Introduction

The nexus of money and politics has always been controversial. Whether with union backers or wealthy businessmen, allegations have long been made that political parties owe allegiance to those that fund them. The likes of George Soros and Robert Mercer have long been used as partisan punchbags for those who oppose their causes in the United States, while in just the last few years a handful of British political donors been identified as: trying to use their weight to replace the Prime Minister (Coates and Waterfield 2018); pledging millions to start up a new political party (Savage 2018); and funding the vast majority of campaigning for both sides of the EU referendum (Transparency International 2016). Concerns over the role of money in politics has helped fuel the rise in populism on both the left and the right of the political spectrum, with the implication underlying big donations always clear and consistent: ‘he who pays the piper calls the tune’.

The upsurge of interest in the role of money in politics by the public has been mirrored by a renewed interest in the subject within academia. Of course, the academic treatment of such subjects is not new: from Michels’ ‘Iron Law of Oligarchy’ (1915) to C. Wright Mills’ ‘Power Elite’, many studies on the pernicious effect of economic inequality on political outcomes can be counted as classics. However, the quantity and quality of recent work is unprecedented, and considerable progress has been made in identifying broad relationships, while much work remains to be done in tracing the whole process of how money leads to policy outcomes. For schools interested in British politics, theory generation and testing has been largely confined to the United States, even when theory might be more easily tested elsewhere.

While political equality is at the very heart of democratic theory (Dahl 1961), recent research has concentrated on how economic inequality can render this notion more fiction than fact. Gilens (2012), for example, uses evidence of the political preferences of people from different income levels to show that government policies in the U.S. are much more responsive to those from wealthy backgrounds. Indeed, he finds that the preferences of the vast majority of Americans have effectively no impact on the policies the government pursues, despite the fact that citizens’ policy attitudes are highly responsive to economic worries (Hacker et al. 2013). This finding of non-responsiveness has been supported by a litany of similar studies (Bartels 2008; Flavin 2012; Gilens and Page 2014; Hacker and Pierson 2010; Winters and Page 2009). Furthermore, comparative studies have stressed the replicability of issues across national boundaries (Rowbottom 2010).

Political scientists have produced mixed evidence that campaign contributions can ‘buy’ votes of policymakers (Ansolabehere et al. 2003; Stratmann 2005) and there is some agreement that available evidence is insufficient to properly assess the impact of contributions on legislators’ behavior (Baumgartner et al. 2009; Fox and Rothenberg 2011; Langbein 1986). Indeed, we might well ask if politicians would be so crass to publicly show blatant favoritism to donors but help return the favor in more subtle ways that are not so easily detected (Hall and Wayman 1990). However, it is true to say that contributors themselves seem to reveal their intentions by behaving in ways consistent with the desire for influence (Fouirnaes and Hall 2014; Stratmann 1992). While there is work linking campaign contributions to legislators’ actions (Kalla 2016), there is still a lot of work that finds scant evidence that contributions systematically affect policymaking (e.g. Bronars and Lott 1997; Chin et al. 2000; McCarty and Rothenberg 1996).

Although work on economic inequality and policy outcomes show striking aggregate shifts towards the preferences of the wealthy, proving links between individual donors or donations and direct influence on laws is much harder. A paper by Kalla and Broockman deservedly garnered headlines for undertaking a randomized field experiment which showed that Congressional staffers were between three and four times more likely to meet with self-identified donors than when the correspondent did not reveal their prior pecuniary support (Kalla and Broockman 2014). However, while their work showed that donations can buy access, it does not show that that access led to bought-and-paid-for policy outcomes. In order to do so one would have to show that: donations led to increased access, and that access led to direct influence over laws.

It is here that the case of the United Kingdom proves interesting to the wider money-in-politics literature. While there has been an upsurge in single-issue groups aiming to limit the influence of big money in US politics, there has been no real equivalent political upsurge in the United Kingdom, despite a similar surge in economic inequality (Piketty 2014). The main vehicles for campaign finance reform in the country, such as Unlock Democracy and openDemocracy, are conspicuous by their lack of political profile outside specialist circles. Is this because the United Kingdom is intrinsically different from a political system in the United States that seems compromised by moneyed interests? Do theories of money in politics largely developed and tested in the United States travel?

