Putting the Economy First: or Does Postmodernization Really Matter?

POLITICAL STUDIES: 2000 VOL 48, 38 50 Putting the Economy First: or Does Postmodernization Really Matter? Francis G. Castles Australian National University In recent years, there has been a strong tendency for the politicians of Western nations to make issues of economic policy making their first priority. Emerging evidence from the economics discipline, which shows a close association between subjective well-being and other macroeconomic variables such as unemployment and inflation, suggests that this priority ordering of politicians may well be one which is functional from an electoral viewpoint. However, recent research by Ronald Inglehart on the development of postmaterialist values challenges the continuing electoral relevance of an economics-first approach to policy making by suggesting that, in advanced societies, mass publics are exercised more by quality of life concerns than by issues of economic affluence. Here we use Eurobarometer data for 12 EU nations to explore the nature of the linkage between economic policy outcomes and levels of popular satisfaction in both economically advanced and less advanced societies. Our findings suggest that affluence is not the only economic determinant of subjective well-being in these countries and that, contrary to Inglehart s thesis, the growth of the economy impacts on subjective well-being in rich and poor nations alike. From Economic Outcomes to Political Consequences More and more, politicians and political commentators appear to put the health of the economy first in their priority ordering of political goals and policy tasks. Bill Clinton s 1992 campaign watchword It s the economy, stupid! states the case most simply, but others have been scarcely less single-minded in concentrating the contemporary political spotlight on the economic fundamentals. The dominant policy paradigm of the late twentieth century neoliberalism, which has had so strong a role in transforming Western societies in the past two decades, sees the exclusive task of government as the creation of a society maximizing economic efficiency. More pragmatically, but with no lesser impact, progress towards European monetary union has made low inflation and debt-reduction the overriding goals of public policy in virtually all EU countries for much of the nineties. With the emergence of a strong social democratic bloc in the EU in the last year or so, that emphasis may be about to change, but it seems more likely to shift to another economic policy objective reducing unemployment than to replace an economic with a non-economic focus. Although the achievement of economic policy goals sometimes causes short-term political pain, it seems reasonable to attribute politicians present economic policy zeal to a belief that, ultimately, those who deliver economic gains to their constituents will, in turn, be rewarded with political support. Such a calculus seems to rest on a rarely articulated but common assumption of a two-step causal process by which favourable economic outcomes translate into feelings of satisfaction or personal well-being on the part of mass electorates, which are then transformed Political Studies Association, 2000. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA

PUTTING THE ECONOMY FIRST 39 into political support for government and regime legitimacy. Arguably, that is why economic policy-making goals, unlike those in most other arenas of public policy, do not differ appreciably from country to country. 1 National policy makers seek to combine high economic growth with low unemployment and inflation because past failures to procure these goals have led politicians to believe that governments will be punished and regimes will become fragile where they are ignored. A two-step process from economic outcomes to subjective well-being and on to political support is not merely the underlying assumption of the policy behaviour of politicians. It is also implicit in much political science and public policy analysis. At a theoretical level, it may be seen as the guiding principle of work in the public choice tradition. At an empirical level, it has informed research on electoral behaviour and the preconditions for political stability. While studies of economic voting do not necessarily distinguish affective states from resulting political support, their focus is on the end points of the process and their objective has been to show the way in which economic outcomes influence the probability of the re-election of incumbent parties and candidates. 2 On the basis of similar assumptions, studies of democratic stability, and of regime legitimacy more generally, have sought to demonstrate the role of economic performance in underpinning continued support for the political system. 3 Inglehart s most recent analysis of changing belief systems in advanced industrial societies in his new book, Modernization and Postmodernization, 4 provides a rare instance where the location of the two-step process is quite explicit. Looking at the full range of data for the 43 countries covered by the 1990 World Values study, both stages in the two-step process leading from economic development to democratic regime stability are separately identified and empirically verified. His analysis shows a strong relationship (r = 0.74) between economic development (1991 GDP per capita) and a measure of subjective well-being (an index summarizing national scores on questionnaire items asking respondents to assess their personal happiness and life satisfaction). 5 The relationship between well-being and the longevity of democratic institutions is shown to be still stronger (r = 0.82). 