banks into four groups, using a two-by-two matrix that measures each bank’s combination of managerial ownership and formal corporate governance score. The four groups are defined as (1) high-managerial ownership and high-formal governance score banks, (2) high-managerial ownership and low-formal governance score banks, (3) low-managerial ownership and high-formal governance banks, and (4) low- managerial ownership and low-formal governance banks.23 Our findings for these groups are reported in Table 14. As predicted, the fourth group (which lacks either a high degree of managerial ownership or formal governance) is riskier. This group was more likely to use borrowed funds, more heavily invested in real estate loans, and had greater expected losses than the other groups. Furthermore, 23 This approach also helps assure us that our earlier results were not driven by outliers in our concentration measure. 38 there is an outside director with more shares than any of the top three managers (individually, not collectively). Such an outside director exists for about 20 percent of the banks in our sample. When a director with a large number of shares is on the Board, we find that the presence of such an individual tends to magnify the prior result of greater risk taking. For instance, the bank tends to have greater shares of loans related to real estate. The tendency for greater risk taking appears to be consequential as banks with large-shareholding directors also are more likely to close during the panic. Section 5.6 Additional Control variables We also tried including a variety of other variables as controls. One such variable was the average score for banks in the same city, which might reflect the best practice of the neighboring banks. This variable tended to have the same coefficient as the bank’s own score variable. Including it did not affect the results about which we are most interested. We also tried including the log of bank assets in the state, which might provide indications of lending opportunities or the banking environment at the city or state level. As an alternative to controlling for specific local factors, we also replaced our local controls with state fixed effects which provide a more general control for things that might be less observable (such as differences in the ability of state banks to offer services prohibited to National banks). Using fixed effects also has little effect on the ownership structure or corporate governance regressions. We also tried including the square of the ownership by the top three managers in case there were diminishing returns to ownership concentration. This variable also did not affect our main results and was largely insignificant. 39 Section 6 Conclusion Our results have interesting, important, and novel implications for how governance differences lead companies manage to attract outside funding sources in an environment where conflicts of interest are important. We find that managerial ownership and formal governance tools are alternative means to resolve conflicts. Each of these alternatives has important and somewhat different implications for rent seeking, the targeting of default risk, and the tools used (cash vs. equity) to achieve the targeted level of default risk. More concentration of ownership leads to less formal structures of governance, more insider tunneling through loans and salaries, more dividend paying, less risk taking (presumably due to risk aversion of manager stockholders), and more reliance on cash (to resolve asset-substitution and adverse-selection problems). Endogenously chosen formal governance structures produce greater risk, and more relative reliance on capital for risk management, but lower managerial salaries. In summary, there are two key corporate governance problems that arise in banking: managerial rent extraction through simple transfers (high salaries and subsidized loans to managers) and the possibility of managers’ undertaking excessive risk (i.e., risk shifting or “asset substitution”). High managerial ownership without formal corporate governance addresses the second of these problems, but permits greater managerial rent extraction than would occur under more formal corporate