The case of the United Kingdom provides an opportunity to test the theory developed in the differing United States context, that suggests donors are rewarded with political favors (Pickard 2011; Lyons 2012; Grice 2012; Monbiot 2017 etc.); in particular allegations that they might be granted a place in the House of Lords (Hope and Swinford 2013). Perhaps the most comparable parallel between the British and American systems is with the long-acknowledged sale of ambassadorships to donors by US administrations of both parties (Torres-Spelliscy 2012; Hollibaugh 2015; Fedderke and Jett 2016)—something even acknowledged on tape and in print by President Nixon (Lardner 1997; Nixon 1990, pp. 35–66). It is this paper’s contention that decades-old journalistic insinuations to this end, or public suspicions that ‘big donors only give money for personal favours’ (Goodrich 2015), cannot be contradicted by the data when it comes to being ennobled into the House of Lords. Furthermore, it provides a unique case of something that has so far eluded the reach of studies in the United States context: money buying access, and that access providing not only influence over laws, but the ability to write and vote on them.

The contrasting ambiguity of the British experience perhaps owes something to the peculiarly organic state of British campaign finance laws. While American political finance legislation is dominated by the lessons learned in the 1970s (Torres-Spelliscy 2012), British political finance laws are still largely framed by the overhauls of the 1870s and 1880s, intended to combat endemic bribery in constituencies, and so are preoccupied with regulating elections to the Commons rather than appointments to the Lords (Pinto-Duschinsky 1981). Even the last major overhaul of political finance legislation, from 2000, inherited this focus on spending limits in Commons elections—though its introduction of mandatory public reporting of donations provided a major new public domain dataset. However, the separate 2000 legislation creating a House of Lords Appointments Commission (HoLAC) has provided a regulator combining many of the powers of the old Political Honours Scrutiny Committee with a statutory underpinning and benefitting from a non-partisan component of commissioners supplementing the traditional nominees of each of the political parties. Additionally, since the 2006 ‘loans for peerages’ scandal, the Commission has built up an increasingly aggressive series of precedents around blocking nominations from most major political parties (HoLAC 2018).

Suggested motivations for responding to donors, for example, in the United States, like having to rely on big donors to fund expensive TV advertising buys (Green and Gerber 2008; Gerber et al. 2011) might not hold in countries where such buys are prohibited. And the potential returns on donations would tend to be less valuable in the U.K. than the U.S. A place in the House of Lords is arguably not as powerful as buying influence in the House of Representatives or the Senate, as the British Upper House has only powers to revise and delay legislation—although it carries more permanence than an ambassadorship. But all of these factors make Britain a ‘least likely’ case where, if such a link between both steps from legislations to laws can be proven, it dramatically increases the likelihood that such a link might also be assumed to exist in the United States and elsewhere.

That being said, it also points to particular avenues for research in other political systems. While (most) representatives might not be so crass so as to directly exchange donations for votes, we might look for more subtle relationships between donations and appointments to important positions where the public are less likely to be informed: everything from diplomatic posts (Keith 2014) to donors and Congressmen exchanging jobs for their families (Vazquez 2018).

The case of the United Kingdom and ‘cash for peerages’

For centuries, the sale of peerages—and an accompanying seat in Britain’s House of Lords—was commonplace, firstly as a revenue-raising device for Kings in past centuries, and more recently, with the bulk of appointments being made by the Prime Minister (and more recently still, the major party leaders), it has been widely alleged as a revenue-raising device for British political parties. A number of journalistic studies have looked at the allegations of ‘cash for peerages’ in twentieth century Britain (Walker 1986; Hollingsworth 1991; Cook 1995; Leigh and Vulliamy 1997; Baston 2000; Cook 2008; Friedman 2013), although scholarly studies have been much rarer and more general in approach (Ridley and Doig 1995; Beetham and Oligarchy 2011). Whilst there have been previous attempts to statistically analyze the phenomenon of ‘cash for peerages’, these have been somewhat partisan in nature—a 1986 trade union-funded study concentrated on the Conservatives (Walker 1986), while a more recent short, two-page analysis by the Conservative Bow Group made the somewhat improbable claim that the Labour Party was the only party embroiled in selling peerages (Philp 2006). In 2006, a large-scale ‘cash for peerages’ scandal erupted, fuelled by such observations as the way that 100% of all Labour party donors of over £1 million since 1997 had been offered either a knighthood or a peerage. Eventually, the police decided that there was insufficient evidence to prosecute, but a whiff of scandal was left in its wake, and HoLAC has been far more aggressive in its peerage vetting since then. Rumors of ‘cash for peerages’ have not abated, with veteran parliamentarian Lord Cormack telling a packed meeting in 2015: ‘The biggest abuse is putting party donors into the House of Lords. This has happened from all three parties, even in the last 5 years’.Footnote 1

Party leaders can dispense peerages for a number of reasons. These can include a genuine desire to strengthen the quality of debate in the Lords, with expert opinion. Since the creation of the HoLAC in 2000, this has removed the pressure on party leaders to make such appointments, as the Commission effectively performs this function for them—though the numbers allocated for such crossbench experts is small, averaging two a year, giving party leaders even more of a roving brief to use party political patronage for the remaining peerages allocated to them.

