How do Policy-Makers Really Understand Globalisation? The Internal Architecture of Anglophone Globalisation Discourse in Europe

Transcription

1 How do Policy-Makers Really Understand Globalisation? The Internal Architecture of Anglophone Globalisation Discourse in Europe Colin Hay γ and Nicola Jo-Anne Smith φ Abstract: There is a growing acceptance in the literature of a potentially significant causal role for ideas about in shaping the trajectory of policy and institutional reform in contemporary Europe. Yet we still know remarkably little about policy-makers understandings of, save those they choose to declare publicly. This paper contributes to the important task of operationalising empirically this key set of ideational variables. Using factor analysis of new survey data collected by the authors it maps and compares UK and Irish policy-maker s understandings of, and orientations towards,. The analysis reveals considerable similarities in the ordering of assumptions and attitudes towards between the two country cases and between civil servants and parliamentarians. Yet it also shows some subtle and intriguing differences between policymakers responses in the UK and Ireland and between elected and unelected officials. Intriguingly, it also suggests a significant disparity between politicians private understandings and public discourses of, with the former less necessitarian in tone than the latter. Above all, it suggests that Anglophone discourse in Europe is principally structured in terms of a number of dimensions which relate to the acceptance or rejection of a series of core neoliberal premises. In effect, the terms and internal architecture of discourse in the UK and Ireland are defined by neoliberal assumptions, to the extent that they provide the core point of reference and orientation for even the most sceptical and critical of views. The ideas of policy-makers, especially those about and European integration, play an increasingly influential role in accounts of the determinants of institutional and policy change in contemporary Europe. 1 Indeed, there now seems to γ Corresponding author: Department of Politics, University of Sheffield, Sheffield, S10 2TU, UK. φ POLSIS, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK. 1 We would like to acknowledge the support of the Economic and Social Research Council for funding this research (project grant RES ). We are also extremely grateful to the Members of Parliament Project team, David Baker, Andrew Gamble and David Seawright, for their collaboration 1

2 be something of a consensus and not just amongst self-styled constructivists and interpretivists - that if we are to understand policy-makers responses to and regionalisation we must pay at least as much attention to their understandings of these processes as we do to the empirical realities such understandings strive to capture. Yet, given the widespread acceptance of the theoretical point, it is remarkable how little we still know empirically about policy-makers understandings of and regionalisation. The contrast between our knowledge of citizens and policy-makers attitudes towards and understandings of such processes is, indeed, stark. Eurobarometer surveys offer an interesting, often intriguing and frequently updated perspective on European citizens attitudes to and understandings of. Yet there is, as yet, no equivalent source of data on policy-makers. And what makes this all the more frustrating is that citizens attitudes to and understandings of feature infrequently, if at all, as key explanatory variables in determining the trajectory of institutional and policy change. 2 In short, we have almost a surfeit of attitudinal data on those to whose views we accord only limited causal weight and very little attitudinal data on those to whose ideas we accord a far greater causal significance. What we do have, of course, are a series of more or less systematic and textual analyses of policy-maker s public discourses of and European integration, augmented in many cases by interviews with political elites (see, for instance, Antoniades 2007; Blyth 2005; Schmidt 2000, 2001, 2002a, 2002b; Verdun 2002; see also Hay and Rosamond 2002; Hay and Smith 2005; Smith 2005; Smith and Hay 2008). Yet, valuable though these studies undoubtedly are, they are simply incapable of providing the detailed picture of policy-makers assumptions and understandings of and regionalisation that we increasingly seem to acknowledge that we need. They are, in short, no substitute for raw attitudinal data. and support. Finally, we would like to thank Akrivi Andreou for the invaluable help she has provided in the administration of the survey and the collation of its results. The analysis and interpretation of those results is, of course, ours alone. 2 Certainly in Europe. There is, however, an interesting North American literature which examines the extent to which public attitudes towards social welfare, immigration and are shaped by exposure to the labour-market effects of international economic integration (Hanson, Scheve and Slaughter 2007; Scheve 2001; Scheve and Slaughter 2007). 2

