Skip to main content Accessibility help

Risk Inequality and the Polarized American Electorate

  • Philipp Rehm

Why has the American political landscape grown more partisan since the 1970s? This article provides a novel account of the determinants of partisanship. The author argues that partisanship is not only shaped by the traditionally suggested socio-economic factors, but also by the uncertainty of future income (risk exposure): rich individuals facing a high degree of risk exposure (or poor people facing low risk exposure) are ‘cross-pressured’; while their income suggests that they should identify with the Republicans, their income prospects make them sympathize with the Democrats. These two traits have overlapped increasingly since the 1970s. Those with lower incomes tend to be also those with higher risk exposure (risk inequality increased). This has led to a sorting of the American electorate: more citizens have become ‘natural’ partisans.

  • View HTML
    • Send article to Kindle

      To send this article to your Kindle, first ensure is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about sending to your Kindle.

      Note you can select to send to either the or variations. ‘’ emails are free but can only be sent to your device when it is connected to wi-fi. ‘’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

      Find out more about the Kindle Personal Document Service.

      Risk Inequality and the Polarized American Electorate
      Available formats
      Send article to Dropbox

      To send this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Dropbox.

      Risk Inequality and the Polarized American Electorate
      Available formats
      Send article to Google Drive

      To send this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Google Drive.

      Risk Inequality and the Polarized American Electorate
      Available formats
Hide All

1 Aldrich, John, Why Parties? The Origin and Transformation of Political Parties in America (Chicago: Chicago University Press, 1995); McCarty, Nolan, Poole, Keith T. and Rosenthal, Rosenthal, Polarized America: The Dance of Ideology and Unequal Riches (Boston, Mass.: MIT Press, 2006), henceforth abbreviated to MPR.

2 Bartels, Larry M., ‘Partisanship and Voting Behavior, 1952–1996’, American Journal of Political Science, 44 (2000), 3550; Jacobson, Gary C., ‘Party Polarization in National Politics: The Electoral Connection’, in Jon R. Bond and Richard Fleisher, eds, Polarized Politics: Congress and the President in a Partisan Era (Washington, D.C.: CQ Press, 2000), pp. 930; Jacobson, Gary C., ‘Partisan Polarization in Presidential Support: The Electoral Connection’, Congress and the Presidency, 30 (2003), 136; Fiorina, Morris P., Abrams, Samuel J. and Pope, Jeremy C., Culture War? The Myth of a Polarized America (New York: Pearson Longman, 2004).

3 Aldrich, John H., ‘Political Parties in a Critical Era’, American Politics Research, 27 (1999), 932.

4 A particularly rich source is Jacobson, ‘Partisan Polarization in Presidential Support’. For polarization at the mass level, see DiMaggio, Paul, Evans, John and Bryson, Bethany, ‘Have Americans’ Social Attitudes Become More Polarized?’ American Journal of Sociology, 102 (1996), 690755; Evans, John H., ‘Have Americans’ Attitudes Become More Polarized? An Update’, Social Science Quarterly, 84 (2003), 7190.

5 Poole, Keith T. and Rosenthal, Howard, Congress: A Political-Economic History of Roll Call Voting (Oxford: Oxford University Press, 1997); McCarty, Poole and Rosenthal, Polarized America.

6 Galston, William A. and Nivola, Pietro S., ‘Delineating the Problem’, in Pietro S. Nivola and David W. Brady, eds, Red and Blue Nation? Characteristics and Causes of America's Polarized Politics (Washington, D.C.: The Brookings Institution and The Hoover Institution, 2006), pp. 147, at p. 19. Reviews can be found in Ono, Keiko, ‘Electoral Origins of Partisan Polarization in Congress: Debunking the Myth’, Extensions (Fall 2005), 18; and Layman, Geoffrey C., Carsey, Thomas M. and Horowitz, Juliana Menasce, ‘Party Polarization in American Politics: Characteristics, Causes, and Consequences’, Annual Review of Political Science 9 (2006), 83110.

7 Brewer, Mark D., ‘The Rise of Partisanship and the Expansion of Partisan Conflict within the American Electorate’, Political Research Quarterly, 58 (2005), 219229.