Nor does this rule out improper ways in which peerages can be ‘bought’ by non-monetary means; for instance, former MP Dr. Tony Wright highlighted how ‘A party leader says to an MP—“You give up your seat to us and we will put you in the Lords”—I am told this happens, and has happened regularly for years … So people can sell their seat [in the Commons], but they cannot donate money without being scrutinised’.Footnote 2

Patronage has long been established as a major factor in appointments to the Lords. The most well represented group in the Lords is former MPs, and yet relatively few former MPs actually proceed to the Lords. The reasons for awarding peerages can thus be diverse. It is our intention here to look at just one possible reason for nominating peers (donations to political parties), particularly in relation to those peers for whom there are no immediately obvious reasons for their nomination.

Data collection

The first task in trying to identify a possible link between donations and elevation to the Lords was to assemble an original dataset of the 303 individuals nominated for peerages between 2005 and 2014. They were nominated under three different governments: the Labour government of Tony Blair, his successor Gordon Brown’s Labour government, and David Cameron’s Conservative/Lib Dem coalition. In order to ascertain whether the peer was nominated through patronage or some other reason, we coded each peer as to whether they fell into the categories most associated with patronage appointments. Looking at the standard reasons why people might be appointed to the Lords, we collectively termed such people the ‘usual suspects’. These ‘usual suspects’ are comprised of the following broad categories:

  • Ex-Parliamentarian Former Members of the British Parliament, European Parliament, Scottish Parliament, Welsh Assembly, Northern Irish Assembly, and Greater London Assembly. This includes heavyweight statesmen of great experience, as well as so-called ‘bed-blockers’ who needed an incentive to retire from a safe seat in the Commons.

  • Former Senior Party Staff This encompasses a number of roles, the key feature having been paid employment in a partisan capacity, so it does not involve full-time professional civil servants. It includes ministerial Special Advisers (even though they are technically on the civil service payroll), as they report directly to a minister, and are traditionally drawn from a party rather than civil service background (Yong and Hazell 2014).

  • Council Leader or ex-Council Leader All three major parties tend to appoint a handful of former council leaders with local government experience.

  • ‘Government of All Talents’ This category was peculiar to the Labour Party. In 2007–2008, Gordon Brown sought to broaden the appeal of his premiership by appointing a number of respected experts to the Lords, from non-partisan backgrounds, but to sit as Labour peers so that they could take up office as Labour government ministers.

  • Joint Interim Peers Panel This category was peculiar to the Liberal Democrats. In 1999, in response to a perception that the party had flagrantly sold peerages in the 1990s, grassroots activist Donnachadh McCarthy instigated a system whereby the party’s activists elected a slate of nominees to be put forward (McCarthy 2014). The procedure has noticeably lacked support from any Lib Dem leader, and only a handful of people from the panel have ever been nominated; a loophole exists whereby the party leader can make his own nominations, and so this has been successively exploited by various leaders to nominate their own candidates over the party’s, with Nick Clegg effectively ditching the system after 2010 (Thévoz 2015).

  • Reserved public-sector post A handful of jobs almost automatically carry a peerage either on retirement, or shortly after retirement.Footnote 3 They do not make up a sizeable proportion of peerages.

  • House of Lords Appointments Commission nominee In addition to their scrutiny of political appointments, HoLAC is also responsible for making independent, non-partisan appointments to the Lords (although one independent nominee subsequently took his seat as a Labour peer). Dubbed ‘The people’s peers’ by the press, these nominees have tended to be the least likely to be engaged in party politics, the least likely to be party donors, and the least likely to speak or vote in debates.Footnote 4 They are a departure from the pre-2000 practice of peerage nominations, in that they do not come from the Monarch, the Prime Minister, or the party leaders.

We then cross-referenced donation data from the Electoral Commission database and checked whether each peer, their immediate family, or a business in his or her control, had made any party political donations, cross-checking this with the peer’s current and former entries in the Register of Lords’ Interests, and their Who’s Who entryFootnote 5. Detecting such information was no easy feat, involving as it did a series of shell companies (this alone involved seeking guidance from journalistic as well as academic colleagues on evolving research in this area), holding companies, wholly-owned subsidiaries, investigating the shareholder composition and boardroom composition of each company, not to mention becoming wise to the minutiae of the Electoral Commission’s data: Different variants on the same person’s middle name occasionally show up, or (in one case) a nominee used a pseudonym from their business career for their donation. Nonetheless, it was vital to do this in detail for all 303 nominees, to fully record all donations made by them.