3 The reasons for this are simple and principally two-fold. First, as Vivien Schmidt (2002b: ; 2006: 39-45) has persuasively argued, there are significant differences between public communicative discourses (such as might be used to legitimate policy development) and private coordinative discourses (as deployed, for instance, by political actors in internal negotiations over the content of policy prior to its public legitimation). We certainly cannot take public statements in which political actors invoke as an indication of the same actors understandings of the extent of or the configuration of constraints and opportunities that it is genuinely held to present (see also Rosamond 2002). Cognitions can rarely if ever be inferred reliably from communicative discourses. Secondly, even policy statements framed clearly, explicitly and unambiguously in terms of tend to make little more than passing references to the term and often reveal little about the conception of upon which they are predicated. Thus, even were we able reliably to infer cognitions from public discourses, we would invariably have too little to go on even if were to assemble exhaustively all relevant speeches, policy statements and proposals. As this suggests, if we are to hope to operationalise empirically policy-makers understandings of, and orientations towards, and to begin to tease out the complex relationship between communicative and coordinative discourses we must take a rather more direct route. That is the aim of this paper. In it we report findings from a larger study, based on an attitudinal survey of 657 civil servants and parliamentarians. We believe this to be the first systematic and comparative survey of elite attitudes towards, European integration and the relationship between the two. Building upon our earlier work on the future of European social models, 3 our aim has been: (i) to develop and trial a comparative survey-based methodology for mapping elite political attitudes to and European integration; (ii) to show how that methodology might be deployed in the United Kingdom and Ireland as a prelude for a larger multi-country comparison in a variety of languages; (iii) to survey and map, in so doing, elite political attitudes to, European integration and the relationship between the two; (iv) to use a 3 Also funded by the Economic and Social Research Council under the auspices of the One Europe or Several Research Programme (Project Grant L ). 3

4 combination of qualitative and quantitative techniques to identify patterns in these discourses; (v) to assess the extent to which (any) such patterns are conserved between the country cases and between civil servants and politicians within each country case; and (vi) to construct a dataset that might be extended in future crosscountry and longitudinal research. In so doing we seek to show how ideational variables in general, and discourses of and European integration in particular, might be operationalised empirically. Here we confine ourselves to policymakers understandings of. 4 Our principal aim in the present paper is to use statistical techniques, principally factor analysis, to map, describe and compare the internal structure of policy-makers understandings of in the UK and Ireland. Yet, if the approach and methodology that we adopt is largely new, this should not lead us to overlook the important inductive work that has already been done on understandings of. In particular, we see our work as building on that of Layna Mosley (see especially 2003). Though her analysis is focused on financial market actors whereas our own is focused on policy-makers, it shows the value of treating actors understandings of as an open and empirical question. For, what Mosley reveals is a considerable disparity between the preferences typically projected on to capital market participants, on the one hand, and the expressed preferences of such actors as revealed in an extensive series of detailed interviews, on the other. The reappraisal of the behaviour of financial market actors that her work prompts has major implications for an assessment of the extent to which constrains domestic policy making autonomy. For if, as she suggests, such actors are far less discriminating or well-informed in their political risk assessment than is conventionally assumed, then the constraints on policy-making autonomy predicted in the existing literature are almost certainly exaggerated. The analysis that follows is a first step to repeating such an exercise, albeit using a rather different methodology, for policy-makers. 4 In two further papers, in preparation, we turn ourselves to an equivalent analysis of policy-makers understandings of European integration and to their understandings of the relationship between and European integration, respectively. The complete data set and copies of the questionnaires used are available for download from Description.asp?sn=

5 In the present paper we analyse the attitudinal data comparatively, considering the extent to which understandings of are conserved between policy-makers in the UK and Ireland and between civil servants and elected politicians. In so doing we seek to draw attention to two largely unexplored aspects of the internalisation of the (perceived) imperatives of namely, the process by which understandings of are translated into substantive policy content and the division of labour, between elected politicians and civil servants, in that process. There is, as yet, virtually no empirical work which seeks to address directly such questions. There are, moreover, limits to what we can infer about the formulation of policy in response to from a consideration of policy-makers attitudes towards and understandings of the constraints it imposes. Yet, if we are to build up a picture of the process in and through which policy-makers both adapt themselves to and translate the imperatives they see as arising from it in the drafting, and implementation of legislation, then this is an important part of the empirical picture. Thus, whilst the analysis of attitudinal data such as this cannot hope to reveal the policy-making division of labour between civil servants and elected politicians, it can show the significance of any such division of labour. It may also allow us to gauge whether the widely reported tendency of elected politicians to emphasise their powerlessness in the face of global constraints in public discourse (in depoliticising and legitimising potential contentious decisions) is reflected, at least relative to their civil servants, in their private understandings of. Methodology The conduct of the survey So as to increase the comparability of findings, our methodology is based on that for the earlier Members of Parliament Project (MPP), funded by the ESRC and founded by Andrew Gamble and David Baker. 5 Our dataset was designed to build upon and extend the MPP by surveying attitudes not only to European integration but also to 5 Project Grants R , R , R , R (see also Baker et al. 1998, 2000; Baker and Seawright 1998). 5

