<<

Competition between Socialist and Nationalist Parties in Established Democracies: The Cases of Britain and

[First Draft: Comments Welcome]

Sonia Alonso Wissenschaftszentrum Berlin [email protected]

Andrew Richards Juan March Institute, Madrid [email protected]

Prepared for delivery at the XVI ISA World Congress of Sociology, Durban, South Africa, 23-29 July 2006. Abstract

The paper examines the dynamics of electoral competition between socialist and nationalist parties in ethnically heterogeneous political contexts. We contend that in such contexts, the historic dilemma of socialist parties, whereby they have sought to retain the loyalty of their core working class constituencies at the same time as seeking votes beyond the working class in order to gain electoral majorities, is exacerbated by competition with nationalist parties. The latter may advance electorally at the expense of socialist parties by mobilising workers on the basis of their ethnic-, as well as their class- based, identities. Through the combined use of aggregate electoral data and individual survey data, we examine such electoral competition in two contrasting cases of ethnically heterogeneous political contexts, those of Britain ( and Wales), where the Labour Party has traditionally dominated regional politics, and Spain (Catalonia and the Basque Country), where electoral competition between socialist and nationalist parties has been much more contested. We find that in all four regions, the lion’s share of vote movements between elections takes place between the socialist and nationalist parties. There are, however, important contrasts between countries and regions. The British Left has a much stronger grip on the working class vote in Scotland and Wales than the Spanish Left in the Basque Country and Catalonia. However, the Scottish and Welsh nationalist parties’ ideological shifts to the left have enabled them to undermine the Labour Party’s longstanding ability to capture both class and nationalist allegiances among working class voters. In Spain, the Left has lost the battle for the working class vote in the Basque Country, while retaining majority working class support in Catalonia only in elections to the national, but not regional, parliaments. We also find that workers with an exclusively Spanish or British identity vote massively for non-nationalist parties, while those with an exclusively regional ethnic identity display more heterogeneous voting behaviour. Further research is required for understanding the role of parties’ strategies in effecting the type of vote movements between nationalist and non-nationalist electoral blocs analysed in the paper.

Key words: Nationalism; Ethnicity; ; Political Identities; Working Class; Voting Behaviour; Political Parties.

2

1. Introduction1

Left-wing political parties have always faced dilemmas in their attempts to gain electoral majorities. Historically, the working class, as their natural constituency, never amounted to an absolute electoral majority in and of itself, thereby defying Marx’s prediction that revolution would be made at the ballot box; besides, a significant proportion of workers voted for other parties. To win electoral majorities, Left-wing parties have always had to construct coalitions of potentially conflicting interests: “whether parties deliberately restrict their appeal to specific groups or attempt to conquer the entire electorate, their opportunities are limited by the heterogeneity of developed capitalist societies. In a heterogeneous society, no party can win the support of everyone without losing the support of someone, because some other party will put in the wedge” (Przeworski and Sprague 1986: 183). The dilemmas of electoral Socialism have largely been conceived in terms of class – that is, the need for Left-wing parties to gain middle-class votes in order to win elections involves, at the same time, a dilution of socialist commitments and the possible alienation of their traditional working-class constituency.2 However, in ethnically heterogeneous contexts, the focus of this paper, these historic dilemmas are even more acute. Where an ethnic cleavage cuts across the class cleavage, Left-wing and Nationalist parties may well compete for the loyalty of the same voters: in a context where ethnic and class allegiances are compatible, what requires explanation is how this is manifested in terms of voting choice. We seek to develop such an explanation based on an analysis of working class voting behaviour in two contrasting ethnically heterogeneous political contexts, those of Britain (Scotland and Wales) and Spain (Catalonia and the Basque Country). The initial historical configuration, and subsequent trajectory, of electoral competition between Nationalism and the Left differed considerably between Spain and Britain. In Catalonia and the Basque Country, Socialism and Nationalism emerged simultaneously as political forces, and none has been electorally predominant until well into the contemporary period. In Wales and Scotland, however, the Labour Party

1 We are extremely grateful to John D. Boy for his invaluable research assistance, especially with respect to preparation of the data. 2 This is precisely how Przeworski and Sprague (1986: 3) formulate the dilemma: “given the minority status of workers, leaders of class-based parties must choose between a party homogeneous in its class appeal but sentenced to perpetual electoral defeats or a party that struggles for electoral success at the cost of diluting its class orientation”.

3 emerged by the 1920s to capture both the class and ethnic loyalties of Welsh and Scottish workers. Only from the mid-1960s onwards was Labour’s political dominance challenged significantly by the emergence of Nationalist parties. The origins of these contrasting initial configurations of Left-Nationalist competition are complex and beyond the scope of this paper. In Spain, in the initial decades of co-existence, Left and Nationalist parties competed electorally, but on the basis of mutually exclusive bases of mobilisation, the former on the basis of class allegiances and the latter on the basis of ethnic and nationalist allegiances. In Britain, given the initial absence of Nationalism as a political force, the Labour Party was able to mobilise and maintain political support as the articulator of Welsh and Scottish working class interests within the wider British polity. Over time, though, we see in both countries the emergence of an explicit battle between the Left and the Nationalists for the same voters. In Spain, what had historically been two separate electoral constituencies became increasingly intertwined, while in Britain, Nationalist parties gradually began to target Labour’s traditional working class constituency as a means of electoral advance. In other words, given an increasingly heterogeneous electorate, the challenge for Left parties has been to attract all workers, irrespective of their ethnic allegiances or origins, while the challenge for Nationalist parties has been to attract all members of an ethnic group, irrespective of their social class. In terms of explicit electoral competition, therefore, Left parties have been forced to take positions on the territorial/regional issue, while Nationalists have been forced to take positions – or, more accurately, accentuate their positions – along the Left-Right dimension. In this paper, we examine the outcome of this battle in the contemporary period. We find that the strength of the class cleavage varies between the four regions studied and within regions over time. Nonetheless, we do not see a clear weakening of class voting per se, rather its transformation into something else, akin to the nationalist class voter. In the Basque Country, this takes the form of voting for one of the nationalist parties, all of which are Left or Centre-Left. In Catalonia, Wales and Scotland, it is still unclear what form it takes, sometimes resulting in support for the Left and sometimes for the Nationalists. What is clear is that Hechter’s claim (2004: 404) is not sustained: “social classes whose members are of different status [cultural/national] groups are less likely to be class-conscious than homogenous ones; by the same token, status groups whose members are of different classes are less likely to have cultural consciousness”. The paper is structured as follows. In Part Two, we provide a brief overview of the historical development of competition between Nationalism and the Left in Britain and Spain, before turning to a more detailed description of aggregate trends in the vote

4 shares of the Nationalists and the British and Spanish Left3 since the early 1970s. In Part Three we describe the data and the variables used in the subsequent analysis of individual electoral data. In Part Four we turn to a more detailed examination of the electoral landscape in the four regions in the contemporary period including a brief description of their parties’ positioning on the ideological and nationalist scales and an analysis of the regions’ electorate in terms of ideology, party identification and strength and nature of nationalist identity. Given this, in Part Five, we look at the movement of votes between the Left and the Nationalists, and in Part Six, we examine the working class vote itself. Finally, we present our conclusions.

2. Left – Nationalist Competition in Historical Perspective

2.1. Labour and Nationalism in Britain.

By the mid-1920s, Labour had displaced the Liberal Party as the principal rival to the Conservative Party in Britain4 and established itself as the dominant political force in Wales and Scotland. The latter regions’ role as Labour strongholds was key to the party’s electoral success at the national level in the postwar period. Labour’s dominance in Wales, compared to elsewhere in Britain, has been especially pronounced. In its landslide victory in the 1945 General Election (GE), Labour’s vote in Wales (58.5%) exceeded that for Britain as a whole (48%) by more than 10% points. In every subsequent GE, Labour has consistently maintained a majority of parliamentary seats and a comfortable plurality of the vote. In its landslide triumph in the 1997 GE, the party’s vote share in Wales (54.72%) again exceeded that for Britain as a whole (44.33%) by more than 10% points. In Scotland, too, Labour dominated postwar politics, though not to the same extent as in Wales. In 14 of the 16 GEs held since 1945, the party gained an absolute majority of parliamentary seats, and in the remaining two (those of 1951 and 1955), a plurality. While its share of the vote in Scotland has exceeded that for Britain as a whole, the margins have been smaller than in Wales (for example, in the GEs of 1997, 2001 and 2005, the margins were 1.22%, 1.31% and 3.68% respectively). Even so, Labour’s continuing grip on Scottish politics was seen in the 2005 GE, when it won 40 of Scotland’s 59 parliamentary seats.

3 We use the terms Spanish Left and British Left in the paper to denote the state-wide, non-nationalist, Left- wing parties. 4 In the 1922 General Election, Labour had obtained 29.5% of the vote, fractionally ahead of the Liberals’ vote share of 29.1%; by the 1929 General Election, Labour’s advantage was emphatic: 37.1% versus 23.4% (Pugh 1993: 237).

5 Labour’s long-term domination of Welsh and Scottish politics has reflected the party’s ability to combine both class-based and nationalist-based sources of electoral support. In fact, a purely class-based explanation of Labour voting in Wales and Scotland has never been able to account for the consistent differences in the size of the Labour vote in these regions compared to Britain as a whole. This is especially the case in Wales - while the Welsh working class has been more strongly pro-Labour than the British working class as a whole, so too has the Welsh middle class. In their analysis of Labour’s historic capacity to capture both “red” and “green” (nationalist) affiliations, Balsom et al (1983: 312) noted that the “Welsh ethnic dimension in Labour voting in Wales suggests that a substantial number of voters are seeking a political expression of Welsh values”. So while 60% of Conservative supporters, predictably, claimed British over Welsh identity, a remarkable 70% of Labour supporters claimed Welsh over British identity (1983: 301-2, 309, 311-312, 322). As such, “the Labour party is not simply the party of the working class in Wales, but is also strongly Welsh in its support. The chosen vehicle of Welsh sentiment has been, for fifty years now, the Labour party” (1983: 323). Nonetheless, such long-term political domination was eventually challenged in both Wales and Scotland by the emergence of Nationalist parties.5 Up until the 1920s, the Liberal Party had been the principle vehicle for the new Nationalisms emerging in Britain. Future Liberal Prime Minister Lloyd George had led the Cymru Fydd (Young Wales) movement in the 1890s which called for the establishment of a Welsh Parliament. In 1912, Prime Minister Asquith pledged the Liberals to “home rule all round” for Ireland, Scotland and Wales, a position with which the Labour Party sympathised. Yet World War One, and the subsequent unemployment and economic depression of the interwar years, pushed Scottish and Welsh claims into the background; in the 1930s, politics was “the language of Socialism or , not Nationalism” (Philip 1978: 158). Though Plaid Cymru (PC) was founded in 1925, and the (SNP) in 1928, it was not until after World War Two that they began to achieve a measure of electoral success. Thereafter, both parties followed a similar trajectory of slow but consistent electoral gain, and increasing membership, in the 1950s and early 1960s, before making significant electoral breakthroughs in the late 1960s and early 1970s, a period which proved to be an important watershed – the vote shares of the PC and SNP in all GEs from 1970 onwards have never fallen back to those obtained in the 1966 GE. Dividing the postwar period in two, 1950-1966 and 1970-2005, illustrates the

5 The literature on Welsh and Scottish Nationalism is voluminous. For a recent account of the rise of PC, see McAllister 2001; for a useful overview of the electoral advance of the SNP between 1966 and 1970, see Kellas 1971. For historical analyses of the relationship between Nationalism and the Labour Party, see Keating and Bleiman 1980 and Howell 1986.

6 point.6 In Wales, Labour’s average vote share in GEs dropped by over 10 points from 58.53% in 1950-1966 to 47.3% in 1970-2005, while PC’s average vote share more than trebled from 3.22% in 1950-1966 to 10.19% in 1970-2005. In Scotland, Labour’s average vote share declined from 47.69% in 1950-1966 to 40.38% in 1970-2005, while the SNP’s average vote share increased twelve-fold from 1.58% in 1950-1966 to 18.82% in 1970- 2005. In GEs, therefore, Wales and Scotland remain Labour strongholds, though to a declining degree over time. In absolute terms, Wales remains more of a stronghold for Labour, vis-à-vis the Nationalists, than Scotland: for the 1970-2005 period, its average advantage over PC was 37.11% compared to 21.56% over the SNP.7 Nonetheless, in relative terms, PC has fared better than the SNP against Labour since the 1970s. In Scotland, the SNP gained its peak postwar share of the vote (30.44%) as early as the October 1974 GE, trailing Labour’s vote share (36.28%) by a mere 5.84%. Since then, the SNP’s vote share has declined unevenly to 17.7% in the 2005 GE, which put it 21.8% points behind Labour’s vote share of 39.5%. In Wales, however, PC gained its peak postwar vote share of 14.3% much later, in the 2001 GE, putting it 34.3% points behind Labour’s vote share of 48.6%, whereas in the October 1974 GE (the SNP’s peak performance in Scotland), PC had trailed Labour by 38.7% (10.82% versus 49.52%). PC’s later surge compared to that of the SNP is reflected in the last four GEs (1992, 1997, 2001 and 2005): in Wales, Labour’s advantage over PC fell from 40.67% in 1992 to 30.14% in 2005, while in Scotland, its advantage over the SNP actually increased from 17.5% in 1992 to 21.8% in 2005. Thus in Wales, PC has been gaining ground on Labour, while in Scotland the SNP has been losing ground. In the 2005 GE, Labour’s advantage in Wales over PC of 30.14% points was nearly 7% points less than its average advantage for the 1970-2005 period as a whole (37.11%); in Scotland, its average advantage over the SNP in the 2005 GE of 21.8% was fractionally higher than its average advantage for the 1970-2005 period as a whole (21.56%).8

6 The divide is not arbitrary: the 1966-1970 period was one of “take-off” for both Nationalist parties, first in various by-election victories and then in the 1970 GE itself, in which PC gained 11.54% of the vote in Wales (compared to only 4.29% in the 1966 GE) and the SNP gained 11.41% of the vote in Scotland (compared to only 5.03% in the 1966 GE) and won its first parliamentary seat in a GE. 7 Taking the entire postwar period as a whole (1950-2005), PC, with an average vote share of 7.58%, is the fourth-placed party in Wales (behind the Liberals, on 11.64%, and the Conservatives, on 27.25%), trailing Labour by an average of 43.93% points (only in the 1970 and 2001 GEs did the PC vote exceed that of the Liberals). In Scotland, though, the SNP, with an average postwar vote share of 12.35%, is the third-placed party (ahead of the Liberals, on 10.57%), trailing Labour by an average of 30.77% points. These relative positions of the parties hold also for the more recent 1970-2005 period. 8 The differential rates of advance of the PC and SNP in the recent period have been confirmed in the 1999 and 2003 elections for the National Assembly of Wales (NAW) and the Scottish Parliament (SP). In both cases, dual voting, in which nationalist parties enjoy greater support in regional rather than national