Hypotheses

Having collected our data, we considered what hypotheses could be tested which, if true, would falsify the notion that money can buy influence through a seat in the House of Lords. Our argument is, in essence, that British party managers are able to use seats in the House of Lords as a tool both of party management and of raising funds. If true, this would mean that a person’s probability of being nominated to the Lords would be a function of their political services to a party and their financial donations to that party, and other variables which might be relevant. So it might look something like the equation,

$$\Pr \left( {\text{Nominated}} \right) = G\left( {\alpha P + \beta F + \gamma X} \right)$$

where \(P\) represents the political services, \(F\) represents the financial services to a party and \(X\) represents other relevant variables, and \(G\left( \cdot \right)\) is an increasing function.

The ideal way to test our hypothesis would be to gather a large random sample of people in the UK, noting their political services and financial donations to political parties, and whether they have been nominated for a seat in the House of Lords. We would ideally also control for various other potentially related variables such as wealth and income. It would then be possible to estimate \(\alpha\) and \(\beta\) directly. Such a method is clearly impractical. In order for a random sample of the UK population to be expected to contain about 100 members of the House of Lords, it would have to contain more than 5 million people.Footnote 6

Our approach has therefore been to consider what relationships we would expect to see in our data concerning those who were nominated to the Lords over 2005–2014 and donations made to British political parties over 2001–2014. The first observation is that comparative advantage among those hoping for a seat in the Lords would suggest that people should specialise either in providing parties with financing or with political services.

People who have had successful careers in business will be of more help to a political party if they donate substantial sums of money rather than switch career and become a political advisor or MP. Conversely, it will be easier for most people who have started providing political services through advising ministers or shadow ministers to continue providing those services rather than to go into business and make large sums of money to donate to the party. The inverse of this observation is null Hypothesis 1, which we aim to falsify.

Hypothesis 1

There is no statistically significant difference between the donation behaviour of nominees from the “usual suspects” and nominees from outside the ‘usual suspects’.

Secondly, if the widespread suspicion that seats in the House of Lords can be purchased through large enough donations to the UK’s political parties is correct, then we should expect to see an unusually large concentration of Lords nominees among the larger donors to political parties. So the second hypothesis we aim to falsify with our data is as follows:

Hypothesis 2

Big donors are just as likely to be nominated to the Lords as anyone else.

Accordingly, we compiled a second database of all ‘big donors’ to British political parties since detailed records began in 2001. In the data gathered for hypothesis 1, we found a noticeable gap between ‘big donor nominees’, who accounted for less than one in ten nominees and typically gave six- or seven-figure sums, and the remaining nine-tenths of nominees who gave under £25,000. ‘Big donors’ seldom made one solitary donation, but typically divided their largesse between smaller sums. Accordingly, to detect the ‘big donors’ who often gave money in large tranches, we defined a ‘big donor’ as any individual or organization which made a single donation of at least £30,000 (Table 1).

Table 1 Nominations by Party (Database of Donations and Loans to UK Political Parties, Electoral Commission website)
Full size table

Finally, we wanted to check that the presence of donors is a substantial difference between the appointed House of Lords and the elected House of Commons. Elections add an element of uncertainty to the appointment of a legislator, for while there is a large body of literature to suggest that money may influence an election, there is no guarantee of ‘buying’ electors’ votes, whereas an appointed legislature inherently lacks the safeguard of an electorate. Nonetheless, it is claimed that HoLAC presides over a ‘clean’ system of appointments, and if this is the case, one would not expect to find any more ‘big donors’ in the appointed Lords than the elected Commons. This leads to the third hypothesis:

Hypothesis 3

Members of the House of Commons are as likely to be big donors as members of the House of Lords.

Data analysis

The party affiliations of the 303 peerage nominations over the period were as follows:

All of the nominees from the smaller parties (with fewer than 10 nominees) were ‘usual suspects’. With no variation in this key variable, they were excluded from the analysis. This left 296 nominees from the three main parties and the Crossbenchers. Table 2 looks at the breakdown of these remaining nominees by party affiliation and membership of the ‘usual suspects’ group. Interestingly, Conservative nominees over the period were the least likely to be drawn from the pool of ‘usual suspects’ (49%), while the Crossbenchers were the most likely to be drawn from them (86%). Labour and the Liberal Democrats both had similar proportions of their nominees drawn from the pool of ‘usual suspects’ (74 and 75% respectively).

Table 2 Nominees by party and ‘usual suspect’ membership
Full size table

Loans

For each nominee to the House of Lords, the total amount of donations in ‘cash and kind’ over £5000 made to national parties and over £1000 to local parties and affiliated party bodies are known from 2001 onwards. However, donations are not the only way in which wealthy individuals are able to financially assist a political party of their choice. One of the revelations of the 2006 ‘Cash for peerages’ scandal was the way in which all major parties were able to avoid divulging some major funding through receiving it in the form of a substantial loan, including several individuals who had been nominated for peerages and had no need to divulge this to HoLAC. Consequently, the party leaders voluntarily divulged their outstanding loans since 2005, and in 2007 this loophole was closed, with loans now having to be declared as well (Friedman 2013; Levy 2008).