8 Recent reviews are Fiorina, Morris P., ‘Voting Behavior’, in Dennis C. Mueller, ed., Perspectives on Public Choice: A Handbook (Cambridge, Mass.: Cambridge University Press, 1997), pp. 391414; Clarke, Harold D. and Stewart, Marianne C., ‘The Decline of Parties in the Minds of Citizens’, Annual Review of Political Science, 1 (1998), 357378; Fiorina, Morris P., ‘Parties and Partisanship: A 40-Year Retrospective’, Political Behavior, 24 (2002), 93115; Johnston, Richard, ‘Party Identification: Unmoved Mover or Sum of Preferences?’ Annual Review of Political Science, 9 (2006), 329351.

9 Fiorina, Morris P., ‘Economic Retrospective Voting in American National Elections: A Micro-Analysis’, American Journal of Political Science, 22 (1978), 426443; Fiorina, Morris P., Retrospective Voting in American National Elections (New Haven, Conn.: Yale University Press, 1981); Fiorina, Morris P., ‘Elections and the Economy in the 1980s: Short- and Long-Term Effects’, in Alberto Alesina and Geoffrey Carlina, eds, Politics and Economics in the Eighties (Chicago: Chicago University Press, 1991), pp. 1738.

10 McCarty, Poole and Rosenthal, Polarized America.

11 Varian, Hal R., ‘Redistributive Taxation as Social Insurance’, Journal of Public Economics, 14 (1980), 4968; Sinn, Hans-Werner, ‘A Theory of the Welfare State’, Scandinavian Journal of Economics, 97 (1995), 495526; Sinn, Hans-Werner, ‘Social Insurance, Incentives and Risk Taking’, International Tax and Public Finance, 3 (1996), 259280; Barr, Nicholas, The Welfare State as Piggy Bank (Oxford: Oxford University Press, 2001); Iversen, Torben and Soskice, David, ‘An Asset Theory of Social Policy Preferences’, American Political Science Review, 95 (2001), 875895; Moene, Karl O. and Wallerstein, Michael, ‘Inequality, Social Insurance, and Redistribution’, American Political Science Review, 95 (2001), 859874; Barr, Nicholas, The Economics of the Welfare State, 4th edn (Oxford: Oxford University Press, 2004). On the importance of social risk for politics, see the seminal contribution by Mares, Isabela, The Politics of Social Risk: Business and Welfare State Development (Cambridge, Mass.: Cambridge University Press, 2003).

12 Kreider, Brent, ‘Income Uncertainty and Optimal Redistribution’, Southern Economic Journal, 69 (2003), 718725. A more general model can be found in Drazen, Allan, Political Economy in Macroeconomics (Princeton, N.J.: Princeton University Press, 2000), pp. 315317. Prominent contributions making use of a similar logic include Piketty, Thomas, ‘Social Mobility and Redistributive Politics’, Quarterly Journal of Economics, 110 (1995), 551584; Alesina, Alberto, Glaeser, Edward and Sacerdote, Bruce, ‘Why Doesn't the United States Have a European-Style Welfare State?’ Brookings Papers on Economic Activity, 2 (2001), 187254; Bénabou, Roland and Ok, Efe A., ‘Social Mobility and the Demand for Redistribution: The POUM Hypothesis’, Quarterly Journal of Economics (May 2001), 447487.

13 There is an interesting connection to the economic voting literature. One could say that people’s egocentric (or pocket-book) calculus of voting (or partisanship) is both retrospective (current income) and prospective (income prospects). The analogy to this article’s argument would be that the expectation of being poor in the future makes one affiliate with the Democrats.

14 Thinking about social policy preferences as shaped by the two traits of income and risk exposure can be useful in studying a range of phenomena. Assume that people fall into one of four categories: (i) low income–high risk (ii) low income–low risk (iii) high income–low risk (iv) high income–high risk. Depending on which cell they inhabit, individuals should have a higher or lower demand for redistribution and/or insurance. Depending on the relative population of these cells (which will vary across policy domain, over time, and across countries), we should observe more or less encompassing support for social policies. In particular, ‘cross-class coalitions’ should be most likely in situations in which the high income–high risk cell (and perhaps the low income–high risk cell) is relatively large. In contrast, social policy should be a more contested issue when the two traits are reinforcing (only the low income–high risk and high income–low risk cells are populated).