governance practices. That outcome may be preferred by the managers who organize banks (i.e., if the potential rents from expanding the size of the bank are limited). If, however, manager/organizers wish to expand their enterprises to a scale that is large relative to their managerial stakes in the bank, then formal corporate governance is likely to become necessary. The formal approach to governance results in higher tolerance for risk (reflecting the greater diversification of holdings of bank stock) and a reduction in the rents that bank managers are able to extract though high salaries subsidized lending. In the presence of formal governance, managers share their privileged access to bank loans with outside directors. 40 References Acharya, Viral, Jennifer Carpenter, Xavier Gabaix, Kose John, Matthew Richardson, Marti Subrahmanyam, Rangarajan Sundaram, and Eitan Zemel (2009). “Corporate Governance in the Modern Financial Sector,” in Restoring Financial Stability, Acharya, Viral and Matthew Richardson, eds., Hoboken: John Wiley & Sons. Anginer, Deniz, Asli Demirguc-Kunt, Harry Huizinga, and Kebin Ma (2013). “How Does Corporate Governance Affect Bank Capitalization Strategies?” World Bank Policy Research Working Paper 6636, October. Berger, Allen, Bjorn Imierowicz, Chrisian Rauch (2012). “The Roles of Corporate Governance in Bank Failures During the Recent Financial Crisis,” European Banking Center Discussion Paper 2012-023. Bodenhorn, Howard (2013). “Large Blockholders, Institutional Investors, Boards of Directors, and Bank Value in the Nineteenth Century,” Clemson University Working Paper. Bruno, Valentina, and Stijn Claessens (2010). “Corporate Governance and Regulation: Can There Be Too Much of a Good Thing?” Journal of Financial Intermediation, 19: 461-482. Calomiris, Charles W., and Mark Carlson (2013). “National Bank Examinations and Operations in the Early 1890s,” Working Paper, October. Calomiris, Charles W., Florian Heider, and Marie Hoerova (2013). “A Theory of Bank Liquidity Requirements,” unpublished mimeo. Calomiris Charles W. and Charles Kahn (1991). “The Role of Demandable Debt in Structuring Optimal Banking Arrangements,” American Economic Review, 81(3): 497-513 Calomiris, Charles W. and Joseph Mason (1997). “Contagion and Bank Failures During the Great Depression: The June 1932 Chicago Banking Panic,” American Economic Review, 87(5): 863- 883. Calomiris, Charles W. and Joseph Mason (2003). “Fundamentals, Panics, and Bank Distress During the Depression,” American Economic Review, 93(5): 1615–1647. Calomiris, Charles W. and Joseph Mason (2008). “Resolving the Puzzle of the Underissuance of National Bank Notes,” Explorations in Economic History, 45(4): pp. 327-55 Calomiris, Charles W.; Mason, Joseph; Wheelock, David (2011), “Did Doubling Reserve Requirements Cause the Recession of 1937-1938? A Microeconomic Approach,” National Bureau of Economic Research Working Papers 16688 Calomiris Charles W. and Berry Wilson (2004). “Bank Capital and Portfolio Management: The 1930s 'Capital Crunch' and the Scramble to Shed Risk,” Journal of Business, 77(3): 421-455 Carlson, Mark (2010). “Alternatives for Distressed Banks during the Great Depression,” Journal of Money, Credit and Banking, 42(2-3): 421-442. Carlson, Mark (2013). “Lessons from the Historical Use of Reserve Requirements in the United State to Promote Bank Liquidity” Federal Reserve Finance and Economics Discussion Series 2013- 11. Cheng, Ing-Haw, Harrison Hong, and Jose A. Scheinkman (2013). “Yesterday’s Heroes: Compensation and Risk at Financial Firms,” unpublished mimeo, September. 