6 However, Inglehart s main interest in this, as in all his previous work, 7 is on what is happening to attitudes in the more economically advanced nations, with the guiding hypothesis being that increasing economic growth together with the security offered by the modern welfare state underpins a shift from materialist to non-materialist or postmodern values. His articulation of the separate stages in a two-step process is essentially a preliminary to the argument that the first stage from economic outcomes to subjective well-being is a contingent one for certain types of societies. In postmodern societies, economic survival has ceased to be the overriding goal and, for many, non-material considerations relating to quality of life have supplanted traditional economic concerns. Thus, the basis of subjective well-being changes as nations become more prosperous. As Inglehart himself puts it: Economic security is still something that everyone wants, but it is no longer a synonym for happiness. 8 Supporting this interpretation, Inglehart notes that, in those countries in his sample with real GDP levels above $6,000 per capita, there ceases to be any clear relationship between economic development and subjective well-being. He sees

40 FRANCIS G. CASTLES this finding as compatible with a hypothesized decline in the subjective returns to economic growth and with the emergence of non-economic concerns as factors progressively shaping individuals feelings of well-being and satisfaction. He concedes, however, that this interpretation is to some degree speculative, given a lack of data resulting from the fact that subjective well-being only began to be measured a few decades ago, and until recently it was measured only in a handful of rich Western societies. 9 The possible contingency of the relationship between economic performance and subjective well-being is potentially of great significance both for political science and policy analysis and for political behaviour. Much political science is political economy in disguise and rests on the assumption that what happens in the economy has important implications for both electoral support and regime legitimacy. Most governments appear to believe that they will be rewarded with political support if they deliver on a standard range of economic promises. If Inglehart is correct, then these assumptions are progressively becoming more and more out of touch with reality. His surveys of the early 1970s suggested that, at that time, materialists outnumbered postmaterialists 4 : 1. He argues now that, by the end of the century, the numbers will be equally balanced in many Western countries. 10 If the subjective well-being of the majority of electors is based on considerations other than economic performance, governments will need to reshape their public policy priorities and transform their electoral agendas on a major scale. Evidence Linking Economic Performance and Well-being Inglehart s argument that there is insufficient data to examine the relationship between economic performance and well-being over time is only partly true. The centrepiece of this paper is an attempt to use data on life satisfaction produced by the Eurobarometer studies of attitudes in the European Union (EU) states to explore linkages between measures of economic performance and perceptions of subjective well-being. This Eurobarometer research is largely what Inglehart has in mind when he refers to the fact that data is restricted to a handful of rich Western societies. For the nine member states that acceded to the Treaty of Rome in the years up to 1973, the time-series begins in 1975. All of these nations bar Ireland can legitimately be regarded as falling within the category of highly advanced post-industrial nations. However, the three new member states admitted to the Community in 1985 the new democracies of Southern Europe were markedly poorer than all existing members other than Ireland. The analysis in the following section uses data for 1986 1993, a period when contrasts within the EC in terms of economic levels were at their greatest, to assess Inglehart s proposition that the subjective perceptions of mass publics in postmodern societies are no longer shaped by the same economic survival goals that motivated poorer and less secure societies. The only well-being findings which go back much further than the Eurobarometer research are US social surveys and poll data asking respondents to categorize their degree of happiness at various time-points since 1946. Easterlin, in various assessments of US time-series, has argued that there is little sign of correspondence between increasing income levels and subjective perceptions of well-being. 11 Other

PUTTING THE ECONOMY FIRST 41 commentators, using data from the US General Social Survey (GSS), and controlling for demographic and compositional changes in the US economy, have noted an increase in happiness over time, but one of very modest dimensions. 12 Since the USA s post-war economic pre-eminence suggests that it is amongst the countries in which we might most expect considerations of economic survival to have been supplanted by quality of life issues, this evidence of a weak or absent link between affluence and happiness does nothing to undermine Inglehart s thesis. More problematic, however, is Inglehart s failure to consider the influence on subjective well-being of aspects of economic performance other than economic affluence. His implicit model is extraordinarily simplistic, without any acknowledgement that factors such as widespread unemployment, high levels of inflation, alterations in the distribution of income and the compensatory actions of the welfare state can all affect the life-chances and, hence, potentially at least, the wellbeing of large proportions of the population. Although only a metaphor, and with no explicit intention of modelling the sources of subjective well-being, economists in recent decades have elaborated the concept of the misery index, which is sum of the inflation and unemployment rates. The very designation of this concept underscores the point that Inglehart neglects: that, for any given level of real per capita income, well-being may vary quite markedly dependent on the presence or absence of a variety of economic ills. The literature in this area has hitherto been substantially underdeveloped, with research input a natural victim of a situation in which the dependent variable has fallen within the domain of psychology and the presumed independent variables have been within the province of economics. 