6 and the relationship between the two, by examining civil servants as well as MPs, and by adding a comparative dimension. As with the MPP, data was acquired through the distribution of a postal survey consisting of 37 closed questions with identical answer formats ( strongly agree, agree, neither, disagree or strongly disagree ). The questionnaire was sent to all Members of Parliament in the UK and to all Members of the Dail in Ireland. It was also sent to five hundred senior and middle-ranking British civil servants (randomly selected using the British Civil Service Yearbook) and five hundred senior and middle-ranking Irish civil servants (randomly selected using the Irish Administration Yearbook and Diary). Access to civil servants was negotiated directly with the office of the head of the appropriate civil service. This allowed us to send out questionnaires accompanied by a letter of support from the head of the UK and Irish civil service. In the UK, this support was conditional upon dropping a number of items, deemed to solicit political responses from civil servants, from the questionnaires they received. To aid comparability between the country cases and between civil servants and parliamentarians, the data relating to these items from other respondents is excluded from the analysis that follows. 6 All respondents were guaranteed anonymity and confidentiality - not least through the use of a coded postcard returned separately from the anonymised questionnaire. 7 The survey itself was conducted via three separate mailings. The offices of non-respondents were contacted via telephone and/or to confirm non-participation. Overall, 657 completed questionnaires were received, a respectable response rate for a postal survey of this kind of 36.2 per cent. The lowest response rate was from UK MPs (31.4 per cent); the highest response rate from Irish civil servants (44.0 per cent). Tests for the representativeness of the sample, with respect to gender, party and, for civil servants, seniority were all satisfied (see Smith and Hay 2008 for details). 6 In fact, where it was available, factor analyses were performed with and without this data. Though a number of the items had reasonably high factor loadings (typically in the range ), the nature and interpretability of the factors extracted was not substantively affected by the inclusion of the additional data. 7 Overall, 85.1 per cent of respondents returned the coded postcard. This allowed us to run tests for the representiveness of the sample surveyed. 6

7 In keeping with our inductive approach, the questionnaire was designed in such a way as to allow responses to reveal their already existing assumptions about - and understandings of -, rather than to impose an externally generated structure upon these. Consequently, we did not presume that policy-makers view in a highly differentiated fashion, discriminating between discrete domains of such as trade, finance, culture and regulation in terms either of the degree of they exhibit or their consequences for public policy. Whilst it might be valuable to explore the extent to which policy-makers do hold a differentiated view of both and its consequences in later work, it would be premature to do so without first considering the internal structure of their understandings of in general. 8 The analysis of the survey data The principal multivariate technique used to generate the findings reported in this paper was factor analysis (strictly speaking, a form of principal component analysis). This is an attractive technique for discerning patterns in attitudinal data. Indeed, it is surprising that it is not used more widely in political discourse analysis. It allows a complex dataset with a large number of respondents offering responses to a range of questions to be expressed more simply in terms of a more limited number of compound variables (or factors ). These are extracted in a stepwise fashion so as best to account for the exhibited variance in the distribution of responses of a specified group of respondents (the entire sample or various sub-populations). In effect, the technique shows up the internal architecture of respondents answers to the questions posed in the questionnaire. As used here, the technique also has the appeal of being entirely inductive, in that the factors extracted are purely statistical constructs which arise naturally from the patterning of variance in the raw data. Accordingly, 8 Moreover, the attitudinal survey data reported here show no evidence that policy-makers understandings of in the UK and Ireland are structured in this sectorally differentiated fashion. 7

8 although the factors identified clearly need to be interpreted, the process by which they are extracted from the data is itself immune to subjective researcher bias. 9 Separate factor analyses were performed for the entire sample, for Irish and UK respondents separately, for all civil servants and parliamentarians, and for each subgroup on its own (UK civil servants, Irish civil servants and so forth). Tables of bivariate correlation coefficients showed no multicollinearity for any of the analyses. However, given the large number of variables, further collinearity tests from calculations of tolerance and variance inflation factors in bivariate regressions were conducted. On this basis a small number of variables were excluded from the analysis of UK civil servants. 10 To determine the number of factors to be extracted in each case a two-step process was adopted. Initially, a principal component analysis was performed in SPSS using the Kaiser-Guttman criterion (all factors with eigenvalues above 1 being included in the solution). The factors, at this stage, were unrotated. Unexpectedly, given the large sample size and number of variables, this typically produced a solution with a large number of factors. Consequently, following Stevens (1986; see also Cattell 1978; Child 2006; Gorsuch 1983), the graphical scree test technique was used to reduce the number of factors to be extracted. A repeat analysis was then performed in SPSS specifying the number of factors. In order to maximise factor loadings and thereby increase the interpretability of the findings, factors were rotated using both varimax (orthogonal) and oblimin (oblique) rotation methodologies (Child 2006: ). Since both techniques produced very similar results, in terms of the principal 9 Of course, we need to be careful here. For the cues which survey respondents receive in the questions posed to them almost certainly do structure patterns of responses to some extent. We must then be careful not simply to assume that any attitudinal structure revealed in the raw data by the extraction of the factors arises naturally out of policy-makers discourse for we do not have direct access to that. The point is that the relevant methodological issue here relates to the representativeness of the raw data that we generate, not the extraction of factors from that data. 10 As such, the factors extracted in this analysis (reported in Table 6) should perhaps be treated more cautiously than those in the other analyses conducted. In dealing with potential problems of multicollinearity, we adopted the rule of thumb that VIF scores for all variables retained in the analysis should not exceed 10 when the factor scores were regressed on the variables in the model (Field 2005: 196). We are grateful to Sean Carey for advise in devising this procedure. 8