43 Table 1 List of variables Variable Source Description Management ownership Exam report The share of stock owned by the top 3 bank managers – the president, vice president, and cashier Ownership of outside directors Exam report The share of stock owned by individuals who were on the board of directors but were not managers Board meets month Exam report Indicator variable for the board of directors meeting monthly or more frequently Outside directors on board Exam report The share of the board of directors that consisted of individuals that were not managers Active discount committee Exam report Indicator variable for having an active independent discount committee President bonded Exam report President posted a surety bond Cashier bonded Exam Report Cashier posted a surety bond Score Derived Sum of governance indicators Turnover Exam reports & bankers magazine Number of changes in the president between 1882 call report and 1892 call report Log assets Call Report Log of assets. Log age Comptroller & Rand McNally Log of the difference between 1892 and the time the bank was established. Salaries to assets Exam report Ratio of salaries of 3 officers to assets Officers loans to insider loans Exam report Ratio of loans made to top 3 officers to loans to all insiders (managers and board members) Dividends to shares Exam report Ratio of dividends paid at last payout to shares outstanding (dollars per share) Used borrowed funds Exam report & call report Indicator that the bank borrowed using interbank certificates of deposit, rediscounts, or bills payable Real estate loans to total loans Exam report Ratio of loans secured by real estate to total loans Other real estate owned to assets Call report Ratio of other real estate owned to assets Troubled loans to total loans Exam report Ratio of “troubled” loans – those past due or suspended – to total loans Losses to assets Exam report Ratio of total losses on all balance sheet items as estimated by the examiner relative to assets Loan losses to assets Exam report Ratio of losses on bad loans, other overdue paper, other loans and overdrafts to assets Other losses to assets Exam report Ratio of losses on securities, bank house, furniture and fixtures, other real estate, cash, and other to assets Individual deposits to total liabilities Call report Share of liabilities consisting of deposits by individuals 44 Checking deposits to individual deposits Exam report Share of individual deposits consisting of checking deposits Net worth to assets Exam report Ratio of capital, surplus, and undivided profits to assets Cash to assets Exam report Cash and legal tender to assets Closed Comptroller reports Indicator that the bank suspended, failed, voluntarily liquidated after filing the Sept. 1892 call report but before Jan 1, 1894. Reserve city Comptroller reports Indicator that the city is a reserve city Log city population 1890 Census Log of city population (city population is not available for El Paso, TX so county population is used) Log distance to New York Log distance in miles to NY Fraction county income from agriculture 1890 Census Value of agricultural products in the county divided by the sum of the value of agricultural products and the value of manufacturing 45 Table 2 Summary statistics Variable Mean Median Std. Dev Min 25th percentile 75th percentile Max Management ownership 0.24 0.17 0.23 0.01 0.08 0.37 0.97 Ownership of outside directors 0.15 0.12 0.11 0.01 0.06 0.22 0.57 Board meets month 0.63 1 0.48 0 0 1 1 Outside directors on board 0.69 0.71 0.13 0.20 0.60 0.78 0.94 Active discount committee 0.60 1 0.49 0 0 1 1 President bonded 0.33 0 0.47 0 0 1 1 Cashier bonded 0.57 1 0.50 0 0 1 1 Score 2.69 3 1.56 0 1 4 5 Turnover 0.67 0 0.81 0 0 1 3 Log assets 14.1 14.1 0.8 12.0 13.5 14.7 15.9 Log age 2.42 2.40 0.74 0.69 1.79 3.14 3.43 Salaries to assets (percent) 0.59 0.46 0.45 0.02 0.33 0.69 3.61 Officers loans to insider loans (percent) 36.7 34.4 29.4 0 8.1 56.2 100 Dividends to shares 4.7 4 6.2 0 3 5 50 Used borrowed funds 0.31 0 0.46 0 0 1 1 Real estate loans to total loans (percent) 3.6 1.1 6.1 0 0 1.2 11.2 Other real estate owned to assets (percent) 0.9 .1 1.6 0 0 1.2 11.2 Troubled loans to total loans (percent) 9.1 5.9 9.9 0 2.5 12.4 71.8 Losses to assets (percent) 1.2 .2 3.8 0 0 1.1 32.1 Loan losses to assets (percent) .95 .10 3.05 0 0 .85 28.6 48 Table 4 Determinants of the Corporate Governance Score Score Management ownership -1.93*** (0.44) Ownership of outside directors 0.73 (0.83) Log age -0.35*** (0.13) Reserve city -0.02 (0.30) Log city population 0.12 (0.18) Log distance to NYC -1.29*** (0.26) Fraction county income from agriculture -0.007 (0.50) Intercept 11.65*** (3.03) Observations 206 Adj R2 0.29 F-statistic 12.7 Notes: The symbols (***), (**), and (*) indicate statistical significance at the 1, 5, and 10 percent level, respectively. Estimated using ordinary least squares. Standard errors in parentheses and italics. 