© Oxford University Press 2007 All rights reserved
Promotion tournaments and white collar careers: evidence from Williams Deacon's Bank, 1890–1941
*Department of Economics, Royal Holloway, University of London, Egham, Surrey, TW20 0EX; e-mail: a.seltzer{at}rhul.ac.uk
Royal Holloway, University of London
 |
Abstract |
|---|
 TOP Abstract 1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
This paper uses a unique historical dataset constructed from the payroll records of Williams Deacon's Bank to examine career structures in a white collar firm. We examine promotions from the level of clerk to branch manager in the context of the Bank's strong internal labour market. There is evidence the Bank used promotion tournaments with returns to promotion that were inversely related to the probability of promotion. In expectation these returns were relatively constant over an individual's career. There is also evidence that the Bank compensated individuals with pay rises for frequent lateral moves and even demotions. Such moves were designed to discourage dishonest behaviour.
Key Words: JEL classifications: N33 • N34 • J41 • J44.
|
1 Introduction |
|---|
|
TOPAbstract1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
Employment in banking and insurance grew more rapidly than in any other sector of the British economy during the late 19th and early 20th centuries (Lee, 1994
). This growth in employment was associated with major organizational changes in the banking industry. Large joint-stock banks consolidated the industry by acquiring small private banks. They expanded branch networks by opening large numbers of smaller branches. These organizational changes had important implications for banking careers. The business of banking increasingly was conducted by salaried staff without the traditional status of private bankers (Bankman, 1919). Staff were located in small branches at a distance from the general managers and directors. Issues of motivating and monitoring staff became much more important than when a bank consisted of only a single branch managed by the proprietor.
Despite the changes to the industry there has been little research on the nature of banking careers over the period. Much of the existing evidence (see Gilbart, 1856; Rae, 1930; and Kynaston, 1994) is descriptive in nature and outlines the general range of practices across the industry, rather than the specific practices of any one bank or changes in those practices over time. The only quantitative study of personnel practices in the industry that we are aware of is Boot (1991), who studies staff at the Head Office of the Bank of Scotland in the earlier period 1730 through 1880. Boot finds that there were several important changes in practices during this period, most notably the rise of tenure-based wage increments from the late 18th century and an increase in the wage premium for better employees. Boot's inclusion of only Head Office workers is an important limitation that grows in significance over time. During the late Victorian period that we study, the majority of staff employed by joint-stock banks were in the branches, not the head offices.
This paper uses newly-collected personnel data from Williams Deacon's Bank (WDB) to examine employment practices in banking between 1890 and 1941. Williams Deacon's was a medium-sized branch bank based in Manchester with offices in London and the North. Its records, housed at the Royal Bank of Scotland Archives in London, cover virtually all staff at the Bank's London offices through 1936 and northern offices through 1941. The records provide the annual nominal wage and branch location of each worker. In addition, they contain information about dates of birth, entry to, and exit from both branch and Bank. Although the records do not contain direct information about position, we are able to infer which employee was the manager of each branch. We also have information on branch size. The data therefore allow us to examine career progression (and associated wage rewards) of staff at the Bank, typically from being a clerk to being manager of a small branch or—for particularly successful employees—managing a large branch.
The theoretical literature on tournaments (Lazear and Rosen, 1981; Rosen, 1986; and Prendergast, 1999) provides a framework for examining how firms can motivate their workers by internal competitions for promotion and associated higher wages. Practices at Williams Deacon's were consistent with the assumptions of the promotions tournament model. Virtually all promotions at Williams Deacon's were internal and even the managers of the largest branches started at the Bank as junior clerks.1 However, promotions to branch manager were not determined by administrative rules such as seniority. For example, in 1893, 1903, 1913, 1923, and 1933 there were a total of 39 promotions. In each case, individuals in a similar salary range with greater seniority than the promoted employee were passed over. Several of those who were promoted had been among the more junior employees likely to have been considered. One employee was promoted in 1903 even though 77% of employees in the salary range that would have made them likely candidates for promotion had greater seniority. Contemporary writers observed that banks were extremely concerned about promoting the right people to the level of manager as poor managerial decisions (such as an inability to attract customers or bad judgement on loans) could be extremely costly (Gilbart, 1856; Rae, 1930; and Kynaston, 1994). As a result, the Bank observed junior clerks’ performance on the job for at least 10 years, and often 20, before their first managerial appointment. Promotion to manager of a small branch was associated with large financial rewards, and these increased substantially for subsequent promotions to managing larger branches.
The promotion tournaments literature contains hypotheses we can test with our data. To maintain comparable incentives throughout an individual's career, salaries need to rise at an increasing rate in the job hierarchy. The reason is that, earlier in the individual's career, a promotion has an option value of increasing the probability of future promotions. Towards the top of the job hierarchy, there are fewer potential promotions and incentives can only be maintained if the salary gain on promotion is higher.2 Under straightforward assumptions we can calculate the numerical returns to effort at different points in the job hierarchy. We find that there are sharply increasing salaries with promotions that are consistent with a roughly equal incentive to supply effort throughout the individual's career.
The expansion of the branch network after the First World War provides a test for another prediction of tournament theory. The period immediately following the war witnessed a considerable increase in the number of bank branches opened across the UK. Demand for banking services was increasing with the rise in incomes. New technologies such as the typewriter, telephone, and adding machine allowed banks to meet this demand by opening new branches (Winton, 1982). These technologies lowered the cost of establishing new branches by improving communication within the Banks and allowing more of the record keeping to be carried out by less skilled back-room staff at the larger offices, rather than at the branch level. In addition, there were restrictions on growth through merger following the Colwyn report in 1918. In the case of Williams Deacon's, the pent up demand for new branches was particularly large because for a decade prior to the war a series of conservative general managers resisted expanding the branch network (Williams Deacon's Bank, 1971). In 1917 R.T. Hindley was made General Manager and immediately set out plans for branch expansion after the end of the war. Much of the growth of Williams Deacon's branch network over the entire period of this study took place immediately after the war, with 52 branches and sub-branches being opened between 1919 and 1922. This created an increase in the number of managerial positions and the probability of promotion.3 Tournament theory predicts that the Bank would maintain optimal incentives by lowering the salary reward to promotion to compensate for the increased probability of promotion. We find evidence of this effect both in the raw data and in our regression analysis of levels and changes in real wages.
