Firm Location

The provincial markets were spread across the United Kingdom, so it might be expected that geography could have played a role in terms of what companies they listed. In Table 3, we cross-tabulate the location of a firm’s head office with its listing type, when it first appeared in the IMM. Those firms headquartered in London and overseas were overwhelmingly chiefly traded on the LSE. By way of contrast, about 81 percent of companies with regional headquarters were chiefly traded on the provincial markets and 66 percent exclusively so.

Table 3

LISTING STRATEGIES OF COMPANIES BY LOCATION OF THEIR HEADQUARTERS

Notes: Analysis is based on when a company was first listed. Company headquarters were reported in the IMM and were coded into regions. The listing strategies were defined as follows: “London Only” are securities that were chiefly traded only on the LSE, “London & Provincial” are securities that were chiefly traded on the LSE and other regional stock exchanges, “Provincial Only” are securities that were chiefly traded just on provincial markets. Northern Ireland only existed as a separate country during the last decade of our sample, but we treat it as a separate region for the entire period.

Source: Authors’ calculations based on data from the IMM.

For many companies, it seems likely that the choice of where to base the headquarters was made first, focused on operating considerations, with the listing decision being made afterward. At the start of our sample period, many companies were based in the provinces for practical reasons. For example, railways, canals, docks, and shipping firms were geographically bound and they served customers in a fixed geographic area. Banks tended to focus on small areas so that they could monitor customers closely. Coal, iron, and steel companies were also often based in the rest of the United Kingdom, with agglomeration economies in the form of transport cost savings helping to explain why these industries were based where they were.

Firms in the traditional sector of spinning and weaving were much more likely to be based in the provinces, particularly Lancashire. Natural and acquired advantages, such as the availability of water power and access to coal, explain why the cotton industry was originally located in this region, but the cotton textile agglomeration explains its locational persistence (Crafts and Wolf 2014). Lancashire’s agglomeration advantages included tacit knowledge, dense labor markets, a network of suppliers, marketing expertise, and the scale of the Liverpool cotton exchange (Broadberry and Marrison Reference Broadberry and Marrison2002; Balderston Reference Balderston2010; Crafts and Wolf 2014).

Once a company was established in a certain area, there may have been a local bias that made it attractive to investors in the surrounding area. Lowenfeld (Reference Lowenfeld1909, p. 121) noted that “if the question of what people think of stocks and what they consider safe is analyzed, we come again to a point of geography pure and simple.” Even in modern capital markets, investors prefer to hold shares in local firms because they have easier access to information about those companies headquartered near to them (Coval and Moskowitz 1999; Gaspar and Massa Reference Gaspar and Massa2007), and this was likely also the case historically (Broadbridge Reference Broadbridge and Reed1968; Campbell and Turner 2012; Reed Reference Reed and Reed1968, 1975; Turner Reference Turner2009; Acheson and Turner Reference Acheson and Turner2011; Newton Reference Newton2010).

However, as time progressed, the foundations given to the provincial markets by regional companies were undermined, partly by merger waves and partly by the emergence of new industries. Of those railways that operated within the British Isles, companies with their head office outside of London represented 49 percent of market capitalization in 1869, but this had fallen to just 2 percent by 1929. Why this dramatic shift? The government’s rationalization of the U.K. railway sector in the Railways Act of 1921 created four railway groups or companies. This rationalization meant the disappearance of nearly 120 railway separate companies, 74 of which, according to the 1920 Stock Exchange Yearbook, had been based outside of London. The four new companies were all headquartered in London, which may have reflected a desire by the government to be able to monitor and control them closely. The railways had been increasingly regulated during the late nineteenth and early twentieth century. They had entered into direct government control during WWI, and this had continued until the Railways Act of 1921 created the new groupings.Footnote ¹⁰ The state would still be influential in the running of the new companies and having them managed from London may have made this easier.

In 1869, about 71 percent of the banking sector’s market capitalization was based outside London, but by 1929, this had fallen to 34 percent. This change was largely due to amalgamations, with 116 bank mergers occurring between 1872 and 1920 (Sykes Reference Sykes1926). Using various issues of the Stock Exchange Yearbook, we found that 48 of these mergers resulted in the disappearance of headquarters that had previously been outside London. One major reason for the movement of the newly merged bank headquarters to London was the concentration of expertise and human capital in the City (Michie Reference Michie1997). In addition, lower communication costs, thanks to the development and widespread use of the telephone and telegraph, made it easier for managers based in London to monitor what was happening in remote branches.