15 This is obviously a radical simplification. Partisanship is a highly complex phenomenon and this article narrowly focuses on some aspects. The language of ‘natural’ independents and partisans is used for the sake of clarity. One immediate complication is that it is conceivable that some voters drop out of this scheme altogether: if individuals do not perceive a link between their policy preferences and the policy options the two parties offer, they may turn their backs on politics altogether (or perhaps classify themselves as ‘Independents’). Conceptually, individuals with extreme policy preference should be most akin to political alienation of that sort (i.e. the ‘natural’ Democrats or the ‘natural’ Republicans). This has two implications. First, we can think of the American electorate as consisting of four groups: ‘natural’ Democrats, ‘natural’ Independents, ‘natural’ Republicans, and the ‘alienated’. Secondly, the group of Independents can be expected to be quite heterogeneous since it may be made up of ‘natural’ Independents as well as individuals with such extreme policy preferences that they do not seem to be served by the two major parties. I will return to this point below.

16 Allan Davis, James and Smith, Tom W., General Social Surveys, 1972–2006 (Machine-Readable Data File, 2nd Release October) (Chicago, Ill.; Storrs, Conn.: Principal Investigator, James A. Davis; Director and Co-Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V. Marsden; Sponsored by National Science Foundation. – NORC ed.– Chicago: National Opinion Research Center (producer); Storrs, Conn.: The Roper Center for Public Opinion Research, University of Connecticut (distributor), 2008). I am using the GSS and not the National Election Studies (NES) for reasons that become clear below. I will discuss robustness checks at the end of this subsection. More details on the data can be found in the Appendix.

17 Respondents were asked the following question(s): ‘Generally speaking, do you usually think of yourself as a Republican, Democrat, Independent, or what?’ If they answered ‘Republican’ or ‘Democrat’, they were asked: ‘Would you call yourself a strong (Republican/Democrat) or not very strong (Republican/Democrat)?’ If, instead, the respondent answered ‘independent’, ‘no preference’, or ‘other’, the follow-up question reads: ‘Do you think of yourself as closer to the Republican or Democratic Party?’ to which they could answer ‘Republican’, ‘Democratic’ or ‘Neither’. The answers to these questions can be coded into a partisanship variable, classifying people into one of the following seven categories (otherwise, they are assigned a missing value and dropped from the analyses): ‘Strong Democrat’, ‘Not very strong Democrat’, ‘Independent, close to Democrat’, ‘Independent’, ‘Independent, close to Republican’, ‘Not very strong Republican’ and ‘Strong Republican.’ Despite some minor coding issues, this variable is comparable across time; see ‘GSS Methodological Report 56’ (

18 More precisely, the binary variable (employed in Figure 1 and the logit estimations below) equals one for answer categories 5 to 7 (and zero for categories 1 to 4) from the following survey item: ‘Some people think that the government in Washington ought to reduce income differences between the rich and the poor, perhaps by raising the taxes of wealthy families or by giving income assistance to the poor. Others think that the government should not concern itself with reducing this income difference between the rich and the poor. Here is a card with a scale from 1 to 7. Think of a score of [7] as meaning that the government ought to reduce the income differences between the rich and the poor, and a score of [1] meaning that the government should not concern itself with reducing income differences. What score between 1 and 7 comes closest to the way you feel?’ [Variable eqwlth in the GSS, reversed].

19 These have been suggested in Rehm, Philipp, ‘Citizen Support for the Welfare State: Determinants of Preferences for Income Redistribution’, in Discussion Paper SP II 2005–02, Wissenschaftszentrum Berlin [] (2005). Applications include Cusack, Thomas, Iversen, Torben and Rehm, Philipp, ‘Risks at Work: The Demand and Supply Sides of Government Redistribution’, Oxford Review Economic Policy, 22 (2006), 365389; Rehm, Philipp, ‘Ballot Boxing: Partisan Politics and Labor Market Risks’, in Katherine Newman, ed., Laid Off, Laid Low: Political and Economic Consequences of Employment Insecurity (New York: Columbia University Press, 2008), pp. 108127; Rehm, Philipp, ‘Risks and Redistribution: An Individual-Level Analysis’, Comparative Political Studies, 42 (2009), 855881; Rehm, Philipp, ‘Social Policy by Popular Demand’, World Politics (forthcoming).

20 It goes without saying that these occupational unemployment rates could be further refined. For example, it would be interesting to have detailed estimates of unemployment duration, or regionally specific unemployment rates. But this runs into data limitation problems.

21 It is therefore critical to have a public opinion dataset that includes detailed occupational information about respondents. The NES only includes detailed occupational variables from 1984 onwards. Therefore, the article relies on the GSS which includes detailed occupational variables from 1972 onward.