7 We will examine these shifts over time in and between the Labour and Nationalist votes in Part 5. For a long time, though, the class-based elements of Scottish and Welsh Nationalism were paid less attention (and hence, the “Leftward” shift of the SNP and PC over the last twenty years or so is seen as something relatively novel). Instead, emphasis was placed on the political nature of Scottish Nationalism, with its longstanding and consistent demands for greater political independence for Scotland, and the cultural basis of Welsh Nationalism, with its greater concern for protection of the Welsh language. There was considerable justification for drawing such a contrast: as Philip (1978: 169- 170) notes, the Scottish nation historically had a stronger separate political identity than the Welsh. The union of the English and Scottish crowns in 1603 was only consolidated in the 1707 Act of Union. Scotland retained (and still does) a separate established church and legal system. A Scottish Secretary of State was first appointed in 1885. In contrast, the political unity of Wales was never clearly established even in medieval times, nor by the Act of Union imposed by Henry VIII in 1536. A Secretary of State for Wales was appointed for the first time only in 1964. As such, Welsh separateness from England has been cultural, based on religious nonconformity (and the rejection of Anglicanism) and, above all, on the survival and influence of the Welsh language. Consequently, when it emerged, PC, while formally committing itself to self-government for Wales (Lutz 1981: 315),9 placed far greater emphasis on safeguarding the language and cultural traditions of Wales. It is no coincidence that when PC eventually succeeded in winning representation at Westminster, it did so (and to date, has only done so) in the rural, non-industrial, heavily Welsh-speaking constituencies of Carmarthen, Ceredigion, Ynys Môn, Caernarfon and Merionydd (the latter two won in the February 1974 GE and held ever since). However, by concentrating exclusively on these contrasting features of Scottish and Welsh Nationalism, it would be easy to ignore the economic and class-based appeals common to both. Appraising the Nationalists’ electoral platforms in the 1945 GE, McCallum and Readman (1947: 121-123, 252) labelled that of PC, with its demands for solutions to the special economic needs of Wales, and public ownership of the coal industry under a Welsh Coal Board, and its fears of Welsh industry being overly dependent on the English midlands, as “radical and perhaps socialistic”, and that of the

elections, is apparent. However, PC’s advance in the NAW elections (29.5% in the 1999 NAW election compared to 9.94% in the preceding 1997 GE, and 20.5% in the 2003 NAW election compared to 14.3% in the preceding 2001 GE) is much more dramatic than that of the SNP in elections to the SP (28% in the 1999 SP election compared to 22.2% in the preceding 1997 GE and 22% in the 2003 SP election compared to 20.1% in the preceding 2001 GE) (Orriols and Richards 2005: 3, 5). 9 Support for outright independence has always been low in Wales, even amongst PC supporters and voters.

8 SNP, with its emphasis on the appalling social and economic conditions then prevailing in Scotland, and the economic dangers of over-dependence on heavy industry, as “broadly Socialist in outlook”. The fact that subsequent postwar British economic development did nothing to alter the status of Scotland and Wales as areas blighted by relatively higher concentrations of poverty, unemployment, and declining heavy industry (Philip 1978: 160, 168) meant that the political benefits of promoting “economic Nationalism” (Cook 1978: 149), and hence the potential for a head-on battle for the loyalty of working class Labour voters, were always present.10 In this regard, the SNP has been generally more successful than PC in making inroads into traditional Labour territory. It gained its first MP in a 1945 by-election, albeit in unusual circumstances, in the normally safe Labour seat of Motherwell (Philip 1978: 158).11 Its major breakthrough in the late 1960s and early 1970s did not come exclusively at the expense of Labour (see below), but it did nonetheless involve making headway in traditional Labour strongholds. In the 1966 GE, the SNP achieved notable advances in the industrial, Labour-held, seats of central Scotland. A near-victory in the Glasgow Pollok by-election in 1967, with 28% of the vote, was followed by its sensational by- election victory later the same year in Hamilton (at the time Labour’s second-safest seat in Scotland) and another strong showing, in Glasgow Gorbals, in 1969 with 25% of the vote (Butler and Duschinsky 1971: 111-112). These successes in urban working class areas continued in the 1968 municipal elections (Mansbach 1973: 186, 197). In November 1973, the SNP took Glasgow Govan in a by-election and in the May 1974 local elections came second in 30 of the 72 Glasgow seats at stake. In the October 1974 GE – to date, as we have noted, the SNP’s peak postwar performance – it came second in 35 of the 41 constituencies won by Labour. District elections in May 1977 saw a further 8% swing from Labour to the SNP compared to the October 1974 GE result (Cook 1978: 142-146). Whether or not, as Mansbach (1973: 200) argues, simple protest voting against an increasingly unpopular Labour government in London lay behind the SNP’s breakthrough in the 1960s (and its later advances in the 1970s), rather than a “secular trend to the SNP itself”, is debatable. What is important is that direct electoral competition between Labour and the SNP was becoming increasingly prominent.12 This

10 In 19 of the 28 by-elections held in Scotland between 1945 and 2000 in which Labour retained the seat, its vote share fell compared to the preceding GE while that of the SNP increased (Leeke 2003, Table 13). 11 It promptly lost the seat in the subsequent 1945 GE. 12 Even if it was protest voting, the question of why disgruntled Labour voters appeared to channel their support to the SNP rather than to other parties would still need to be addressed. In any case, the SNP’s 1973 victory in Glasgow Govan, when Labour was no longer in power in London, does not square with arguments in terms of protest voting. (Moreover, after it lost Glasgow Govan to Labour in the February

9 pattern has held ever since. For example, 4 by-elections were held in Scotland during the 1997-2001 parliament, all of them in safe, urban, Labour seats. Labour held on to all of them but, compared to the 1997 GE, suffered a sharp drop in its vote share at the same time as the SNP increased its vote share, while the vote shares of the Conservatives and Liberal Democrats fluctuated only moderately. Each seat therefore recorded substantial swings in the vote from Labour to the SNP of 11.2% (in Paisley South, 1997), 22.6% (in Hamilton South, 1999), 6.7% (Glasgow Anniesland, 2000) and 16.2% (Falkirk West, 2000) (Morgan 2003: 5). In Wales, too, PC’s dramatic victory, with 39% of the vote, in the July 1966 by- election in the largely rural seat of Carmarthen tended to overshadow the even larger vote share (40%) it subsequently gained in the March 1967 by-election in Rhondda West, a mining constituency which epitomised a declining working class community in which unemployment was double the average rate for Britain as a whole. The huge increase in PC’s vote share in what was “a bedrock of the ” made it in some ways an even more significant result than Carmarthen. PC went on to almost snatch Caerphilly, “another old mining stronghold”, from Labour in a by-election the following year, again with 40% of the vote (Cook 1978: 149; Butler and Duschinsky 1971: 111-112).13 The question of whether or not these were protest votes is, again, less important than the fact that “the success of [PC] between 1966 and 1970 changed the character of the Party. The hard core of [PC] was joined by a new influx of working-class supporters – men uncertain of their future in the declining heavy industries of the South” (Cook 1978: 149). Moreover, such newcomers to the party had little interest in the issue of the Welsh language: “their aims (...) were much more an economic Nationalism” (Cook 1978: 149). While subsequently, PC, unlike the SNP, has never succeeded in winning from Labour an urban or industrial working class parliamentary constituency, the inroads it did make into Labour’s traditional strongholds should not be underestimated. In the 1992, 2001 and 2005 GEs, PC’s vote share in five14 of the 11 traditional industrial constituencies held

1974 GE, the SNP retook the seat from Labour in a 1988 by-election when, again, Labour was not in power in London). 13 In the 1966 GE, Labour had held Rhondda West with 76.12% of the vote, and a majority of 16,888; in the 1967 by-election, this was cut to 2,301. Caerphilly had also been held by Labour in the 1966 GE with another enormous majority of 21,148, and 74.25% of the vote. 14 The constituencies concerned were Cynon Valley, Caerphilly, Neath and Rhondda (1992, 2001 and 2005 GEs), plus Islwyn (2005 GE), Merthyr Tydfil (2001 GE) and Pontypridd (1992 GE). In the February 2002 by-election in Ogmore, another traditionally safe, industrial Labour stronghold, Labour retained the seat, but there was an 8.5% swing in the vote, compared to the 2001 GE, from Labour to PC. While the former’s vote share had dropped by 10.1%, the latter’s had increased by 6.8% (Leeke 2003: 35; Young 2005: 10). Moreover, in the 1999 NAW elections, PC won 3 of its 17 seats in the traditionally Labour-dominated coalfield areas.

10 continuously by Labour since 1945 exceeded its vote share for Wales as a whole.15 In this period, PC has undoubtedly accentuated its (Left-wing) class appeal, and softened its nationalist tone. In the 1997 GE campaign, for example, Butler and Kavanagh (1997: 151) note that PC’s election broadcasts “came across as Old Labour with a Welsh accent and a Green [nationalist] coat”. The party also switched its name in English in the late 1990s from “Welsh Nationalist” to the “Party of Wales”. For the 2001 GE campaign, it “preserved its traditional style of trying to outflank Labour as the most Left-wing party in Wales”, arguing that Labour had turned its back on Wales and on its traditional values of equality and social justice (Butler and Kavanagh 2002: 71, 152; emphasis added).16 In the British case, therefore, there is longstanding precedent for direct competition for votes between the Left and the Nationalists, though studies of PC and the SNP written in the immediate aftermath of their breakthrough period in the late 1960s and early 1970s, while acknowledging their incursions into traditional Labour territory, nonetheless tended to emphasise the diverse bases of Nationalism’s advance in the first three postwar decades. Cook noted, for example, that all 3 parliamentary seats gained by PC in the October 1974 GE had been Liberal-held in 1945: “in many ways, [PC] and the Liberals are fighting for the same radical heritage” (1978: 151). In Scotland, Mansbach (1973: 194, 200) argued similarly that the “SNP and the Liberals are competing for the same votes”, yet also noted that in 34 of the 42 parliamentary constituencies which the SNP had contested between 1950 and 1967 there was a strong inverse relationship between the SNP’s performance and that of the Conservatives. However, what is probably true is that in the contemporary period, since the early 1970s, direct competition between Labour and the Nationalist parties for the working class vote has become increasingly accentuated. Indeed, in the case of Wales, if we look at all the parliamentary seats won and lost by PC, such direct competition with Labour has always been prominent.17 Regardless of the strong Liberal traditions in the seats it has actually represented at Westminster, the majority of PC’s wins and defeats in these seats have involved gains from, or losses to, Labour: in its by-election victory in 1966, PC took Carmarthen from Labour (it subsequently lost it to Labour in the 1970 GE, regained it in

15 In the 1997 GE landslide, when Labour gained its highest vote share in Wales since 1966, this was true in 3 of the 11 constituencies. 16 In his appraisal of the 1970 GE results in Wales, Steed (1971: 402 [in Butler and Duschinsky 1971]) notes that the 8 seats where PC’s vote had risen dramatically fell into two groups: five rural seats, overwhelmingly Welsh-speaking and until 1960 the main bastion of the Liberal Party in Wales, and 3 seats in Labour’s declining industrial heartland: Aberdare, Caerphilly and Rhondda East (where PC’s vote had risen on average by 18.5%). Steed labels the latter an “aberrant pocket”, but the subsequent trends we have outlined here suggest they were not so aberrant. 17 In 13 of the 15 by-elections held in Wales between 1945 and 2002 in which Labour held on to the seat, its own vote share fell, compared to the previous GE, while that of PC increased (Leeke 2003, Table 13).

11 the October 1974 GE from Labour, lost it, once again to Labour, in the 1979 GE, and regained it from Labour in the 2001 GE); Caernarfon and Merioneth were both taken from Labour in the February 1974 GE (and have been held ever since); A partial exception is Anglesey/Ynys Môn, taken from the Conservatives in the 1987 GE but lost to Labour in the 2001 GE; the only full exception is that of Ceredigion and Pembroke North, won from the Liberals in the 1992 GE, and lost to the Liberals in the 2005 GE.

PC Wins and Losses, 1966-2005:

1966 (by-election): Carmarthen gained from LAB. 1970 GE: Carmarthen lost to LAB. Feb.1974 GE: Caernarfon gained from LAB; Merioneth gained from LAB. Oct.1974 GE: Carmarthen gained from LAB. 1979 GE: Carmarthen lost to LAB. 1987 GE: Anglesey/Ynys Môn gained from CON. 1992 GE: Ceredigion and Pembroke North gained from LIB. 2001 GE: Angelsey/Ynys Môn lost to LAB; Carmarthen East and Dinefwr gained from LAB. 2005 GE: Ceredigion and Pembroke North lost to LIB

In Scotland, there is a more mixed pattern, with the SNP gaining seats from, and losing them to, both the major parties. Tracing all seats won and lost by the SNP since 1945, we see that its gains from, and losses to, the Conservative Party (15 gains and 7 losses, totalling 22) marginally outnumber its gains from, and losses to, Labour (9 gains and 9 losses, totalling 18).

SNP Wins and Losses, 1945-2005:18

1945 (by-election): Motherwell gained from LAB. 1945 GE: Motherwell lost to LAB. 1967 (by-election) Hamilton gained from LAB. 1970 GE: Hamilton lost to LAB; Western Isles gained from LAB. 1973 (by-election): Glasgow Govan gained from LAB. Feb.1974 GE: Argyll gained from CON; Aberdeen East gained from CON; Banffshire gained from CON; Dundee East gained from LAB; Glasgow Govan lost to LAB; Moray and Nairn gained from CON; Stirlingshire East and Clackmannan gained from LAB. Oct.1974 GE: Angus South gained from CON; Dunbartonshire East gained from CON; Galloway gained from CON; Perth and East Perthshire gained from CON.

18 We have counted Dunfermline West in the 1992 GE as an SNP loss to Labour, though the case is unusual: Dick Douglas won the seat for Labour in the 1987 GE, but subsequently resigned the Labour whip and joined the SNP in 1990. He lost the seat to Labour in 1992 as the SNP incumbent.

12 1979 GE: Angus South lost to CON; Argyll lost to CON; Aberdeen East lost to CON; Banffshire lost to CON; Dunbartonshire East lost to LAB; Galloway lost to CON; Moray and Nairn lost to CON; Perth and East Perthshire lost to CON; Stirlingshire East and Clackmannan lost to LAB. 1987 GE: Angus East gained from CON; Banff and Buchan gained from CON; Dundee East lost to LAB; Moray gained from CON; Western Isles lost to LAB. 1988 (by-election): Glasgow Govan gained from LAB. 1992 GE: Dunfermline West lost to LAB; Glasgow Govan lost to LAB. 1995 (by-election): Perth and Kinross gained from CON. 1997 GE: Galloway and Upper Nithsdale gained from CON; Perth gained from CON; Tayside North gained from CON. 2005 GE: Dundee East gained from LAB; Na h’Eileanan an Iar gained from LAB.

Nonetheless, given the distortions produced by the first-past-the-post electoral system, the accentuation over time of direct competition between Labour and the Nationalist parties is best illustrated (albeit at this point by aggregate data) by comparing once again trends in vote shares in the pre- and post- watershed periods. In Wales, over the course of the earlier period, PC improved its standing vis-à-vis Labour only fractionally, by a rate of 0.98%, from a deficit of 56.95% points in the 1950 GE to one of 56.39% points in the 1966 GE. In contrast, there was significant movement in terms of the Left-Right battle: Labour improved its standing vis-à-vis the Conservatives by a rate of 9.67%, from an advantage of 30.72% points in the 1950 GE (a poor year for Labour) to one of 33.69% points in the 1966 GE (a landslide year for Labour). In the later period, however, PC improved its standing vis-à-vis Labour dramatically, by a rate of 24.69%, from a deficit of 40.02% points in the 1970 GE to one of 30.14% points in the 2005 GE. The rate at which the Conservatives also gained on Labour in the later period, by 10.77%, while significant, was clearly outstripped by that of PC. In Scotland, too, in the earlier period, the SNP gained on Labour only modestly, by a rate of 2.01%, from a deficit of 45.8% points in the 1950 GE to one of 44.88% points in the 1966 GE. Again, in contrast, there was much more significant movement in terms of the Left-Right battle: Labour improved its standing vis-á-vis the Conservatives by a factor of 12, from an advantage of 0.95% points in the 1950 GE to one of 12.27% in the 1966 GE. In the later period, the trend is distorted by the sensational collapse of the Conservative Party in Scotland, in which its standing vis-á-vis Labour deteriorated by a factor of four, from a deficit of 6.56% points in the 1970 GE to one of 23.7% points in the 2005 GE. Even so, the SNP, compared to its performance in the earlier period, gained ground massively on Labour by a rate of 34.18%, from a deficit of 35.12% points in the 1970 GE to one of 21.8% in the 2005 GE. Moreover, the rate of increase in Left-Nationalist ‘movement’ in the later

13 period compared to the earlier period is clearly much greater than that for Left-Right ‘movement’.19

2.2. Nationalism and the Left in Spain

In the last decade of the 19th century, Socialist and Nationalist parties began to emerge simultaneously in the Basque Country and Catalonia in a moment of rapid industrialization and, consequently, of massive immigration of workers from the rest of Spain. The strength of the Spanish Socialist and Communist parties in the Basque Country and Catalonia relied on the support of this mass of immigrant workers. The strength of Nationalist parties was a response to the negative reaction by the autochthonous population (emerging middle-classes, petit bourgeoisie, rural dwellers) to this double invasion of Spaniards (“foreigners”) and of increasingly organized and politically active workers. For a while, there was no confusion concerning each other’s camp: the Socialists defended the interests of the working class which, at that time, happened to be mainly Spanish; the Nationalists defended the interests of the autochthonous petit bourgeoisie, the rural constituencies and, in Catalonia, of the big industrialists as well. However, with increased industrialization and modernization the socio-demographic profile of Basque and Catalan society changed rapidly. Second- and third-generation Spaniards lived side by side with an emerging autochthonous working class; urban development diminished the rural constituencies that moved in increasing numbers to feed the needs of manpower in the cities; the interests of the middle-classes clashed with those of the big industrialists, who had more to gain by keeping a close relationship with Spain. Given an increasingly heterogeneous electorate, splits started to take place within the Nationalist camp and in the first two decades of the 20th century we can already find in both the Basque Country and Catalonia nationalist Left-wing parties side by side with nationalist Right-wing parties20.

19 Put differently, while the standing of the Conservative Party vis-à-vis Labour deteriorated twelve-fold in the earlier period, it did so only by a factor of four in the later period. In comparison, the rate at which the SNP gained on Labour in the later period was some 16 times greater than that for the earlier period. In point of fact, in the 1970-2005 period, in both Wales and Scotland, it is the Liberal Party that has gained most ground on Labour, followed by the Nationalist parties which, in turn, are followed by the Conservatives (the latter in Scotland, of course, have lost ground on Labour). Again, though, the increase in the rate at which the Nationalists gained on Labour in Wales and Scotland in the 1970-2005 period compared to the 1950-1966 period exceeds that for the Liberals. The dynamics of the four main parties’ changing vote shares in Scotland since the early 1970s (a moderate increase for Labour, a very significant increase for the Liberals, a significant decline for the SNP, and a highly significant decline for the Conservatives) requires further examination. 20 In 1910 the followers of Francisco de Ulacia split from the ranks of the PNV in disagreement about the theocratic clericalism defended by the party. They formed a nationalist left-wing party. In 1922 there was a

14 It is difficult to keep track of the electoral trajectories of the Spanish Left and the Nationalists in the Basque Country and Catalonia given the continuous interruptions to electoral competition. Since the emergence of Socialism and Nationalism as electoral competitors at the turn of the 19th century, there were two periods of dictatorship in Spain: between 1923 and 1929, and between 1939 and 1977. Neither the Spanish Left nor the Nationalists were electorally strong in the Basque Country or Catalonia until the Second Republic (1931-1936). It was then that the Basque Nationalist Party achieved its best electoral results in general elections in the Basque Country and the Left-wing nationalist party, ERC, became the incumbent party in the Catalan autonomous government, the Generalitat, with votes from both the middle- and working class. It was also during the Second Republic when the Spanish Left started to be increasingly sympathetic to the autonomist ideas defended by the moderate nationalists in the Basque Country and Catalonia and when the Nationalists strategically moved into an alliance with the Spanish Left, given the full opposition of the Spanish Right to any idea of autonomy or decentralization for the Basque and Catalan regions. It was precisely due to the dual threat of Socialism and Separatism (the rojoseparatista alliance) that Franco initiated his coup d’etat against the Republican government. The immediate effect of Franco’s dictatorship was to make this -in a way unnatural- alliance between the Spanish Left and the Nationalists a stronger one and to strengthen the Nationalist case in favour of the national liberation of Basques and Catalans from the Spanish yoke.

The Basque Country after Franco

Since the return to democracy in Spain, politics in the Basque Country have been characterised by the electoral domination of the principal Nationalist parties. With the exception of the 1977 and 2000 GEs, such domination has been consistent across all election types, though the extent of the Nationalists’ advantage (in terms of vote share) over the Spanish parties increases from national to municipal level. The average vote share of the Nationalist parties increases from 49.83% in GEs (1977-2004) to 63.94% in elections to the Basque Parliament (AEs; 1980-2005) and 62.05% in municipal elections (MEs; 1979-2003). Conversely, that for the Spanish parties over the same periods declines from 42.22% in GEs to 34.12% in AEs and 32.65% in MEs. The average

split from the conservative Lliga in Catalonia. The new party, Acció Catalana, was clearly left-wing. For a comprehensive analysis of the evolution of Basque nationalism in the early 20th century, see Pérez-Nievas 2002, chapter 2.

15 advantage of the Nationalist over the Spanish parties of 7.6% in GEs is massively exceeded by that of 29.82% in AEs and 29.4% in MEs. Nonetheless, the vote share of the Nationalist parties has declined over time across all election types, with the rate of decline increasing from national to municipal levels. In GEs, their vote share increased from 35.42% in the 1977 GE to 50.69% in the 1979 GE, thereafter declining unevenly and, in the end, modestly, to 49.07% in the 2004 GE. In contrast, their vote share in AEs declined from 64.47% in 1980 to 58.81% in 2005, and in MEs from 59.47% in 1979 to 52.58% in 2003. Put differently, the Nationalist vote share declined over time by a rate of 3.2% in GEs21, 8.78% in AEs and 11.59% in MEs. These trends suggest that electoral volatility, measured in terms of changes over time between the Nationalists’ and the Spanish parties’ respective vote shares, increases from national to municipal levels. Regardless of “intra-bloc” trends, there has been a narrowing of the gap between blocs over time, and this narrowing has been more acute at the autonomous and municipal levels compared to the national level, even though the extent of Nationalist domination in the former has continued to exceed the latter. Notably, these trends across election type took place broadly in tandem, with the Nationalist vote share peaking in generally the same period (the 1989 GE, the 1994 AE, and 1991 ME) and declining thereafter. Conversely, the vote share of the two principal Spanish parties (PSE-EE/PSOE and PP) tended to increase steadily from each of these three elections onwards.22 If we focus on competition between the PNV and the PSE23 as the respective principal parties of the Nationalist and Class-based blocs, we see that with the exception of the 1993 GE, the vote share of the PSE has never exceeded that of the PNV in any type of election since 1977. Again, though, as for the Nationalist bloc as a whole, the extent of the PNV’s electoral domination differs considerably between, on the one hand, GEs, and on the other, AEs and MEs. The average vote share of the PNV in all GEs since 1977 is 28.35% compared to 24.8% for the PSE, giving it an average advantage of 3.55% points. In contrast, its average vote share in all AEs since 1980 is 33.94%, compared to 19.32% for the PSE, giving it an average advantage of 14.62% points, and in all MEs since 1979 it stood at 30.98%, compared to 19.75% for the PSE, giving it an average advantage of 11.23% points.

21 Comparing the 2004 GE with that of 1979, not 1977. 22 Though the PP’s vote share fell back slightly in the 2005 AE (compared to the 2001 AE) and sharply in the 2004 GE (compared to the 2000 GE). 23 Although the PP gained a higher vote share than the PSOE-PSE in the 2000 GE and the 2001 and 1998 AEs, its average vote share in all three election types since the return to democracy is well below that of the PSOE-PSE.

16 In relative terms, the PNV’s vote share has remained remarkably stable across all election types, increasing in GEs from 29.34% in 1977 to 34.19% in 2004, increasing fractionally in AEs from 38.1% in 1980 to 38.67% in 2005, and decreasing slightly in MEs from 37.75% in 1979 to 35.58% in 2003. The PSE’s vote share, at least in GEs, also remained stable, increasing marginally from 26.45% in 1977 to 27.59% in 2004. However, in both AEs and MEs, its vote share increased significantly, from 14.21% in the 1980 AE and 15.21% in the 1979 ME to 22.68% in the 2005 AE and 22.44% in the 2003 ME. In the process, it narrowed the gap between itself and the PNV from 23.89% points in the 1980 AE to 15.99% points in the 2005 AE, and from 22.54% points in the 1979 ME to 13.44% points in the 2003 ME. In relative terms, therefore, while the PSE has gained ground on the PNV in all election types, it has done so most significantly at the municipal level, where it narrowed the gap in MEs by a rate of 41.7% between 1979 and 2003, followed by a rate of 33.07% in AEs between 1980 and 2005 and 22.14% in GEs between 1977 and 2004. In sum, in the Basque Country in the contemporary democratic period the Nationalist bloc has maintained a consistent advantage over the Class-based bloc across all three election types. While the extent of the Nationalists’ advantage has been most emphatic at the autonomous and municipal arenas, it is precisely in the latter that they have suffered a relatively faster rate of decline over time. Nonetheless, such higher rates of decline in AEs and MEs have not been a function of trends in the vote share of the bloc’s principal party, the PNV, which, as we have seen, has been remarkably stable. In AEs, while the Nationalists’ vote share as a whole declined by a rate of 8.78% between 1980 and 2005, that of the PNV increased by a rate 1.5%. In MEs, the Nationalists’ vote share as a whole declined by a rate of 11.59% between 1979 and 2003, while that of the PNV declined by a rate of only 5.75%.24 This relatively stable performance over time of the PNV and, in contrast, the PSE’s improved showing over time, particularly at the autonomous and municipal levels, will be examined further in Parts 5 and 6.

Catalonia after Franco

In general terms, the Nationalist parties’ grip on Catalonian politics in the contemporary democratic period has been weaker than that of the Nationalists in the Basque Country. The Nationalist bloc has performed best at the autonomous level, where the CiU governed continuously until 2003, and where its average vote share for the 1980-

24 Though in both the 2003 ME and the 2005 AE, we are talking of PNV-EA.

17 2003 period of 48.02% gave it an extremely modest average advantage of 0.91% points over the Spanish parties’ average vote share of 47.11%. In contrast, the Nationalist bloc has performed considerably worse at the municipal and national levels. In MEs, its average vote share of 32.95% for the 1979-2003 period placed it, on average, 24.09% points behind the Spanish parties’ average vote share of 54.07%. In GEs it has performed worst of all, never coming close to a majority vote share: its average vote share for the 1977-2004 period of 32.48% lags well behind that of the Spanish parties, at 62.73%, with an average gap between the two blocs of 30.25% points. Conversely, as can be seen, the Spanish parties have performed best in GEs, less well (though still gaining a comfortable majority share of the vote) in MEs, and worst of all in AEs. Nevertheless, in contrast to the Basque Country, the Nationalist bloc of parties has increased its vote share over time across all election types. This was most notable in MEs, where it increased its vote share by a rate of 44.2%, from 22.85% in 1979 to 37.19% in 2003. In GEs, it increased its vote share by a rate of 34.37%, from 27.27% in 1977 to 36.37% in 2004. Finally, in AEs, it increased its vote share by a rate of 29.0%, from 36.73% in 1980 to 47.38% in 2003. Conversely, the Class-based bloc vote has declined most in GEs, by a rate of 9.54%, from a 67.31% vote share in 1977 to one of 60.89% in 2004, followed by MEs, in which its vote share declined by a rate of 8.51%, from 60.64% in 1979 to 55.48% in 2003, and then by AEs, in which its vote share declined only marginally, by a rate of 1.48%, from 51.81% in 1980 to 50.33% in 2003. If we focus on the CiU and PSC as the respective principal parties of the Nationalist and Class-based25 blocs, the CiU has been most dominant in AEs, gaining an average vote share for the 1980-2003 period of 39.45%, compared to 29.11% for the PSC. On average, therefore, CiU has enjoyed an average advantage over the PSC in AEs of 10.34%. This position is reversed in GEs and MEs. In the former, the PSC’s average vote share is 36.5% compared to 26.82% for the CiU, giving it an average advantage of 9.68% points over the nationalist party; in the latter, its average vote share is 35.01%, compared to 27.36% for the CiU, giving it an average advantage of 7.65% points. Notably, both parties have increased their vote share in all election types over the course of the entire contemporary democratic period. In GEs, the PSC’s vote share increased by a rate of 38.53%, from 28.55% in 1977 to 39.47% in 2004, while that of CiU increased by a rate of 26.86% from 16.38% in 1979 to 20.78% in 2004. In AEs, the PSC increased its vote share by a rate of 38.92%, from 22.43% in 1980 to 31.16% in 2003, while that of CiU increased more modestly, by a rate of 11.17%, from 27.83% in

25 Within the Class-based bloc, the average vote shares of the second party, the PP, in GEs (14.2%), AEs (8.91%) and MEs (9.34%) are well below those of the PSOE-PSC.

18 1980 to 30.94% in 2003. In MEs, the PSC increased its vote share by a rate of 26.66%, from 26.86% in 1979 to 34.02% in 2003, while that of CiU increased by a rate of 28.54%, from 18.99% in 1979 to 24.41% in 2003. Nonetheless, the relative standings of both parties, particularly that of CiU, have been affected by the more recent electoral advance of the Left-wing nationalist party, ERC. While ERC’s vote share up until the early 1990s typically lagged well behind those of the CiU and PSC, since then it has closed the gap significantly. Moreover, this advance has coincided in the 1990s with a relative and significant decline in the CiU’s vote share in all election types, and a slower rate of advance in the PSC’s vote share in GEs and AEs,26 and a relative decline in MEs. Thus in MEs, while the CiU’s vote share fell by a rate of 26.83%, from 33.36% in 1991 to 24.41% in 2003, and that of the PSC by a rate of 8.15%, from 37.04% in 1991 to 34.02% in 2003, that of ERC increased almost four-fold (or 281.49%) from 3.35% in 1991 to 12.78% in 2003. In GEs, while the CiU’s vote share fell by a rate of 34.7%, from 31.82% in 1993 to 20.78% in 2004, that of the PSC rose by a rate of 13.19%, from 34.87% in 1993 to 39.47%, while that of ERC rose more than three-fold (211.57%) over the same period, from 5.1% to 15.89%. In AEs, again, the vote share of CiU fell by a rate of 33.02%, from 46.19% in 1992 to 30.94% in 2003, while that of the PSC rose by a rate of 13.1% over the same period, from 27.55% to 31.16%, and that of ERC more than doubled (106.53%), from 7.96% to 16.44%. It is at the autonomous level, of course, where the rise of ERC has had the greatest political repercussions, as a coalition involving itself and the PSC finally dislodged the CiU from power in the 2003 AE.27 The vote share of the ousted CiU in this election, at 30.94%, represented a very considerable decline from its peak vote share of 46.8% in the 1984 AE. Meanwhile, the PSC, ironically, finally entered government at the autonomous level, despite having suffered a loss of more than 6% points on its vote share in the previous (1999) AE. Thus at the aggregate level at least, it appears that on the one hand, there is a fairly clear inverse relationship since the early 1990s between the respective vote shares in all election types of CiU and the ERC. In this sense, in terms of intra-bloc movement, the ideologically more moderate nationalist party (CiU) has suffered at the hands of its ideologically more Left-wing counterpart (ERC). On the other hand, in terms of inter-

26 In GEs, the PSOE-PSC’s rate of advance of 26.66% in the 1977-1989 period (from 28.55% of the vote in 1977 to 35.59% in 1989) slowed to one of 13.19% in the 1993-2004 period (from 34.87% of the vote in 1993 to 39.47% in 2004). Similarly, in AEs, its rate of advance of 32.77% in the 1980-1988 period (from 22.43% of the vote in 1980 to 29.78% in 1988) slowed to one of 13.1% in the 1992-2003 period (from 27.55% of the vote in 1992 to 31.16% in 2003). 27 ERC left the coalition in May 2006.

19 bloc movement, the principal Spanish Left-wing party (PSC) also appears to have suffered at the hands of its Nationalist Left-wing counterpart (ERC), as seen in its slower rate of advance in GEs and AEs, and its relative decline in MEs, since the early 1990s. We examine to what extent, therefore, ERC, as a Left-wing and Nationalist party, has advanced electorally at the expense of both CiU and the PSC in Parts 5 and 6.

3. The Individual Dataset

Until now we have used aggregate electoral data to demonstrate that there is an exchange of votes between the Spanish/British Left and the Nationalists. The next part of the paper presents a descriptive analysis of individual electoral data, where we try to show that there is congruence between the macro electoral data and the individual data. The dataset we use has been created by merging together survey data collected in the four regions under analysis during the last 30 years. We have used, on the one hand, the existing CIS surveys for the Basque Country and Catalonia between 1979 and 2005 (whenever there was more than one survey per year, we have included in the merged dataset the first one and last one of the year). On the other hand, we have used the existing British surveys (only respondents from Wales and Scotland) and Welsh and Scottish surveys between 1974 and 2003. A detailed list of the surveys merged is provided in the Annex. From each survey, we have used only those questions of interest for our research. These questions refer to voting behaviour, party identification, national identity, language use, government evaluation, socio-demographic characteristics, employment and social class. Unfortunately, the questions do not appear consistently in all the surveys through time. For some of these questions there is a good time perspective; for others, we have only two points in time or, in the worst cases, none. From the original questions in the surveys we have created a number of derived variables. The main ones are vote stability, the movement of votes between electoral blocs and the working class status of respondents. Vote stability is measured as the percentage of respondents who do not change their vote between different types of elections: (a) between two consecutive national elections, (b) between two consecutive regional elections, and finally (c) between one national election and the consecutive regional election. We have created this variable based on retrospective voting whenever possible (i.e., if the vote in national or regional election n and the vote in national or regional election n-1 coincide then there is vote stability). In the Basque and Catalan surveys this is not always possible because we do not have questions about two or more retrospective national elections. In this case, we

20 have used a combination of retrospective and prospective voting (i.e., if the vote in the last national election and the vote intention in the next national election coincide, then we assume there is vote stability)28. The volatility of the vote is the reverse of stability: the percentage of respondents who change their vote between elections. The movement of votes between electoral blocs is measured as the percentage of respondents that move from one electoral bloc to another between different types of elections (the same ones mentioned above). We have defined three electoral blocs: the Spanish/British Right, the Spanish/British Left and the Nationalists. The Spanish/British Right represents the set of state-wide parties that are Right-wing. The Spanish/British Left represents the set of state-wide parties that are Left-wing. The Nationalist bloc represents the set of parties that are regionally based and that defend a nationalist ideology, irrespective of their ideological position, that in turn can also be Right-wing or Left-wing29.

28 We are aware of the problems involved in using retrospective voting when we do not have panel data, namely the increased probability that people forget or get confused about their vote the more into the past they are asked. In order to test the reliability of respondents' self-reported voting behaviour in elections prior to the one immediately preceding the survey, we undertook some reliability tests. For the British surveys, we made use of a three-wave panel dataset available from the UK Data Archive (British Panel Election Study #1614). In it, we compared self-reported past voting behaviour of respondents in Wave n with self-reported vote in Waves n-1 and n-2. The rationale behind this reliability test may be easier to understand by presenting the following table concerning the answers about retrospective vote in past general elections: Wave GE1 GE2 GE3 GE4 GE5 III + * # II + * # I + * # Comparing the self-reported retrospective "vote in previous election" from Wave III with vote from Wave II, we discovered an incongruence of 5% - that is, 5% reported "incorrect" past voting behaviour based on their answer in the previous survey. Similarly, comparing answers from Waves II and I, the incongruence was also approximately 5%. Comparing the respondents’ answers to "vote two elections ago" from Wave III with vote from Wave I, we found incongruence to be up to 24%. Following the table above, we compared * with # twice, yielding the result of 5% incongruence, and we compared + with # once, yielding the result of 24% incongruence. The results indicate that making use of "previous vote" is tenable, whereas "vote two elections ago" is too inaccurate to assess respondents' vote stability or volatility. Therefore, we have only used the answers to the questions “vote in the immediately preceding general election” and “vote in the general election before the immediately preceding one”. The questions about vote beyond that point have been discarded. For the Spanish surveys, we compared the respondent’s self-reported vote in regional election n and in regional election n-1 with the respondent’s self-reported vote in regional election n and the respondent’s vote intention in regional election n+1. The incongruence amounted to 4% of answers. That is, if we compare the stability of the vote calculated on the basis of retrospective voting with the stability of the vote calculated on the basis of a combination of retrospective and prospecting voting, we obtain a different answer 4% of the time. We think that 5% and 4% incongruence are not high figures and therefore we think it is tenable to use retrospective voting as a way to calculate the stability of the vote. 29 Right-wing bloc: Conservative Party (Con), Partido Popular (PP). Left-wing bloc: Labour Party (Lab), Liberal Democrats (Libdem), Partido Socialista (PSOE), Izquierda Unida (IU-EB in the Basque Country and ICV-IU in Catalonia).

21 The paper is mainly interested in studying the voting behaviour of the working class. We have defined working class status based on the classification of occupations used in the Spanish and British surveys. In the Spanish case, the surveys use the National Classification of Occupations (CNO); in the British case, they use the International Standard Classification of Occupations (ISCO). Respondents’ occupations are codified following these schemes. We define working class as all those respondents whose occupation belongs to one of these groups: - CNO, Gran Grupo 7/8/9: “Personal de la extracción de minerales, preparación y tratamiento de materiales, fabricación de productos, del montaje y manejo de maquinaria e instalaciones, de la construcción y de los transportes. Trabajadores no clasificados en otros subgrupos (peones)”. - ISCO, Major Groups 7, 8, and 9: craft and related trades workers, plant and machine operators and assemblers, elementary occupations, respectively. Because the codification of respondents’ occupations according to the CNO did not start until the surveys collected in the year 1992, our analysis of the working class vote in Spain is limited to the period 1992-200530.

4. The electoral landscape of the Basque Country, Catalonia, Wales and Scotland

Electoral competition in the Basque Country, Catalonia, Wales and Scotland takes place along two dimensions, those of class and nation. Electors’ political allegiances may therefore be based on their class identity, their national identity or on a combination of both. National identity is strong in all four regions under analysis. According to survey data collected during the last 30 years (see above), 51% of Basque respondents, 46% of Catalans, 47% of Welsh and 67% of Scottish claim to feel only Basque, Catalan, Welsh or Scottish or more so than Spanish/British. The number of respondents with dual national identity is also large in all cases: 35% in the Basque Country, 36% in Catalonia, 24% in Wales and 26% in Scotland. If we look at the self-placement of Basque and

Nationalist bloc: Plaid Cymru (PC), Scottish National Party (SNP), Partido Nacionalista Vasco (PNV), Euskadiko Eskerra (EE), Eusko Alkartasuna (EA), Herri Batasuna/ Euskal Herritarrok (HB), Convergència i Unió (CiU), Esquerra Republicana de Catalunya (ERC). We are aware of the fact that within the nationalist bloc in Spain parties are also distributed along the left- right scale. However, we want to focus here on the nationalist-non nationalist distinction. Therefore, we assume that nationalist parties in the Basque Country and Catalonia are first and foremost defined by their position on the territorial issue and, only then, by their ideology. 30 Similarly, because the ISCO codification is only to be found in some of the British surveys after 1974, the N is considerably reduced in the analysis of the working class vote in Wales and Scotland.

22 Catalan respondents on the nationalism scale31, we see that in the first years of the period between 1980 and 2005 the distribution of the electorate was skewed to the right, towards more nationalism (see Tables 1 & 2). Over time, however, the situation has centered on a more moderate position and, at the same time, has become more polarized, with the non- nationalist extreme becoming larger. According to survey data for the years 1983, 2001 and 2003, the distribution of the Welsh and Scottish respondents is also clearly skewed towards more nationalism, especially in Scotland32. Between 1997 and 2001 in Wales, the only temporal series we are able to produce with the data available, there has been no important change in this distribution (see Table 3). With respect to ideology, the respondents follow more closely the shape of a normal distribution than is the case for the nationalism dimension: a majority of them cluster around the centre position on the scale, which goes from 1 (extreme left) to 10 (extreme right)33, with the extremes looking similarly small. However, there are important differences across regions and across time. The Basque respondents show a distribution more skewed to the left than the other regions, with the extreme right considerably smaller than the extreme left (see Table 4). On the other hand, the Scottish respondents seem to have moved towards a distribution more skewed to the left between 1983 and 2001 (see Table 6). This is confirmed by the electoral trends we described in Part 2 showing the important losses of the Conservative Party in Scotland during the period and the large combined vote of the Labour party, Liberal Democrats and SNP, all of which are on the centre-left of the ideological scale. The distribution of the Welsh respondents has also changed, with people beefing up the extremes, both to the left and to the right, while the centre moves slightly to the left and becomes considerably smaller (see Table 634). This overall shift of the Welsh electorate slightly towards the left is also in keeping with the aggregate electoral data we examined in Part 2, above all the significant decline in the Conservative Party’s vote share over the last 25 years.

31 The nationalism scale goes from 0 (non nationalist) to 10 (very nationalist). 32 In the British surveys there is no question in which respondents are asked to place themselves along a nationalism scale. Therefore, we have to rely on the national identity question in order to compare the situation with that in Catalonia and the Basque Country. Therefore, the scale goes from 1 (I feel only Welsh/Scottish) to 5 (I feel only British). 33 In the British surveys, the ideological space starts at value 0 and not at value 1, as in Spain. Therefore, the centre of the scale is at value 5 in Wales and Scotland and at value 5.5 in Spain. 34 For the Welsh case we present a graph based on surveys from 1983 and 1997 where the left-right scale was measured from 1 to 5. We have preferred this scale to that of 0-10, to be found in only one survey, because it offers us a temporal perspective.

23 Table 1: Evolution of self-placement on the Nationalism scale in the Basque Country

1983 1984 1987 1988 .3 .2 .1 0

1989 1992 1994 1998 .3 .2 .1 Fraction 0

0 5 10 0 5 10

2001 2005 .3 .2 .1 0

0 5 10 0 5 10 Self-placement on nationalism scale (strong to weak) Graphs by year

Table 2: Evolution of self-placement on the Nationalism scale in Catalonia

1984 1987 1988 .3 .2 .1 0

1991 1992 1999 .3 .2 on .1 0 Fracti 0 5 10 2001 2003 .3 .2 .1 0

0 5 10 0 5 10 Self-placement on nationalism scale (strong to weak) Graphs by year

24 Table 3: Self-placement on the Nationalism scale in Scotland (above) and Wales (below) (Scale: 1- Exclusive Scot/Welsh Identity 5-Exclusive British identity)

Relative strength of natl id, Scotland 2001 .4 .3 on i .2 act r F .1 0

1 2 3 4 5 Relative strength of nat'l id

Relative strength of natl id, Wales 1997 2001 .4 .2 0 on

i 1 2 3 4 5 act

r 2003 F .4 .2 0

1 2 3 4 5 Relative strength of nat'l id Graphs by Year of survey

25 Table 4: Evolution of self-placement on the ideological scale in the Basque Country

Ideological self-placement in the Basque Country 1983 1984 1986 1987 .3 .2 .1 0

1988 1989 1990 1992 .3 on .2 .1 Fracti 0

1994 1998 2001 2005 .3 .2 .1 0

0 5 10 0 5 10 0 5 10 0 5 10 Self-placement on ideological spectrum (left to right) Graphs by year

Table 5: Evolution of self-placement on the ideological scale in Catalonia

Ideological self-placement in Catalonia 1984 1987 1988 1989 .4 .2 0

1991 1992 1995 1999 .4 on .2 Fracti 0

0 5 10 0 5 10

2001 2003 .4 .2 0

0 5 10 0 5 10 Self-placement on ideological spectrum (left to right) Graphs by year

26 Table 6: Self-placement on the ideological spectrum in Scotland (above) and Wales (below)

Ideological self-placement Scotland 1983 1997 .6 .4 Fraction .2 0

0 5 10 0 5 10 Self-placement on ideological spectrum Graphs by Year of survey

1983 1997 .6 .4 Fraction .2 0

1 2 3 4 5 1 2 3 4 5 Self-placement on ideological spectrum Graphs by Year of survey

The party systems of the Basque Country and Catalonia, where the electoral rule is proportional and based on the D’Hondt formula for the allocation of seats, are characterized by high levels of fragmentation and polarization, especially if we compare them to those in Wales and Scotland, where the electoral system is the first-past-the-post majoritarian type35. In the Basque Country and Catalonia there is more than one nationalist regional party and their combined vote share, as we saw in Part 2, is

35 This is the electoral system for elections to Westminster. The elections to the Scottish and Welsh Assemblies are based on a mixed proportional-majoritarian electoral system. However, it is still too early to evaluate its effects on the party system.

27 frequently much larger than that of state-wide parties, especially in the Basque Country. In the latter, there are three main parties competing for votes within the nationalist bloc: the Basque Nationalist Party (PNV), ideologically in the centre; Eusko Alkartasuna (EA), a party that split from the PNV on the centre-left of the ideological dimension; and the radical Nationalists Herri Batasuna (HB), later renamed Euskal Herritarrok (EH) and ideologically on the extreme left. During the first years after the fall of the Franco dictatorship there was a fourth nationalist party, Euskadiko Ezkerra (EE), also Left-wing, which eventually merged with the Spanish Socialist Workers’ Party (PSOE). In Catalonia, there are two main parties competing for votes within the nationalist bloc: Convergència i Unió (CiU), on the centre-right of the ideological spectrum, and the radical Left-wing Nationalists, Esquerra Republicana de Catalunya (ERC). On the other hand, there are three main Spanish state-wide parties competing in both the Basque Country and Catalonia: the Socialists (PSOE), the conservative Popular Party (PP), and the former Communists (IU). In both Wales and Scotland, however, there is only one nationalist party competing for votes with the two main British parties, Labour and Conservative: PC in Wales and the SNP in Scotland. As we saw in Part 2, their vote share in national elections is relatively modest compared to that of the largest party, Labour: an average of 10% for PC and 18% for the SNP for the 1970-2005 period. However, these figures are remarkable given the electoral system and the limits it creates for smaller parties. Both PC and the SNP are seen by surveys’ respondents as being Left-wing parties. In ideological terms, the PNV and the PSOE in the Basque Country are not far from each other. According to survey respondents, both parties occupy the centre space of the ideological scale: their mean positions on the ideological scale between 1980 and 2005 are 5 and 6 respectively. The self-declared position of respondents is more to the left, with a mean position of 4, closer therefore to the Socialists than to the Nationalists. On the Nationalism scale, however, both parties are far apart. According to respondents, the PSOE clearly occupies a non-nationalist position while that of the PNV is clearly nationalist: their mean positions on the Nationalism scale are 3.3 and 8 respectively. The respondents’ self-declared position, at value 6, is closer to the PNV than to the PSOE. In Catalonia, in ideological terms, the two main parties, the PSOE and CiU, are far from each other. According to survey respondents, the PSOE in Catalonia occupies the centre-left while the CiU occupies the centre-right (their mean positions on the ideological scale between 1980 and 2003 are 4.3 and 6.5 respectively). The Catalan respondents’ position, at value 4.7, is clearly closer to the Socialists. In terms of Nationalism, both parties are also far apart, even more so than in their ideological

28 positions. The PSOE clearly occupies a non-nationalist position while CiU occupies a clearly nationalist one, slightly more moderate than that of the PNV (their mean positions on the Nationalism scale are 3.3 and 7.7 respectively). In terms of Nationalism, the position of Catalan respondents is closer to CiU than to the PSOE (value 6.3). According to survey respondents, the Labour Party in Scotland is more to the left than the SNP on the ideological scale: their mean values are 4 and 4.7 respectively. Yet both parties are clearly placed on the centre-left of the scale. This is congruent with some additional data. In a 1992 survey, 73% of those who had an opinion said that the SNP was a Left-wing party (the choices were the following: “right-wing”, “left-wing”, “neither” and “both”). According to the same survey, 63% of Scottish respondents said the SNP is closer to Labour than to the Conservatives and only 14% said the opposite. The electorate, with a value of 4.8 on the scale, is closest to the SNP. In Wales, the Labour Party is considered to be more to the left than in Scotland. It has a mean value of 3.2 on the ideological scale in the period between 1983 and 2003. Hence the Labour Party is quite distant from PC, which occupies the centre of the scale, with a mean value of 5.2. The electorate, with value 4.9, is closer to PC than to the Labour party. The Conservative Party, on the other hand, occupies a centre-right position, with a value of 6.5 on the left-right scale in Scotland and a value of 5.8 in Wales. In Wales, the nationalist party is, on average for the period as a whole, ideologically closer to the Conservatives than to Labour, though as we described in Part 2, it has moved significantly to the left since the 1980s. The Conservatives in Wales are closer to the ideological centre than their counterparts in Scotland. In the Basque Country, the two axes of competition, class and nation, are not independent. There is a statistical correlation between them of -0.23, significant at 0.01. This correlation indicates that being Left-wing is associated with being more nationalist. This is congruent with the ideological and nationalist position of parties in the regional party system. In the Basque Country there is only one Right-wing party, the PP, which defends an exclusive Spanish identity and whose average vote share has been 18% in national elections and 15% in regional ones. All nationalist parties in the Basque Country are placed on the centre-left or the left side of the ideological spectrum and the majority of Left-wing parties are nationalist-oriented, including the Spanish former Communists, IU-EB. The exception is, of course, the Spanish Socialists, PSOE-PSE. However, the Spanish Socialists are more sympathetic to the Basque nationalist cause than the Spanish Right-wing party. Therefore, a clear majority of the population, in terms of voting behaviour, are Left-wing and nationalist and we are unlikely to find a Basque nationalist who is simultaneously Right-wing.

29

5. The Movement of Votes between the Left and the Nationalists

As we have just shown, Left-wing and Nationalist parties are not far from each other in ideological terms: they all seem to share a centre or centre-left space, except for the Catalan CiU which is more to the centre-right than the other nationalist parties. At the same time, they are not far from the electorate’s ideological position either. In terms of nationalism, however, their respective positions are far apart in the Basque Country and Catalonia (unfortunately, we do not have the same kind of data for Wales and Scotland). Does the distance in the degree of nationalism preclude the competition for votes between Left-wing and Nationalist parties? In this section we analyse the extent to which there is an exchange of votes between Nationalist and Left-wing parties. Individual vote stability is slightly higher in the British regions than in the Spanish ones: in the former, 63% of voters stick to the same party between two consecutive national elections and 73% of voters between national and regional elections; in the latter, however, 59% of voters stick to the same party between two consecutive national elections, 61% of voters between regional and national elections and 49% of voters between two consecutive regional elections. Taken separately, there is no difference in terms of individual vote stability between Wales and Scotland, irrespective of whether the elections take place at the same level (both national or both regional) or at different levels (see Table 8). In Spain, vote stability is always higher in Catalonia than in the Basque Country. In fact, the figures for Catalonia are relatively similar to those for Wales and Scotland. The lowest level of vote stability takes place in the Basque Country between regional elections: on average, only 43% of voters stick to the same party (see Table 7).

Table 7: Vote stability/volatility in the Basque Country and Catalonia Basque Country Catalonia (1979-2005) (1980-2001) % of resp. National- Regional- National- National- Regional- National- Regional Regional National Regional Regional National Vote stability 60 43.6 55 62.4 63.7 61.9 Vote change 40 56.4 45 37.6 36.3 38.1 N 17893 18812 7989 18812 3563 8833 (100) (100) (100) (100) (100) (100)

30 Table 8: Vote stability/volatility in Wales and Scotland Wales Scotland (1974-2003) (1974-2003) % of respondents National- National- National- National- Regional National Regional National Vote stability 73.2 66.5 73.3 67.1 Vote change 26.8 33.5 26.7 32.9 N 1160 2648 957 3094 (100) (100) (100) (100)

Table 9: Percentage of voters who change electoral blocs between two consecutive regional or national elections % of resp. Basque C. Catalonia Basque C. Catalonia Wales Scotland that move36: (regional) (regional) (national) (national) (national) (national) From left to 12.8 2.8 8.9 14.6 29.7 17.8 right From right to 7.1 2.2 2.4 0.7 23.7 23.7 left From Nat to 44 44.7 27.4 20.6 7.7 13.1 left From Nat to 8.3 11 13.7 12.1 1.3 3.6 right From left to 24.9 36.5 47.6 42 30.5 24.9 Nat From right to 3 2.8 0 10 7 16.9 Nat N 266 181 124 281 387 473 (100) (100) (100) (100) (100) (100)

Given these levels of individual vote volatility, where do voters move to and from? The lion’s share of vote movements between elections of the same level (either regional or national) takes place between the Spanish or British Left and the Nationalists. This is so for all four regions, although in Wales and Scotland the movement of votes between the British Left and Right is also important, nearly as high as that between the Nationalists and the British Left, and much higher than that in Catalonia and the Basque Country. The net movement of votes in national elections between the Spanish or British Left and the Nationalists for the whole period under analysis has benefited the

36 These percentages are calculated for the total number of respondents who have voted for the parties included in the definition of the electoral blocs in both elections and whose vote has changed from one bloc to another between elections. Movements to and from abstention and to and from other minor parties are not included. Missing values in one of the two elections are also excluded.

31 Nationalists in all four regions: 20% of net vote change has been in favour of the Nationalists in the Basque Country; 24% in Catalonia; 23% in Wales and 12% in Scotland (see Table 9). In regional elections, however, things are quite the opposite. It is the Spanish Left that has benefited from the movement of votes in Catalonia and Basque Country: by 20% in the former and 19% in the latter. (Unfortunately, we have no reliable data for Scotland and Wales, because to date there have only been two sets of regional elections37). The Nationalists have also benefited from the exchange of votes with the British Right in national elections in Wales and Scotland, although to a lesser extent than from the exchange with the British Left: 8.1% of net vote change in favour of the Nationalists in Wales and 13.3% in Scotland. In Catalonia and the Basque Country, however, the opposite holds true. In national elections in Catalonia the Spanish Right has benefited by 2% and in the Basque Country by nearly 14%. In regional elections these figures are 6% and 5% respectively. The main competition for votes therefore takes place between the Spanish or British Left and the Nationalists in all cases. This is followed by British Left-Right competition in the cases of Wales and Scotland, around 20%, in contrast to the low levels of Left-Right exchange in Catalonia, around 2%, where the Spanish Left and the Spanish Right seem to be very closed electoral blocs. In third place comes Nationalist-Spanish Right competition in Catalonia and Spanish Left-Right competition in the Basque Country. The net movement of votes between the Spanish Left and the Spanish Right in regional elections has benefited the Right by very low margins in Catalonia (1% of net vote change in favour of the right) and in the Basque Country (nearly 6% of net vote change in favour of the right). In national elections, the net movement of votes between Spanish Left and the Spanish Right has benefited the Right in Catalonia by 12%, in the Basque Country by 6% and in Wales by 6%, whereas in Scotland it has benefited the British Left, also by 6%. These results seem to contradict the expectation that PR electoral systems encourage “within bloc” shifts of allegiance rather than “between blocs”. In both the PR and the majoritarian systems under analysis there are “between blocs” shifts of allegiance, even slightly higher in the case of PR. Clearly, the nationalist and the class electoral blocs are more permeable than expected, especially in Catalonia and the Basque Country. In Spain, the competition in national elections between the Spanish Left and the

37 Though Orriols and Richards (2005) argue that a pattern of differential voting in both Scotland and Wales is at least suggested by the results of the 1999 and 2003 regional elections, in which the Labour vote has declined sharply and that of the SNP and PC has increased significantly. The trend in Britain at regional level, therefore, appears to be opposite to that in Spain.

32 Nationalists has clearly benefited the Nationalists while in regional elections it has clearly benefited the Spanish Left. In Britain, the competition in national elections between the British Left and the Nationalists has also benefited the Nationalists. So far, we have shown that there is competition for votes between the Spanish/British Left and the Nationalists to a much greater extent than between the Spanish/British Right and the Nationalists, and that this competition is the largest one, irrespective of the national or regional nature of the elections.

6. The Working Class Vote

To be effective in elections parties must win votes. Left-wing parties organize, mobilize, and seek to obtain the votes of, workers. However, as we noted in the Introduction, this strategy may be a poor one for achieving electoral success, for two main reasons. First, working class voters never constitute a majority of the electorate. Second, Left-wing parties, most of the time, have not achieved a monopoly of the working class vote. How successful, therefore, are Left parties in attracting the working class vote in contexts where workers have conflicting class and national identities? Do workers prefer Left parties to Nationalist parties? Is it incompatible to be Left-wing and vote for a Nationalist party? Conversely, is it incompatible to identify with a particular national identity and vote for a Socialist or ? We start with the question of how Left is the working class in the Basque Country, Catalonia, Wales and Scotland, irrespective of their vote38. In order to do this, we have to rely on the position that blue-collar workers adopt on the left-right scale. In the Basque Country and Catalonia the left-right scale goes from 1, representing the extreme left position, to 10, representing the extreme right position. The centre is, therefore, the value 5.5. The mean value for Basque workers on the scale is 4.2 (n=6216) and that for Catalan workers is 4.5 (n=3747). In Wales and Scotland the left-right scale goes from 0, representing the extreme left, to 10, representing the extreme right. The centre is, therefore, the value 5. The mean value for Welsh workers on the left-right scale is 4.7 and that for Scottish workers is 4.6 (n=78 and 342 respectively39). The working class in all four regions takes a centre-left position.

38 The percentage of blue-collar workers in our datasets is the following: Basque Country, 51%; Catalonia, 46%; Wales, 42%; and Scotland, 41%. 39 Unfortunately, there are only two British surveys in which this question was asked, one from 1983 and one from 1997. This is why we have a small n.

33 The Welsh and Scottish working classes show more awareness of their objective class positions than Basque workers40. When questioned about the social class they think they belong to, 20% of blue-collar workers in Wales and 11% in Scotland claim to belong to the middle-class as opposed to 80% in Wales and 88% in Scotland who claim to belong to the working class. In the Basque Country these figures are quite different: 52% of Basque blue-collar workers claim to belong to the middle-class as opposed to 45% who claim to belong to the working class. This may in part be an effect of the particular phrasing that this question has in the Basque surveys. The choice is not between upper-, middle- and working class but between upper-, middle-, and lower- class. And yet we believe that the phrasing alone cannot account for such a large difference. Blue-collar voters do not necessarily vote for Left-wing parties, especially in those situations in which they have national allegiances alongside their class ones. To what extent they do is mainly an empirical question. In Tables 12, 13 and 14 we present the vote of blue-collar workers and compare it with the average vote share that parties have received in national and regional elections. This comparison allows us to see the extent of blue-collar workers’ over- or under-representation among the parties’ electorates. In this sense, negative figures indicate an over-representation of the working class in a given party’s electorate. In general, as expected, Left-wing parties show an over-representation of working class voters in their electorates. However, there are wide and unexpected differences among parties across regions and across time. In regional elections, in the Basque Country, Spanish Left-wing parties have lost the battle for the working class vote. Basque workers vote, in the first place, for the Basque Nationalist Party, PNV (44% of the total working class vote), followed, a long way behind, by the , PSOE-PSE (19%) and, closely after the Socialists, by the radical Nationalist party, HB (16%). The former Spanish Communist party, IU-EB, only obtains 6% of the blue-collar vote. Together, therefore, the Spanish Left captures only 25% of the working class vote. Moreover, the PNV shows an over-representation of blue-collar workers among its electors of 12%, while neither the Spanish Socialists (PSOE-PSE) nor the Spanish Communists (IU-EB) seem to attract working class voters more than they attract other types of voters. In national elections, the competition between the PNV and the PSOE- PSE for the working class vote is tighter, with 39% and 24% of the total blue-collar vote respectively, followed by the Spanish Communists, with 7%. The difference, though, still singles out the Basque nationalist party as a favourite among the Basque working class.

40 Unfortunately, we have no data for Catalonia because there is no question on subjective social class in the Catalan surveys.

34 In regional elections in Catalonia working class voters prefer, paradoxically, the Catalan Nationalist party, CiU, despite being a centre-right party: 35% of the total working class vote goes to this party. In contrast to the Basque Country, however, the Spanish Socialists (PSOE-PSC) follow closely behind, with 32% of the blue-collar vote. The competition for the working class vote is therefore very tight between these two parties in regional elections. The Spanish Socialists show a slight over-representation of blue-collar workers among its electorate, while the Catalan Nationalists show a slight under-representation. In national elections, however, the working class vote behaves in the opposite direction: the Spanish Socialists are the preferred party for blue-collar workers, gathering 50% of the total working class vote. Nevertheless, the centre-right Catalan party, CiU, still manages to gain a considerable share of the blue-collar vote, 27%. The Spanish Socialists in national elections show a clear over-representation of the blue-collar vote. Together, the Spanish Left obtains 48% of the working class vote in regional elections and 57% in national elections. Hence the Spanish Left in Catalonia captures a much larger share of the working class vote than in the Basque Country.

Table 12: The vote of blue-collar workers in regional elections and their under- and over-representation in the parties’ electorates in the Basque Country and Catalonia (1990-200541) Basque Country Catalonia % of blue- av. % vote in % of blue- av. % vote in collars who regional collars who regional Parties voted for party elections Dif. Parties voted for party elections Dif. PNV 44.3 34.9 -9.4 CiU 35.4 38.3 2.9 PSE 19.4 19.4 0.0 PSC 31.8 29.7 -2.1 PP 6.6 15.8 9.2 PP 4.9 9.1 4.2 HB 14.5 14.3 -0.2 ERC 7.9 11 3.1 IU 6.5 5.4 -1.1 ICV-IU 16.5 5.4 -11.1 EA 4.6 10.9 6.3 EE 1.2 7.8 6.6 Others 2.9 3.2 0.3 Others 3.4 7.3 3.9 3077 3036 N (100) N (100) Abstn. 12.9 33.1 20.2 Abstn. 16.2 41.1 24.9

41 The selected period of time is determined by the year in which surveys start to classify respondents following the national classification of occupations (CNO-Clasificación Nacional de Ocupaciones), in the Spanish surveys, and the ISCO classification in the British surveys.

35 Table 13: The vote of blue-collar workers in national elections and their under- and over- representation in the parties’ electorates in the Basque Country and Catalonia (1990- 2005) Basque Country Catalonia % blue-collars av. % vote in % blue-collars av. % party in who voted for national who voted for national Parties party elections Dif. Parties party elections Dif. PNV 39.6 27.6 -12 CiU 27 30.1 3.1 PSE 24.4 24.3 -0.1 PSC 49.6 36.1 -13.5 PP 7.5 18.2 10.7 PP 0.5 19.3 18.8 HB 15.6 14.8 -0.9 ERC 5.1 5.0 -0.1 IU 7.5 8.8 1.3 ICV-IU 7.5 6.2 -1.3 EA 5.6 8.8 3.2 EE 6.5 8.8 2.3 Others 3.2 6.8 3.5 Others 1.2 1.7 0.5 3478 2143 N (100) N (100) Abstn. 27.1 30.6 3.5 Abstn. 19.1 27.9 8.8

Table 14: The vote of blue-collar workers in national elections and its under- and over- representation in the parties’ electorates in Wales and Scotland (1983-2003) Wales Scotland

% blue-collars av. % vote in % blue-collars av. % vote in who voted for national who voted for national Parties party elections Dif. Parties party elections Dif. Con. 18.6 25.2 6.6 Con. 14.8 21.2 6.4 Lab. 66.2 46.4 -19.8 Lab. 53.6 40.8 -12.8 Lib. 5.8 15.1 9.3 Lib. 7.4 18.1 10.7 P. C. 7.1 10.1 3 S.N.P. 22.7 17.9 -4.8 Others 2.2 3.2 1 Others 1.62 2 0.38 769 1359 N (100) N (100) Abstn. 26 Abstn. 17

Since the early 1990s, the Spanish Socialist party in the Basque Country has been losing the working-class vote in one regional election after the other (see Annex, Tables 19 and 20). The blue-collar vote share decreased from 23% in the 1990 regional elections to a mere 13% in 2001. The year 2005, however, seems to show a change of tendency. The PSOE-PSE obtained 27% of the working-class vote, surpassing the best levels of 1990. For the first time since 1990 the PSOE-PSE shows an over-representation of the

36 working-class vote among its electorate (4%). The evolution has been similar in national elections. In 1989 the PSOE-PSE had a 10% over-representation of workers among its electorate. Since then, it has been losing the blue-collar vote with every national election from 32% in 1989 to 20% in 2000. The year 2004 signals a change of tendency. The blue-collar vote rises to 32%, the same level achieved in 1989, and for the first time in the last 15 years the PSOE-PSE shows an over-representation of the working-class vote among its electorate (4%). The tendency in the working-class vote for the PNV has been opposite to that of the Socialists. The PNV received increasing numbers of blue-collar votes from 29% in 1989 to a peak of 55% in the national elections of 2000. In the 2004 national elections, however, it lost 10% of its blue-collar vote, which probably moved to the Socialists. Since the beginning of the 1990s, CiU has been increasing its working-class vote in regional elections and has moved from having an under-representation of blue-collar workers to having an over-representation of 6% in 2001 and 4% in 2003. In national elections, however, the working-class vote has remained with the Spanish Left for the last 13 years, and there are no signs of any other tendency at work (see Annex, Tables 21 and 22). In national elections in Wales the working-class vote goes massively to the Labour party. This party shows a 20% over-representation of blue-collar workers among its electorate. In Scotland, workers vote first and foremost for the Labour party as well; however, the SNP receives 23% of the working-class vote and is therefore a highly significant rival to the Labour party. The SNP shows a 5% over-representation of blue- collar workers among its electorate. Despite the low working-class vote for PC, it is important to note that since the beginning of the 1980s its working-class vote has increased from 4.5%, its lowest level in the period 1974-2001, to 12.2% in 2001 (see Annex, Table 23). This is very likely the result of an intentional strategy, as we outlined in Part 2, on the part of PC, to mobilize explicitly the working-class vote in order to compete with Labour. The strategy seems to be working well. In a 2003 survey, 71% of those who had an opinion said that PC looks after the working class very closely or fairly closely. In contrast, only half of Welsh respondents consider New Labour to be looking after working class interests! Moreover, the position of PC on the ideological scale has been moving to the left according to the views of the surveys’ respondents. In 1983 its position was 5.2; by 1997 this had moved to 4.5; finally, according to the latest data from the 2005 British Election Study, PC is placed by respondents at 4.3. Simultaneously, Welsh respondents place the Labour party increasingly to the right: from value 3.3 in 1983 to value 4.1 in 1997 to value 5.1 in 2005.

37 With respect to the national identity of the Basque, Catalan, Welsh and Scottish working class, we present in Table 15 the distribution of blue-collar workers according to the categories of the following question: “Do you feel (a) only Spanish/British, (b) more Spanish/British than Basque/Catalan/Welsh/Scottish, (c) equally Spanish/British and Basque/Catalan/Welsh/Scottish, (d) more Basque/Catalan/Welsh/Scottish than Spanish/British, (e) only Basque/Catalan/Welsh/Scottish?”

Table 15: The national identity of the working class in the Basque Country, Catalonia, Wales and Scotland between 1983 and 2005 (%) National identity Basque C. Catalonia Wales Scotland Only Spanish/British 7 14.4 8.6 1.7 More Sp/Brit than B/C/W/S 6.5 23.9 9.1 2.8 Equally Sp/Brit and B/C/W/S 39.4 29 31 23.9 More B/C/W/S than Sp/Brit 21.7 23.8 27 36.4 Only Basque/Cat/Welsh/Scot 25.4 8.8 24.4 35.2 N 6246 983 872 528 (100) (100) (100) (100)

The number of blue-collar workers who feel an exclusive nationalist identity (only Basque, Catalan, Welsh, or Scottish) is highest in Scotland (35%), followed by the Basque Country and Wales, (25% and 24% respectively). In Catalonia the number of workers who feel only Catalan is very low compared with the other three cases, a mere 8%. The number of those who feel only Spanish is larger in Catalonia than in the other regions. The contrast is especially stark between Scotland, with only 1.7% of workers feeling only British, and Catalonia, with more than 14% of workers feeling only Spanish. The distribution of Catalan workers along the national identity spectrum is the most centred one, with a shape that most resembles that of a normal distribution. In the other three regions, the distribution is clearly skewed to the right, that is, towards an exclusive nationalist identity, especially in Scotland. Paradoxically, despite the high number of workers with an exclusive nationalist identity, blue-collar workers vote mainly for the Labour party in Wales and Scotland (see Tables 16, 17 and 18, where we present the working-class vote according to national identity). On the opposite side, in Catalonia, with only 8% of the working class showing an exclusive nationalist identity, blue-collar workers vote equally for the Socialists and for the Nationalist Right-wing party, CiU. The behaviour of the Welsh and Scottish workers can be explained in two different ways: either they vote more according to their ideological position than to their nationalist identity, or the Labour party has retained its

38 historic ability (described in Part 2) to make both identities compatible and to combine the Left and the Nationalist vote. Whichever is true, probably a combination of both, it is clearly not applicable to the Spanish Left in the Basque Country.

Table 16: Vote of those blue-collar workers whose national identity is: “ONLY BASQUE/CATALAN” Basque Country Catalonia Parties Regional National Parties Regional National elections elections elections elections PNV 51.5 55.7 CiU 44.5 50 PSE 2.1 2.6 PSC 13.5 16.1 PP 0.1 0.2 PP 1 3.6 HB 35.8 22.6 ERC 37 23.2 IU 2.5 3.4 ICV 3 7.1 EA 6 8.5 EE 0.1 0.2 Other 1.8 6.6 Others 1 0 728 760 200 56 N (100) (100) N (100) (100) Abstention 7.4 23.8 Abstention 0 13.8

Table 17: Vote of those blue-collar workers whose national identity is: “ONLY SPANISH” Basque Country Catalonia Parties Regional National Parties Regional National elections elections elections elections PNV 14.9 9.2 CiU 13.6 3.6 PSE 46.3 51.3 PSC 66.7 62.5 PP 26.4 32.9 PP 14.7 23.2 HB 0 0 ERC 0 0 IU 4.2 3.3 ICV 2.8 8.9 EA 0.8 0.6 EE 0.8 0 Other 2.5 2.6 Others 2.3 1.8 121 152 177 56 N (100) (100) N (100) (100) Abstention 18.2 26.2 Abstention 0 14.3

As we see in Table 16, those blue-collar workers with an exclusive Basque identity only vote for Nationalist parties, in both regional and national elections. The total vote for Spanish parties is 4.7% in regional elections and 6.2% in national elections.

39 Given the structure of the party system in the Basque Country, these voters do not have to give up their ideological position in order to vote for a Basque Nationalist party: they are voting Nationalist and Left when they vote for any of the Basque Nationalist parties, which are all located between the centre and the extreme left of the ideological spectrum. The Catalan case seems slightly more complex. A majority of workers who feel only Catalan vote for Catalan parties, especially the CiU (44%), a centre-right party, and ERC (37%), a Left-wing party. Nevertheless, the total vote for Spanish parties is much larger than in the Basque case: 17% in regional elections and 27% in national elections. The explanation for the large working class vote that goes to a centre-right party such as CiU may, again, be twofold: either these workers decide their vote solely according to their national identity, hence leaving their ideological position aside, or the nationalist party, CiU, has managed to make both identities compatible through regional policies that do not alienate the working class and that, in fact, may even attract them or somehow contribute to their interests as a way to compete for votes with their closest competitors, the Spanish Socialists. It is important to note that in Catalonia there are a considerable number of votes for Spanish Left parties in regional (16%) as well as national elections (23%) by workers who claim an exclusive Catalan identity. This may be an indication that feeling only Catalan does not mean necessarily that workers vote according to their national identity. Alternatively, it can also mean that voting for the Spanish Left is also a way of casting a Nationalist vote. Blue-collar workers in Catalonia with an exclusive Catalan identity are a heterogeneous group of people, whose voting behaviour follows different rules. In the Basque Country, in contrast, the “only Basque” group seems to follow only one rule: if you want to vote Basque and Left you can only vote for Basque parties. The Spanish Left is not considered to be, as in Catalonia, Wales and Scotland, a way of casting a Nationalist vote. With respect to workers who hold an exclusive Spanish identity, they vote massively for Spanish parties, both Left-wing and Right-wing. In the Basque Country, 77% of the working class vote in regional elections and 87% in national elections goes to Spanish parties. In Catalonia, 84% in regional elections and 95% in national elections goes to Spanish parties. The attraction power of the Spanish parties among those workers who feel only Spanish is stronger in Catalonia than in the Basque Country. If we take into account that the number of workers feeling only Spanish in Catalonia is nearly double that in the Basque Country, then it is clear that exclusive Spanish identity offers the Spanish Left in Catalonia a relevant pool of secure votes, whose strength seems as much based on national identity as it is based on class.