The mergers in the railway and banking sectors are plausibly exogenous events. In both industries, the shift of headquarters to London happened because of the mergers and not because of the desire of the newly merged companies to list on the LSE.

Campbell, Grossman, and Turner (Reference Campbell, Grossman and Turner2019) estimated that railways and banks were responsible for 76 percent of market capitalization in 1869, but this had fallen to 39 percent by 1929. This was mainly due to structural changes within the economy. There had been ongoing technological innovation, opening up new fields, such as oil, electricity, chemicals, and consumer products such as beer and tobacco. These newer industries tended to be based in London for several reasons. Improvements in telecommunications made it much easier for companies to be controlled from a central point. Similarly, London offered transportation advantages that made it possible to travel across the country quickly, with a mature railway network, shipping links, and a growing road system. There had also been the development of local railways, tramways, the underground, and omnibuses, which made travel around and from London relatively easy (Michie Reference Michie1997; Scott Reference Scott1998).

London was also a global distribution and trading hub (Scott Reference Scott1998). It became a more dominant political center from 1914 onward (Michie Reference Michie1997, p. 78) and was the base of the government, the Civil Service, the military, the legal system, and the British Empire. This may help to explain the strong propensity for companies that had an international focus to base themselves in London. Of the 1,651 companies in our sample that operated mainly abroad,Footnote ¹¹ only 5 percent had their offices in the provinces of the United Kingdom, about 10 percent were based in a foreign country, and an overwhelming 85 percent chose to have their headquarters in London.

London also had a large and flexible workforce, which had been enlarged by the arrival of commuter trains (Crafts and Leunig Reference Crafts and Leunig2014). It also had a dense and unique clustering of support services for businesses (e.g., marine insurance, merchant banks, accountants, and brokers), which facilitated commerce (Michie Reference Michie1997, p. 76; Mollan and Michie Reference Mollan and Michie2012).

These changes in industry composition resulted in fewer large companies being based in the provinces. The decline in the proportion of companies in our sample that had their headquarters in the rest of the United Kingdom, outside of London, is shown in Figure 1. There is a visible connection with the proportion of companies that were chiefly traded only on provincial markets. To test empirically the relationship, we calculated the changes in each of the variables to deal with non-stationarity. We then regressed the monthly change in the proportion of companies with provincial-only listings against the change in the proportion of companies with headquarters in the rest of the United Kingdom, and found a highly significant relationship, with a t-statistic of 15.2 and an R-squared of 63.4 percent.

Figure 1

PROPORTION OF COMPANIES WITH HEADQUARTERS OUTSIDE LONDON OR CHIEFLY TRADED ONLY ON PROVINCIAL MARKETS

Notes: “HQ: Rest of UK” shows the percentage of companies listed in the United Kingdom whose headquarters were based in regions outside London. “Markets: Only Provincial” shows the percentage of companies whose securities were traded only on provincial markets.Source: Authors’ calculations based on data from the IMM.

It is possible that there was also some reverse causality involved in this. It could be argued that companies increasingly chose to base themselves in London to make it more likely that they could get listed there. However, this does not seem plausible, as it was still very possible for companies based elsewhere to list on the LSE. Only a minority of the provincial companies in our sample chose to do so, but they are clear evidence that basing themselves in the regions did not prevent a listing in London. The preferred listing venue, therefore, should not have been the main determinant of where the headquarters was based.

Industry Sector and Firm Size

The industry sector that a company was involved in may also have played a role in terms of whether a company was solely traded on London or the provincial exchanges, or whether they were cross-listed on both. Patterns in terms of this for each industry are shown in Table 4.

Table 4

INDUSTRY VERSUS WHERE CHIEFLY TRADED

Notes: Industries are generally consistent with how they were originally reported in the IMM, but we have split companies that had been included in the miscellaneous “Commercial and Industrial” into more detailed subcategories. Includes companies that operate mainly overseas, explaining why railways are reported as being frequently listed on London Only.

Source: Authors’ calculations based on data from the IMM.

