The Liberalisation of Incorporation in the United Kingdom

Prior to the liberalisation of incorporation law in the middle of the nineteenth century, the right to incorporate in the United Kingdom was controlled by Parliament and the Crown. From 1844 onwards, there were a series of acts, which made incorporation, and limited liability, easier to obtain. The 1862 Companies Act saw a consolidation of existing pieces of legislation and was the capstone on the liberalisation process. The ownership and control of the companies that incorporated under the 1862 Act, and raised share capital from the market, are the focus of this study.

Manager-Owner Conflict

Many of the companies incorporated under the 1862 Companies Act had diffuse ownership (Acheson et al. Reference Acheson2015). Such a separation of ownership from control allows specialisation in share ownership (or residual risk-bearing) and in managerial control (Fama and Jensen Reference Fama and Jensen1983). With the liberalisation of U.K. incorporation law, this specialisation was available to every firm. However, this specialisation may give rise to an agency problem, in that managers pursue their own interests and do not act in the interests of shareholders (Berle and Means Reference Berle and Means1932; Jensen and Meckling Reference Jensen and Meckling1976). Examples of managerial inefficiency in the late Victorian era have been highlighted by a number of authors looking at the railway industry (Arnold and McCartney Reference Arnold and McCartney2005; Crafts, Leunig, and Mulatu Reference Crafts, Leunig and Mulatu2008; Mitchell, Chambers, and Crafts Reference Mitchell, Chambers and Crafts2011).Footnote ⁶ A possible solution to the managerial agency problem may be provided by monitors who have the incentive to collect information about the firm, and the power to influence managers.

In this study, we categorise the key monitors into three groups: large active family blockholders, large passive blockholders, and directors who are not blockholders. A blockholder is a shareholder controlling more than 10 percent of the company. We hypothesise that the presence of large active blockholders will reduce the conflict between managers and owners, and raise profitability as measured by ROA. These large active blockholders have the incentive and the power to reduce managerial agency costs and raise operating profits. They may run the firm on a day-to-day basis, or at least be closely involved in it, ensuring that all other managers and employees maximise their efforts, and minimise costs. On the other hand, large passive blockholders, with less involvement in the monitoring of managers will not have such an effect.

Directors who are not blockholders often have little impact on the manager-owner conflict. In dispersed-ownership companies, Eugene F. Fama and Michael C. Jensen (Reference Fama and Jensen1983) have argued that it is natural that the most influential members of the board will be internal managers as they have the most information about the firm, and they are also well placed to nominate outside directors. Alternatively, the directors may be independent of management, but they do not have the day-to-day involvement in the firm, which would be required to reduce the agency problem. Their real influence would be at a strategic level, on major issues that came before the board, not on everyday issues.

To summarise, we use ROA as our measure of operating performance to show how efficiently a company uses its assets to earn a profit. The greater the managerial expropriation, the lower the return on assets. If large active blockholders can reduce managerial expropriation, then, ceteris paribus, ROA will be higher. On the other hand, large passive blockholders and directors who are not blockholders should have had no impact on ROA.

Blockholder-Minority Shareholder Conflict

Although large active blockholders may be able to discipline managers, they may focus on their own interests and attempt to maximise these at the expense of minority shareholders (Shleifer and Vishny Reference Shleifer and Vishny1997). This introduces an additional risk factor that investors must consider in their valuations. If they are concerned about expropriation, then the high profitability earned in companies with large active blockholders may not necessarily translate into higher share prices. However, if investors feel protected from expropriation then they may reward the firm with a relatively higher valuation. The blockholder-minority shareholder conflict generally concerns strategic decisions, rather than day-to-day management, and would be typically dealt with at board level. Consequently, having an independent board could help promote good practice and increase shareholder value. Boards with more directors who are not blockholders could be viewed as a commitment that there is a low risk of blockholder expropriation, as collusion would be much more difficult. The result is that, ceteris paribus, valuations (as measured by Tobin's Q) should be higher.

There were numerous ways in which blockholders could put their own interests before those of minority shareholders First, the active blockholder could engage in tunnelling (or related party transactions),Footnote ⁷ by getting the company to lend money at low rates, or buy assets at inflated prices. These would typically affect only the balance sheet, rather than profitability, in the short term. Another form of tunnelling can occur when the blockholder gets the company to trade with other firms that he owns. The company is then charged inflated prices and the benefits are earned by the blockholder's other firms, at the expense of the minority shareholders. Even if this did not immediately occur but investors understood that it could occur in the future, it would mean that current profitability could remain high, but the share price would be discounted to reflect the additional risk. An example of this type of tunnelling was seen in the case of the Grosvenor Hotel; R. C. Drew was the largest shareholder and exercised control over a small board, who colluded with him to obtain most of the supplies for the hotel from a butcher, grocer, and laundry all owned by Drew.Footnote ⁸

Blockholders could also put themselves first by withholding information. As insiders, they had detailed knowledge about the performance of the firm, which they could use to inform their decisions on whether to buy or sell stock in it (insider trading). They could deny much of this information to others, placing the minority shareholders at a disadvantage. For example, The Financial Times remarked about the Globe Cotton Spinning, a company in our sample, that it paid a dividend by forwarding “a cheque to each shareholder for whatever amount may be due to him at a rate the Directors fix on … Apparently no accounts are circulated … No information is obtainable beyond the capital subscribed and a few office particulars.”Footnote ⁹ This secrecy would tend to raise the risk for minority shareholders, and lower valuations, despite the profitability of the company.

Active blockholders could also continue in their position within the company, even when it was no longer performing well. The removal of a large active blockholder, as managing director, would likely have been much more difficult than the removal of someone who was not a blockholder. The Financial Times reported on the Aylesbury Dairy implying that the severe difficulties it faced were related to the past decisions of its active blockholder, G.M. Allender, who it referred to as the “Life Managing Director.”Footnote ¹⁰ A similar problem could arise if the active blockholder used his power to appoint some of his family as his replacement. In a similar vein, if there was a risk that the blockholder would engage in empire-building, it could be detrimental to shareholder value. Foreman-Peck and Hannah (Reference Foreman-Peck and Hannah2015, p. 20) note the example of Samuel Waring who engaged in a disastrous expansion and acquisition plan with the furniture company Waring & Gillow. Finally, and most blatantly, the active blockholders could attempt to deceive the minority shareholders by falsifying financial statements, to attract capital.

The potential for these abuses by large active blockholders, even if they did not materialise, raised the risks faced by minority shareholders. We hypothesise that even though large active blockholders may raise profitability (ROA), they do not necessarily raise firm value (Tobin's Q). Large passive blockholders, who have no effect on profitability, should have no effect on value, because they do not exercise enough power to engage in expropriation. However, having a board with many directors who are not themselves blockholders will significantly increase the value of the firm, as collusion on expropriation would be much more difficult. If their focus is on major strategic decisions, rather than the minutiae of day-to-day decisions, such directors may have little influence on profitability, but they can provide safeguards against systematic expropriation.

Empirical Framework

We examine the relationship between the governance structure and performance of non-statutory public companies in Victorian and Edwardian England. Our estimating equation takes the form

(1)

We use ROA and Tobin's Q as performance measures, and the dependent variables, in our post-1880 performance sample. ROA is measured as:

(2)

and Tobin's Q is calculated as follows:

(3)

where MVEq = year-end market value of firm's common stock; Pref = year-end value of firm's preferred stock; Debt = year-end value of firm's total debt; and BV Assets = total assets of firm. Controls on age, size, location, growth, industry, ownership structure, and corporate control measures are described in the next section.

The main threat to validity that has been identified in studies of ownership structure is that of causality: does ownership structure affect performance or performance ownership structure (Cho Reference Cho1998; Demsetz and Villalonga Reference Demsetz and Villalonga2001). Harold Demsetz and Belén Villalonga (Reference Demsetz and Villalonga2001, p. 221) argue that “compensation plans, insider trading possibilities, and corporate takeovers suggest that firm performance may influence ownership structure as well as be influenced by it.”

However, there are several reasons to believe that reverse causality in a Demsetz and Villalonga world was not a major issue in the Victorian context. First even if some related-party transactions or tunnelling could be seen as affecting compensation, directors in our firms were paid said salaries and performance related bonuses, none of which resulted directly in changes of ownership, unlike modern equity-based plans. Second, hostile takeover bids and accumulation of stakes by potential acquirers were very rare in this era (Cheffins Reference Cheffins2008, p. 79; Hannah Reference Hannah1974, Reference Hannah1976, p. 150, Reference Hannah, Green, Owens, Maltby and Rutterford2011, p. 241). Third, Fabio Braggion and Lyndon Moore (Reference Mayer2013) find that while insider trading was not illegal, it rarely happened. Fourth, institutional investors were very small players and typically only invested in fixed-interest securities such as corporate and government bonds, not equities (Cheffins Reference Cheffins2008, p. 190; Raynes Reference Raynes1928).Footnote ¹¹

Indeed, in the Victorian context, it was difficult for changes in ownership structure to occur quickly, as stock was often not as actively traded.Footnote ¹² This meant that large one-off changes in ownership structure were not possible. We are not suggesting that blockholder changes were uncommon. However, it would have taken investors a long time to accumulate a block-holding stake in a company to fundamentally alter the ownership structure.

The absence of such mechanisms does not completely eliminate the reverse causality issue. One potential source of reverse causality is where a blockholder may be more likely to take a position on the board when companies are more profitable. In a robustness check, we find very little evidence of large passive blockholders rotating in or out of directorships, but we cannot conclusively rule this possibility out. However, it could be argued that blockholders might be more likely to take a seat on the board if profitability was low and the company was not performing well, in the hope of improving performance. A finding that active blockholders are associated with high profitability runs counter to what would be expected from results driven by endogeneity.

As an additional robustness check, we use a different measure of performance, which should be less prone to the reverse causality problem, namely firm outcome. It seems plausible that ownership structure could affect the final status of the company. However, it is unlikely that the future state of the company would affect current ownership. Firm outcome differs from our other measures of performance, ROA, and Tobin's Q, in that conceptually, we are measuring performance or success purely in terms of survival (Alchian Reference Alchian1950). We are posing the following two questions: (a) Was a specific type of ownership structure a survivable trait of firms over the long run? (b) Did a certain type of ownership structure make it more likely for a firm to fail for explicit performance reasons?Footnote ¹³

Although reverse causality should not be a major issue in this analysis, there may be unobserved factors that jointly cause performance and ownership structure. There may also be issues in terms of whether certain conditions led to different ownership configurations. For example, a firm founder who was likely to obtain high rents, and therefore high ROA, may keep a large block of shares to extract as much value as possible. The founder of a firm with high capital expenditures, and who needed to repeatedly raise more capital from minority shareholders, would be more likely to initially configure the firm with independent directors. For firms that were expected to be more successful, there may have been stronger demand for shares from subscribers, so that ownership became less concentrated. These issues mean that, ultimately, the most that we can say about any relationships we find is that they are correlations, and do not necessarily imply causality.

Sample

Our main sources of ownership data are the Companies Registration Office files held at the National Archives in London (BT31 series) and the National Archives of Scotland (BT2 series). Companies registered under the 1856 and the 1862 Companies Act were required annually to return a list of their ordinary and preference shareholders to the Registrar of Companies. Up until 1970, whenever a company was dissolved in the United Kingdom, its ownership records were placed within the Companies Registration Office files.Footnote ¹⁴

Because we are interested in the ownership of publicly-listed companies, we collected the names of all U.K. companies issuing common stock between 1862 and 1900 from the Course of the Exchange (COE) and the Investor's Monthly Manual (IMM). Unfortunately ownership returns were extensively weeded by archivists so that, for any one firm, only a small sample of years was preserved. We collected ownership returns for 1865, 1870, 1880–1884, 1890, and 1900 or one or two years either side of these sample years if the return existed. If a company had ownership returns that fell outside the selected sample years, we collected a return for each decade between 1860 and 1900, where available.

The ownership returns report the cash-flow rights of shareholders, but not the voting rights of shareholders or the names of directors. We therefore hand collected data on each company's voting rights for each year for which we had an ownership return from Burdett's Official Intelligence (BOI), Stock Exchange Official Intelligence (SEOI), or a company's articles of association, usually available with the ownership returns. The names of directors for the relevant years were obtained from the same sources. In order to ascertain the share ownership of each director, we manually checked each set of ownership returns.

As noted previously, we use three performance measures. The accounting data to calculate Tobin's Q and ROA comes from Burdett's collection of company accounts stored in the Guildhall Library in London, while the market value of the firm's securities is obtained from the December issue of the relevant IMM.Footnote ¹⁵ As Burdett's accounting data is only available from 1880 onwards, our first sample is restricted to the post-1880 period and contains 252 firm-years, consisting of 186 unique firms, 11 with three observations, 44 with two observations, and 131 with one observation. Within our sample, 34.1 percent of firm years come from the 1880s, 40.5 percent from the 1890s, and 25.4 percent from 1900–1901. The four most represented industry sectors, as shown in Appendix Table 1, are industrial and commercial firms comprising 25.8 percent of our sample, banks comprising 15.5 percent, mortgage and finance companies comprising 13.9 percent, and iron, coal, and steel companies comprising 12.3. The remainder of the sample is spread across eight other industry sectors. We obtain survival/failure date from the Register of Defunct Companies, which lists any company delisted from the stock exchange from 1875 onwards with the reason for delisting.Footnote ¹⁶ We are able to trace the year of delisting as well as the ultimate fate of 344 companies.

Table 1

SUMMARY STATISTICS

Notes:

Data is winsorised at the 1^st and 99^th percentiles.

Source:

See text.

Compared to the total population of publicly-quoted companies listed in BOI, our sample is a relatively small proportion. According to Essex-Crosby's data, there were 1,585 and 2,581 non-railway companies listed in 1885 and 1895, respectively (Jefferys Reference Jefferys1977, p. 458). In terms of the number of unique companies quoted in the COE and IMM between 1862 and 1901, we estimate that there were 2,664 companies excluding railways, quoted at some point during this period. In Appendix Table 1, we present a detailed breakdown of how our sample compares to non-railway companies listed in the IMM. The median par value of IMM firms in 1880 was £150,000 and in 1900 was £250,000. Our performance data sample firms have a median par value of £150,000, while our final status sample firms have a median par value of £133,099, implying that our sample is more weighted towards small- and medium-sized public companies. When compared to all of the non-railway companies in BOI, which included more of the small firms, our sample is fairly representative. The average paid-up capital of our companies for the ROA and Tobin's Q sample is £256,700, and for the final status sample is £216,560, which are similar to the average paid-up capital of non-railway companies in BOI in 1885 and 1895, which were £209,752 and £262,340, respectively (Jefferys Reference Jefferys1977, p. 458).

There were some substantial changes in industrial composition during this period, as seen from Appendix Table 1, but the only industry which seems to be under-represented in our sample in both 1880 and 1900 is mining. The commercial and industrial sector, which consisted mainly of manufacturing and processing companies, is the largest sector in our sample, which is unsurprising since this was a growth sector in the stock market (Grossman Reference Grossman2002, p. 130). In terms of those listed in BOI, our sample under-represents the commercial and industrial sector and over-represents banking. This is again unsurprising because banks were more likely than other companies to have a large shareholder base and have their shares traded on public markets, whereas industrial and commercial public companies were more likely to be small and not regularly traded.

Ownership Variables

The richness of our data permits a wide variety of measures of ownership structure. Following Demsetz and Kenneth Lehn (Reference Demsetz and Lehn1985) and Demsetz and Villalonga (Reference Demsetz and Villalonga2001), we looks at the ownership of the five largest shareholders as this may proxy the ability of shareholders to control managers. We can also construct a Herfindahl Index of ownership. We describe ownership by the board of directors as a whole, and of insiders that embraces both directors and blockholders. We then split this insider category into three types of individuals. Large active blockholders (i.e., individuals who owned more than 10 percent of capital or voting rights and were on the board), large passive blockholders (i.e., individuals who owned more than 10 percent of capital or voting rights and were not on the board), and directors who were not blockholders. The variables are described in Appendix Table 2.

Table 2

SUMMARY STATISTICS

Notes:

The propensity for a director to be a large blockholder is calculated as (ActiveBlockholder)/(Solely Director + ActiveBlockholder) and the propensity for a large blockholder to be a director calculated as (ActiveBlockholder)/(Passive Blockholder + Active Blockholder).

Source:

See text.

Summary statistics are presented in Table 1, VoteLargest5 and VoteHHI variables both indicate that corporate ownership was relatively diffuse in our sample. Notably, the average of the VoteLargest5 variable at 21.44 percent is very similar to the capital owned by the five largest shareholders in modern studies (e.g., Demsetz and Villalonga Reference Demsetz and Villalonga2001; Davies, Hillier, and McColgan Reference Davies, Hillier and McColgan2005; Maury Reference Maury2006). In our regression analysis, we take log values of VoteLargest5 and VoteHHI to obtain symmetric distributions. Directors, on average, control a fairly small proportion of voting rights in our sample, with the averages of VoteDir being 11.53 percent, not dissimilar to that reported by Foreman-Peck and Hannah (Reference Foreman-Peck and Hannah2012) for the 337 largest British firms in 1911. We also summarize the ownership of large active blockholders who were directors, the ownership of large passive blockholders who were not directors, the ownership of directors who were not blockholders, and the ownership of all insiders i.e., directors plus large blockholders. The mean and median of VoteInsiders, which measures the votes of both directors and large shareholders, are 14.38 and 9.06 percent, respectively. ActiveBlock and PassiveBlock averages imply that a minority of firms had someone who was a large blockholder. However, the mean votes controlled by those blockholders was 24.87 percent, and for those firms that had a passive blockholder, the mean votes controlled by those blockholders was 24.90 percent. The DirOnlyVote variables show that directors who were not large blockholders, on average, held a total of 7.78 percent of voting rights between them.