49 Table 5 - Determinants of Manager Salaries Relative to Assets Spec 1 Spec 2 Spec 3 Spec 4 Management ownership 0.32* (0.18) Score 0.00 (0.03) Score – predicted -0.16* (0.09) Score – residual 0.01 (0.02) Ownership of outside directors -0.01 -0.04 0.09 -0.04 (0.30) (0.30) (0.34) (0.25) Log age -0.20*** -0.18*** -0.25*** -0.18*** (0.05) (0.05) (0.07) (0.05) Reserve city 0.03 0.03 0.02 0.03 (0.11) (0.11) (0.12) (0.07) Log city population -0.04 -0.06 -0.03 -0.06 (0.07) (0.07) (0.08) (0.05) Log distance to NY 0.06 0.09 -0.15 0.09 (0.09) (0.10) (0.18) (0.08) Fraction county income from agriculture 0.10 0.07 0.07 0.07 (0.19) (0.19) (0.21) (0.16) Intercept 1.06 1.12 3.09* 1.06 (1.14) (1.20) (1.77) (1.61) Observations 172 172 172 172 Adj R2 0.12 0.10 F-stat 4.19 3.64 Notes: The symbols (***), (**), and (*) indicate statistical significance at the 1, 5, and 10 percent level, respectively. Specifications 1 and 2 estimated using ordinary least squares; specifications 3 and 4 estimated using two-stage least squares. Standard errors in parentheses and italics. Standard errors in specifications 3 and 4 have been adjusted to reflect the use of generated regressors. 50 Table 6 – Determinants of Loans to Management as a Share of Insider Loans Spec 1 Spec 2 Spec 3 Spec 4 Management ownership 33.63*** (8.96) Score -4.97*** (1.37) Score - predicted -17.40*** (5.53) Score - residual -3.78*** (1.41) Ownership of outside directors -37.56** -36.18** -24.88 -40.70** (16.75) (16.82) (20.57) (17.32) Log age 1.77 1.19 -4.30 3.39 (2.72) (2.76) (4.03) (2.87) Reserve city 1.17 0.35 0.91 0.13 (5.99) (6.00) (7.13) (6.92) Log city population 0.23 -0.02 2.36 -0.97 (3.54) (3.55) (4.34) (3.69) Log distance to NY 13.52*** 10.75** -8.87 18.59*** (5.18) (5.46) (10.57) (4.98) Fraction county income from agriculture 20.75** 18.81* 20.63* 18.08* (10.09) (10.09) (12.02) (10.32) Intercept -73.94 -27.98 128.92 -90.74 (61.16) (63.54) (100.75) (55.05) Observations 206 206 206 206 Adj R2 0.18 0.17 F-stat 7.28 7.11 Notes: The symbols (***), (**), and (*) indicate statistical significance at the 1, 5, and 10 percent level, respectively. Specifications 1 and 2 estimated using ordinary least squares; specifications 3 and 4 estimated using two-stage least squares. Standard errors in parentheses and italics. Standard errors in specifications 3 and 4 have been adjusted to reflect the use of generated regressors. 53 Table 9 – Other measures of bank risk taking Management ownership Score Score – predicted Score - residual Real estate loans to total loans -3.5* 0.1 1.8* -0.0 (1.9) (0.3) (1.1) (0.4) Other real estate owned to assets -1.1** 0.1 0.5* 0.0 (0.5) (0.1) (0.3) (0.1) Troubled loans to all loans -5.0* 0.4 2.6 0.1 (3.0) (0.5) (1.6) (0.6) Estimated losses to assets -2.2* -0.4** 1.1 -0.5** (1.3) (0.2) (0.7) (0.2) Bank closed its doors -0.88* .06 -.18 .09 (.49) (.07) (.54) (.08) Notes: The symbols (***), (**), and (*) indicate statistical significance at the 1, 5, and 10 percent level, respectively. Rows 1-4 are estimated using ordinary least squares while row 5 uses probit analysis. Columns 3 and 4 using a two-step procedure. Standard errors in parentheses and italics. Standard errors in columns 3 and 4 have been adjusted to reflect the use of generated regressors. All regressions include the controls used in the previous regressions (such as those shown in the preceding table). 54 Table 10 –Determinants of the Ratio of Net Worth to Assets Spec 1 Spec 2 Spec 3 Spec 4 Management ownership -15.55*** (3.38) Score 1.01* (0.54) Score - predicted 7.82*** (2.39) Score - residual 0.34 (0.53) Ownership of outside directors -19.76*** -19.02*** -24.70*** -18.14*** (6.31) (6.59) (9.1) (5.93) Individual deposits to total liabilities -15.97*** -16.82*** -15.32** -17.01*** (4.78) (4.99) (6.73) (6.22) Checking deposits to individual deposits 11.63*** 9.45** 3.87 9.94** (4.20) (4.40) (6.20) (4.44) Log age -6.10*** -6.45*** -3.75** -6.86*** (1.07) (1.13) (1.77) (1.14) Reserve city -5.12** -5.10** -6.11* -4.99** (2.37) (2.47) (3.35) (2.83) Log city population -1.85 -1.16 -1.57 -1.04 (1.49) (1.55) (2.09) (1.66) Log distance to NY 1.76 0.92 