While the structural changes in the banking industry provided the Bank with a greater need to motivate employees, it also needed to increase monitoring for honesty. Rae (1930) observed: ‘A bank ought not to appraise the value of an officer's services merely by what they would fetch in the clerk market. He may not be a man of capacity; he may be little more than an honest and willing drudge; but when it is considered how much the Directors have to trust the honesty, integrity, and honour of its staff, they will not lightly part with those who have proved themselves by long service, the possessors of those essential qualifications, even if they have little else.’ Promotion tournaments were unlikely to have a strongly positive effect on honesty because a ‘willing drudge’ was unlikely to be promoted. A direct way of lessening the opportunity for dishonest behaviour was to move staff frequently across branches, compensating them for the disutility of moving by significant pay increases. There is evidence in our data for this effect.
Our main interest in this study is in understanding banking employment practices during this period of time. However, the confirmation of some of the predictions of promotions tournament theory is of broader interest. The historical payroll data we use in this paper has considerable advantages for examining career structures and promotion tournaments, compared to existing studies based on recent data. Most studies specifically examining tournaments concentrate upon chief executives and professional athletes (Prendergast, 1999), given the difficulties in finding well-defined hierarchies in payroll records. Modern studies of payroll records of more representative white-collar employees have explored more general issues in career dynamics. However, these records have necessarily been of a duration that cannot cover the majority of an individual's career, particularly given current mobility across firms. For example, Baker et al. (1994) have payroll records of 20 years duration, typically covering only a small portion of the career of individuals who enter and exit their dataset during the period. In contrast, we observe the whole career at the Bank for 36% of our sample. Another advantage of our data is that the workers were extremely homogeneous in terms of background and education, being almost uniformly middle class and possessing some secondary education.4 Further, they were on a well-defined hierarchical structure. All staff started as junior clerks, typically at age 15–18 and almost always before age 21, and thus all of the employees in the sample potentially could have progressed to manager of branches of different sizes.
The plan of the paper is as follows. Section 2 outlines the nature of the English banking industry and the position of Williams Deacon's within the industry. We also summarize the Williams Deacon's data set. Section 3 shows the career structure at the Bank, and in particular the probabilities of promotion to different ranks and the typical wage gains from promotion. The raw data support the tournament model, but also show evidence of significant gains to wages when moving branch without a promotion. These results continue to hold in Section 4 when we control for a number of different factors in regressions on the level and change in wages. We are also able to control for unobservable heterogeneity and to examine the time pattern of the returns to promotion. Section 5 presents our conclusions.
|
2 The UK banking industry and the Williams Deacon's bank data |
|---|
|
TOPAbstract1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
The UK banking industry underwent large-scale changes in a period of consolidation that began about 1870 and intensified after 1890. By 1913, virtually all of the private banks were gone, typically merged into joint-stock banks. The total number of banks in the United Kingdom declined from 303 in 1890 to 75 in 1920 (Capie and Wood, 1994
). Another important change to the industry during this time was the growth in branch networks. Between 1890 and 1920, the total number of banking offices in the United Kingdom increased from 3,478 to 9,668, implying a thirteen-fold increase in the average number of branches per bank and nearly a doubling of bank offices per capita (Capie and Webber, 1985).
Williams Deacon's was a medium sized branch bank that, in common with much of the industry, arose as the product of a series of mergers, most importantly the acquisition of the London-based Williams, Deacon's & Co. by the Manchester and Salford Bank in 1890.5 Williams Deacon's itself was absorbed by the Royal Bank of Scotland in 1930, but continued to trade separately under its own name until 1969. Like most surviving banks, Williams Deacon's was expanding during the period of this study. In 1896, the first year for which we have full records, it operated 68 branches and sub-branches with 411 staff. In 1936, the last year for which we have full records, it operated 202 branches and sub-branches with 968 staff. The branches opened after 1890 were universally small. In 1896, 56.1% of branches had fewer than 6 staff and 90.2% had fewer than 20. By 1930, 69.18% and 91.94% of branches had fewer than six and 20 staff, respectively. The growth in the number of small branches had important implications for the staff. New branches typically were staffed by existing employees, and thus the expansion of the network resulted in increased mobility across branches. In addition, the smaller size of the new branches meant that there was a rise in the proportion of employees at the level of manager. In 1900 10.7% of sample staff held managerial positions, in 1930 this figure was 14.9%. With the increased distance (both geographically and hierarchically) of staff from the Head Office, issues of motivating and monitoring staff became more important. This was well understood by contemporaries. Gilbart (1856) noted, ‘Numerous branches require a peculiar mode of government and a rigid system of discipline’.
The primary source of data for this paper is the unusually comprehensive Williams Deacon's Bank (WDB) wage records, collected from the Royal Bank of Scotland's Archive in London. The records, which are written in eight sets of ledgers, are organized by branches and time period.6 The records contain the date of birth, date of entry to the Bank, date of entry to the branch, annual wage, date of exit from the branch, and reason for exit (transferred to another branch, left the Bank, died, dismissed, or retired on a pension) for staff at each branch during the period covered by the ledger. They cover virtually all employees at the Bank's London branches between 1890 and 1936 and at the northern branches between 1890 and 1941.7 We have recorded this information for all male employees, entering wage and branch information annually, using values as of October 1 in each calendar year.8 All totalled, the sample contains 2,116 male employees and 34,977 man-years of data.9 The wages in the records are denominated in nominal pounds, and we have deflated them using Feinstein's price series indexed to 1890 values (Feinstein, 1972).