By their nature, some industries operated over several regions, and companies with their headquarters in London could still be regarded as local in other areas. For example, the network of the London and North Western Railway spread over hundreds of miles and included stations in major cities such as London, Liverpool, and Manchester. As well as operating in multiple regions, the major railways also tended to be traded on multiple exchanges in these areas. In contrast, there were some industries whose firms tended to work in very small areas for monitoring purposes, such as banks, and would have been more likely to list on just a single exchange.

Those firms that were listed jointly on London and the provincial markets tended to be much larger, on average, as shown in Figure 2. This may have been a spillover from some of the railways having a dual listing, but large blue-chip firms would also have been naturally more capable of attracting investors across multiple venues. Other large firms may then also have wanted to cross-list across several exchanges, to achieve a certification effect that would boost their status. This may help to explain the rising proportion of firms on provincial markets that also had a listing on London.

Figure 2

MEDIAN SIZE OF COMPANIES BY LISTING TYPE

Notes: The market capitalization of each company was calculated each month by multiplying the number of shares by the price and then summing across all of the securities listed for that company. The median company size was then calculated each month for companies that were cross-listed on both the London and the provincial markets (“London & Provincial”) or solely on the London exchange (“Only London”) or solely on the provincial exchange (“Only Provincial”).Source: Authors’ calculations based on data from the IMM.

Size, therefore, may have played a role in the listing decisions of large firms. However, it may also have been a factor at the other end of the distribution, as some exchanges imposed listing requirements that mandated that only companies of a certain size could be traded. The provincial markets tended to be quite liberal, with the minimum for Liverpool set at £50,000 and the guideline for Manchester being only that they would not quote small companies “insufficiently in the public hands” (Thomas Reference Thomas1973).

For an official quotation on London, there was a general policy that companies should be of “sufficient magnitude and importance” (Burhop, Chambers, and Cheffins Reference Burhop, Chambers and Cheffins2014, p. 63). In 1902, it was suggested that firms should have a subscribed capital of at least £100,000 (Gore-Browne Reference Gore-Browne1902, p. 454).Footnote ¹² However, it was possible that if a company could not meet the conditions for an official quotation„ they could still be traded if they applied for a Special Settlement, which required only very basic conditions to be met. These companies became more visible in 1916 when a Supplementary List was introduced by the LSE, which showed the prices of these securities (Burhop, Chambers, and Cheffins Reference Burhop, Chambers and Cheffins2014, p. 64; Michie Reference Michie2001, p. 272). This may have attracted smaller firms to also trade there.

Firm Characteristics and Listing Strategy

To analyze the importance of firm characteristics as a determinant of listing strategy, we use a multinomial logit regression. In this specification, the dependent variable can take on three values depending on where a company was chiefly traded: (1) Only London, (2) Jointly London and Provincial Markets, and (3) Only Provincial Markets. We choose Only London as the base group, to analyze what factors were associated with making it more or less likely that a company had some form of provincial listing rather than being traded only on London.

To analyze these factors within our multivariate analysis, we test whether a company having its headquarters in the regions (HQRestUK) affected listing strategy. We also control for differences between the two principal industry sectors (IndTransport and IndFinancial) and for differences in size (MarketCap). We repeat the regressions for the last month of each decade, with the results shown in Table 5.Footnote ¹³

Table 5

MULTINOMINAL LOGIT REGRESSIONS EXPLAINING DETERMINANTS OF WHICH MARKETS COMPANIES WERE LISTED ON

Notes: Multinomial logit analysis with listing strategies as the dependent variable, with the following specification:

$${ListingType = \beta_0 + \beta_1(HQRestUK) + \beta_2(IndTransport) + \beta_3(IndFinancial) + \beta_4(MarketCap).}$$

“London Only” is the base ListingType group. Robust standard errors are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. When companies had multiple securities, the returns and liquidity are averaged across securities. Definitions of variables are provided in Appendix Table 1.

Source: Authors’ calculations based on data from the IMM.

The first variable to focus on is HQRestUK, which equals one if a company had its headquarters anywhere in the United Kingdom, outside of London. As was suggested from the cross-tabulation in Table 3, this was a very important predictor of having a provincial listing and was significant for both Joint London and Provincial listings and for Only Provincial listings. This finding is maintained in every decade of our sample. This implies that companies that were based in the regions were significantly more likely to have some form of provincial listing, rather than an Only London listing.