In the data, there are six recorded loans from House of Lords nominees to political parties. Loans are sometimes not repaid, but later converted into donations. So our strategy for dealing with loans was to include them as though they were donations if they remained outstanding, but to exclude them if they had already been repaid. Most of the excluded repaid loans were for small amounts (less than £5000) and had been repaid within 6 months. The smallest loan that remained outstanding (and so was included) was for £1 m and had been initiated in 2005.

Inflation

The data spans a decade, so it is prudent to allow for inflation. All reported numbers below have been converted into 2014 lb sterling using the annual ONS consumer price index.

Hypothesis 1

The average donation to UK political parties by nominees to the House of Lords split out by party affiliation and membership of the ‘usual suspects’ is shown in Table 3. The table shows that, on average, nominees from ‘The Others’ outside the ‘usual suspects’ donate substantially more than nominees from among the ‘usual suspects’, and that this is true overall and within all party affiliations, including the Crossbenchers.Footnote 7 Exactly how much more the average nominee from ‘The Others’ donates than one from within the ‘usual suspects’ varies across the political parties. While these differences are large, it is still necessary to check whether they are statistically significant before rejecting hypothesis 1.

Table 3 Average Donation per Nominee
Full size table

As an aside, it is also worth commenting on the skewed distribution of ‘big donors’ within the totals in Table 3. Whilst ‘The Others’ account for some 97.9% of party funds donated by nominees to the Lords (compared to just 2.1% from the ‘usual suspects’), it would be misleading to attribute this to all 92 of ‘The Others’—just 27 of the 92 ‘Others’ have donated 95% of all the money donated by the ‘Others’. None of the 27 were Crossbenchers. These 27 data points noticeably contrast with the bulk of ‘The Others’, in having donated at least £200,000, and often sizably more. Among the ‘usual suspects’, only 1 nominee donated over £200,000, with all other nominees donating well under £100,000, and most donating under £25,000. Thus 28 individuals account for over 95% of all party political donations to have emerged from all Lords nominees since 2005.

Difference in means

To test the statistical significance of these differences, a series of standard difference in means t-tests were conducted under the null hypothesis that the means were in fact the same. The tests conducted did not assume the same variance in donations from ‘usual suspect’ nominees, and nominees from ‘The Others’ outside the set of ‘usual suspects’.

As has been noted already, the donation data is skewed by the presence of a small number of individuals who have donated large sums of money. Overall, we believe it is legitimate to keep these individuals in the data, but we allow for the possibility of excluding them as outliers to check that our conclusions are robust. We allow for two possible definitions of outliers. The definition with the smallest impact on the data treats anyone donating more than £2 m in 2014 money as an outlier. An alternative definition is anyone donating more than £1 m in 2014 money.Footnote 8

Table 4 shows the one-sided P-Values of these tests and confirms that these differences are statistically significant overall and for nominees from all three of the major political parties. The only case where the differences are not statistically significant is the case of the Crossbenchers. Since the Crossbenchers are not nominated by a party leader, this is exactly what we would expect to be the case if political parties are using peerage nominations as a means of raising revenue.Footnote 9

Table 4 P-Values for one-sided t tests of statistical significance of the difference in mean donations between nominees from the “usual suspects” and nominees from outside that group
Full size table

While the test results reported in Table 4 appear conclusive, there may be concerns about the nature of these tests and the sample size. The donation data is quite clearly not normally distributed and these tests are designed to test the difference in the means of two normal distributions. The Central Limit Theorem shows that the sample mean drawn from any distribution is normally distributed with a large enough sample, but in some of these tests, it is questionable whether the sample is large enough to apply the theorem. The small number of Labour, Liberal Democrat and Crossbenchers nominees from outside the ‘usual suspects’ is a particular problem reducing the number of degrees of freedom in those tests.

To overcome this problem and check that our conclusions remain robust, we generated a series of binary indicator variables as to whether a nominee had donated more than a fixed sum of money.Footnote 10 Binary indicator variables have a binomial distribution, so that the variance is determined by the mean. So when testing the null hypothesis of the same mean across two different samples, we can assume the same variance and use a pooled sample variance estimate. This allows for more degrees of freedom when testing for a difference in the means. The one-tailed p-values on these tests are reported in Table 5.

Table 5 P-values of t-tests for a difference in means between nominees from the “usual suspects” and nominees from outside that group of binary indicators for whether nominees donated more than the given threshold amount
Full size table

Among Labour and Conservative nominees, all nominees and all nominees excluding crossbenchers, the difference is always significant at the 1% level. Among Liberal Democrat nominees, it is significant at the 1% level at most thresholds, but only significant at the 5% level at a threshold of £5k and (just) not significant at the 5% level at a threshold of £10k. However, caution should be attached to the Liberal Democrat tests at thresholds of £30k and above, as there are only 8 Liberal Democrat nominees who fall into this category, which is rather low. However the significance at the £5k threshold and the near significance at the £10k threshold should be sufficient to conclude that Liberal Democrat nominees donate more if they have been nominated from outside the ‘usual suspects’.

Similarly, although the statistical tests look highly significant for Crossbenchers donating more than £10k, there is actually only one Crossbencher nominee who has done this. At this point distributional assumptions behind the t-tests have broken down and we remain unable to conclusively say that Crossbencher nominees from outside the ‘usual suspects’ donate more than those from within the ‘usual suspects’.

Allowing for other factors

While we can conclusively say that nominees from ‘The Others’ outside the ‘usual suspects’ donate more than nominees from within this group, it does not necessarily follow that nominations to the Lords are being purchased. It could be that donations and membership of the ‘usual suspects’ are negatively correlated among nominees to the House of Lords because of their relationship to some third, unknown variable.

To test this possibility, we conducted a series of logit and probit regressions with being ‘The Others’ as the dependent variable and a series of independent variables, including donation data. The goal was to see if the relationship between higher donations and not being a ‘usual suspect’ remained allowing for the potential influence of other variables. As the results of the logit regressions in Table 6 show, donation behavior is positively related with being outside the group of ‘usual suspects’ among nominees, even after allowing for the influence of other factors that we know about. Similar results are obtained via the probit model, and when potential outliers are excluded from the regressions. However, we must admit that leaves open the possibility of some influence from conflating factors we don’t know about.

Table 6 Logit Regression models
Full size table

Model 1 records that some elements of donation behaviour have unexpected signs and are insignificant. This is most likely a feature of the colinearity of the various measures of donation behaviour and our sample is not large enough to disentangle the different effects of different elements of donation behaviour accurately. Of the different facets of donation behaviour, the best predictor of a nominee coming from outside the group of ‘usual suspects’ based on model fit seems to be the total amount of real donations.

Among the alternative characteristics, being nominated by the Conservative Party has a large and significant positive impact on the probability that a nominee is not from the ‘usual suspects’, being male and born in the UK has a large and significant negative impact on the probability of a nominee coming from outside the group of usual suspects. The former observation reflects what was already known from Table 2 above, that the Conservative party is nominating relatively large numbers of people from outside the group of ‘usual suspects’. The latter observations may simply be reflective of the way in which the UK’s political class, from which the ‘usual suspects’ are drawn is dominated by men born in the UK.

Finally, we are able to confirm via a series of tobit regressions that the relationship is maintained in the other direction, meaning that being a nominee from outside the ‘usual suspects’ has a significant and positive impact on donation behaviour. By creating a series of interaction dummies for each party, we are also able to see how this impact varies across the parties. These interaction dummies take the value 1 if the nominee is both a nominee from the party in question, and a nominee from outside the group of ‘usual suspects’. So for example, the independent variable in Table 7 called ‘Con_NU’ takes the value 1 when a nominee has been nominated by the conservative party and is not from the group of usual suspects, and is zero in all other cases. Table 7 records the results of these tobit regressions.

Table 7 Tobit regression results
Full size table

A key feature of these regressions is the large and significant impact of excluding outliers. This is because the outliers have given so much more to political parties than others that when the amount donated is the dependent variable, their inclusion skews the results. The average total donations of these outliers is £3.7m, and their exclusion reduces the average amount donated by all nominees from £116k to £55k.

The important results to take away from the results in Table 7 are that all of the interaction effects are positive and significant. So nominees from ‘The Others’ are donating significantly more money to political parties than nominees within the group of ‘usual suspects’, with the notable exception of the Crossbenchers, the only group for whom the interaction effect is not significant. If these results are indeed caused by the sale of peerages, then these coefficients would represent the average price being charged by the three main parties. The emphasis is on the word ‘average’ here, because we should bear in mind that this would be a market in which the sellers have ample opportunity to gather data about potential nominees that could help them discern their willingness to pay. So the political parties, if they do sell peerages, have ample scope for engaging in price discrimination. Furthermore, it should be pointed out that a number of further variables could affect the scope for price discrimination, such as the number of people engaged in any peerages sales at any one time, and their tenure in office.

The second point of interest is the significant discount for nominees who were born in the UK, or, looked at from the other side, the significant premium in donations from people born outside the UK, with those born outside the UK being significantly more generous donors. Again, if the data represents the sale of peerages, there could be several reasons for this. It could be that political parties perceive that, with the scrutiny of their peerage nominations, there are higher political costs to nominating large donors born outside the UK. It could also be the people born outside the UK place a higher value on being a member of the House of Lords in terms of a status symbol and so are willing to pay more for the privilege. Additionally, some of these nominees could come from cultures where the offering of donations is considered a standard business practice. However, it is important to remember that there could also be more innocent explanations for these features of the data.

Hypothesis 2

We noted that our dataset of big donors contains 779 big donors to the three main political parties, 26 of whom were nominated for a peerage over the period 2004–2015.Footnote 11 To test hypothesis 2, we estimated the probability of taking a random sample of 779 people without replacement from the pool of those eligible to be nominated to the Lords, and finding 26 or more nominees from 2005 to 2014 in that sample.

The calculation is then a relatively simple use of the hypergeometric distribution. We know the sample size (779) and the number of ‘successes’ (2005–2014 Lords nominees) that should be in that sample (26 or more). We know the number of ‘successes’ available to be drawn (the people who’ve been nominated to sit in the Lords), the only element of the calculation which remains to be estimated is the population from which this sample is to be drawn. The larger we estimate this population to be, the lower will be the probability of drawing so many nominees for a peerage in the sample.

One potential estimate for the size of this population is to take the 44,655,226 individuals who were on the UK electoral roll at the beginning of the period in 2004. This would probably be an underestimate of the true number of people eligible to sit in the House of Lords as any Irish or Commonwealth citizen may sit in the Lords provided they are domiciled in the UK for tax purposes. However, this number seems improbably large for our purposes as someone would presumably only actually be nominated if they had some demonstrable interest in politics or policy. For that reason, we took as our pool of available nominees the reported membership of the UK’s three main political parties in this period, which stood at 383,800.Footnote 12 Since we are only looking at the population who are members of the three main political parties, we restrict the number of available successes to the 233 peers nominated by those three parties.

The answer to the question we have posed is then astonishingly small. The probability of seeing at least 26 people nominated for a peerage in 2005-14 in a random sample of 779 people from the 383,800 who are members of the three main UK political parties is of 9.05 × 10−37. This is approximately equivalent to entering the National Lottery and winning the jackpot 5 times in a row.Footnote 13

We are therefore confident in rejecting hypothesis 2 for the alternative hypothesis that large donors differ systematically from the rest of the UK population in such a way that they have a much higher probability of being nominated to the House of Lords. While this is conclusive evidence that big donors differ systematically from the rest of the population in the characteristics that affect one’s probability of being nominated to the House of Lords, it is not necessarily conclusive evidence that peerages are being purchased. See below.

Hypothesis 3

The third hypothesis suggests that one is no more likely to find big donors in the House of Lords than in the House of Commons. In order to test this hypothesis, we cross-referenced the 928 individuals elected to the House of Commons over the period 2005–2014 with our information about big donors. Only four of these individuals were big donors at any one time. Of these big donors, only two actually donated to their local constituency party. The other two were former Prime Ministers who could be excluded from consideration due to a number of factors.Footnote 14

So the probability of somebody being elected to the House of Commons over this period also being a big donor was 0.2%, while the probability of a nominee to the House of Lords being a big donor was 8.6%. To test Hypothesis 3, we tested the hypothesis that nominees to the House of Lords and those elected to the House of Commons had the same probability of being big donors, but that the differences observed in the data were simply a result of random variation. We were able to reject such a null hypothesis with a P Value of 1.4 × 10−17. This is another astronomically small probability roughly equivalent to the probability of winning the National Lottery twice in a row.

One shortcoming of this approach might be that by considering only those entering each chamber, we artificially reduced the probability of big donors in the commons relative to the Lords. The reason being that everyone who sits in the Commons over this time period must have been elected to the Commons at some point over this period. So everyone who sits in the Commons over the period is included in the denominator when calculating the probability of big donors in the Commons, while only a small number of Lords are included in the equivalent calculation.

To check that our estimate is robust to this criticism, we re-ran the calculations comparing the probability that someone who sat in the Lords over this period is a big donor to the probability that someone who sat in the commons over this period is a big donor. On this basis, the probability of a member of the Lords being a big donor falls to about 5.7%, but the difference is still significant at a P Value of 2.3 × 10−12.

What is unfortunately missing from our data, and may provide an avenue for future research is information on the donations of individuals who put themselves forward for election but were defeated. However the highly statistically significant differences on the basis of the data available lead us to confidently reject the third hypothesis that big donors are as likely to be found in the elected House of Commons as in the appointed House of Lords.

Omitted variable bias

While we are able to convincingly reject all three hypotheses, it must be admitted that this does not amount to a proof that peerages are being systematically sold by political parties. However, what has been shown is that nominees to the Lords can be divided into those who have been drawn from the ‘usual suspects’ or what might otherwise be called a ‘political class’ and ‘The Others’ who have been drawn from outside this group. The latter have donated, on average, substantially more to party funds than those drawn from the political class.

We have also shown that big donors are substantially more likely to be nominated to the House of Lords than the rest of the population, and that it is far more common to see big donors in the appointed House of Lords than in the elected House of Commons.

However these results may not be a consequence of the corrupt sale of peerages, but may be a consequence of a correlation between donations being correlated with some unobserved variable and that unobserved variable in turn being correlated with an individual’s probability of being nominated for a peerage. For example, a frequent justification for the system of life appointments to the House of Lords is that it is a revising chamber which draws on the expertise of its members in scrutinising legislation. Not all experts on topics of relevance to such detailed scrutiny of legislation developed that expertise within the political class. Those from outside the political class may well have excelled in some area and become wealthy in the process. It is no great stretch of the imagination to suggest if such people are willing to serve as legislators, they must have some interest in politics. If they do, there will be a political party which advocates policies they believe to be the correct ones more frequently than do other political parties. Such people may well wish to support that political party’s activities financially. Finding data to empirically test this explanation remains an area of ongoing research.

Conclusions

This has been the first full-scale analysis of the relationship between cash and peerages for all parties in Britain across a sustained period of time. While rumours of ‘cash for peerages’ have long dogged the reputation of the House of Lords, the lack of a strong evidence base has hindered efforts to test the validity of such assertions. This article seeks to allow some greater precision in the continuation of debates around this claim.

Crucially, all three hypotheses have been disproved, and the relationship between donations and nominations has been found to be significant. This is thus wholly in keeping with the argument that lifetime appointments to Britain’s Upper House are being sold to wealthy donors.

The ‘usual suspects’ of nominees one might logically expect to find in the House of Lords for the most plausible reasons—former parliamentarians with legislative experience, major public officeholders, people nominated by an internal party election—make up a tiny proportion of all party political donations generated by Lords nominees—just 2.1% (£735,000) between the 204 of them, despite making up 68.9% of nominees. By contrast, ‘The Others’, who make up just 31.1% of Lords nominees, account for 97.9%, or £33.8 million. Furthermore, just 28 individuals out of the 303 nominees—27 of them not being ‘usual suspects’—donated over 95% of the money generated from Lords nominees. Clearly, those peers nominated outside the ‘usual suspects’ are far more likely to be big donors, with major variations between the parties (although Crossbenchers donate negligible amounts, either as ‘usual suspects’ or as ‘Others’). Interestingly, those nominees born outside the UK are disproportionately likely to be big donors.

As is frequently claimed by all parties accused of selling peerages, it is of course perfectly possible that it is pure coincidence that ‘big donors’ are disproportionately likely to be nominated for peerages. However, the odds of it being pure coincidence are roughly the same as those of entering Britain’s National Lottery five consecutive times and winning the jackpot on each occasion. Whilst coincidence is theoretically possible, this explanation does stretch the limits of credulity.

Finally, we looked at whether an appointed chamber such as Britain’s House of Lords was more prone to ‘big donors’ gaining seats than the wholly-elected House of Commons. With 28 out of 303 Lords nominees being ‘big donors’, one might logically expect a similar ratio to yield some 86 ‘big donor’ MPs elected to the House of Commons. In fact, there were just 2. This suggests that an elected chamber is much less likely to see ‘big donors’ gaining seats.

The implications of these findings are considerable. Whilst much work, particularly focused on US legislative studies, has looked at how campaign contributions can skew an election, or can skew a representative’s vote, Britain’s relatively unusual system of selection to the Upper House, on a large scale, nominated by leaders of all three parties, has long left it vulnerable to the accusation that such a system is prey to the sale of lifelong seats for money. The evidence presented here is entirely consistent with that. It also suggests possible avenues of investigation for other political systems where patronage takes place largely outside of public scrutiny and might be particularly vulnerable to donor pressure.

We acknowledge that innocent explanations such as the one offered in the ‘Omitted Variable Bias’ section are logical possibilities. In other words, the relationships we have uncovered, although they are consistent with peerages serving a dual purpose of party management and fund raising, do not prove in themselves that this is what is happening beyond a reasonable doubt. It remains the case that large donations to political parties is one way in which nominees from outside the ‘usual suspects’ differ from nominees within the ‘usual suspects’. Unless alternative explanations can be supported by the data, a cloud of suspicion will continue to hang over the major political parties and their nominees with no record of public service.