In considering career structures at the Bank, we examine promotions to branch manager. While the records do not contain direct information about position, it is possible to infer the name of the manager of each branch. We assume that the first person listed for each branch was the manager at the start of the period covered by the record. If he left the branch and another senior person arrived at approximately the same time, we assume that he was the new manager. If the manager left and there wasn’t another senior person arriving at the time, we then assume that there was an internal promotion to manager. In this case we consider the structure of wages of the other branch staff (e.g. high wages and large increases at the time of the original manager's departure) to infer the new manager.10 A bank document lists all branch managers prior to 1900 and there is a 100% correspondence between this list and the list inferred through the method above.11
Another important piece of information that can be determined from the records is the size of the branch, measured by the number of staff.12 Banking scholars have noted that the responsibilities of the branch manager were roughly proportionate to the size of his branch. Blackburn (1967) stated, ‘[T]here are many promotions possible within the status of branch manager, for branch sizes vary considerably.’ We follow this approach by considering managers of different branch sizes as different levels within the hierarchy. In the regressions on wages in Section 4 we use a continuous variable to characterise branch size. However, in the analysis of job changes in Section 3 we construct transition matrices between ranks in the hierarchy and it is necessary to divide branch sizes into discrete categories. We classify branches as small (five or fewer staff), medium (6–19 staff), large (20–50 staff), and the two very large branches (over 50 staff) at Mosely Street, Manchester and Birchin Lane, London. Any size division is necessarily somewhat arbitrary. We chose five staff as a cut-off because this was normally the largest size of branch that would typically not have an appointed accountant. The upper cut-off of 20 was chosen to ensure a reasonably populated cell size of large branch managers (0.83% of the sample). With these cut-offs, employees were distributed fairly evenly in the different sized branches (27.7% of sample employees were at small branches, 31.4% were at medium branches, 15.0% were at the large branches, and the remaining 26.0% were at the two very large branches).
There is one way in which the absence of position information is problematic. While all branches had a number of clerks, larger branches also had at least one specialized teller and a branch accountant. The Mosely Street and Birchin Lane offices also housed a few very senior staff such as the Bank's secretary and treasurer. Because of the limitations of the data, we do not attempt to identify different types of non-managers.13 In approximately 0.5% of observations, non-managerial employees had annual wages over £1100, approximately 4 standard deviations above the mean. In all of these observations, the employees were based at either Mosely Street or Birchin Lane, and were thus almost certainly holding fairly senior positions at the Bank. We have conducted our analysis in Sections 3 and 4 excluding these observations from the sample.
Table 1 presents summary statistics on the WDB data used in this paper for the whole sample period, and then for the two period sub-samples covering the pre-World War I period (1890–1913) and the post-war period (1919–1941). There are several important changes in the later period. With the opening of new branches, the average branch size declined, staff were moved more regularly, a higher percent were employed at the level of manager, and a lower percent were employed at the Head Office. In addition, in the later period staff were less likely to resign early in their careers, resulting in higher average tenure and a higher proportion remaining through to retirement.
|
|
3 Promotions, demotions, and lateral moves |
|---|
|
TOPAbstract1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
The Bank's workplace displayed the characteristics of an internal labour market defined by Doeringer and Piore (1971
). Very few individuals were hired from outside beyond the initial entry age. Among the sample entrants, 74.2% and 93.1% began prior to their 18th and 21st birthdays, respectively.14 Further, if individuals departed from employment at the Bank, they generally did so in the first few years. Approximately 47% of employees in the sample left within seven years of starting, but of those reaching the 7th year approximately 69% stayed until retirement or death. Virtually all promotions to branch manager were internal, with the exception of those who started their careers at a bank acquired by Williams Deacon's. However, promotion tended to be a slow process: the mean time to the first promotion was 18.63 years. Individuals who failed to be promoted to branch manager were often eventually moved to a more senior clerking position at a large branch or the Head Office. Among employees remaining at the Bank for at least 20 years, 81.4% of those who never rose to manager finished their career at a large branch, compared to 28.8% of those who at some point in their career held a managerial post. Upon retirement, employees received final salary pensions. These policies were commonplace throughout the industry prior to the Second World War (Gilbart, 1856; and Rae, 1930+ Blackburn, 1967). The internal labour market does not mean that the Bank was not concerned about motivating and monitoring its employees. In fact, since workers tended to remain for life, the Bank had strong reasons to encourage effort and honesty. In this section, we argue that the Bank set up promotion tournaments to motivate workers and engaged in frequent reassignment of workers across branches, in part to monitor for honesty. In total, the sample contains 2,636 job changes. Of these, 2,044 were lateral moves between branches below manager level, 115 were lateral moves at manager level, 273 were promotions to manager involving a move, 86 were promotions to manager without a move, 23 were demotions without a move, and 95 were demotions involving a move. These job reassignments, whether promotions (or demotions) or lateral transfers, were facilitated by the changing structure of the Banking industry over the period of the sample. The Bank opened 161 new branches and sub-branches, creating 107 additional managerial positions (since sub-branches did not typically have their own managers) and requiring the regular transfer of experienced clerks.
We first focus on changes in rank. Table 2 presents a year-on-year transition matrix between ranks at the Bank, and the associated wage gains, for workers remaining in the Bank's employment. We define the ranks to be clerk, manager of a small branch (5 or fewer staff), medium branch (between 6 and 19), large branch (between 20 and 49), or one of the two major branches (Mosely Street and Birchin Lane).15 The bulk of staff (over 98%) remained in the same rank each year, with a small number promoted (cells below the diagonal) or demoted (cells above the diagonal). In an average year, about 1.3% of clerks were promoted to branch manager, typically to managing a small branch, although a few to managing a medium or large branch.16 Managers of small branches had comparable probabilities of being demoted to clerk or being promoted to medium branch manager, although the great majority (93%) remained as manager of a small branch. Interestingly, more medium and large branch managers were demoted each year than promoted. Table 2 also shows the real wage gains to remaining in rank, being promoted, and being demoted. As would be expected, promotions were typically accompanied by an above-normal wage increase, compared to remaining in the given post. The wage gains were generally larger with moves further up the hierarchy and for those who skipped one or more levels. Although demotions from large branch manager had a wage penalty, demotions from small and medium branch managers to clerk came with an above-normal pay increase.
|
The Bank could have filled its new managerial posts by administrative rules such as seniority rather than competition, and it could have attached relatively small wage rewards to promotion. The lack of an external market meant that relatively few individuals would leave if not promoted or if the wage increase was viewed as insufficient. However, it would seem surprising if the Bank ignored the incentive potential of rewarding employees who were sufficiently able or hard working to attain promotion. The environment facing the Bank was ideal for setting up promotion tournaments. The homogeneity of entrants in terms of background probably tempered one of the well known drawbacks of promotion tournaments, namely that individuals would have had little incentive to supply effort once they realized they were unlikely to win promotion. At the start of their careers, both the Bank and an individual employee would have had little information about the suitability of the individual to a career in banking and thus their ability relative to their peers. Thus ex ante employees at the Bank would have been relatively evenly-placed in a promotions tournament, with their future career achievements largely determined by their choice of effort. Furthermore, the frequency of managerial openings would have meant that individuals passed over for a particular promotion would have been soon considered again. As such, a promotions tournament would have well-directed effects on inducing effort. Rae (1930
) noted exactly this sort of practice, stating, ‘A bank cannot give high wages to all its officers; but if its highest offices are open to every one in its employment, who shall prove his fitness for the same, there will be no apathy for its staff. It will be the object of every one to devote his best abilities to the practice and study of his profession; and thus an able, zealous and loyal staff will be developed.’ Tournament theory emphasizes the lifetime gains from a promotion, part of which arise from a higher wage at the new position and part of which arise from the increased probability of subsequent promotions and still higher wages. A small branch manager was more likely to be promoted to managing a medium or large branch than a clerk was to gain a direct promotion to managing a medium or large branch. In Table 3, we show the probabilities that a clerk (or a holder of any of the given ranks) will have ended up being promoted to each more senior rank as a terminal position, where we have amalgamated the categories of large branches and the two very large branches (due to the small number of observations). For example, the conditional probability of becoming a large branch manager starting from the position of clerk is 3.24%, while the conditional probability of becoming a large branch manager—having already been promoted to small branch manager—is 5.49%. We also show in Table 3 the average wage over his career for an individual whose highest position was that shown in the left hand column. For this table, we only use entrants from 1890 to 1905 and restrict attention to employees remaining at the Bank for at least 20 years. These restrictions allow us to consider individuals who spent all or nearly all of their working career at the Bank during our sample period.
|
Table 3 can be used to illustrate the expected gains from promotion (starting from a given rank) and therefore the incentives to provide high effort. In the tournament model, a clerk who does not supply effort beyond the level required to avoid dismissal will never be promoted and has an average annual real wage over their career (lifetime wage) of WC, which from Table 3 is £177.92. By providing effort, he could have hoped to be promoted to higher ranks following the conditional probabilities in the Table. The average lifetime expected wage VC for a clerk putting in effort is the probability that he finally ended up in each of the higher ranks (
CS for example is the probability that a clerk finishes his career as a small branch manager, which from Table 3 is 24.07%), multiplied by the average lifetime wage of someone attaining that terminal rank. Similar calculations hold for individuals starting from the other ranks (S for small branch manager, M for medium branch manager, and L for large branch manager): |
|
|
|
|
|
|
|
Using the data from Table 3, the values of VC, VS, VM, and VL are £201.92, £221.77, £250.12, and £501.00. At each career point, the worker has to decide whether to remain in the promotions tournament by supplying high effort, or to settle for the lifetime average wage were his current post to be the highest he ever achieved. An approximation of the gains to supplying effort at each rank is given by the difference Vi–Wi, multiplied by expected tenure at the Bank (44 years for an employee entering at age 16 and retiring at 60).17 The respective lifetime gains are £1,056, £1,518, and £789 for clerks, small managers, and medium managers. These are significant gains compared to the average annual wage of £177.92 for a clerk or £187.26 for a small branch manager who does not progress. Interestingly, these gains are fairly comparable at different points in an individual's career—a clerk has roughly the same gain to putting in sufficient effort for being considered for promotion as does a medium branch manager.18 An important result in tournament theory (Rosen, 1986) is that relatively equal incentives for effort over the entire career are sustained by sharply increasing salaries through the job hierarchy. This is evident from the average real wage by position. The gain to promotion at each level was sharply increasing in absolute (although not necessarily percentage) terms, at £115.39 (84%) from clerk to small branch manager, £156.37 (62%) to medium branch manager, £381.61 (93%) to large branch manager, and £1,715.11 (217%) to managing the largest branches.
A second test of the promotions tournament hypothesis is to compare the pre- and post- World War I periods. The appointment of R.T. Hindley as General Manager in 1917 led to significant growth in the Williams Deacon's branch network after the war, with 52 branches and sub-branches being opened between 1919 and 1922. Tournament theory predicts that, when the probability of promotion exogenously increases in this way, the wage return will go down (to maintain incentives at roughly the initial level). As predicted by the theory, the size of the gains declined after the First World War. In the pre-war period, non-managers (sample 9,634) averaged real wages of £139.58 and managers (sample 1,234) £436.46. In the post-war period, non-managers (sample 16,768) averaged real wages of £144.55 and managers (sample 3,166) £349.12. This provides additional support for the hypothesis that the Bank was using promotion tournaments to induce effort. An alternative hypothesis to promotion tournaments is that the Bank set wage levels as compensating differentials for the extra responsibilities of being a manager. Rae (1930) noted that promotion to manager carried—‘a serious heritage of unaccustomed duties and anxieties’. But if this—rather than promotion tournaments—was the basis for the wage differentials, these should have increased after the war. The greater the number of managers out of a given pool of employees, the greater the marginal disutility that needed to be compensated. The Bank would have needed to promote not just those individuals who welcomed the additional responsibilities and social status, but additional staff who suffered greater ‘anxieties’ and who therefore needed greater compensation.
Promotions were not the only form of job transition in the payroll data. About 82% of all transitions in the sample were lateral moves between non-managerial positions. Lateral moves were associated with large wage increases—for current clerks, promotion led to a £14.76 wage increase, but moving branches with no change in position led to a nearly comparable £14.25 increase. One explanation is that lateral moves were used to expose promising young employees to new tasks or to be reviewed by a different manager (Blackburn, 1967). Among staff entering as a clerk in the sample period and remaining in the data set for at least 20 years, the average number of observed moves for those eventually promoted to manager was 4.42 compared to 2.52 for those never promoted. Under this explanation, the pay increase could be associated with the Bank's recognition of the high ability of the clerk.19 However, comparable pay increases occurred for managers moving between branches without a promotion. An alternative explanation is that lateral moves were primarily to help the Bank monitor its employees. Staff remaining in the same branch indefinitely had greater opportunities to collude with each other or with customers in order to defraud the Bank. Moving staff between branches helped to prevent such coalitions and to uncover them, as when an employee moved, the first task of his replacement was to check the books for errors or fraud. In this case, the Bank may have offered staff moved to a new location a wage increase to compensate them for the disutility of moving.
Another possible transition involves demotion, which occurred 110 times in the sample. Anecdotally, demotion was used in two distinct types of situations. First, when an employee committed some sort of offence or proved incapable of handling the responsibilities of their position, demotion was used as a punishment. The Bank's records show that demotions often followed poor annual appraisals (GB1502/WD/475; see Williams Deacon's Bank, various years). Second, as discussed previously, movements between branches were an important part of the Bank's employment strategy, and demotion may have simply been substituted for a lateral move when there were no suitable openings at an equivalent level. The evidence of a pay increase following demotion supports the view that many demotions were simply routine moves between branches. This is further supported by the fact that about a third of demoted employees were later promoted back up to manager.
|
4 The determinants of real wage levels and changes |
|---|
|
TOPAbstract1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
In this section, we examine wage levels and changes after controlling for a number of variables. The main control is seniority, since wages will typically rise with job tenure even in the absence of any job transitions. We re-examine the tests of the tournament hypothesis in Section 3, and then look at additional issues such as unobservable heterogeneity—‘ability’—and the time pattern of the returns to job changes.
In Table 4, we report the results of a series of regressions on the natural log of real wage, using a standard Mincer model of the determinants of earnings. We model the relationship between wages and tenure at the Bank as a polynomial of fourth order, rather than the standard quadratic, in order to allow for the possibility that—even after controlling for promotion—the Bank used high expected end-of-career wage rises as a way of inducing effort and cutting turnover.20 In addition to the tenure variables, we include the following independent variables: age at entry to the Bank (and its square), dummies for whether the employee was at the head office or in a London branch, a dummy for a war year (1914–19), a dummy for whether the employee came to WDB as a result of an acquisition of another bank, a time trend, the national inflation rate, the size of the branch (measured by number of staff), a dummy for whether the employee was the branch manager, the interaction of the manager dummy and branch size variables, the number of moves between branches made by the employee prior to the period of the observation, and a dummy variable for whether the employee had been previously demoted.
|
The estimated coefficients show that wages rose rapidly at the beginning of workers’ employment but also continued to rise throughout their career. Entry age (a proxy for experience or education when joining the Bank) had the expected positive (but diminishing) effect. Being at the Head Office or in London had a positive effect. Real wages were lower during the war. Even after controlling for the very high war-time inflation rates, World War I had a strong negative effect since many younger staff were conscripted and received no increments while on active duty.21 In general, inflation had a negative effect on real wages. The Bank did not typically respond to inflation by raising nominal wages, although there were catch-up rises after the war in 1919, 1920, and 1925. The size of branch also has a positive effect on employment. For example, a clerk in a branch with 50 staff earned about 1.5% more than a clerk in a branch with two staff. However, this is dwarfed by the effect for managers; a manager of a branch with 50 staff earned approximately double that of manager of a branch with 2 staff. Finally, the number of moves prior to the date of the observation has a positive effect on wages, as does a prior demotion, consistent with the raw data.
The regressions allows us to look again at the effects of job transitions, after controlling for tenure at the Bank and the other variables. All employees could have expected to receive wage increases with seniority, whether or not they were promoted, transferred between branches, or even possibly demoted. In Section 3, we took account of this by looking at lifetime wages of 1890–1905 entrants. With the regression results, we can look at the entire sample since we can—by choosing specific values for the other variables in the regression—trace out the predicted wages for each year of seniority for an individual with those other values. To illustrate, we calculate predicted values of wages for an individual who directly entered Williams Deacon's at age 16 in 1890; whose career as a clerk was spent at branches with five staff; who moved branch in years two, five, and 10; who did not work in London or at the main office; and who retired at age 60. An individual with these characteristics who remained as a clerk their entire career would have predicted real wages (at 1890 prices) of £95.16, £171.49, £177.81, and £281.51 in their 10th, 20th, 30th, and 40th years of seniority respectively. By contrast an individual promoted to small branch manager (five staff) in their 20th year, but no further, would earn £95.16, £222.55, £230.75, £365.33 in their 10th, 20th, 30th, and 40th years of seniority.
We can use the predicted values from the regression to estimate the expected returns to effort by summing up the predicted annual earnings under a specific set of assumptions about potential promotions. We consider four career paths: an individual who remained at the level of clerk their entire career, one who was promoted in their 20th year to the level of small branch manager (five staff) but received no further promotion, one who received one further promotion in their 25th year to the level of medium branch manager (10 staff), and one who received a final promotion in their 30th year to the level of large branch manager (70 staff). Using the predicted wages from the regression for each of these career paths, we calculate the predicted lifetime average wages WC, WS, WM, and WL for someone ending their career in each level as £164.31, £200.41, £225.13, and £536.02. Doing the same calculation as in Section 3, to find lifetime non-discounted gains to effort at each rank, we estimate the gain to putting in effort as a clerk (and benefiting from the probability of promotion to the various higher ranks) as £1,469, that of a small branch manager £1,229, and that of a medium branch manager £912. These values are similar to the figures (£1,056, £1,518, and £789) computed from the raw data in Table 3. To provide some sensitivity analysis, we have also estimated the values assuming that individuals have a 2% annual discount rate and a 5% annual discount rate.22 Using these discount rates we have returns to effort at each stage in the job hierarchy as (£737, £595, £437) and (£277, £210, £151) respectively. Discounting lowers the absolute return to promotion, as would be expected. It also lowers the relative rewards to promotions further up the hierarchy. This implies that individuals with higher discount rates had less incentive to supply effort and earn promotion and that this disincentive increased with seniority.
In the 3rd and 4th columns of Table 4, we compare the pre-war and post-war periods. Tournament theory predicts that the gain in ‘winning the tournament’ would be set at a lesser value if the probability of winning was greater, to maintain given incentives. The sharply increased percentage of staff at the level of manager in the post-war period shown in Table 1 would suggest a smaller return to promotion. As with the raw data in Section 3, we find significantly lowered returns to being a manager in the post-war period compared to the pre-war period. Interestingly, the magnitude of the decline differed across branch size. The regression results show that the premium for managing a branch with two staff declined by 42.9% (from 29.1% to 16.6%), whereas the premium for managing a branch of 50 declined by a more modest 24.3% (from 138.6% to 105.0%). Since the bulk of the increase in branches in the latter period was in small branches, this is consistent with the tournament model.
An alternative hypothesis on the decline of the promotion premium after the war is that the position of manager was deskilled due to new technologies, such as the phone and the motor car, which improved communications between the branches and the Head Office and thus allowed senior managers to make decisions that would have been previously made by local branch managers. If this explanation is correct then one would expect that the managerial premium would decline most for relatively remote smaller branches. Even in 1900 a manager in Manchester or London could consult the Head Office or London manager relatively easily. However, it would have been more difficult for a manager in other northern branches to consult the Head Office in 1900 than in the decades after the war. Thus these new technologies would have deskilled northern managers of small branches outside Manchester more than similar managers in London or Manchester. We have tested this by re-running the regression from the first column of Table 4, adding dummy variables for (i) being in a branch outside London or Manchester, (ii) being the manager of a small branch outside London or Manchester, and (iii) being the manager of a small branch outside London or Manchester after the war. If the deskilling explanation is correct the coefficient on three should be negative. However, the regression results (not reported in the paper) indicate that it is close to zero and statistically insignificant. In addition, this hypothesis predicts that salaries of managers of the largest branches—who would acquire additional supervisory responsibilities—would increase following the war. In fact, they also decreased.
In the whole sample (column 1 of Table 4), moves to date and prior demotions have a positive effect on wages. As discussed in the previous section, this is consistent with a model where the Bank moves individuals as part of the monitoring process, and needs to reward them for the disutility of moving. Interestingly, when we divide the sample, lateral moves have a significant positive impact only in the post-war years and demotions have a larger effect in the pre-war years. Prior to the war there were relatively few managerial positions and thus it may have been frequently necessary for the Bank to demote competent managers as part of the routine monitoring process. Because the pre-war managerial premium was relatively large, a manager who was demoted as part of the monitoring process and did not take a pay cut at the time of demotion would have earned substantially more than other clerks. After the war, the larger number of branches meant that it would have been easier for the Bank to find openings at a similar level. This implies that the in post-war period mangers were more likely to be moved laterally, rather than demoted, as part of the monitoring process. Thus, an increased proportion of demotions were likely to have been employees who were unsuitable for their previous position.
Although the intake to the Bank was relatively homogeneous, particularly in education levels, individuals still differed in ability. If ability was at least partially observable to the Bank, it might pay higher wages to more able individuals whether or not they were promoted, and independently of their level of seniority. Not taking account of these individual-specific effects (since ability is also a determinant of promotions) would lead to an over-estimate of the return to being promoted. The standard way to control for unobserved heterogeneity is to use a fixed effects estimator as reported in column 2 of Table 4. As expected, both the estimated return to being a branch manager, and to managing a larger branch, decline when we control for fixed effects. A similar effect holds for employment at Head Office, suggesting that this was an assignment given to the Bank's more capable employees. In contrast, moves to date has a higher estimated return, possibly because more able individuals were not moved as often—the Bank may have moved less able individuals to monitor them more closely and to teach them banking processes in a variety of environments.
While the fixed effects estimator is one way of controlling for unobserved heterogeneity, it is also possible that the results are driven by selective exits. It is likely that lower ability or unsuitable workers leave early in their career, and this may bias the regression coefficients because tenure will be correlated with ability. We control for this in the fifth column of Table 4 by considering only employees in their 30th year of tenure, thereby eliminating individuals who exited early.23 It can be seen that the coefficients on manager and the manager-branch size interaction are only slightly smaller than in the OLS specification, thereby suggesting only a limited role for selective exits.
In the promotions tournament model, wage levels differ from marginal product and are attached to ranks. A worker who puts in high effort will not receive an immediate pay reward, but will increase his chances of promotion and will gain a large pay increase consequent upon promotion. Alternatively, if wages are more closely connected with marginal product, a worker who puts in high effort will gain higher wages even before promotion, as well as increasing his chances of promotion, and will gain a smaller amount upon promotion. We examine this by estimating changes in real wages in Table 5.24 We find in column 1 that a promotion has a significant effect on pay (a 3.55% rise) in the year of promotion with weak evidence of a substantially smaller further effect in the subsequent year. There is no evidence that wage increases were higher in the year prior to promotion. This supports the tournament approach, that wages are attached to ranks rather than directly to productivity. The second column of the Table adds terms for promotion to manager of a large branch, with similar but stronger results—in this case, the initial pay rise of over 15% (including both the coefficient for promotion to manager and that for promotion to manage a large branch) is followed by a subsequent pay rise of 4%. Another interesting result is that spending a long time in the same position (i.e. failure to be reassigned or promoted), carried a small wage penalty. Finally, Table 5 also provides insights into the effects of a demotion. Unlike Table 4, which shows a positive long-run impact from demotion, Table 5 shows a negative short-run impact of a demotion.
|
One explanation that is consistent with the negative impact of time spent at the same position and the negative short-run, but positive long-run impact of demotion is that there were binding pay scales for many of the positions at the Bank. An individual below the top point of the scale for their position would receive an annual increment, but once they had reached the top point for their position they could only receive a higher wage through reassignment. Gilbart (1856
) notes that this practice was used by some banks. He states, ‘[A] bank may have a fixed minimum salary [for each post]. Each clerk holding a post for a certain period has an annual advance for that period. Then he stops, and receives no further advance until he is promoted to the next post, where again he becomes entitled to annual advances.’ A binding pay scale would result in a negative coefficient on time spent at the same position because at some point, which may have differed across posts, an individual would have reached the top of the salary scale for their position. A demotion would have had a negative short-run impact on salary if the individual was placed onto the salary scale for the lower ranked position. However, if the demotion was for routine monitoring purposes and the individual was moved back onto a managerial or higher ranked clerical position as soon as one opened up, the original demotion would have no lasting negative consequences for salary. There is evidence for this phenomenon in the data. Individuals who were demoted the previous year were 47% more likely to move branches than the rest of the sample. These moves came with a substantial wage premium; the average real wage increase for employees who moved locations in the year following a demotion was £24.42, compared to £12.57 for other employees moving branch and £7.01 for employees staying at the same branch. |
5 Conclusions |
|---|
|
TOPAbstract1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
In this paper, we have used a unique data-set of historical payroll records from Williams Deacon's Bank to examine the career structure at a white collar firm. By using historical data, we have a long series of wage figures and can trace many workers through their entire careers. As such, we can examine not only impact effects of job transitions—lateral moves, promotions, and demotions—but the longer-term career effects. There was a strong internal labour market at the Bank with virtually all promotions coming from within. Those workers who made it through an initial seven-year period typically remained with the Bank until retirement. Given the internal labour market, the Bank did not need to follow marginal productivity and outside opportunities in determining wages. It could instead establish promotion tournaments with wage rewards set to induce the optimal level of effort.