Companies in the transport sector were also significantly more likely to list jointly or solely on provincial markets than to be listed only on London. However, as can be seen from Table 5, this relationship broke down in the 1920s, after the grouping of the railways into a small number of very large firms. Financial companies were significantly more likely to be traded solely on provincial markets.

The size of companies had an effect at both extremes, as was suggested from the patterns observed in Figure 2. There was a positive relationship, generally significant, with having a Joint London and Provincial listing. This suggests that larger companies were more likely to be cross-listed on multiple exchanges, rather than just be listed on London alone. In contrast, there was a negative relationship with an Only Provincial listing. This implies that smaller companies tended to be more often quoted on the provincial markets than on London. However, this significance breaks down toward the end of our sample period, possibly reflecting growth in the London Supplementary List, which may have allowed smaller companies to trade on London.

These results suggest that firm characteristics, namely, location, industry, and size, affected listing strategy. We now go on to analyze exchange characteristics, by considering how listing strategy was connected with governance requirements and liquidity.

Liquidity of the Exchanges

The liquidity of stock exchanges matters to investors who want to convert their shares into cash. The rise and attractiveness of the provincial markets, therefore, may have been because they offered superior liquidity to London. Concomitantly, their demise may have been partially attributable to the cessation of this liquidity advantage.

In order to test this conjecture, we need to measure the liquidity of the various stock markets. The generally preferred approach to looking at liquidity is to use bid-ask spreads. Unfortunately, the IMM did not report separate bid and ask prices. To overcome this, we use a proxy, based on the approach of Bekaert, Harvey, and Lundblad (2007) who analyze liquidity in modern emerging markets. This measure implicitly assumes that a security traded if its price changed during a particular month and that it did not trade if its price remained constant. This may understate liquidity, as some trades may have taken place and prices just remained the same, but this should not lead to bias between markets or over time. Goyenko and Ukhov (2009) find that this measure performs well against other liquidity measures in modern markets. Campbell, Turner, and Ye (2018) also find that it is a good measure historically and is significantly related to bid-ask spreads when they look at the LSE between 1825 and 1870. We begin by evaluating whether a security traded in a particular month and then aggregate this for each market by calculating the proportion of stocks that traded.

Liquidity on each market may have been affected by several factors. First, there were some regulations imposed by the exchanges that were aimed at promoting trading. London did not allow firms to purchase their own shares and required companies to allot two-thirds of their capital to the public (Melsheimer and Laurence Reference Melsheimer and Laurence1879, 1884, 1905). In 1887, a court decision made it illegal for any company to purchase their own shares, meaning this rule also applied to the provincial markets. The two-thirds requirement was also put forward by the Council of Associated Stock Exchanges for the provincial markets, but this was objected to by the Manchester exchange (Thomas Reference Thomas1973, p. 198).Footnote ¹⁴

Another difference between the exchanges, also related to liquidity, could have to do with the market microstructure. In London, a dual system was used that separated brokers from jobbers, whereas the division was not explicit on the provincial exchanges (Lavington Reference Lavington1921, p. 221). If this affected the ability to execute trades quickly, it may have led to differences in liquidity that could have been influential.

Our results, reported in Table 6, show that liquidity was as high, if not higher, on provincial markets than London. For the overall period, an average of 50 percent of securities traded on London each month, and for provincial markets, it was 53 percent. While these averages may seem quite low, they are consistent with the qualitative evidence. Lavington (Reference Lavington1921) cites the investment writer Lowenfeld as saying that about 5,000 securities were officially quoted in British stock exchanges, but less than 400 had a free market.

Table 6

AVERAGE LIQUIDITY ON EACH MARKET BY DECADE

Note: Liquidity was measured as the proportion of securities in the IMM that traded in a particular month, and this was estimated by looking at the proportion whose price changed.

Source: Authors’ calculations based on data from the IMM.