9 factors identified, and the more complex and contentious oblimin process did not produce more readily interpretable factors, we here report only the varimax rotated factor loadings. Factor scores were calculated to allow the comparison of means between sub-groups of the total sample. Results and analysis The results of the factor analysis are, we suggest, illuminating and suggestive providing, in effect, a first opportunity to explore the internal architecture of policymakers understandings of, and orientations towards, the process of in the UK and Ireland. But before turning directly to this multivariate analysis it is important first to consider some simpler univariate summaries of the data. This we provide in Tables 1 and 2. UK Ireland Overall Significance of difference in country means economic dynamics behind (n=366) (n=284) (n=650) political dynamics behind p <.001 (n=331) (n=253) (n=584) extent of exaggerated (n=369) (n=284) (n=653) benefits of outweigh p <.001 costs (n=366) (n=282) (n=648) good for economic prosperity (n=366) (n=280) (n=646) benefits the poor p <.001 (n=355) (n=271) (n=626) threatens social justice p <.001 (n=362) (n=275) (n=637) threatens policy-making autonomy (n=362) (n=277) (n=639) makes neoliberalism more persuasive (n=339) (n=263) (n=602) makes social democracy more persuasive (n=350) (n=268) (n=618) increases need to reduce personal taxation (n=357) (n=266) (n=623) increases need for public investment in skills (n=357) (n=271) (n=628) increases need for R&D spending (n=357) (n=271) (n=628) increases need to be in EMU p <.001 (n=359) (n=266) (n=625) can be regulated effectively (n=352) (n=268) (n=620) should be regulated more p <.001 effectively (n=354) (n=270) (n=624) trade liberalisation good for p <.05 9

10 prosperity of developing (n=364) (n=284) (n=648) countries developed countries responsible to developing countries 1.65 (n=364) 1.57 (n=284) 1.62 (n=648) my country influential in shaping (n=364) (n=269) (n=625) anti-globalisers are misinformed (n=359) (n=270) (n=629) anti-globalisers seek to reverse the irreversible (n=357) (n=273) (n=630) makes European integration more important (n=367) (n=283) (n=650) p <.001 p <.001 p <.001 Table 1: Comparison of means UK and Irish policy-maker s views of In the first of these, we show simple comparisons of means between the UK and Irish respondents, for civil servants and parliamentarians combined. In order to interpret the findings, it is important be recall that responses to questionnaire items are recorded in terms of a common scale, where 1 = strongly agree and 5 = strongly disagree. Consequently, a higher mean equates with a higher level of disagreement with the proposition. Some caution is, of course, required in interpreting a table like this. For if, say, we take a 5 per cent threshold of statistical significance and perform, as in this case, over twenty such tests, we should expect to find at least one spurious correlation. Nonetheless, where a series of such findings pull in a common direction as, we would contend, is the case here - we can be pretty confident that the picture they present is not a statistical artefact. That picture is a very interesting one. A number of points might be made. First, given the relatively large sample sizes, it is interesting that so few of the differences in means on these key items prove statistically significant. Policy-makers in the UK and Ireland, it seems, are united in seeing as a process driven economically which has a series of generally benign economic effects and which is associated with a series of political imperatives (such as increased investment in skills and in R&D). It is a process with considerable momentum, yet one driven to a greater extent by the political agency of developed countries who must, as a consequence, be responsible to developing countries for its effects. 10