22 McCarty, Poole and Rosenthal, Polarized America.

23 See, for example, Edlund, Lena and Pande, Rohini, ‘Why Have Women Become Left-Wing? The Political Gender Gap and the Decline in Marriage’, Quarterly Journal of Economics, 117 (2002), 917961.

24 The substantive effects for the other variables are (simulated changes in parentheses): Redistribution (see Table 1, Model 2): age (min→max): −0.12; female (0→1): 0.053; some college (0→1): −0.007; college (0→1): −0.029; African-American (0→1): 0.13; South and non-black (0→1): −0.047; frequent church attendance (0→1): −0.037. Partisanship (see Table 1, Model 5): age (min→max): 0.005; female (0→1): −0.049; some college (0→1): 0.019; college (0→1): 0.0202; African-American (0→1): −0.296; South and non-black (0→1): 0.022; frequent church attendance (0→1): 0.086.

25 The predicted probabilities to be in favour of redistribution and to affiliate with the Republican party are, respectively: ‘Natural’ Democrats: Prob (Pro redistribution): 0.55; Prob (Republican): 0.28; ‘Cross-pressured’/‘Natural’ Independents: Prob (Pro redistribution): 0.47; Prob (Republican): 0.36; ‘Natural’ Republican: Prob (Pro redistribution): 0.34; Prob (Republican): 0.47. The simulations are based on Models 2 and 5 of Table 1, and averaged over three groups (natural Democrats, cross-pressured voters, and natural Republicans). To arrive at these groups, both income and risk exposure are divided into tertiles (for each year). Those with lowest income and highest risk are coded as ‘natural’ Democrats; those with the highest income and lowest risk are coded as ‘natural’ Republicans, while all others are coded as ‘natural’ Independents. The results would be similar if the tertiles were calculated across all years, or if different cut-off points were used.

26 Numerous robustness checks were carried out. The results are robust to a wide range of changes, the following factors in particular: (i) Datasets: Results are very similar if NES data are used instead of the GSS (because of the absence of a detailed occupational variable for earlier years, the analysis can only be carried out on data from 1984 onwards). (ii) Dependent variables (DVs): To ease the presentation of substantive effects, the DVs are recoded into dummy variables. Estimates on the seven-category versions of the dependent variables, using ordered logit models, lead to similar results. Furthermore, the results also hold if vote choice in presidential elections is the dependent variable. Moreover, the results of the article remain basically unchanged if a stricter definition of Democrat or Republican is chosen. (iii) Measurement of risk exposure: Occupational unemployment rates are measured with a detailed classification. The results are similar if more aggregated or different classifications are used (such as the major occupational groups displayed in Table 2 or the ISCO88 classification at various levels). To avoid results that are driven by outliers, the occupational unemployment rate variable is top-coded at its 98th percentile. This makes no difference. More details, including the thorny issue of dealing with occupational classifications that change over time, can be found in the Appendix. (iv) Control variables: Robustness checks reveal that the results hold up if the regressions include an additional variable containing occupational wages or if other occupational concepts are included (such as class or prestige scores). (v) Income concept: Real family income is the income variable used in the micro-analysis. One could also follow MPR and employ relative income rather than absolute income, where relative income is simply the ratio of a respondent’s real family income (reported in the GSS survey) to the average income in a given year. The results do not depend on this choice. Alternatively, one could use wages at the occupational level, again leading to the same conclusions (unsurprisingly, occupational wages have less explanatory power than family income). (vi) Sample: I assign retired respondents the risk exposure level of their previous occupation. Alternatively, one could restrict the sample to those in the labour force; one could also assign a respondent with a missing value on occupation of her or his spouse’s occupation. Neither of these changes would change the results meaningfully.

27 I prefer to use the CPS over the GSS for the macro-analysis because the GSS is not sampled representatively with respect to the labour market indicators of interest in this article. For example, it is implausible that the GSS is representative with respect to the 380 or so occupations distinguished for the risk exposure variable.

28 These are all but one major group from the Standard Occupational Classification (SOC) (see As in the rest of the analysis, I left out ‘Military Specific Occupations’ (major group 55). Within these 23 major groups, there are 96 minor groups, 449 broad occupations and 821 detailed occupations.

29 The years 1971 and 2002 span the longest possible time-range without a major change in the occupational classification. The economy-wide unemployment rate in these years is also very similar (about 5.8 per cent).