13 In the past, there has been some interest in the link between suicide and/or attempted suicide and levels of unemployment, 14 but economists have only just begun to pay explicit attention to subjective well-being measures. The most recent and econometrically most sophisticated contribution to the literature has used time-series data from both the GSS and the Eurobarometer over a period of approximately two decades from the early 1970s to the early 1990s to test for the effects of economic performance variables after controlling for personal characteristics, year dummies and country fixed effects. 15 The findings of this research are not readily compatible with the thrust of the postmaterialism thesis, since a series of microeconomic models of both perceived happiness and life satisfaction demonstrate that the unemployment rate is strongly negatively associated with subjective well-being in the USA, in Europe and in the individual nations of the European Community. Pooled time-series regression using Eurobarometer data also point to a very marked negative association between life satisfaction and inflation and more modest positive ones with both the level of affluence and the replacement rate of unemployment benefits. If mass publics in Western democracies base their support of governments and their views on the legitimacy of political regimes on their perceptions of subjective well-being, as presupposed in the second step of the process linking economic outcomes and political consequences, the evidence offered by this study suggests that materialist motivations are likely to be continuingly relevant in even the most affluent of nations.

42 FRANCIS G. CASTLES Some Methodological Considerations The research which follows also relies on Eurobarometer data. However, the methodology we employ is different from that used in the Di Tella study and, indeed, different from that employed in most research on the link between subjective well-being and public opinion. Such studies generally use large scale sample surveys to determine the population characteristics associated with particular outcomes, allowing the researcher to match up variation in the factors impinging on individuals with the behaviour and opinions they manifest. Inglehart, however, does not utilize such a methodology. Rather his focus is on data aggregated at a national level. In this study we follow Inglehart s lead. National level research has costs in terms of reducing the degrees of freedom available for statistical analysis, but is quite natural and, indeed, required where one or more of the variables under examination have a reality which is best understood or only understood at a national level. Obvious variables with such characteristics are political stability, which is an important focus of Inglehart s 1997 study, and inflation, which is one of the economic conditions we strongly suspect may influence cross-national differences in subjective well-being. Moreover, it seems quite probable that the link between the perception of economic conditions and well-being is no less socio-tropic in nature than the link between perceptions of economic conditions and voting behaviour. If that is so, the response of national communities to both adverse and positive economic conditions will not simply be the sum of the responses of those directly affected by those conditions in the manner assumed by studies which examine only individual behaviour, attitudes and opinion. It is arguable that, apart from the small-n problem resulting from a decision to focus on data aggregated at a national level, there are further methodological difficulties which result from linking macro variables, like national levels of affluence, inflation or unemployment with micro variables, such as levels of subjective wellbeing. In our view this is not necessarily the case where linkages are stipulated carefully and where the possibility of differences in findings at individual and aggregate levels is always borne in mind. If it is the case that individual well-being derives in some part from the satisfaction of individual s material needs (which is Inglehart s micro-level assumption), it would seem to follow axiomatically that macro variables standing unambiguously for higher aggregate levels of need satisfaction (more GDP, more jobs, less threat to living standards from inflation, etc) will be associated, all other things being equal, with higher aggregate levels of need satisfaction at a national level. This is what Inglehart tests for in the case of affluence, and our only quarrel with him and the point of this article is that the test he provides is not a sufficient basis for dismissing the possible impact of other sources of material need satisfaction. A final methodological point should be noted. The aim here is only to explore a specific aspect of Inglehart s argument. It is not to provide a complete model of the determinants of (aggregate level) subjective well-being in particular nations. Clearly, there are many other factors than economic ones likely to be associated with levels of well-being or life satisfaction. Some, like individual proneness to