Although not reported separately, there is little difference between the concentration of capital and voting, and, if anything, voting is slightly less concentrated than capital. Many companies had voting regimes with caps limiting the total votes of each investor or which skewed the voting scale in favour of small investors and discriminated against large shareholders (Campbell and Turner Reference Campbell and Turner2011).Footnote ¹⁷

In Table 2 we describe how ownership by directors and large blockholders varies across industry. Four things are worthy of note. First, very few firms in any sectors had large blockholders. Second, there was a very small propensity for directors to be large blockholders, but a higher propensity (60.5 percent) for large blockholders to be directors, which implies that the majority of such shareholders wanted to maintain control over the company. Third, the vast majority of directors were not blockholders, with the average board having 6.1 directors and, on average, 5.9 of these were not large blockholders. Fourth, there is some variation across industries in terms of the percentage of voting rights controlled by large blockholders, with breweries, iron, coal, and steel companies, and industrial and commercial companies having higher blockholder ownership than other industries.Footnote ¹⁸

Performance Variables

As noted previously the main performance measures are ROA and Tobin's Q. In our regression analysis, we winsorize each of the variables at the 1^st and 99^th percentiles to deal with outliers and use industry-adjusted Tobin's Q and industry-adjusted ROA whereby Tobin's Q, and ROA are adjusted using industry averages as per the industrial sectors in Table 2.Footnote ¹⁹

Our robustness check on performance is firm status. The Register of Defunct Companies explains why firms delisted. We use this information to create variables about the ultimate fate or final status of our sample companies. We find that 41 percent of firms merged with other companies and disappeared as separate entities; 5 percent of firms were wound up by court orders; 15 percent of firms were removed from the stock exchange yearbook; 24 percent of firms were voluntarily wound up by their shareholders; 16 percent of firms were reconstructed (see Table 1).

In terms of ex ante optimality for shareholders, if a firm were to disappear, the worst outcome would have been a court order to wind up, followed by a voluntary winding up. In the former, larger losses would usually have been incurred by shareholders, whereas the losses may not have been as severe in the case of a voluntary winding up. Removal from the Stock Exchange Yearbook usually occurred because a firm had failed, which again suggests poor performance. The best outcome would be a merger. The extensive rationalisations and merger movements in many of the industries at the turn of the twentieth century (Sykes Reference Sykes1926; Supple Reference Supple1970, pp. 273–96; Hannah Reference Hannah1974, Reference Hannah1976) suggest that few mergers were for explicit performance reasons. Reconstruction of companies also does not necessarily imply performance issues with firms, and would lie between the best outcome (i.e., merger) and the worst outcomes (i.e., winding up or removal).Footnote ²⁰

We include two variables to capture governance and control, Board size (NumDir) ranged from 3 to 23, with a mean and median of 6.0.Footnote ²¹ Independence is captured by the number of directors who are not also large blockholders (DirOnlyNum) and ranges from 2 to 23, with a mean of 5.8 and median of 5.0 (See Table 1).

We use the establishment year of the company to capture the maturity of the firm (EstDate). We control for cyclical variations in ROA and Tobin's Q by using dummies for the years in which the data was obtained, and capture a time effect (OwnDate) for the final status of the company. We control for firm size by including the natural logarithm of a firm's paid-up capital in our regressions (Size) and a leverage ratio (LTDebt).Footnote ²²

Since directors in this era were usually required to own a certain number of shares, which was usually low relative to the capital of the firm (Campbell and Turner Reference Campbell and Turner2011), we include a director qualification variable (DirectorQual) because this may have affected the relationship between ownership and performance. The mean, median, and range suggest that director qualifications were not a high proportion of the firm's par (paid-up) capital. Indeed, on average, 29.3 percent of shareholders had enough shares to qualify as a director.

We also control for whether or not a firm had London headquarters because such firms may have had greater access to capital markets and oversight by professional investors (HeadLondon). As our data comes from two different archives, we control for any possible effect this might have on the relationship between ownership and performance by including a binary variable which takes the value of one if our firm was Scottish, zero otherwise. In our regressions where ownership is regressed on Tobin's Q, we also control for the ROA, allowing us to analyse the determinants of firm value after controlling for differences in operating performance.