30 When using the 1971 income-risk distribution with 2002 weights, the correlation coefficient is −0.45, which suggests that the change in risk inequality is the result of both a change in the income-risk characteristics of occupations as well as a change in the size of occupations. Note that the correlation coefficients in Figure 5 differ (they are −0.32 (1971) and −0.47 (2002), respectively). The differences arise for three reasons. First, the results in Figure 5 are based on a more detailed occupational classification. Secondly, it uses mean (not median) wages (which makes no qualitative difference). Thirdly, the weights are employed and unemployed within each occupation, while above they are only the employed.

31 As detailed in the Appendix describing the CPS data, the occupational classification changes several times. One break in the series happens in 1983; even the standardized occupational classification used in this article cannot completely correct for that. However, it is possible to employ more detailed occupational classifications, such as the twenty-two ‘major groups’ listed in Table 2, or the twenty-seven ISCO88-2d groups. Replicating Figure 5 with these more aggregated data leads to a much smoother and even stronger pattern.

32 Occupations (at the major group level) are assigned into one of three combinations of risk exposure and income. Based on data from 1971, occupational groups are grouped in income noviles and occupational unemployment rate noviles. Adding these two variables up leads to a measure of income-risk combinations which ranges from 2 to 18 (highest income novile, lowest unemployment novile). Low income/high risk occupations are values 2 to 6; middle income/middle risk occupations are values 7 to 13; high income/low risk occupations are values 14 to 18. Employing the same definitions of ‘natural’ partisans and Independents as above as a measure of job quality would lead to similar patterns.

33 These developments are likely to be interrelated. How exogenous they are to politics is an interesting question for further research.

34 Job polarization has been studied by Autor, David H., Levy, Frank and Murnane, Richard J., ‘The Skill Content of Recent Technological Change: An Empirical Exploration’, Quarterly Journal of Economics, 118 (2003), 12791333; Wright, Erik Olin and Dwyer, Rachel, ‘The Patterns of Job Expansions in the United States, a Comparison of the 1960s and 1990s’, Socio-Economic Review, 1 (2003), 289325; Goos, Maarten and Manning, Alan, ‘Lousy and Lovely Jobs: The Rising Polarization of Work in Britain’, Review of Economics and Statistics, 89 (2007), 118133; Goos, Maarten, Manning, Alan and Salomons, Anna, ‘Job Polarization in Europe’, American Economic Review Papers and Proceedings, 99 (2009), 5863. To be sure, these authors focus only on income as a covariate of occupations, while I am interested in income and risk exposure.

35 Goos, , Manning, and Salomons, , ‘Job Polarization in Europe’, p. 58.

36 Although one could also argue that lower minimum wages may decrease unemployment rates in these jobs.

37 The figure is based on the CPS, but the GSS would lead to similar results.

38 Note that Figure 8 defines the ‘natural’ partisans based on tertiles that are calculated on a year-by-year basis. The tertiles could also be computed for the entire time period. The trends would be similar, but less smooth and with a stronger increase of ‘natural’ Rs compared to Ds.

39 It is not unproblematic to assess the mapping of ‘natural’ into ‘actual’ partisans. First, the language of ‘natural’ partisans is meant to communicate the main message (namely that the tails of the income-risk distribution became fatter; that we should and do find more partisans in these tails; and that therefore more citizens experience pure partisan traits) – the aim is not to maximize the statistical fit between ‘natural’ and ‘actual’ partisans. Secondly, the available data are not really appropriate for a thorough assessment (the GSS is better than the NES when it comes to occupational information; the reverse is true regarding partisanship measures; none is representative regarding income-risk distributions). Keeping these caveats in mind, the percentage of correctly predicted partisans is about 60 per cent, and fairly constant over time. The percentage of correctly predicted Republicans increased somewhat over time, while the percentage of correctly predicted Democrats decreased somewhat over time.

40 Non-voters constitute between 40 and 50 per cent of those classified as ‘natural’ Democrats; the proportion of ‘natural’ Republicans who do not vote is between 10 and 20 per cent.

41 Jacobson, ‘Partisan Polarization in Presidential Support’.

42 Hacker, Jacob S., The Great Risk Shift: The Assault on American Jobs, Families, Health Care, and Retirement (Oxford: Oxford University Press, 2006).

43 King, Miriam et al. , Integrated Public Use Microdata Series, Current Population Survey: Version 2.0. (Machine-Readable Database). (Minneapolis: Minnesota Population Center (producer and distributor), 2009).