11 Yet, the statistically significant differences in the pattern of responses are no less intriguing for this. Indeed, what makes them all the more interesting is that they are so consistent with Irish policy-makers consistently less inclined to see as a politically-driven process, less convinced of the influence of their own country in steering that process, more sceptical about its benign effects and those associated with trade liberalisation more specifically, more sensitive to its potentially sociallydislocating consequences, and more likely to see EMU and European integration as domestic imperatives in the context of. Overall, then, there is evidence of a relatively highly conserved Anglophone discourse of in Europe. Yet Irish policy-makers would seem to be consistently less convinced of the neoliberal interpretation of both s benign economic consequences for all and the market-conforming policy implications with which it has tended to become associated. MPs and TDs Civil servants Overall Significance of difference in means economic dynamics behind (n=265) (n=384) (n=649) political dynamics behind (n=229) (n=354) (n=583) extent of exaggerated p <.05 (n=267) (n=385) (n=652) benefits of outweigh costs (n=264) (n=383) (n=647) good for economic prosperity p <.05 (n=263) (n=382) (n=645) benefits the poor (n=249) (n=376) (n=625) threatens social justice (n=259) (n=377) (n=636) threatens policy-making autonomy p <.05 (n=259) (n=379) (n=638) makes neoliberalism more persuasive (n=244) (n=357) (n=601) makes social democracy more persuasive (n=254) (n=383) (n=617) increases need to reduce personal p <.05 taxation (n=247) (n=375) (n=622) increases need for public investment in skills (n=249) (n=378) (n=627) increases need for R&D spending p <.05 (n=249) (n=378) (n=627) increases need to be in EMU p <.001 (n=250) (n=374) (n=624) can be regulated effectively p <.001 (n=249) (n=370) (n=619) should be regulated more effectively p <.05 (n=250) (n=373) (n=623) trade liberalisation good for p <.05 11

12 prosperity of developing (n=263) (n=384) (n=647) countries developed countries responsible to developing countries 1.59 (n=263) 1.64 (n=384) 1.62 (n=647) my country influential in shaping (n=252) (n=372) (n=624) anti-globalisers are misinformed (n=251) (n=377) (n=628) anti-globalisers seek to reverse the irreversible (n=250) (n=379) (n=629) makes European integration more important (n=264) (n=385) (n=649) p <.001 p <.001 Table 2: Comparison of means parliamentarians and civil servants views of In Table 2 we present a similar comparison of means for parliamentarians and civil servants. The findings are no less illuminating. Again, there is a fair degree of communality amongst respondents, with parliamentarians and civil servants seemingly united in viewing as an economically-driven process with considerable momentum for which developed countries are responsible, which contributes, at least in aggregate terms, to economic prosperity and which is associated with a series of political imperatives. 11 Yet they are rather more divided on the nature of those political imperatives, with civil servants rather more sceptical than their political masters about the need to reduce rates of personal taxation and the need to increase R&D expenditure, yet more convinced of the merits of European intregration and EMU in a context of. Interestingly, they are also more likely to view as a threat to policy-making autonomy, less likely to see their country as influential in shaping and less likely to see as a process that can be governed effectively. In other words, they tend to see less room for political agency than parliamentarians, although they are rather more sceptical of the (neoliberal) association of with the need to reduce 11 It is interesting also to note that the standard deviation for the civil servants was lower than that for the parliamentarians for each item. There are two potential explanations for this that UK and Irish parliamentarians are more divided in terms of their understandings of than are their civil servants, or that, in both polities, parliamentarians are divided on party lines in terms of their attitudes towards and understandings of. Limits of space prevent a resolution of this issue here. It is, however, a theme to which we intend to return in future work examining the salience of political parties in accounting for variance in parliamentarians attitudes to and European integration 12

13 taxation and with the idea that trade liberalisation is the key to economic prosperity for developing nations. This is interesting, especially when set alongside the suggestion in much of the existing literature that elected politicians are often keen to depoliticise contentious decisions by invoking (publicly) non-negotiable external constraints (see, for instance Antoniades 2007; Watson and Hay 2003). In particular, it suggests quite a strong disparity between parliamentarians private understandings of, which tend (at least relative to their civil servants) to emphasise their agency, and their public discourse which tends to emphasise the non-negotiable character of the constraints they face. Yet, such differences notwithstanding, the overall picture is of a loosely-conserved Anglophone conception of shared by civil servants and politicians alike, but with a more consistently neoliberal inflection given to that shared understanding by the latter. This provides some interesting context for the factor analysis. Consistent with the broadly inductive approach to the analysis of discourse adopted in this paper, we do not assume the existence of an Anglophone discourse of conserved between UK and Irish policy-makers. Consequently, it is appropriate to start by considering factor analyses performed on the attitudinal data for UK and Irish policy-makers considered separately. Table 3 shows the results of such an analysis performed on the UK data for civil servants and politicians combined. 12 components extracted cultural dynamics behind.649 is spread of.646 democracy can be regulated effectively.504 is international.451 governance UK influential in shaping.402 good for employment.797 good for economic prosperity.796 benefits workers.785 benefits outweigh costs It is customary to show all factor loadings greater than 0.4 However, to aid the interpretation of the factors and given the large number of items in the questionnaire, we show here (and for all other analyses) only factor loadings greater than 0.5 or, where no more than 5 variables would otherwise be identified, those greater than

14 good for quality of life.692 good for public services.687 benefits firms.672 benefits the poor.639 good for UK s international.544 profile trade liberalisation good for.507 prosperity threatens national sovereignty.844 threatens democracy.739 is erosion of.737 national sovereignty threatens national identity.711 threatens social justice.661 threatens policy-making.587 autonomy makes European integration.698 more important increases need to be in EMU.522 makes social democracy more.427 persuasive developed countries responsible.422 to developing countries should be regulated more.411 effectively makes neoliberalism more persuasive increases need to reduce social spending increases need to reduce corporate taxation increases need to reduce personal taxation makes corporatism more.630 persuasive makes nationalism more.522 persuasive increases need for unrestricted.462 migration increases need for environmental regulation increases need for R&D.782 spending increases need for public.609 investment in skills economic dynamics behind.434 extent of exaggerated means personal taxation will.613 means corporate taxation will.577 means immigration policy will.563 means education policy will.499 means monetary policy will.494 eigenvalue (of unrotated factor) % of variance explained total variance explained Table 3: The internal structure of UK policy-maker s understandings of 14

15 Using the method described in the preceding section, a rather complex seven-factor solution emerges. Together the factors extracted account for some 43 per cent of the variance amongst UK respondents. Yet, despite the complexity of the solution, the factors are readily interpreted. Factor 1 (accounting for an impressive 15 per cent of the total variance) shows how UK respondents attitudes to are structured in terms of their views of its economic effects with those accepting an essentially conventional neo-ricardian view of the benefits of trade liberalisation seeing it as an economic process with a series of evenly distributed benign economic consequences. Factor 2 shows how respondents views are structured in terms of more obviously political considerations with those seeing as increasing the salience of neoliberalism and decreasing the salience and appeal of social democracy also seeing it as associated with a series of neoliberal policy imperatives. Such respondents are also, interestingly, less likely to see as in need of more effective regulation, less likely to see developed countries as responsible to developing countries, and less likely to see the advantages, in a context of, of EMU and European integration. Factors 1 and 2 thus seem to structure respondents views in terms of their acceptance or rejection of distinct economic and political dimensions of neoliberalism. Factor 3 is also easily interpreted, relating to respondents views as to the threat to political autonomy associated with. Those who see as eroding national sovereignty also see it as a threat to democracy, national identity, social justice and policy-making autonomy. This dimension seems to indicate that pessimism about amongst UK policy-makers is associated with a perceived loss of political efficacy and, by the same token, that those less convinced that does pose a threat to sovereignty are more optimistic about its effects. Factor 4 again relates to political efficacy, but here in a rather more practical way with respondents who see as a benign politically and culturallydriven process also more likely to see the UK as influential in shaping its content and more optimistic about the capacity to regulate effectively. Factor 5 relates to the imperatives issuing from. Respondents who conceive of as an economic process tend not to see claims of as exaggerated and tend to associate it with a series of imperatives for the delivery of collective goods (environmental security and high levels of investment in skills and 15

16 R&D). This perhaps suggests the influence of endogenous growth theories in structuring UK policy-makers understandings of the imperatives of. Factor 6 is very simply interpreted pulling out, as it does, those variables relating to as a source of policy nce. In effect, in shows that, in policymakers understandings, nce comes as a package which transcends policy domains. Those who view as a source of policy nce do so consistently. Factor 7 is the most difficult to interpret, with few high factor loadings. What it perhaps suggests is that those who associate with rising levels of migration and with environmental degradation also see collective political responses at the national-level (such as corporatism and nationalism) as more persuasive as a consequence. The above analysis is certainly interesting, not least because it does suggest that UK policy-makers understandings of are structured in a relatively coherent, consistent and readily interpretable way. Yet it may well also hide potentially significant differences between MPs and civil servants. There are two ways to examine this proposition further. The first is to look for differences in mean factor scores between MPs and civil servants for the seven factors identified above. Table 4 shows the results of such an analysis. Mean factor score MPs Civil service significance p <.01 p <.01 p <.01 Table 4: UK civil servants and MPs attitudes to compared These results need to be interpreted with some caution. For a factor score to be calculable for any given respondent we need that respondent to have answered all questions in the survey. Consequently, although the overall sample size for the factor analysis is high, the comparison of means is based on a far smaller sub-set of complete responses. Accordingly, differences in means need to be quite large to show up as statistically significant. Yet even taking this account, the results are 16

17 suggestive. 13 MPs, it seems, are perhaps predictably rather more optimistic than their civil servants about their capacity for governing effectively, are more inclined to see as necessitating the provision (by themselves) of a series of collective public goods, and yet somewhat less likely to see as driving a process of policy nce. In essence, they are more likely than their civil servants to see themselves as politically efficacious. As noted above, this suggests a certain contrast between their private understandings of, on the one hand, and their public discourse, on the other. The former, it seems, accentuates agency, whereas the latter accentuates constraint lending credence to the view that public discourse may be used, particularly in the UK context, to legitimate and depoliticise otherwise contentious policy choices (Hay and Rosamond 2002; Hay and Smith 2005). This is all very well, but the above analysis rests on the premise that UK civil servants and MPs have understandings of that are structured in a similar way such that it is useful to perform a single factor analysis for both groups combined. The interpretability of the factors generated in such an analysis certainly adds credence to such a view, but the only way to test the proposition directly is to perform factor analyses for each group considered separately. The pattern matrices for these are shown in Tables 5 and 6. Limits of space prevent a detailed consideration of each, but what is remarkable about these is how similar the findings they summarise are to those reported above. If we take Table 5, for instance, which shows the pattern matrix for MPs, factors 1 and 2 are essentially identical to those identified in the combined analysis relating to the attitudes of respondents towards the economic and political dimensions of the neoliberal view of. Similarly, factors 3 (relating to sovereignty and political efficacy), 5 (relating to the imperatives of ) and 6 (relating to nce) are the same as those extracted in the combined analysis - and they are even extracted in the same order, accounting for similar proportions of the total variance. It is only in the composition 13 To interpret them it must be recalled that negative factor scores imply greater agreement with the items with positive factor loadings in a pattern matrix such as Table 3 (since responses are recorded on a scale in which 1 = strongly agree, 5 = strongly disagree). 17

18 of factor 4 where we see any significant difference, and even here that difference is very slight. components extracted threatens national sovereignty.866 threatens national identity.766 threatens democracy.745 threatens policy-making.745 autonomy is erosion of.710 national sovereignty threatens social justice.676 cultural dynamics behind.595 is spread of.576 technology is international.556 governance is spread of.444 democracy benefits workers.750 good for employment.743 good for economic prosperity.738 good for quality of life.738 benefits outweigh costs.719 good for public services.667 benefits the poor.631 good for UK s international.514 profile increases need for R&D.695 spending economic dynamics behind.503 increases need for public.481 investment in skills benefits firms.450 meaning of unclear makes European integration.850 more important increases need to be in EMU.657 should be regulated more.616 effectively makes social democracy more.584 persuasive developed countries responsible.510 to developing countries increases need for environmental.505 regulation trade liberalisation good for prosperity increases need to reduce social spending increases need to reduce corporate taxation Increases need to reduce personal taxation means corporate taxation policy.684 will means personal taxation will.665 means education policy will.621 means immigration policy will

19 means monetary policy will.413 Eigenvalue (of unrotated factor) % of variance explained total variance explained Table 5: The internal structure of MP s understandings of Table 6, for UK civil servants, again presents a very similar picture albeit with one fewer factor extracted. Factors 1 and 4 are identical to those identified above; factor 2 maps directly onto factor 3 in the preceding two analyses, factor 3 onto factor 5, and factor 5 (with some modest variations) onto factor 7 in the combined analysis. In sum, there is a strong evidential basis here for thinking that UK policy-makers understandings of, and orientations to, are structured in a very similar way and that this internal structure is highly conserved between civil servants and politicians. components extracted threatens national sovereignty.725 threatens democracy.662 increases need to reduce social.623 spending increases need to reduce.617 corporate taxation is erosion of.614 national sovereignty threatens national identity.575 Increases need to reduce.540 personal taxation threatens social justice good for employment.792 good for economic prosperity.776 benefit outweigh costs.766 benefits workers.743 benefits firms.684 benefits the poor.621 good for public services.615 good for quality of life.614 trade liberalisation good for.603 prosperity good for UK s international.599 profile is being regulated effectively.525 anti-globalisers are likely to be effective makes socialism more persuasive increases need for R&D.751 spending increases need for public.589 investment in skills increases need for environmental.521 regulation developed countries responsible.487 to developing countries extent of is

20 exaggerated can be regulated effectively.649 UK is influential in shaping.423 is spread of technology means corporate taxation policy will means personal taxation will is international.514 governance makes corporatism more.507 persuasive increases need for unrestricted.449 migration cultural dynamics behind.416 Eigenvalue (of unrotated factor) % of variance explained Total variance explained Table 6: The internal structure of UK civil servants understandings of Might we draw the same conclusion from an equivalent analysis of the Irish data and, indeed, is there evidence of a shared basic structuration of understandings of conserved between the UK and Irish cases? Consider first the pattern matrix from the combined factor analysis for Irish civil servants and parliamentarians (see Table 7). components extracted benefits workers.782 good for economic prosperity.765 good for employment.761 benefits firms.718 benefits outweigh costs.665 benefits the poor.612 good for Ireland s international.610 profile good for public services.523 trade liberalisation good for.518 prosperity is integration of.507 world markets good for quality of life.500 threatens national identity.717 threatens national sovereignty.685 threatens democracy.679 threatens policy-making.670 autonomy is erosion of.666 national sovereignty threatens social justice.664 threatens national security.608 can be regulated effectively is being regulated effectively is spread of