40 This difference between the Basque Country and Catalonia is further confirmed by the voting behaviour of immigrant workers, that is, of those workers who were born outside the regions under analysis. It is to be expected that these workers are more likely to claim an exclusive Spanish or British identity than autochthonous workers and that they vote, to a very large degree, for Spanish or British Left parties. The number of immigrant workers who claim an exclusive Spanish and British identity is, however, relatively low, especially in Scotland: 13.7% in the Basque Country; 19.7% in Catalonia; 27% in Wales and none in Scotland. If we include those who claim to feel more Spanish/British than Basque, Catalan, Welsh or Scottish the figures increase: 26% in the Basque Country; 27% in Catalonia; 53% in Wales and 25% in Scotland. Meanwhile, 29% of immigrant workers in the Basque Country, 36% in Catalonia, 19% in Wales and 41% in Scotland claim to feel either more Basque, Catalan, Welsh or Scottish than Spanish/British or only Basque, Catalan, Welsh or Scottish. These are large numbers that underscore the success of nationalist ideology in permeating those categories of voters who would be expected to be least sensitive to the nationalist appeal. It also highlights the malleability of national identities: national identity seems to be an exclusively subjective phenomenon. To feel exclusively Basque, Catalan, Welsh or Scottish it is enough to want to; there is no objective criterion that can keep you away from it. Again, this may be in part the result of parties’ strategies: they have moved from the mobilization of a nationalist identity determined by birth to the mobilization of a nationalist identity determined by choice. The Socialists in Catalonia have successfully attracted the immigrant blue-collar vote in national elections, in which 70% of immigrant workers vote for the PSOE-PSC, and in regional elections, where they obtain 52% while the Spanish Communists obtain another 17%. This is in clear contrast to the situation in the Basque Country, where the immigrant blue-collar vote in national elections goes, in the first place, to the PNV (39%), and only secondly to the Spanish Socialists (28%). In regional elections, the largest share of the blue-collar immigrant vote is for the PNV, 44%, followed again by the Spanish Socialists, 22%, and, finally, by the Basque radical Nationalists, 11%. This is further confirmation that the Spanish Left in Catalonia has a larger pool of secure votes among those workers who feel only Spanish or those with an immigrant background than the Spanish Left in the Basque Country. On the other hand, it is interesting to see the evolution of the immigrant working class vote in national elections in the Basque Country since 1990. A large part of the immigrant working class vote that the Spanish Socialists has been losing since 1990 has

41 been captured by the Spanish Right-wing party, the PP. The immigrant working class vote for the PP has increased from 4% in the 1989 national elections to 15% in 2005. Twenty-three per cent of workers with an exclusive Spanish identity in Catalonia and 33% in the Basque Country vote for the PP in national elections. In Catalonia, only 0.5% of all blue-collar workers vote for the PP in national elections; 7% in the Basque Country. The mean value on the ideological scale of blue-collar workers who feel only Spanish is, in Catalonia, 5, and in the Basque Country, 4.9. In other words, ideologically, they are centre or centre-left, and yet a large number of them vote for the PP which, at 8.6, occupies a clearly Right position. This is probably an indication that for those workers with an exclusive Spanish identity, their national identity is much more important than their class position when deciding their vote at the national level. In fact, these blue-collar PP voters in national elections are closer to the PP on the Nationalism scale than on the ideological scale. In this sense, an exclusive Spanish identity seems to have a more significant effect on the vote than an exclusive Basque or Catalan identity. Workers with an exclusive Spanish identity are more likely to vote for a party far from their position on the ideological scale to the extent that this party’s position is more congruent with their national identity. This does not seem to be the case with workers who claim an exclusive Basque or Catalan identity. Clearly there is a difference between regional and national elections. Blue-collar workers who feel only Spanish are more likely to vote for a nationalist party in regional than in national elections. In fact, in regional elections, nearly 14% of blue-collar workers who feel only Spanish vote for the nationalist party CiU, on the centre-right of the ideological scale. This figure is 15% for blue-collar workers with an exclusive Spanish identity who vote for the main nationalist party, the PNV, in the Basque Country. Those blue-collar workers with an exclusive Scottish or Welsh identity vote massively for the Labour party. It is obviously compatible to feel exclusively Scottish or Welsh and to vote for the Labour party. Does this mean that they assign more relevance to their ideological position than to the nationalist one? We do not know with these data. It could be that they also agree with the Labour party’s position on the territorial issue.42 It could well be, as we have emphasized in the paper, that voting Labour is a way of casting a nationalist vote. It is also interesting to note that a vote for the Conservative

42 This could well be the case. As we noted in Part 2, support for the radical option of independence has always been low amongst the Welsh electorate in general, and even amongst PC supporters. Moreover, since the 1960s, Labour legislation on devolution (as the more moderate form of political decentralisation) and on protection and promotion of the Welsh language has accommodated those issues of greatest concern to the Nationalists. Thus with respect to those elements of the territorial issue gaining majority support in Wales, Labour has succeeded in capturing the median voter.

42 Party is considerably more likely among blue-collar workers with an exclusive British identity. Again, as happens in the Spanish regions, this behaviour could be as much based on national identity as it is based on class: voting for the Right-wing party may be a way of casting a British vote despite class allegiances. If this is the case, exclusive British identity supersedes class as a source of political allegiance and not an exclusive Welsh or Scottish identity, as is often assumed in the literature on ethnic allegiances. However, the Ns are too small in Table 18 to be able to conclude anything of much consequence.

Table 18: Vote of blue-collar workers in national elections according to national identity in Wales and Scotland “ONLY BRITISH” “ONLY WELSH/SCOT.” Parties Wales Scotland Wales Scotland Conservative 43 33 14.3 2.9 Labour 43 67 60.3 73.5 Libdems 7.9 2.9 PC/SNP 7 16 20.6 Others 7 1.6 14 3 63 68 N (100) (100) (100) (100) Abstention 43.2 33.3 37.3 20.2

7. Conclusions

In this paper, we have sought to demonstrate that in political contexts in which the class cleavage is intersected by an ethnic or nationalist cleavage, the key political conflict, in terms of electoral competition, is that between class-based and nationalist parties and, in the four specific regions we have studied, that between Left-wing class- based parties and the Nationalists. We have shown that in the contemporary period, the great majority of vote movement between elections at the same level (whether national or regional) in the Basque Country, Catalonia, Scotland and Wales takes place between these two political blocs. In this sense, the working class, as the traditional bastion of Left-wing parties, has increasingly become an electoral target for Nationalist parties as well. The extent to which Left-wing parties are able to appeal to workers’ ethnic or national identities and, conversely, the extent to which Nationalist parties are able to

43 capture workers’ class allegiances, determines, by and large, the political fortunes of each bloc. If this defines the nature of contemporary electoral competition in all four regions analysed here, it was not always the case. In historical terms, as we have described, there were important differences between the Spanish and British regions in terms of the initial configuration, and subsequent trajectory, of such competition. In Spain, Socialism and Nationalism emerged simultaneously as political forces. Even though initially their respective electorates remained apart, over time they intersected and overlapped; as such, contemporary competition in the Basque Country and Catalonia between Nationalist and Left parties for the same votes has a longstanding precedent. In Britain, though, the Labour Party had emerged as the dominant political force in Scotland and Wales by the 1920s; only much later was its political hegemony challenged by Nationalism. As such, contemporary competition between Labour and the Nationalists is a relatively recent phenomenon (and that for working class votes extremely recent). Our empirical analysis of electoral behaviour in the four regions since the mid- 1970s strongly supports the contention that the key battle between the Left and the Nationalists is that for the allegiance of working class voters, though how this battle is actually resolved clearly differs between regions. In the Basque Country, we have seen the PNV’s extraordinary, continuing, capacity to retain a majority share of the working class electorate, thereby combining very successfully class and nationalist allegiances, while not losing support among the middle classes. In contrast, the PSOE-PSE has failed to combine such allegiances, partly due to the moderate Left position of the PNV, but also to its own relatively moderate nationalist profile (especially when compared to its counterpart in Catalonia). In Catalonia, while the PSC has continued to dominate the working class vote in national elections, this has been challenged at the regional level by the ability of CiU and, above all, ERC, to attract working class votes. (In fact, as both a Left-wing and Nationalist party, ERC has made significant electoral advances, as we described, at the expense of both CiU and the PSC). In Scotland and Wales, meanwhile, we have seen how the Labour Party’s historic longstanding ability to capture both class and nationalist allegiances amongst working-class voters – the key to its political domination since the 1920s - has been challenged by the ability of the SNP and PC to

44 make inroads into its traditional electorate. PC has been particularly successful in doing so. These inter-regional differences make it difficult to determine whether it is the Left or the Nationalists that, in the long-term, is winning the battle (though in broad terms, Nationalism has lost some ground to the Left in Spain, while the reverse is true in Britain). What our data do underscore, however, is the continuing, and possibly increasing, saliency of this battle. Common to both the Left and the Nationalists in these regions is that their appeal to the working class is based on the fundamental compatibility of the pursuit of class interests and the maintenance of nationalist or ethnic identity. Our evidence with respect to the movement of votes back and forth between the Left and Nationalist blocs indicates strongly that a highly significant proportion of working class voters agrees that such interests are compatible; hence the increasingly important phenomenon of what we term the “nationalist class voter”.43 Nevertheless, we have made the case that the Left and the Nationalists are locked in battle for the allegiances of an increasingly common pool of voters only by looking at the behaviour of the voters themselves. We have said little, if anything, about the role of the political parties in determining who wins the battle. Why, for example, do the Nationalists prevail in the Basque Country, while the Left prevails, albeit to a declining degree, in Scotland and Wales? The next logical step is therefore to complement the study of the movement of votes between blocs with an analysis of the role of parties’ strategies in effecting those movements, thereby explaining whose appeal – that of the Left or that of the Nationalists – wins the allegiance of the nationalist class voter in a given regional context. We thus share the central thesis of Przeworski and Sprague (1986: 9) that “the voting behaviour of individuals is an effect of the activities of political parties”.

43 As we have noted in the course of the paper, the Left-Right battle remains an important axis of political competition. This remains especially the case in British GEs, though it may be being eclipsed by Left- Nationalist competition in regional elections. Moreover, we found that those workers who maintain an exclusively Spanish or British identity in the 4 regions studied are more prone to supersede their class identity as a source of political allegiance and, as such, may become the focus of electoral appeals from the Right as well as the Left (as is the case in the Basque Country, for example). Nonetheless, as we have pointed out, such workers account for only a small proportion of workers in the 4 regions studied, while those workers maintaining an exclusively Nationalist identity do combine this with class voting.

45 A brief look at the case of Wales illustrates the importance of taking party strategies into account, though the issues that arise are equally applicable to the other regions that we have studied. Labour’s domination of Welsh politics over the course of the 20th century was based in no small measure on its ability to defuse the potential threat of Welsh Nationalism as an independent political force. It did so through a series of measures which can reasonably be seen as highly strategic: the establishment of a Secretary of State for Wales in 1964; legislation later in the 1960s to promote and protect the Welsh language; the 1974-79 Labour government’s proposals, ultimately defeated, for devolution; and the eventual successful establishment of a National Assembly of Wales by the Labour government elected to power in 1997. At the same time, the Nationalist party’s strategy hardly remained static. As we have seen, PC over time shifted Left ideologically, precisely as a means of broadening the basis of its nationalist appeal beyond the rural, heavily Welsh-speaking areas of West Wales and into Labour’s traditional working class strongholds in the industrial southeast. In doing so, it made significant inroads into Labour’s electorate, inroads which it has managed to defend by and large ever since. If, as we have seen, PC has moved to the Left on the ideological scale in the perceptions of the voters (and Labour to the Right), this is very much the result of action on the part of the parties themselves. Why has Labour in Wales responded in an apparently inadequate manner, losing votes to PC and not recapturing them? Here we offer, in tentative and speculative fashion, three possible explanations, though with relevance, again, for the other regions studied in the paper. First, we have emphasised the significance of the initial, historical configuration of competition between the Left and the Nationalists. In this sense, Labour’s longstanding “historical advantage” may have equipped the party particularly poorly to respond to the Nationalist threat once that threat had emerged. (In comparison, the PNV in the Basque Country, long accustomed to direct and constant competition with the Left, has proven much more adept at maintaining its traditional advantage). Second, the institutional context may also play a role; in this case, the perverse effects of the British electoral system enabled Labour to retain its dominance in terms of seats despite a declining vote share, thereby mitigating the concrete effects of the Nationalist threat. Third, parties are not unitary actors – this is not the case even in highly centralised

46 polities, and they are even less so in regionalised polities like those of Britain and Spain. As such, their ability and willingness to make strategic choices and responses are constrained.44 For example, the Labour Party in Wales and Scotland has tended to be to the left of Labour in England, reflecting the ideological profile of their respective electorates. Labour in Scotland was always more enthusiastic than Labour in Wales about the need for, and desirability of, political decentralisation. Once the NAW was established, Labour in Wales was constrained by its own identification with New Labour in London; this, in turn, hindered its response to the threat posed by PC’s move to the Left. Only after its disastrous showing in the 1999 NAW elections did Labour in Wales seek to put “clear red water” between itself and New Labour by re-establishing both its Welsh and Left credentials, and outflank PC once again (Orriols and Richards 2005: 17). This sketch of the Welsh case, though speculative, nonetheless indicates the importance of examining across regions the strategies and responses of both challengers and challenged and the constraints on those strategies and responses. Only then can the shifting vote shares of Left and Nationalist parties and the actual choices made by the nationalist class voter that we have examined in this paper be fully explained.

44 Przeworski and Sprague (1986: 119) point to “internal relations within each party” as an important constraint on the ability to exercise strategic alternatives.

47 REFERENCES

Balsom, Denis; Madgwick, P.J., and Denis van Mechelen. 1983. “The Red and the Green: Patterns of Partisan Choice in Wales”. British Journal of Political Science 13: 299-325.

Butler, David, and Michael Pinto-Duschinsky. 1971. The British General Election of 1970. London: Macmillan.

Butler, David, and Dennis Kavanagh. 1997. The British General Election of 1997. Basingstoke and New York: Macmillan / St. Martin’s Press.

Butler, David, and Dennis Kavanagh. 2002. The British General Election of 2001. London: Macmillan,

Cook, Chris. 1978. “The Challengers to the Two-Party System”. In Trends in British Politics Since 1945, eds. Chris Cook and John Ramsden. London and Basingstoke: Macmillan.

Hechter, Michael, 2004. “From Class to Culture”. American Journal of Sociology 110, 2: 400-445.

Howell, David. 1986. A Lost Left: Three Studies in Socialism and Nationalism. Chicago: University of Chicago Press.

Keating, Michael, and David Bleiman. 1980. Labour and Scottish Nationalism. Atlantic Highlands, N.J.: Humanities Press.

Kellas, James G: 1971. “Scottish Nationalism”. Appendix V in David Butler and Michael Pinto-Duschinsky, The British General Election of 1970. London: Macmillan. (pp. 446- 462).

Leeke, Matthew. 2003. “UK Election Statistics: 1945-2003”. House of Commons Research Paper 03/59, 1 July 2003.

Lutz, James L. 1981. “The Spread of the Plaid Cymru: The Spatial Impress”. The Western Political Quarterly 34, 2: 310-328.

Mansbach, Richard W. 1973. “The Scottish National Party: A Revised Political Profile”. Comparative Politics 5, 2: 185-210.

McAllister, Laura. 2001. Plaid Cymru: The Emergence of a Political Party. Bridgend, Wales: Seren Books.

McCallum, R.B., and Alison Readman. 1947. The British General Election of 1945. London and New York: Oxford University Press.

48

Morgan, Bryn. 2001. “By-election results: 1997-2000”. House of Commons Research Paper 01/36, 29 March 2001.

Orriols, Lluis, and Andrew Richards. 2005. “Nationalism and the Labour Party: Differential Voting in Scotland and Wales since 1997”. Estudio/Working Paper 2005/213. Madrid: Instituto Juan March de Estudios e Investigaciones.

Pérez-Nievas Montiel, Santiago. 2002. Modelo de Partido y Cambio Político. El Partido Nacionalista Vasco en el Proceso de Transición y Consolidación Democrática en el País Vasco. Madrid: Instituto Juan March de Estudios e Investigaciones.

Philip, Alan Butt. 1978. “Devolution and Regionalism”. In Trends in British Politics Since 1945, eds. Chris Cook and John Ramsden. London and Basingstoke: Macmillan.

Przeworski, Adam, and John Sprague. 1986. Paper Stones. A History of Electoral Socialism. Chicago and London: University of Chicago Press.

Pugh, Martin. The Making of Modern British Politics 1867-1939. Second Edition. Oxford: Blackwell.

Steed, Michael. 1971. “An Analysis of the Results”. Appendix II in David Butler and Michael Pinto-Duschinsky, The British General Election of 1970. London: Macmillan. (pp. 386-415).

Young, Ross. 2005. “By-election results 2001-05”. House of Commons Research Paper 05/34, 11 May 2005.

49

ANNEX

Table 19: Basque Country (regional elections): working class vote and the parties’ electorate Survey number 2040 2096 2120 2282 2414 2601 Year of survey 1992 1994 1994 1998 2001 2005 Work Work Work Work Work Work Class Class Class Class Class Class Vote AEs Vote AEs Vote AEs Vote AEs Vote AEs Vote AEs Party % 1990 Dif % 1990 Dif % 1994 Dif % 1994 Dif % 2001 Dif % 2005 Dif PNV 32,5 28,5 -4 42,3 28,5 -13,8 38,4 29,8 -8,6 45,5 29,8 -15,7 57,3 42,7 -15 40,9 38,7 -2,2 PSE 23 20 -3 21,4 20 -1,4 12,9 17,1 4,2 17,2 17,1 -0,1 13,7 17,9 4,2 26,7 22,7 -4 PP 1,3 8,2 6,9 3,8 8,2 4,4 5,4 14,4 9 4,7 14,4 9,7 12,9 23,1 10,2 6,1 17,4 11,3 HB 17,9 18,3 0,4 17,1 18,3 1,2 18,3 16,3 -2 13,3 16,3 3 8,6 10,1 1,5 15,4 12,4 -3 IU 4,2 1,4 -2,8 3,8 1,4 -2,4 11,5 9,2 -2,3 5,1 9,2 4,1 6,7 5,6 -1,1 7,5 5,4 -2,1 EA 8,5 11,4 2,9 6,5 11,4 4,9 8,6 10,3 1,7 13,3 10,3 -3 EE 7 7,8 0,8 1,8 7,8 6 Other 5,4 4,4 -1 3,3 4,4 1,1 4,9 4,2 1,2 4,2 3 0,6 0,6 0 3,4 3,4 0 N (100) 446 100 397 100 349 100 571 100 758 611 100 Abstention 28,2 39 10,8 27,2 39 11,8 21,8 40,3 18,5 19,9 40,3 20,4 31,3 21 -10 17,5 32 14,5

50

Table 20: Basque Country (regional elections): working class vote and the parties’ electorate Survey number 2040 2096 2120 2282 2308 2414 2601 Year of Survey 1992 1994 1994 1998 1998 2001 2005 Work Work Work Work Work Work Work Class Class Class Class Class Class Class Vote GEs Vote GEs Vote GEs Vote GEs Vote GEs Vote GEs Vote GEs Party % 1989 Dif % 1993 Dif % 1993 Dif % 1996 Dif % 1996 Dif % 2000 Dif % 2004 Dif PNV 28,7 22,9 Dif 34,1 24,4 -9,7 37,1 24,4 -12,7 41,5 25,4 -16,1 37,9 25,4 -13 55,4 31,3 -24 42,8 34,2 8,6 PSE 31,6 21,2 -5,8 24,8 24,9 0,1 21,6 24,9 3,3 19 24 5 22 24 2 19,9 24 4,1 31,8 27,6 4,2 PP 2,4 9,4 -10,4 4,6 14,9 10,3 5,7 14,9 9,2 8,9 18,6 9,7 10,5 18,6 8,1 13,4 29 15,6 7,1 19,1 12 HB 14,7 17 7 16,5 14,8 -1,7 16,9 14,8 -2,1 14,4 12,5 -1,9 15,7 12,5 -3,2 0 IU 6,3 2,3 8,3 8,2 7,8 9,3 1,5 6,6 9,3 2,7 7,5 7,6 8,3 0,7 EA 6,5 11,2 4,7 6,2 10 3,8 8,5 10 1,5 7,4 8,3 0,9 6,1 8,3 2,2 1,1 7,8 3,1 6,6 3,5 EE 6,5 8,8 2,3 Other 3,1 9,3 6,2 5,4 11 5,6 1,9 11 9,1 1,1 1,7 0,6 1,2 1,7 0,5 2,6 7,7 5,1 7,4 4,19 3,21 N (100) 414 366 387 564 504 612 631 Abst. 30,1 33,1 3 26,6 30,3 3,7 32,6 30,3 -2,3 23,5 28,5 5 21,5 28,5 7 30,5 36,2 5,7 25,1 25 0,1

51

Table 21: Catalonia (regional elections): working class vote and the parties’ electorate Survey number 2137 2192 2374 2410 2546 Year of survey 1995 1995 1999 2001 2003 Work Work Work Work Work Class Class Class Class Class Vote AEs Vote AEs Vote AEs Vote AEs Vote AEs Party % 1992 Dif % 1992 Dif % 1999 Dif % 1999 Dif % 2003 Dif CiU 43 46,2 3,2 23,4 46,2 22,8 35,9 37,7 1,8 44 37,7 -6,3 34,8 30,9 -3,9 PSC 38,3 27,5 -10,8 17,3 27,5 10,2 43,1 30,3 -13 34,9 30,3 -4,6 39,3 31,2 -8,1 PP 4,7 6 1,3 3,4 6 2,6 5,3 9,5 4,2 6,6 9,5 2,9 5 11,9 6,9 ERC 4,9 8 3,1 3,5 8 4,5 9,6 8,7 -0,9 10,7 8,7 -2 14,7 16,4 1,7 ICV 6,5 6,5 47,9 6,5 -41,4 3,9 2,5 -1,4 2,9 -2,9 6 7,28 1,28 Others 9,1 5,8 -3,3 4,5 5,8 1,3 2,1 13,8 11,7 0,8 13,8 13 0,2 2,32 2,12 N (100) 514 914 376 830 402 Abstention 23 45,1 22,1 17,2 45,1 27,9 26,9 40,8 13,9 22,9 40,8 17,9 24,4 37,4 13

Table 22: Catalonia (national elections): working class vote and the parties’ electorate Survey number 2137 2192 2308 2546 Year of survey 1995 1995 1998 2003 Work Work Work Work Class GEs Class GEs Class GEs Class GEs Party Vote % 1993 Dif. Vote % 1993 Dif. Vote % 1996 Dif. Vote % 2000 Dif. CiU 27,1 31,8 4,7 26 31,8 5,8 27,4 29,6 2,2 28,3 28,8 0,5 PSC 49,3 34,9 -14,4 52,1 34,9 -17,2 49,2 39,4 -9,8 45,6 34,1 -12 PP 6,8 17,0 10,2 7,8 17,0 9,2 11,9 18 6,1 12,5 22,8 10,3 ERC 5,7 5,1 -0,6 3,7 5,1 1,4 5 4,2 -0,8 7,2 5,6 -1,6 ICV 9,5 7,5 -2,0 9 7,5 -1,5 5,5 7,6 2,1 3,8 3,5 -0,3 Others 1,6 3,0 1,4 1,4 3,0 1,6 1 0,6 -0,4 0,2 1,5 1,3 N (100) 27,1 26 27,4 28,3 Abstention 21,7 24,1 2,4 18,7 24,1 5,4 17,1 23,5 6,4 17,9 36 18,1

52

Table 23: Wales (national elections): working class vote and the parties’ electorate

Year of Survey 1974 1979 1983 1992 2001 Work Work Work Work Class Work Class Class Class Vote GEs Class GEs Vote GEs Vote GEs Vote GEs Party % 1974 Dif. Vote % 1979 Dif. % 1983 Dif. % 1992 Dif. % 2001 Dif. Con. 13 25.8 12.8 23.5 23.8 0.3 19.7 31 11.7 24.4 28.6 4.2 13.1 21 7.9 Labour 73.1 46.8 -26.3 65.9 49.5 -16.4 60.5 37.5 -23 64.4 49.5 -14.9 65.5 48.6 -16.9 Libdems 5.6 16 10.4 6.1 15.5 9.4 1.3 23.2 21.9 6.7 12.4 5.7 7 13.8 6.8 Plaid C. 8.3 10.7 2.4 4.5 10.8 6.3 3.9 7.8 3.9 2.2 8.8 6.6 12.2 14.3 2.1 Other 14.5 0.5 -14 2.2 0.7 -1.5 2.2 2.3 0.1 N (100) 108 311 76 45 229 Abstn. 9 11.6 8 15.1 14.7

Table 24: Scotland (national elections): working class vote and the parties’ electorate

Year of Survey 1974 1979 1983 1992 1997 Work Work Work Work Work Class Class Class Class Class GEs Vote GEs Vote GEs Vote GEs Vote GEs Party Vote % 1974 Dif. % 1979 Dif. % 1983 Dif. % 1992 Dif. % 1997 Dif. Con. 15.6 24.7 9.1 21.9 31.4 9.5 16.1 28.4 12.3 15.2 25.6 10.4 4.6 17.5 12.9 Labour 49.4 36.3 -13.1 51.9 41.5 -10.4 62.9 35.1 -27.8 47.9 39 -8.9 66.1 45.6 -20.1 Libdems 3.1 8.3 5.2 9 9 0 2.1 24.5 22.4 9.6 13.1 3.5 6.4 13 6.6 SNP 31.2 30.4 -0.8 16.8 17.3 0.5 9.8 11.7 1.9 26.6 21.5 -5.1 21.6 22.1 0.5 Other 9.1 0.3 -8.8 0.7 0.8 0.1 1.3 0.5 N (100) 160 256 143 564 236 Abstn. 10.8 13 17.9 18 18.5

53 Description of the Dataset

Basque Country and Catalonia: surveys merged.

source N name of survey year 1217 2009 Pre-electoral Regional Elections in Catalonia (I) 1980 1366 2198 Pre-electoral Regional Elections in Catalonia 1984 (I) 1983 1377 1680 Pre-electoral Regional Elections in Catalonia 1984 (II) 1983 1412 2281 Pre-electoral Regional Elections in Catalonia 1984 (VII) 1984 1706 2895 Social and Political Situation in Catalonia (III). Pre-electoral (I) 1987 1719 2891 Social and Political Situation in Catalonia (IV). Pre-electoral (II) 1987 1735 2889 Social and Political Situation in Catalonia (VI). Pre-electoral (IV) 1988 1750 2899 Post-electoral Regional Elections in Catalonia 1988 1988 1829 2474 Pre-electoral General Elections 1989. Catalonia. 1989 1978 2500 Social and Political Situation in Catalonia (XII). Pre-electoral (I) 1991 1987 2484 Social and Political Situation in Catalonia (XIII). Pre-electoral (II) 1992 1998 2489 Post-electoral Regional Elections in Catalonia 1992 1992 2137 1995 Perception of the Socio-political situation in Catalonia (I) 1995 2192 2492 Perception of the Socio-political situation in Catalonia (II) 1995 2374 1368 Post-electoral Regional Elections in Catalonia 1999 1999 2410 2778 Social and Political Situation in Catalonia (XVI) 2001 2546 1373 Post-electoral Regional Elections in Catalonia 2003 2003 1195 1200 Basque Statute of Autonomy 1979 1378 1500 Pre-electoral Regional Elections in the Basque Country 1984 (II) 1983 1402 1602 Post-electoral Regional Elections in the Basque Country 1984 1984 1513 1997 Social and Political Situation in the Basque Country (IV) 1986 1565 2969 Post-electoral Regional Elections in the Basque Country 1986 1986 1721 2087 Social and Political Situation in the Basque Country (VI) 1987 1757 2080 Social and Political Situation in the Basque Country (VII) 1988 1795 2387 Social and Political Situation in the Basque Country (VIII) 1989 1873 2052 Social and Political Situation in the Basque Country (IX) 1990 1903 1605 Post-electoral Regional Elections in the Basque Country 1990 1990 2040 1615 Public Opinion and Political Culture in the AACC. Basque Country 1992 2096 1579 Social and Political Situation in the Basque Country (XI) 1994 2120 1481 Post-electoral Regional Elections in the Basque Country 1994 1994 2282 2099 Social and Political Situation in the Basque Country (XII) 1998 2308 2099 Post-electoral Regional Elections in the Basque Country 1998 1998 2407 2482 Social and Political Situation in the Basque Country (XIII) 2001 2421 2489 Post-electoral Regional Elections in the Basque Country 2001 2001 2598 1495 Pre-electoral Regional Elections in the Basque Country 2005 2005 2601 2466 Post-electoral Regional Elections in the Basque Country 2005 2005 Total 76979

54 Wales and Scotland: surveys merged.

source N name of survey year of survey 359 2462 BES (Feb) 1974 666 2365 BES (Oct) 1974 1533 1893 BES 1979 1591 858 BES: Welsh election study 1979 1604 1647 BES: Scottish booster study 1979 1852 4146 BBC Election Survey 1983 2005 3955 BES 1983 2278 4886 BBC Election Survey 1987 2981 5232 BES 1992 3171 957 General Election in Scotland 1992 3889 882 Scottish Election Survey 1997 3952s 676 Scottish Referendum Study 1997 3952w 686 Welsh Referendum Study 1997 4546 1085 Welsh Election Study (Wales Life and Times Study) 2001 4766 7251 Devolution and Constitutional Change 2001 5052 988 Wales Life and Times Study (Welsh Assembly Election Study) 2003 Total 39969

55