44 Most directly, I follow Eckstein, Zvi and Nagypál, Éva, ‘The Evolution of U.S. Earnings Inequality: 1961–2002’, Federal Reserve Bank of Minneapolis Quarterly Review, 28 (2004), 1029. See their Appendix A.

45 Following Katz, Lawrence F. and Murphy, Kevin M., ‘Changes in Relative Wages, 1963–1987: Supply and Demand Factors’, Quarterly Journal of Economics, 107 (1992), 3578.

46 Meyer, Peter B. and Osborne, Anastasiya M., ‘Proposed Category System for 1960–2000 Census Occupations,’ BLS working paper, no. 383 (2005). I am very grateful to the authors for providing me with various syntax files.

47 Occupational unemployment rates can be calculated using different classifications, and at different levels of detail. The following classifications were employed for robustness tests: ILO’s ‘International Standard Classification of Occupations 1988’ (ISCO88) at the 1-, 2-, 3- and 4-digit level, distinguishing 9, 26, 98 and 238 different types of occupations, respectively; an aggregated version of the standardization due to Meyer and Osborne, distinguishing twenty-two different occupations; and the occupational classification employed in the NES, distinguishing seventy-one different occupations. Furthermore, the unemployment rates were calculated both in total and by gender. This partially builds on a ‘crosswalk’ constructed by Paul Lambert (see

48 The following survey items tap perceived job security: (1) Thinking about the next 12 months, how likely do you think it is that you will lose your job or be laid off – very likely, fairly likely, not too likely, or not at all likely? [JOBLOSE; N = 13,960] (2) (a) Would you please look at this card and tell me which one thing on this list you would most prefer in a job? (b) Which comes next? (c) Which is third most important? (d) Which is fourth most important? B. No danger of being fired [JOBSEC; N = 9,916] (3) Now, I'm going to read you another list of statements about your main job. For each, please tell me if the statement is very true, somewhat true, not too true, or not at all true with respect to the work you do. K. The job security is good. (Answer categories: very true; somewhat true; not too true; not at all true) [JOBSECOK; N = 1,658] (4) Please respond to the following statements based on your experience during the past 12 months unless otherwise specified, with reference to your current place of employment only. L. At work, job security is good. (Answer categories: very true; somewhat true; not too true; not at all true) [GDJOBSEC; N = 1,604] (5) For each statement about your main job below, please circle one code to show how much you agree or disagree that it applies to your job. A. My job is secure. (Answer categories: strongly agree; agree; neither agree nor disagree; disagree; strongly disagree) [RSECJOB = 1,480. (6) On the following list, there are various aspects of jobs. Please circle one number to show how important you personally consider it is in a job: A. Job security? (Answer categories: very important; important; neither important nor unimportant; not important; not important at all) [SECJOB; N = 1,543]. The samples were restricted to those in employment.

49 I broadly follow Goos, Manning and Salomons, ‘Job Polarization in Europe’. They, in turn, build on the following two contributions: Autor, Levy and Murnane, ‘The Skill Content of Recent Technological Change: An Empirical Exploration’; Autor, David H. and Dorn, David, ‘Inequality and Specialization: The Growth of Low-Skill Service Jobs in the United States’, (unpublished paper, 2007).

* Department of Political Science, The Ohio State University (email: ). Earlier versions of this article have been presented at various occasions. The author thanks participants of the ‘Research Workshop in Political Economy’ (Harvard University), the ‘Methods of Political Analysis’ class at Harvard University, a seminar at the Max Planck Institute for the Study of Societies, the Post-Doc as well as the Politics Seminar at Nuffield College, Oxford University, the ‘Permanent Seminar series’ at the Juan March Institute, and the Department of Government’s seminar series at the University of Essex. Comments from John Aldrich, Jim Alt, Michele Belot, Ray Duch, Geoff Evans, Ben Goodrich, Peter Hall, Herbert Kitschelt, Scott Moser, Katherine Newman, David Rueda, David Soskice, Jim Stimson, Vera Troeger and Inés Valdez are gratefully acknowledged. Several of the Journal’s reviewers have made excellent suggestions that have improved this article.

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

British Journal of Political Science
  • ISSN: 0007-1234
  • EISSN: 1469-2112
  • URL: /core/journals/british-journal-of-political-science
Please enter your name
Please enter a valid email address
Who would you like to send this to? *


Altmetric attention score

Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed