The possibility that politicians might act to further their private interests or those of their backers, as opposed to the general public interest, has led to the creation of conflict-of-interest rules in modern democracies. For instance, the House of Commons Code of Conduct requires MPs to register financial interests and disclose other private interests related to their public duties (House of Commons 2015). In the US, the Ethics Committees of the House and the Senate maintain similar codes of conduct. Moreover, since the Lobbying Act of 1946, there is a mandatory publicly accessible registry of lobbyists at the federal level.
Ongoing research confirms the advantages of this type of regulation, as the behaviour of elected politicians and public officials clearly changes when under the public eye. Several studies have singled out transparency as one of the most effective bulwarks against political corruption and rent-seeking (Olken 2007, Reinikka and Svensson 2011). And yet, even if the exposure and outlawing of conflicts of interest are necessary to limit corruption, they are not sufficient to eliminate it.
In the 1840s, speculation in railway shares in the UK prompted the creation of hundreds of new railway companies. Each company needed to petition Parliament for a Parliamentary Act that would allow the company to build a line of railway. Faced with the necessity of picking winners amongst the hundreds of railway applicants, Parliament improvised a new regulatory system in which many small committees of five MPs would be responsible for judging the merits of competing lines and deciding on which to approve. This procedure, largely inspired by Gladstone, contained an important proviso, namely, that no committee member could have a vested interest in the railways under consideration by the committee.
Despite the careful institutional design, we show in a new paper that Parliamentarians avoided the restriction on allowing vested interests to influence the selection of companies by coordinating with their colleagues to trade votes across subcommittees (Esteves and Geisler Mesevage 2017). We uncover one of the earliest instances of ‘logrolling’, a term first used in the context of US politics in the 1820s. We confirm that logrolling was both prevalent and significant among UK MPs in the 1840s – our estimates suggest that it could have influenced the approval of at least a quarter of the railway companies that received an Act of Parliament. Moreover, these same companies underperformed in the stock market during the railway bubble and after its final crash in 1847.
The political economy of railways
The setting for this study is the British railway mania of the mid-1840s. Over 1,200 railway projects were registered in 1845 alone (see Figure 1) and this resulted in an unprecedented expansion of the British railway network, which formed the basis of the railway system that endured into the 20th century (Dyos and Aldcroft 1969, Campbell and Turner 2015, Casson 2009).
Figure 1 Number of railways companies promoted, October 1844 to December 1845
Source: Geisler Mesevage (2016)
At the height of the mania (1844-45), over 700 railway companies petitioned Parliament for an Act that would allow them to construct a railway. Rather than having the whole House of Commons vote on each project – which would have been infeasible given the number of applications – Parliament classified competing railway bills into 113 separate groups and entrusted the decision on which bills to approve in each group to committees of five MPs. To prevent conflicts of interest, Parliament required that railway companies publish shareholders’ lists and proposed routes. This publicity enabled the exclusion from each committee of all MPs who held shares in a project, or who represented one of the constituencies crossed by the lines under examination by the committee. In this way, the British government hoped to ensure that the general interest prevailed over private and local interests in the approval of railway bills.
Despite these precautions, prior research has emphasised the inefficiency of the resulting network (Casson 2009). We contend that this resulting inefficiency can, in part, be attributed to the effective collaboration of MPs, who traded votes to ensure the passage of railway companies they had invested in or that passed through their constituencies.
Logrolling in Parliament
Vote trading has been extensively modelled in political theory, the first seminal contribution being that of Buchanan and Tullock (1965), who considered logrolling to be welfare-enhancing – at least for the parties involved. Riker and Brams (1973) later argued that even if mutually advantageous, vote trades generate externalities to the parties not involved in the trades. They showed that in aggregate vote trading harms everybody, including the traders – a result they called the ‘paradox of vote trading.’
If the theoretical advantages of logrolling are dubious, testing for its presence is also notoriously difficult. As in the case of corruption, in the absence of direct evidence on logrolling, researchers have resorted to inferring the presence of vote trading from particular patterns of voting. This is difficult for two main reasons. First, there is indeterminacy in interpreting a politician’s vote, which may be motivated by personal interest, ideology, or logrolling (Kau and Rubin 1979). Second, the set of possible trades may not be well defined. Politicians can enter into a form of implicit collusion where they vote for the interests of others in expectation of compensation in the future whenever a project they are interested in comes up for vote.
The institutional structure designed to approve railway bills allows us to approach the problem of measuring logrolling with a novel perspective. We do so by drawing on methods from social network analysis. In essence, the assignment of small groups of MPs to the oversight of small groups of railway companies – taken in combination with data on the political and pecuniary interests of MPs in railway companies – enables us to map out which MPs were regulating the interests of their colleagues as a directed social network. Within this network, we identify the set of possible trades between MPs and use it to test for the extent of logrolling. The method is depicted in Figure 2 using a simple artificial example to highlight the intuition.
Figure 2 Finding the set of feasible logrolls
Notes: Panel (a) depicts the affiliation of MPs with companies via their interests. Panel (b) shows MP affiliation with companies by presence on the sub-committee that regulated the company in question. Panel (c) is the matrix product of (a) and the transpose of (b), and shows which MP is regulating the interests of which of their colleagues. The diagonal of (c) is hollow (all zeros) because no MP oversees a railway they are themselves interested in. Panel (d) is the network representation of matrix (c), with row/column names equivalent to nodes, and arrows corresponding to the non-zero entries of the matrix. Panel (e) shows that a logroll is a cycle on the directed network – Adams oversees the company Fell wants approved and Fell oversees the company Adams wants approved. Panel (f) shows that logrolling need not be limited to two participants, and illustrates with a cycle of length 3.
Source: Esteves and Geisler Mesevage (2017).
If MPs were trading votes across subcommittees to ensure the passage of their favoured projects, then their success rates – measured as the fraction of projects an MP favours that got approved – should be correlated. We estimate a peer-effects model to test for the presence of logrolling, using our network of feasible trades (Bramoullé et al. 2009). We were able to draw on a very rich dataset about MPs (education, social status, investments, political affiliation, and ideology) to control for confounding factors driving their votes in committees. Our findings reveal significant levels of vote trading. As a lower bound, we estimate that a quarter of all approved projects were likely to be the result of logrolling.
Apart from establishing the significance of logrolling, our study also quantifies the implications of this behaviour by politicians. We collected the daily stock prices of the railway companies approved by Parliament for the period 1846 to 1848 and found that the stock market systematically discounted the prices of companies that could have been approved through logrolling. The discounts on tradable companies averaged half of the standard deviation of railway share prices, a very large quantity during a period of significant market volatility. Figure 3 summarises the data and illustrates the price difference between the two types of companies.
Figure 3 Stock prices by logroll status
Source: Esteves and Geisler Mesevage (2017)
If we aggregate the losses to investors, they add up to £4-5 million, or 1% of UK GDP at the time. This figure is possibly a lower bound itself, as the social welfare losses from the selection of an inferior railway network may be large but are difficult to quantify – a problem that would compound until the famous Beeching cuts of the early 1960s (Casson 2009).
Our research highlights the ability of politicians to influence outcomes in favour of their private interests, even when they are blocked from regulating those interests directly. Our analysis illustrates how merely acknowledging conflicts of interest, and abstaining from voting when conflicted, may not resolve the problem of vested interests if politicians are able to collude. Similar to the ‘golden goose’ effect defined by Niehaus and Sukhtankar (2013), we need to be alert to the dynamic nature of corruption – particularly as politicians may be willing to vote for political reforms that limit their ability to extract rents today in expectation of using the new rules to gain larger rents in the future. Our work also confirms the disadvantages inherent in Riker and Brams’ (1973) ‘paradox of vote trading’, as the private and social losses of an inefficiently regulated railway network were large by any measure.
Bramoullé, Y, H Djebbari, and B Fortin (2009) “Identification of peer-effects through social networks”, Journal of Econometrics 150 (1), 41-55.
Buchanan, J M, and G Tullock (1965), The calculus of consent: logical foundations of constitutional democracy, Ann Arbor: University of Michigan Press.
Campbell, G, and J Turner (2015), “Managerial failure in mid-Victorian Britain?: Corporate expansion during a promotion boom”, Business History, 57(8), 1248–76.
Casson, M (2009), The World's First Railway System, Oxford: Oxford University Press.
Dyos, H J, and D H Aldcroft (1969), British transport: an economic survey from the seventeenth century to the twentieth, Leicester: Leicester University Press.
Esteves, R, and G Geisler Mesevage (2017), “The Rise of New Corruption: British MPs during the Railway Mania of 1845”, CEPR Discussion Paper 12182.
Geisler Mesevage, G (2016), "Information Bubbles: The Market for Financial Information and the Railway Mania of 1845", PhD Dissertation, Geneva: Graduate Institute of International and Development Studies.
House of Commons (2015), The Code of Conduct together with The Guide to the Rules Relating to the Conduct of Members (HC 1076), London: The Stationery Office.
Kau, J, and P Rubin (1979), “Self-Interest, Ideology, and Logrolling in Congressional Voting”, The Journal of Law & Economics, 22 (2), 365-84.
Niehaus, P, and S Sukhtankar (2013), “Corruption Dynamics: The Golden Goose Effect”, American Economic Journal: Economic Policy, 5 (4), 230-69.
Olken, B (2007), “Monitoring Corruption: Evidence from a Field Experiment in Indonesia”, Journal of Political Economy, 115(2), 200-249.
Reinikka, R, and J Svensson (2011), “The power of information in public services: Evidence from education in Uganda”, Journal of Public Economics, 95(7–8), 956-966.
Riker, W, and S Brams (1973), “The Paradox of Vote Trading," The American Political Science Review, 67 (4), 1235-47.