11.16** -0.62 (2.01) (2.18) (4.51) (2.05) Fraction county income from agriculture 1.65 3.15 3.14 -3.21 (3.88) (4.03) (5.42) (4.06) Intercept 66.30*** 60.65** -27.77 73.62*** (23.07) (25.01) (44.79) (24.20) Observations 206 206 206 206 Adj R2 0.41 0.33 F-stat 15.19 12.19 Notes: The symbols (***), (**), and (*) indicate statistical significance at the 1, 5, and 10 percent level, respectively. Specifications 1 and 2 estimated using ordinary least squares; specifications 3 and 4 estimated using two-stage least squares. Standard errors in parentheses and italics. Standard errors in specifications 3 and 4 have been adjusted to reflect the use of generated regressors. 55 Table 11 – Determinants of the Ratio of Cash to Assets Spec 1 Spec 2 Spec 3 Spec 4 Management ownership 2.30** (1.06) Score -0.24 (0.16) Score - predicted -1.15** (0.58) Score - residual -0.15 (0.17) Ownership of outside directors -0.43 -0.47 0.29 -0.68 (1.98) (1.99) (2.20) (1.81) Individual deposits to total liabilities 4.46*** 4.56*** 4.37*** 4.61*** (1.50) (1.51) (1.63) (1.61) Checking deposits to individual deposits 0.82 1.21 1.96 1.16 (1.32) (1.33) (1.50) (1.35) Log age 1.18*** 1.19*** 0.83* 1.29*** (0.34) (0.34) (0.43) (0.34) Reserve city 0.04 0.05 0.19 0.04 (0.74) (0.75) (0.81) (0.89) Log city population 1.51*** 1.41*** 1.47*** 1.37*** (0.47) (0.47) (0.51) (0.47) Log distance to NY 1.63*** 1.63** 0.25 2.00*** (0.63) (0.66) (1.09) (0.60) Fraction county income from agriculture -0.94 -1.17 -1.16 -1.19 (1.22) (1.22) (1.31) (1.01) Intercept -26.99*** -25.02*** -13.10 -28.03*** (7.24) (7.57) (10.86) (6.79) Observations 206 206 206 206 Adj R2 0.25 0.24 F-stat 8.75 8.36 Notes: The symbols (***), (**), and (*) indicate statistical significance at the 1, 5, and 10 percent level, respectively. Specifications 1 and 2 estimated using ordinary least squares; specifications 3 and 4 estimated using two-stage least squares. Standard errors in parentheses and italics. Standard errors in specifications 3 and 4 have been adjusted to reflect the use of generated regressors. 58 Table 14 Mean bank characteristics by Management Ownership and Governance Score High ownership high governance High ownership low governance Low ownership high governance Low ownership low governance Test for differences in means [1] [2] [3] [4] 1 vs 4 2 vs 4 3 vs 4 Salary to assets 0.67 0.75 0.51 0.54 * (0.55) (0.49) (0.45) (0.20) Officer loans to insider loans 36.3 53.3 24.7 33.9 *** * (26.2) (32.2) (22.8) (28.3) Dividends per share 4.4 7.0 3.3 4.1 * (3.4) (10.4) (1.5) (2.9) Used borrowed funds 20.9 27.1 33.8 45.8 ** * 41.2 44.8 47.6 50.9 Real estate loans to all loans 3.2 3.5 2.5 7.3 ** ** *** (4.4) (4.3) (4.8) (11.7) Other real estate owed to assets 0.8 0.7 0.8 1.2 (1.9) (1.6) (1.5) (1.7) Troubled loans to total loans 8.8 10.4 7.3 11.5 ** (9.4) (11.0) (9.0) (10.1) Losses to assets 0.7 1.1 0.8 3.6 ** ** *** (1.5) (4.1) (1.3) (7.6) Loan losses to assets 0.58 0.90 0.58 2.73 ** * *** (1.28) (3.72) (1.00) (5.66) Other losses to assets 0.10 0.21 0.21 0.86 ** ** ** (0.28) (0.60) (0.52) (2.24) Closed 30.2 28.3 25.0 33.3 (46.5) (45.4) (43.6) (48.0) Cash to liabilities 7.9 8.3 7.6 8.1 (4.3) (3.8) (3.3) (3.1) Net worth to assets 30.4 30.2 35.8 34.8 (11.2) (11.4) (13.1) (15.5) Notes: The symbols (***), (**), and (*) indicate statistical significance at the 1, 5, and 10 percent level, respectively. Standard errors in parentheses and italics. 59 Chart 1 Distribution of ownership by top 3 managers Chart 2 Distribution of ownership by outside directors 0 2 4 6 8 10 12 14 Sh ar e o f sa m p le ( p er ce n t) Ownership share of top 3 managers (percent) 0 5 10 15 20 25 30 Ownership share of outside directors 60 Chart 3 Distribution of ownership by non-managers, non-board members Chart 4 Manager ownership and Board Composition 0 2 4 6 8 10 12 0 -2 4 -6 8 -1 0 1 2 -1 4 1 6 -1 8 2 0 -2 2 2 4 -2 6 2 8 -3 0 3 2 -3 4 3 6 -3 8 4 0 -4 2 4 4 -4 6 4 8 -5 0 5 2 -5 4 5 6 -5 8 6 0 -6 2 6 4 -6 6 6 8 -7 0 7 2 -7 4 7 6 -7 8 8 0 -8 2 8 4 -8 6 8 8 -9 0 9 2 -9 4 9 6 -9 8 Ownership share of non-managers/non-board members 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 F ra ct io n o f sh a re s o w n e d b y t o p 3 m a n a g e rs Share of board consisting of outside directors