Our measure of liquidity for each market relies on estimating the proportion of securities traded on a market compared to all of the securities on that market. Our estimates of average liquidity for each exchange, therefore, could have been affected if our data source was more likely to include only the most liquid assets for some exchanges. However, our findings are consistent with some qualitative contemporary observations on bid-ask spreads. The Economist noted that compared to London, if a broker “went to a provincial stock exchange he would probably be able to buy or sell on rather better terms as regards perhaps a couple of hundred shares.”Footnote ¹⁵ In other words, liquidity for small trades was generally as good, if not better, on the provincial markets than London.

To test this empirically, we regress binary variables for listing strategies (London&Provincial and ProvincialOnly) on liquidity to test if they are correlated. We also control for other variables used in our previous regressions. The results in Table 7 suggest that for most of the period, firms that were listed jointly on London and the provincial markets, or only on provincial markets, were more liquid than those listed only on London.

Table 7

REGRESSIONS EXPLAINING LIQUIDITY

Notes: Robust standard errors are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Definitions of variables provided in Appendix Table 1. Ordinary least squares regressions, where the dependent variable is the proportion of months that the securities of a company traded over the past ten years, with the following specification:

$${Liquidity = {\beta _0} + {\beta_1}\left( {London \& Provincial} \right) + {\beta_2}\left( {ProvincialOnly} \right) + {\beta_3}\left( {HQRestUK} \right) + {\beta_4}\left( {IndTransport} \right) + {\beta_5}\left( {IndFinancial} \right) + {\beta_6}\left( {MarketCap} \right).}$$

Source: Authors’ calculations based on data from the IMM.

The problem the Economist saw with the provincial markets was not the size of the bid-ask spread, it was the depth of the market: “it is excellent to be able to deal a threepenny margin between buying and selling, but if the price extends only to 500 shares, it is obviously better to have another market where the broker can deal at a shilling margin in 2,000 shares if he wants to.”Footnote ¹⁶ London was likely to offer more depth of trading in large companies, but it did not necessarily ensure that small companies would be very actively traded.

The results in Table 7 suggest that any liquidity advantage to having a provincial listing had disappeared by the 1920s. One factor that may have affected liquidity over time was the trading links between the provinces and London. For many years, provincial brokers were able to deal directly with London jobbers. This helped to facilitate “shunting,” which meant that any arbitrage opportunities involving the prices of companies cross-listed on multiple exchanges were quickly exploited and prices were consistent in each market. Some London shunters may also have effectively acted as jobbers in provincial stocks (Michie Reference Michie1985, p. 74). However, in 1909, the LSE passed a rule that brokers could not quote prices for non-members and no jobber could deal for non-members, and another rule was passed in 1912 that established minimum commissions (Michie Reference Michie1985, p. 77; Michie Reference Michie2001, pp. 120–22).

This made it more difficult for provincial brokers to access the London market. If investors who lived in the provinces still wanted to trade in LSE securities, they would have to find a London broker, which would have been possible given improvements in communication technology. Alternatively, they could use a provincial broker and trade on the provincial markets. If they chose the former, this would increase business for the London brokers; if they chose the latter, it would reduce liquidity on London but increase it in the provinces. The results in Table 6 suggest that LSE liquidity barely changed after the rules were introduced, suggesting that the additional restrictions and minimum commissions did not reduce trading there.Footnote ¹⁷ Interestingly, provincial liquidity declined during the 1920s. There may have been other factors involved in this, but it could have been indirectly connected to the rule changes if the reduced links between London and provincial brokers also made it more difficult for London-based investors to trade in provincial stocks.

Overall, the results on liquidity suggest that the provincial markets offered superior liquidity than the London market until the 1920s. Therefore, the decline in the provincial markets, which had started around 1900, cannot be attributed to the loss of this liquidity advantage. Nevertheless, the relative decline of the provincial markets accelerated during the 1920s, possibly due to the loss of this liquidity advantage.