PUTTING THE ECONOMY FIRST 43 psychiatric states, may not vary significantly in cross-national terms. Others, such as housing conditions, family structures and life events like divorce, clearly do differ markedly from country to country. 16 In order to ensure that factors of this kind do not unduly bias our subsequent findings, we use the lagged dependent variable (i.e. life satisfaction at time t 1) wherever possible in order to control for unknown sources of national difference. Some Comparative Explorations That there are very substantial differences in national levels of subjective wellbeing is clearly demonstrated by the data to be found in the first column of Table 1. The life satisfaction scores reported there denote the percentages of national respondents reporting that they are either very or fairly satisfied with the lives they lead at a given time-point. The data as presented here are averages for 12 EU countries over the 8-year period, 1986 93, and come from the Eurobarometer surveys published by the European Commission. 17 The number of Eurobarometer surveys conducted in each of these eight years was between two and three, so that annual life satisfaction scores are themselves averages. There is, of course, some variability in national relativities from survey to survey and year to year, but, on the whole, the rank ordering of the countries does not vary appreciably over the period in question. Data for six economic performance variables are also presented in Table 1. Real GDP per capita standardized on a purchasing power parity basis at constant 1985 Table 1: Average Life Satisfaction Scores in 12 European Union Nations, 1986 1993, and Average Lagged Values of Economic Performance Variables Country Average Average Average Average Average Average Average life RGDP change in annual rate of level of unemploysatisfaction per RGDP per inflation employ- ment score capita in per capita ment rate $US capita growth rate Denmark 95.1 13,655 206 1.5 3.6 76.9 7.4 Netherlands 92.6 12,741 255 2.0 1.5 58.8 8.7 Luxembourg 92.5 16,490 468 3.1 2.4 68.0 na Germany 87.2 13,062 360 2.8 2.1 64.1 7.3 UK 85.2 12,656 219 1.9 5.7 68.5 8.8 Belgium 84.3 13,074 289 2.2 2.6 55.8 10.4 Ireland 82.5 8,672 368 4.4 3.5 51.2 16.2 Spain 74.6 9,213 286 3.3 6.6 47.1 18.5 Italy 74.2 13,020 303 2.4 6.1 54.6 11.2 France 73.7 13,867 239 1.8 3.4 59.4 9.9 Portugal 67.5 7,203 271 4.0 12.1 67.2 6.0 Greece 59.2 7,535 105 1.4 17.7 54.3 7.6 Sources and definitions as in text.

44 FRANCIS G. CASTLES prices and annual change in real GDP per capita operationalized on the same basis are calculated from OECD, National Accounts. 18 All other variables come from OECD, Historical Statistics. 19 They include the annual growth rate of real GDP per capita, the inflation rate (i.e. annual change in the consumer price index), the level of employment (total employment as a percentage of the population from 15 to 64) and the unemployment rate (unemployment as a percentage of the total labour force). Given that our data for life satisfaction consist of survey data collected at different times over the year and then annualized, it is necessary to lag all the independent variables by one year to obviate the danger of reversed causation. Of the independent variables listed above, three require comment. Two change in real GDP per capita in absolute terms and economic growth measured more conventionally as the annual percentage change in per capita income can be dealt with together. Extraordinarily, Inglehart does not consider the possibility that subjective well-being might be in some part a response to short-term changes in economic conditions, although he does devote considerable time and attention to the alternative proposition that subjective well-being is amongst the factors determining the long-term economic growth rate. Yet, in commonsense terms, the former seems the most obvious link between happiness or life satisfaction and economic performance, with individuals responding positively to the realization that the conditions of economic survival are becoming easier and negatively when their income levels purchase less than previously. Whether individuals assess their economic wellbeing in absolute dollar terms or relative to existing standards is a matter which can only be established empirically, which is why we investigate the link between life satisfaction and both kinds of measure of the growth of the economy. The other variable requiring comment is the level of employment. The crucial point to note is that this variable is very far from being the simple obverse of the unemployment rate that the term sometimes suggests to the economically unversed. Leaving aside the differing extent to which national economies fail to fulfil the demands of those who seek work (i.e. unemployment), national economies also vary markedly in the extent to which they provide opportunities to participate in paid employment, with great differences in the extent of full-time higher secondary and tertiary education, in early retirement practices, in the availability of parttime jobs and in the availability of government employment. These last two factors strongly influence the level of female employment, which is probably the strongest single influence on the overall employment rate. 20 All of these differences seem a priori relevant to both the economic survival of individuals and to their subjective well-being. Looking at the data in Table 1, which reports average values across the period as a whole, certainly confirms Inglehart s point that there is no automatic correspondence between life satisfaction and economic affluence. Both France and Italy manifest far lower levels of subjective well-being than might be expected on the basis of their real GDP per capita, and the Netherlands exhibits a far higher level. Given that the four poorest countries are to be found in the bottom half of the well-being distribution, while disjunctions between well-being scores and affluence occur only amongst the richer nations, is also compatible with Inglehart s view that countries above a certain level of affluence are randomly distributed in life satisfaction terms.