We find evidence that the Bank used promotion tournaments in this way. There were above normal wage increases in the year of promotion and, to a lesser extent, in the subsequent year, showing the tie between wage rates and rank. If the promotion went to the most productive worker, and the Bank was setting wages to match productivity, wages would have been higher prior to the actual promotion, but there is no evidence that this was the case. Tournament theory suggests that the wage gains on promotion should increase at an increasing rate higher up the hierarchy. This maintains a relatively constant inducement to effort throughout the individual's career, since later promotions do not contain the same option value as earlier promotions. For example, the promotion to general manager is the final possible promotion and the gain is just the immediate wage gain, while an earlier promotion increases the likelihood of an eventual promotion to general manager. We find, in both the raw data and in wage regressions, evidence that wages increased at an increasing rate up the hierarchy, and that the calculated expected return to effort remained fairly constant. Another prediction of tournament theory is that the return to a promotion should be less if the probability of achieving the promotion is higher. By breaking the sample into two periods, pre-war (when relatively few individuals were promoted to manager) and post-war (when branch expansion meant that many more individuals were promoted), we test this hypothesis and find support for it.
In addition to promotions, there were frequent lateral moves across branches, and less frequent demotions. Honesty from staff was particularly important in this industry, and moving staff across branches lessened the opportunities for dishonest behaviour and increased the monitoring since the new holder of the post would report on any irregularities discovered. As such, the Bank treated moves and even some demotions as a normal part of career progression. There is strong evidence that wages increased with lateral moves and that, after an initial negative impact on wage, demotions too had a long-run positive impact on wages. This suggests that staff were being compensated with wage increases for the disutility of moving between branches.
|
Acknowledgements |
|---|
|
TOPAbstract1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
We wish to acknowledge the staff of the Royal Bank of Scotland Archives in Islington (Derek Hammond, Lucy Wright, Jenny Mountain, Alison Turton, and particularly Philip Winterbottom) for their enormous help with the records used in this paper. We have also benefited from comments by Michael Sykuta and participants in the Economic History Congress in Helsinki. Remaining errors are entirely ours.
|
Notes |
|---|
1 Over 91% of those who rose to the level of branch manager began at Williams Deacon's before their 21st birthday.
2 There are two other reasons for increasing pay raises further up the hierarchy. The probability of further promotion may decline higher up the hierarchy because of smaller numbers of openings relative to competitors. Thus a greater prize must be offered in order to maintain the same expected value to competition. Second, if individuals have diminishing marginal utility of income, a higher increment upon promotion will be needed to provide the same incentives.
3 In the short-run, the creation of new managerial posts would create an immediate rise in the probability of promotion to manager for eligible staff (with sufficient experience). Since the new branches were relatively small, the ratio of managers to total employees rose and the steady state probability of promotion was also higher.
4 While the WDB records do not contain information about personal background, the practise of hiring those from a middle class background with secondary education was universal in the industry. See Klingender (1935), Blackburn (1967), and Hill (1982). The homogeneity of educational attainment was further ensured by the requirement that new recruits had to pass a banking entry exam covering geography, arithmetic, and English as a condition of employment.
5 Although it was by far the smaller partner, the name Williams Deacons was retained because of its seat on the London Clearing House. The official name after the merger was Williams Deacon and Manchester and Salford Bank. This was shortened to Williams Deacon's Bank in 1901. For the sake of simplicity, we use the name Williams Deacon's Bank throughout the study. In addition to Williams Deacon's and Manchester and Salford, sample employees began their careers at Heywood Brothers & Co. and Hardcastle Cross & Co. (two northern banks that were absorbed by Manchester and Salford Bank in the 1870s) and at Sheffield and Rotherham Bank (which was absorbed in 1906). Among the employees in the sample 6.6% started their careers at one of the banks that was absorbed by Manchester and Salford or Williams Deacon's (Williams Deacon's Bank, 1971).
6 Williams Deacon's Bank Limited, Staff Registers (see Williams Deacon's Bank, various years).
7 There are a few exceptions to this. Any employee who left the Bank before 1896 is absent from the records. The records for the Bank's head office at Mosley St. in Manchester are complete only through 1936. We do not believe that these omissions have much effect on the results presented in sections 3 and 4 as the number of missing observations is likely very small and the results of this paper are very robust. As a further check on the data were have used the overlap between the ledgers to check for missing or inconsistent information, and found no evidence of inaccuracy.
8 In the cases of first-year employees who joined after October 1 and last-year employees who left before October 1, we have recorded the information as of the latest date available.
9 From 1915 the records include female as well as male staff. Women had very different career prospects than men; for example, during this period they could not be promoted to branch manager and they typically had much shorter careers. Because of these differences in career structures, the sole focus of this paper is on male staff.
10 One issue that arises in the identification of managers is the treatment of employees at the sub-branches. Our approach to classification of managers at the sub-branch is as follows. If the branch and sub-branch are listed on the same page in the wage books, we treat the most senior employee at the sub-branch as a non-manager. If the sub-branch is listed separately from the branch, we treat the most senior employee at the sub-branch as a manager.
11 Williams Deacon's Bank Limited, Particulars of Branches. Although this volume continues through 1940, only the ledgers prior to 1900 list branch managers (Williams Deacon's Bank, various years).
12 We are fairly confident that this provides a reasonable measure of branch size. The volume Particulars of Branches provides information on credit transactions, debit transactions, and number of accounts at each branch (Williams Deacon's Bank, various years). The correlation between these variables and the number of staff at the branch ranges from 0.74 to 0.89.
13 The method used to identify branch managers would not help to identify different types of non-managerial employees because the number of tellers, accountants, and other senior staff differed across branches and within branches over time. It also would be unhelpful to identify position based on assumptions about wage because one of our main objectives in this paper is to identify the relationship between position and wage.
14 Many of those entering after their 21st birthday came to Williams Deacon's as a result of a merger, were hired during or immediately after the First World War, or were on temporary contracts.