Stock Exchange Governance Requirements

The initial popularity of the provincial stock exchanges may simply have been due to them having less onerous listing requirements with regard to firm governance than the London market. When a company wanted to list on an exchange, it would typically issue a prospectus showing details of the offer and would invite applications for shares. They usually included a note similar to the one by Henry Pooley and Son (The Times Book of Prospectuses 1900, p. 47), which said “it is intended to apply in due course for an official quotation for the Debenture Stock and Shares on the London, Glasgow, Liverpool, Manchester and Birmingham Stock Exchanges.”Footnote ¹⁸ A consideration for companies that wanted to get listed was whether it would imply any changes to their governance structure. From 1862 to 1899, there were few legal restrictions imposed on companies in terms of shareholder protections, but the exchanges could impose their own requirements. To examine this further, we have obtained data from Acheson, Campbell, and Turner (Reference Acheson, Campbell and Turner2019), who record in detail which shareholder protection provisions were included by companies in their articles of association during the period before 1900.Footnote ¹⁹ They include 18 provisions that could potentially have protected shareholders, such as whether the company committed to having its accounts audited, proxy voting, and preemptive rights on new share issues.

From the sample provided by Acheson, Campbell, and Turner (Reference Acheson and Turner2019), we have data on 189 companies listed only on London, 131 companies only on the provincial markets, and 26 listed on both. In Table 8, we show the average number of provisions included in their articles of associations (SPITotal) by each of these types of companies. We find that firms only on the provincial markets actually had slightly higher shareholder protections overall. When looking at the 11 provisions recommended by the government in Table A of the 1862 Companies Act (Default1862), the firms only on provincial markets also had more than the London listings.

Table 8

AVERAGE NUMBER OF GOVERNANCE PROVISIONS OF COMPANIES ON LONDON VERSUS PROVINCIAL MARKETS

Notes: “SPI Total” refers to all 18 governance provisions examined by Acheson, Campbell, and Turner (Reference Acheson and Turner2019). These included whether accounts were audited, accounts were mailed before the Annual General Meeting (AGM), access to company books, the auditor was a shareholder, 10 percent or less of capital was needed for an EGM, proxy voting, five or less shareholders required to force poll, graduated voting scheme, upper limit on votes, more than one AGM, preemptive rights on new share issues, limits on directors’ borrowing powers, director cannot vote if any conflict of interest, director cannot profit directly from contracts, ban on repurchases, no approval of transfer, transfer books not closed, capital loss triggers AGM to liquidate, and “Default 1862” refers to the subset of 11 provisions that were recommended by the government in Table A of the 1862 Companies Act. “LSE 1902” refers to the subset of six governance provisions that would normally be expected of companies listing on the LSE, according to Gore-Brown (Reference Gore-Browne1902). Further details of these variables are given in Appendix Table 1.

Source: Data from Acheson, Campbell, and Turner (Reference Acheson and Turner2019).

Six of the provisions would go on to be required by the LSE in 1902 (LSE1902), according to Gore-Brown (Reference Gore-Browne1902). Of these, the provincial companies had slightly fewer, although notably, the LSE firms also had less than the maximum, confirming that these requirements were not imposed until later.

This univariate analysis suggests that by some measures, companies only on London had higher shareholder protections, but by other measures, companies listed only on the provincial markets were higher. These conflicting results make it difficult to conclude that governance requirements imposed by the exchanges had a major impact during this period. In Table 9, we go on to regress dummy variables for listing strategy against our measures of shareholder protections, controlling for other variables. However, none of the governance measures had a significant relationship with the listing decision.

Table 9

REGRESSIONS EXPLAINING GOVERNANCE PROVISIONS

Notes: Robust standard errors are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Definitions of variables provided in Appendix Table 1. Ordinary least squares regressions, where the dependent variable is the number of shareholder protection provisions that were adopted by a company, with the following specification:

$${{Shareholder Protections = }\beta _0{ + }\beta _1{ (London \& Provincial) + }\beta _2{ (ProvincialOnly) + }\beta _3{ (HQRestUK)}}\,{{ + }\beta _4{ (IndTransport) + }\beta _5{ (IndFinancial) + }\beta _6{ (MarketCap).}}$$

Sources: Authors’ calculations based on data from Acheson, Campbell, and Turner (2019) and IMM.

It, therefore, seems unlikely that burdensome regulations on governance would have been the reason for companies avoiding London before 1900, when the provincial markets were at their height. In the twentieth century, the provincial markets followed London in terms of listing requirements (Thomas Reference Thomas1973, p. 198). It was agreed that there should be general uniformity with London in most matters, but local committees could waive certain rules (Thomas Reference Thomas1973, p. 178). If the provincial markets did exercise discretion in waiving some provisions, they do not seem to have benefitted much, as London became ever more dominant.