PUTTING THE ECONOMY FIRST 45 However, these linkages between affluence and life satisfaction are, by no means, the only correspondences to be encountered in the data presented in Table 1. With the exception of Portugal, high employment levels tend to be associated with high satisfaction scores. Moreover, the story for inflation is, if anything, more clear-cut than that for affluence, with only France departing from the seeming rule that a low inflation level is associated with high satisfaction scores. The story for unemployment is less clear, although there is a clustering of high levels towards the bottom of the distribution. The only two variables which seem to be randomly associated with subjective well-being are those measuring short-term changes in economic conditions. This is scarcely surprising. Without controlling for the factors influencing the very substantial differences in subjective well-being revealed in the first column of Table 1, we would not expect to capture such effects. Later in our analysis, we use a simple pooled time-series design to investigate whether such effects are present. The second stage of our analysis moves beyond these rough and ready comments on possible correspondences in the data by seeking to locate whether there are statistically significant relationships between subjective well-being and various aspects of economic performance in this group of 12 countries. In Table 2, we Table 2: Correlation Coefficients between Life Satisfaction Scores and Lagged Values of Economic Performance Values in 12 European Union Nations Years RGDP per Change in Per capita Rate of Level of Unemploycapita in RGDP per growth inflation employment ment rate $US capita rate 1986 0.65** 0.70** 0.36 0.77*** 0.41 0.08 (0.36) (0.66*) (0.60) ( 0.67*) (0.54) ( 0.41) 1987 0.67** 0.56* 0.31 0.77*** 0.52* 0.18 (0.38) (0.45) (0.41) ( 0.55) (0.55) ( 0.55) 1988 0.72*** 0.21 0.39 0.77*** 0.47 0.20 (0.21) ( 0.53) ( 0.55) ( 0.59) (0.55) ( 0.70*) 1989 0.73*** 0.03 0.57* 0.79*** 0.49 0.05 (0.20) ( 0.41) ( 0.43) ( 0.34) (0.68*) ( 0.81**) 1990 0.67** 0.25 0.09 0.82*** 0.47 0.00 (0.23) (0.08) (0.02) ( 0.41) (0.68*) ( 0.62) 1991 0.71*** 0.51* 0.35 0.88*** 0.40 0.08 (0.20) (0.21) (0.18) ( 0.39) (0.61) ( 0.39) 1992 0.66** 0.13 0.06 0.83*** 0.57* 0.01 (0.22) (0.25) (0.23) ( 0.53) (0.68*) ( 0.54) 1993 0.71*** 0.28 0.20 0.85*** 0.55* 0.01 (0.16) (0.03) (0.00) ( 0.28) (0.73**) ( 0.67) Notes: Calculated from the data summarized in Table 1. Values in parentheses are correlation coefficients excluding the four poorest EU countries (Spain, Ireland, Greece and Portugal). Significance levels: * 0.10; ** 0.05; *** 0.01.