15 Since these size divisions are somewhat arbitrary, we performed sensitivity analysis by trying different boundaries for the size categories. The results of all of the tests in Sections 3 are qualitatively very similar, regardless of the boundaries.
16 In three cases, a non-manager was promoted directly to managing either the Mosely Street or Birchin Lane branches. These individuals had very high wages prior to promotion and almost certainly occupied senior positions such as accountant or secretary.
17 This actually understates the total gains over the employee's lifetime because the Bank's pension was based on the final wage.
18 The gains are roughly equal in absolute amounts. This is the appropriate calculation if the worker compares the gain to a given disutility of effort. An alternative approach would be to compare the gains relative to his current salary—in this case, the incentives are declining over the individual's career.
19 It is also possible that the pay increases given to clerks for lateral moves were in part due to annual increments on the Bank's pay scales. The salary scale tended to give the largest increments to younger employees. The average tenure for clerks moving between branches was 9.1 years, compared to 13.5 years for clerks not moving. Thus it is possible that these clerks would have received above average increments, whether or not they moved branches.
20 The quartic specification originally comes from Murphy and Welch (1990). Seltzer and Merrett (2000) find that the quartic specification describes wages in the Australian banking industry better than the standard quadratic specification.
21 The Bank continued to pay staff who were on military leave (usually at the rate of the difference between their bank wage and military wage). We do not have information as to which staff were on leave and which were fully-employed at the Bank during the war, so we cannot examine—for example—whether pay increases varied between these two groups.
22 Since we are using real wages, the appropriate discount rate is the real interest rate. For that reason 2% is probably more accurate, but we include 5% as an extreme value. We have also estimated the gains incorporating the Bank's pension and find that they are 16%-40% larger than the values reported.
23 The lack of an effective instrument in the data for the exit decision precludes the approach of jointly modelling exits and wages.
24 The independent variables in Table 5 differ somewhat from those in Table 4. Many of the variables in Table 5 (moved branches, change in branch staff, promoted) are first differences of the variables in Table 4. Several of the remaining variables from Table 4 (entry age, entry age squared, entered through merger) have first differences of zero and thus were not included in the regressions in Table 5. We use a quadratic tenure specification rather than a quartic specification for simplicity, and further regressions (not reported) show that the main results of Table 5 are not sensitive to this specification. We also include dummy variables for 1919, 1920, and 1925—years when the Bank had across-the-board wage adjustments.
|
References |
|---|
|
TOPAbstract1 Introduction2 The UK banking...3 Promotions, demotions, and...4 The determinants of...5 ConclusionsAcknowledgementsReferences |
|---|
-
Baker G, Gibbs M, Holmstrom B. The wage policy of a firm. Quarterly Journal of Economics (1994) 109:921–55.[CrossRef][ISI]
Bankman. The bankman's status. The Bank Officer (1919) 1:6.
Blackburn RM. Union Character and Social Class: A Study of White-Collar Unionism (1967) London: B.T. Batsford Ltd.
Boot HM. Salaries and career earnings in the Bank of Scotland, 1730–1880. Economic History Review (1991) 44:629–53.[CrossRef][ISI]
Capie F, Webber A. A Monetary History of the United Kingdom, 1870–1982 (1985) London: George Allen &Unwin.
Capie F, Wood G. Money in the economy, 1870–1939. In: The Economic History of Britain, Volume 2: 1860–1939—Floud R, McClosky D, eds. (1994) Cambridge: Cambridge University Press.
Doeringer P, Piore M. Internal Labor Markets and Manpower Analysis (1971) Lexington, MA: D. C. Heath and Company.
Feinstein CH. National Income, Expenditure and Output of the United Kingdom 1855–1965 (1972) Cambridge: Cambridge University Press.
Gilbart JW. A Practical Treatise on Banking (1856) 6th. London: Bell & Daldy.
Hill J. From Subservience to Strike: Industrial Relations in the Banking Industry (1982) St Lucia: University of Queensland Press.
Kynaston D. The City of London, Volume II 1890–1914 (1994) London: Random House.
Klingender F. The Condition of Clerical Labour in Britain (1935) London: Martin Lawrence.
Lazear E, Rosen S. Rank-order tournaments as optimum labor contracts. Journal of Political Economy (1981) 89:841–64.[CrossRef][ISI]
Lee C. The service industries. In: The Economic History of Britain—Floud R, McCloskey D, eds. (1994) Cambridge: Cambridge University Press.
Murphy K, Welch F. Empirical age-earnings profiles. Journal of Labor Economics (1990) 8:202–29.[CrossRef][ISI]
Prendergast C. The provision of incentives in firms. Journal of Economic Literature (1999) 37:7–63.[ISI]
Rae G. The Country Banker: His Clients, Cares, and Work from an Experience of Forty Years (1930) 7th. London: John Murray.
Rosen S. Prizes and incentives in elimination tournaments. American Economic Review (1986) 76:701–15.[ISI]
Seltzer AJ, Merrett D. Personnel practices at the Union Bank of Australia: panel evidence from the 1887–1900 entry cohorts. Journal of Labor Economics (2000) 18:573–613.[CrossRef][ISI]
Williams Deacon's Bank Limited (various years). Staff Registers. The Royal Bank of Scotland Archives, London, GB1502/WD/480/1 (covers 1890–1901), GB1502/WD/480/2 (covers 1901–1910), GB1502/WD/481 (covers most northern branches, 1907–1918), GB1502/WD/46/1 (covers London branches and Mosley St. 1911–1921), GB1502/WD/482/2 (covers most northern branches 1919–27), GB1502/WD/482/3 and WD/482/4 (covers most northern branches 1924–1934), GB1502/WD/482/5 and GB1502/WD/482/6 (covers most northern branches 1931–1941), GB1502/WD/46/2 (covers London branches and Mosley St. 1921–36).
Williams Deacon's Bank Limited. Particulars of Branches (various years) London: The Royal Bank of Scotland Archives. GB1502/WD/377/1.
Williams Deacon's Bank Limited. Staff Record Cards (various years) London: The Royal Bank of Scotland Archives. GB1502/WD/475.
Williams Deacon's Bank Limited. Williams Deacon's 1771–1970 (1971) Manchester: McCorquodale and Co.
Winton JR. Lloyds Bank, 1918–1969 (1982) Oxford: Oxford University Press.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||