21 technology is spread of.574 democracy increases need for environmental.516 regulation Ireland influential in shaping.513 political dynamics behind.427 technological dynamics behind anti-globalisers seek to reverse.495 the irreversible increases need for R&D.492 spending increases need to be in EMU.487 anti-globalisers are likely to prove effective increases need to reduce.707 personal taxation increases need to reduce.640 corporate taxation increases need to reduce social.612 spending Should be regulated more.621 effectively developed countries responsible.615 to developing countries makes European integration.587 more important Ireland should influence.549 more means personal taxation will.651 means corporate taxation will.625 means labour market policy will.551 means immigration policy will.534 means education policy will.456 eigenvalue (of unrotated factor) % of variance explained total variance explained Table 7: The internal structure of Irish policy-maker s understandings of Once again the graphical scree test leads to a rather complex seven factor solution. Yet, once again, the factors are themselves readily interpretable not least because many of them are familiar from the analysis of the UK data. Taken together the factors account for some 45 per cent of the total variance in respondents responses. Factor 1 is almost identical to that identified in the combined UK analysis the only difference being that, in addition to the 10 items with high factor loadings in the analysis of the UK, an economic definition of also receives a high factor 21

22 loading in the Irish case. The factor is readily interpretable in terms of respondents orientation towards an economic neoliberal depiction of (as a benign process arising from trade and capital liberalisation). Factor 2 is rather interesting. It would appear to map fairly directly on to the third factor extracted in the equivalent analysis of the UK relating to respondents views of the threat to political autonomy associated with. Yet there are some subtle differences here which are worthy of note. As in the UK, it would seem, pessimism about amongst Irish policy-makers tends to associate the process with a loss of political efficacy. Those who see as eroding national sovereignty also see it as a threat to democracy, national identity, national security, social justice and policy-making autonomy. But, interestingly, and in contrast to their UK counterparts, such respondents are also more likely to view as a process suffering from a governance deficit - and one that is unlikely to be filled, since the process itself is not seen to be amenable to effective governance. By the same token, those more optimistic about the capacity for to be regulated effectively tend not to see it as a threat to democracy, national identity or policy-making autonomy. As in the UK, Irish policy-makers understandings of are structured in terms of a key dimension relating to the acceptance or rejection of a fatalistic view of the capacity for political agency. Factor 3 equates very closely to factor 4 identified in the UK analysis. Like their UK counterparts, Irish policy-makers who see as driven principally by noneconomic forces (cultural, political and technological) are more likely than the rest of their peers to see Ireland as influential in shaping its content. Yet they are also more likely to see as necessitating the enhanced capacity for governance at the trans-national level (in response to environmental degradation, in particular). Factor 4 is interesting, and does not seem to have a direct analogue in the UK data. Those who see as a technologically-driven process are, perhaps unremarkably, less likely to see as reversible and more likely to see it as necessitating higher levels of R&D investment. Yet they are also more likely than their peers to see as increasing the need for European monetary integration. 22

23 By contrast, factor 5 is readily interpretable and has a direct analogue in the UK data. Again, it seems, Irish policy-makers have highly conserved attitudes with respect to the relationship between and policy nce. Factor 6 is, however, further suggestive of differences in the internal structure of policy-makers understandings of in the UK and Ireland. It would seem to relate to the capacity of, and need for, greater political agency to regulate. Those who see as a process in need of more effective regulation are also the most likely to hold developed countries responsible for the we have, more likely to want to see a greater influence exerted on the process of by their own government, and the most likely to see as necessitating regional political and economic cooperation (in the form of European integration). Finally, factor 7 would seem to equate closely with attitudes towards a neoliberal view of the political consequences of. In this respect it is very similar to the second factor identified in the UK analysis. But what is of course interesting is that, in the Irish case, it accounts for a much smaller proportion of the total variance and is the final factor to be identified. As for the UK data, it is important not simply to leave the analysis at the point at which readily interpretable factors have been extracted from the combined Irish data. For this may well serve to hide significant differences between civil servants and politicians. As before, these may be of two kinds differences in the factor scores of these two sub-groups with respect to the factors already identified, and more fundamental differences in the internal structuring of the discourse between political actors of different types. We consider each in turn. Table 8 looks for differences in mean factor scores for the seven factors identified in the combined analysis of the Irish data. Mean factor score TDs Civil service significance p <.005 p <.005 p <