46 FRANCIS G. CASTLES present correlations between annual life satisfaction scores and the lagged values of each of our performance variables. Given that Inglehart fully concedes that economic well-being is a crucial determinant of well-being below a certain threshold value of real GDP per capita, we also provide correlations excluding the four poorest countries in the group. The very low number of cases for each of these calculations is an important reason for underlining the tentative character of the findings presented in Table 2. By the same token, it should be noted that where a correlation proves to be statistically significant, that is because the underlying relationship is, presumably, rather a strong one. Table 2 presents a picture which confirms aspects of both Inglehart s thesis and our own suppositions concerning the wider relevance of economic performance. Beginning with the latter, it is clear that economic affluence is not the only economic factor associated with subjective well-being in the full 12-nation sample. Affluence is significantly linked with life satisfaction throughout, but, in every single year, the negative association with inflation is still greater. Moreover, there are also signs of a strengthening positive relationship between life satisfaction and employment level. There are, however, no significant relationships between well-being and unemployment in the full sample and significant links between well-being and the growth of the economy are scattered rather than systematic. Moreover, the 1989 finding of a negative and significant relationship between life satisfaction and the economic growth rate goes completely against the whole thrust of the economic performance argument. Given the small numbers of cases with which we are dealing, however, it is probably more appropriate to base conclusions on patterns apparent over a number of years rather than on single aberrant findings. Inglehart s thesis is bolstered by the very clear signs that the most significant relationships identified in Table 2 result from threshold effects. The only significant result which remains for the findings concerning affluence and inflation when the poorer nations are moved from the sample is that relating to the negative impact of inflation in 1986. The disappearance of these very strong relationships is, at least, broadly compatible with the view that subjective well-being amongst the group of more advanced nations is less influenced by the factors which distinguish between the well-being levels of rich and poor nations. On the other hand, these findings cannot be read as indicating that the conditions of economic survival suddenly become irrelevant in the more advanced nations. A further point to emerge from Table 2 is that, with the exclusion of the poorer nations, issues of employment and unemployment become more important. In the case of both of these variables, the correlation coefficients for the more advanced group of nations are greater than for the wider sample, being significantly positive four times in the case of employment and twice in the case of unemployment. Given that we lack unemployment data for Luxembourg, a country with a high level of life satisfaction and an unquestionably tight labour market, it may well be that the latter relationship is understated. As already noted, these comparisons of gross differences in national well-being scores make it unlikely that we will be able to pick up the presence of short-term fluctuations caused by the growth of the economy. In Table 3, we seek to locate the presence of such effects by examining the influence of each economic performance

PUTTING THE ECONOMY FIRST 47 Table 3: Coefficients and Standard Errors for Relationships between Life Satisfaction and Lagged Values of Economic Performance Variables in 12 European Union Nations, 1986 1993 (controlling for life satisfaction at time T 1) Countries RGDP per Change in Per capita Rate of Level of Unemploycapita in RGDP per growth inflation employment ment rate $US capita rate 12 2.771E-4 0.007** 0.625** 0.301* 0.054 0.012 nations (2.047E-4) (0.002) (0.214) (0.125) (0.058) (0.117) Rich 8.429E-8 0.005* 0.623* 0.167 0.07 0.404 nations (3.145E-4) (0.002) (0.257) (0.213) (0.072) (0.3) Poor 1.277E-4 0.014* 1.062* 0.627* 0.009 0.262 nations (0.001) (0.005) (0.405) (0.274) (0.134) (0.21) Notes: Calculated from the data summarized in Table 1. Values in parentheses are standard errors. The poor nations are Spain, Ireland, Portugal and Greece. Except for unemployment, there are 96 cases for each of the 12 nation equations (88 in the case of unemployment). Except for unemployment, there are 64 cases in each of the rich nation equations (56 in the case of unemployment). All poor nation equations are based on 32 cases. Significance levels: * 0.05; ** 0.01. variable separately in pooled time-series regressions in which the lagged dependent variable serves as a control for such differences. Because this procedure increases the number of cases available for comparison, we present coefficients for rich and poor nations separately as well as for the 12-nation sample as a whole. Table 3 shows that, removing gross national differences from the picture, three economic performance factors influence changes in the level of life satisfaction. As expected, two of them are our measures for the growth of the economy. The findings for these variables are very similar. Irrespective of which measure is used, the growth of the economy is very strongly and positively associated with life satisfaction scores in the full 12-nation sample and only marginally less so in the two sub-samples. This finding is of considerable interest, since it provides unequivocal evidence of effects common to rich and poor nations. The third factor associated with outcomes is inflation, which is moderately and negatively linked to life satisfaction in both the full sample and in the poorer nations, but not in the richer grouping. The fact that inflation is the one variable which shows up as consistently important in both Tables 2 and 3 is presumably due to the fact that our measure of annual change in the consumer price index simultaneously picks up on gross differences in the price stability of different nations as well as annual fluctuations around distinct national profiles. Our explorations so far suggest that the major differences in life satisfaction as between countries are strongly influenced by such factors as affluence and inflation and that shorter-term fluctuations over time are, at least in part, a function of the impact of the growth of the economy. Table 4 seeks to complete the story by elaborating models including both types of variable simultaneously. Given the relatively small number of cases for such an

48 FRANCIS G. CASTLES Table 4: Economic Performance-based Models Accounting for Life Satisfaction Scores in 12 European Union Nations, 1986 1993 12 Rich Poor 12 Rich Poor nations nations nations nations nations nations Intercept 12.219 8.661 36.02 11.086 7.932 35.165 T 1 0.759*** 0.811*** 0.51*** 0.772*** 0.816*** 0.543*** (0.065) (0.071) (0.167) (0.066) (0.07) (0.168) Change in 0.006*** 0.005** 0.012** RGDP (0.002) (0.002) (0.005) Per capita 0.582*** 0.627** 0.885* growth (0.21) (0.259) (0.432) rate Inflation 0.326** 0.224 0.477 0.352*** 0.22 0.463 rate (0.131) (0.224) (0.282) (0.129) (0.224) (0.288) Level of 0.128** 0.116 0.014 0.134*** 0.119 0.014 employment (0.058) (0.075) (0.123) (0.058) (0.075) (0.128) Adj. R 2 0.883 0.851 0.751 0.878 0.852 0.746 Durbin 2.399 2.419 2.404 2.439 2.417 2.408 Watson Cases 96 64 32 96 64 32 Notes: Calculated from the data summarized in Table 1. Values in parentheses are standard errors. The rich nations are those in Table 1 apart from Spain, Ireland, Portugal and Greece. The poor nations are Spain, Ireland, Portugal and Greece. Significance levels: * 0.10; ** 0.05; *** 0.01. exercise, we drop real GDP per capita, the effect of which is negligible when we control for lagged life satisfaction, and unemployment, for which there is missing data for Luxembourg. In addition, because our two measures of the growth of the economy are strongly interrelated, we use them in alternative specifications of the models. As in Table 3, we also offer alternative specifications for rich and poor nations separately as well as for the 12-nation sample as a whole. What Table 4 adds to our account is evidence that the most important economic performance effects identified in both Tables 2 and 3 are simultaneously present in both 12-nation comparisons. For the European Union nations as a group, it would seem that inflation, employment and the growth of the economy are all factors making a substantial difference to levels of life satisfaction. Looking to the subsample models, the story is identical to that told by Table 3, with the disappearance of all structural effects, but strong evidence that growth of the economy influences subjective well-being in rich and poor nations alike. Of course, in terms of a critical appraisal of Inglehart s postmodernization thesis, what matters most is what happens in the advanced countries in which postmaterialist values have supposedly superseded economic survival as the basis of

PUTTING THE ECONOMY FIRST 49 subjective perceptions of well-being. Estimating the real impact of change in lagged real GDP per capita in the rich-country model in the second column of Table 4 suggests that, for every $200 increase in real GDP, a rich country obtains a one percentage point higher level of life satisfaction. This compares with an estimate of one percentage point of life satisfaction for every $83 derived from the poorcountry model in the third column of Table 4. One way to think about this is in the terms Inglehart himself uses of declining returns to economic growth. Another, less in tune with the thrust of his argument, would be simply to note that subjective well-being becomes more expensive as countries grow richer. Either way, Inglehart is surely incorrect in his basic supposition that, in the advanced nations of Western capitalism, money can no longer buy one happiness. Conclusion This paper has attempted to assess Inglehart s claim that economic performance is no longer the key to subjective well-being by exploring Eurobarometer data on life satisfaction for the period 1986 1993. Our findings suggest that Inglehart is correct in identifying a threshold value of economic affluence above which a country s level of economic resources ceases to be relevant to its level of subjective well-being. However, they also suggest that his analysis is mistaken on at least two major counts. First, it is by no means clear that affluence is the only, or even the main, economic performance variable relevant to understanding differences in well-being between rich and poor nations. For the group of EC countries under examination here, inflation appears to have been a more significant factor than economic affluence and there are also indications of an emerging employment effect. Second, even when a country s level of affluence has ceased to influence degrees of life satisfaction, the evidence presented here strongly suggests that fluctuations in the growth of economy continue to influence short-term changes in life satisfaction. It is important that the basis of this critique of Inglehart s argument is properly understood. We are not arguing against the propositions that postmaterialist values are increasing in Western mass publics or that, potentially, such values may contribute to perceptions of subjective well-being, although Inglehart nowhere demonstrates that this latter is actually the case. All that we are pointing out is that affluence is not the only economic performance variable that may be relevant to cross-national differences in well-being and that there is compelling evidence that growth of the economy influences the way in which individuals in rich countries assess the extent of their subjective well-being. We are not contending that the trend to postmodernization is a myth; only that the comparative evidence indicates that the process is by no means as complete as suggested in Inglehart s most recent work. On the basis of the analysis presented here, there seems little reason for contemporary political leaders to abandon their standard assumption that what happens in the economy is crucial for their political survival. A political agenda setting process that puts the economy first is likely to be with us for some time to come yet. It may be worth stressing, however, that there is no simple formula for running an economy or for maximizing the economic outcomes required to obtain political support. Our reading of the Eurobarometer data suggests that inflation, high levels of

50 FRANCIS G. CASTLES employment and economic growth all matter. Obviously, though, they matter to different groups of individuals within the economy in different ways and potentially involve complex patterns of policy trade-off. The recognition that the economy is and remains important to Western mass publics does not make it any simpler for governments to choose sensibly amongst the wide range of alternative policy options open to them. (Accepted: 21 December 1998) About the Author Francis G. Castles, Research School of the Social Sciences, The Australian National University, Canberra, ACT 0200, Australia. Notes The author wishes to record his intellectual debt to Professor Andrew Oswald of Warwick University, whose work on the macroeconomics of happiness was the inspiration for this paper. 1 The commonality of economic policy goals is argued in A. J. Heidenheimer, H. Heclo and C. T. Adams, Comparative Public Policy. New York: St Martin s Press, 1990, 3rd ed. 2 For a comparative treatment, see M. Lewis-Beck, Comparative economic voting: Britain, France, Germany and Italy. American Journal of Political Science, 30 (1986), 315 46. 3 The most famous account is, of course, that of S. M. Lipset, Some social requisites of democracy, American Political Science Review, 53 (1959), 69 105. 4 R. Inglehart, Modernization and Postmodernization. Princeton NJ: Princeton University Press, 1997. 5 Inglehart, Modernization and Postmodernization, p. 62. 6 Inglehart, Modernization and Postmodernization, p. 177. 7 See R. Inglehart, The Silent Revolution: Changing Values and Political Styles. Princeton NJ: Princeton University Press, 1977 and R. Inglehart, Culture Shift in Advanced Industrial Society. Princeton NJ: Princeton University Press, 1990. 8 Inglehart, Modernization and Postmodernization, p. 36. 9 Inglehart, Modernization and Postmodernization, p. 63. 10 Inglehart, Modernization and Postmodernization, p. 35. 11 See R. Easterlin, Does Economic Growth Improve the Human Lot? Some Empirical Evidence in P. A. David and M. W. Reder (eds), Nations and Households in Economic Growth: Essays in Honour of Moses Abramowitz. New York: Academic, 1974 and R. Easterlin, Will raising the income of all increase the happiness of all?, Journal of Economic Behaviour and Organization, 27 (1995), 35 48. 12 D. G. Blanchflower, A. J. Oswald and P. B. Warr, Well-being over time in Britain and the USA, paper presented at an Economics of Happiness Conference. London School of Economics, 1993. 13 For this argument, see A. J. Oswald, Happiness and economic performance, Economic Journal, 107 (1997), 1815 31. 14 See S. Platt and N Kreitman, Para-suicide and unemployment amongst men in Edinburgh 1968 82, Psychological Medicine, 291 (1985), 1563 6; and J. Charlton, S. Kelly, K. Dunnell, B. Evans, R. Jenkins and R. Wallis, Trends in suicide deaths in England and Wales, Population Trends, 69 (1992) parts I and II. 15 See R. Di Tella, R. J. MacCulloch and A. J. Oswald, The macroeconomics of happiness. Economics Department Working Paper, University of Warwick, 1997. 16 For an account of cross-national differences in these areas, see F. G. Castles, Comparative Public Policy: Patterns of Post-war Transformation. Cheltenham: Edward Elgar, 1998, pp. 248 93. 17 See Commission for the European Community, Eurobarometer. Brussels, various years. 18 OECD, National Accounts. Paris, 1997. 19 OECD, Historical Statistics. Paris, 1997. 20 See Gösta Esping-Andersen, The Three Worlds of Welfare Capitalism. Cambridge: Polity, 1990.