The public opinion–foreign policy paradox in Germany: integrating domestic and international levels of analysis conditionally

Abstract

Does public opinion influence foreign policy? International relations theory is divided on whether foreign policy outputs follow public opinion in advanced democratic countries. Using the case of cold war and post-cold war Germany, I offer an integrated realist theory of the effect of public opinion on foreign policy. I test the theory and the generalizability of the hypothesis of a public opinion–foreign policy nexus using process tracing as well as a time series analysis between the years 1973 and 2002. Using new measures, results here contradict literature on expected public opinion and policy outputs in the cold war period yet are supported after. I find that the predicted effect of public opinion on foreign policy outputs to be confounded by such factors as security threats.

Keywords:

• Germany
• foreign policy formation
• public opinion
• Western Europe

Notes

1. See Brooks (1990, p. 508–529). The UN Security Council resolution 1973 (2011) shows Germany continues to be a reliable foreign policy partner for the United States, though abstaining from the vote.

2. In 2002, the UN Security Council (SC) asserted that Iraq was in ‘material breach’ of its obligations under previous resolutions. The SC instructed the resumed inspections to begin within 45 days and warned Iraq that it would face ‘serious consequences’ as a result of continued violations. UN SC resolution 1441 passed unanimously. UN inspectors found that Iraq had not fully accepted its obligation to disarm, triggering the serious consequences required under this resolution. These serious consequences were seen to include the use of military force by the US. This interpretation was rejected by some SC members with vigorous support of the German Government of Gerhard Schroeder.

3. Support among various governments despite vast popular disapproval for the Iraq war policy demonstrates that public opinion and foreign policy are not always congruent. Spain's Partido Popular President Jose Maria Aznar, UK's Labor PM Tony Blair, US Republican President George Bush, et cetera, are examples.

4. See also Page and Barabas (2000, p. 340) who further cite Arrow (1951), Downs (1957), Davis and Hinich (1966), and Davis, Hinich and Osterhook (1970).

5. Studies focused specifically on Germany have addressed the ideological differences between East and West Germany and the generalizability of a Left/Right political spectrum (Shikano and Pappi 2004) or have considered the effect of exogenous variables such as international events or ‘rally round the flag’ effects (Frey and Schneider 1979, 1980, Rattinger 1981, Kirchgassner 1985) but have not directly looked at the effect of security threat on German foreign policy formation.

6. See Bouton and Page (2002) and Rielly (1991).

7. Powlick and Katz also develop a ‘followership model’ where he shows how elites influence public opinion (Powlick and Katz 1998).

8. Moravcsik (1993) builds on Putnam's model to show that skillful leaders can create symbiotic connections between domestic and foreign policy to enlarge their win-sets and strengthen the state's policy autonomy.

9. This is particularly true of neo-realists who argue all states perform the same function but differ only with regard to ‘capabilities’ (Waltz 1979, p. 93–97).

10. Ray attempts integration with regard to regime change and threats (2001).

11. SPD = Sozialdemokratische Partei Deutschlands. Installation of the Pershing II missiles by the US in the early 1980s in response to similar Soviet missile installation in Eastern Europe brought crisis. Schmidt decided to proceed with installation despite vast public opposition and within his own party.

12. Also, Spain's Partido Popular President Jose Maria Aznar, UK's Labor PM Tony Blair, US Republican President George Bush, and so on have all suffered electoral defeats individually or for their party to varying degrees due to public discontent with the Iraq war policy.

13. See Bouton and Page (2002) and Rielly (1991), in addition to (Almond 1960, Cohen 1973, Brooks 1990, Powlick 1995).

14. Especially since every state has the primary state interest or ‘function’ to survive (Waltz 1979).

15. See Lippmann (1922), Cohen (1973), Almond (1950, p. 53, 1960), and Holsti (1987, p. 23).

16. See Fearon (1994, 1997), Page and Shapiro (1983).

17. Rarely do unpopular actions incur immediate sanctions on behalf of the domestic audience resulting in the removal of power. However, this can occur, as was the case mentioned earlier with the Schmidt case in Germany. A more recent example would be the loss of power in Spain by the 2003 Partido Popular following a terrorist attack. However, this was made possible due to the timing of elections. Had elections been scheduled for a more distant time period, the fate of the Partido Popular may have been different.

18. Kagan (2003) also suggests differences between ‘Kantian’ European and ‘Hobbesian’ American views on war may be irreconcilable; see chapter 1.

19. See Michael Mandelbaum (2005), or a shortened version of this argument in his (2006), ‘David's Friend Goliath’, Foreign Policy, January/February.

20. Morrow suggests that in asymmetric alliances the more powerful state gains greater autonomy, while the weaker partner gains security and that there is a trade-off between autonomy and the degree/freedom to which a state can determine and maintain its own policy (Morrow 1991, p. 908, 914).

21. Weitsman expands on Walt's (1987) Balance of Threat (BOT) arguing that the relationship between threat and alliance formation presents four potential state strategies, adding hedge and tether to Walt's balance and bandwagon strategies (Weitsman 2004, p. 4, 20).

22. The theory fits within a framework defined by the grand theories of liberalism and realism and this implies certain positivist assumptions. Additionally, the before and after the cold war periods offer the opportunity for a quasi-experiment, and the rationalist approach is ideally suited to this type of analysis. I make no claims to theoretical or methodological exclusivity and recognize that alternate approaches, such as constructivist, interpretivist, or critical theory, may have advantages with regard to explaining some of the dynamics assessed here.

23. I create a separate measure for the security specific votes for the analysis (Model 2). Based loosely on Voeten's (2000) use of the six Main Committees of the United Nations for: (1) Disarmament and International Security (DISEC), (2) Economic and Financial (ECOFIN), (3) Social, Cultural, and Humanitarian (SOCHUM), (4) Special Political and Decolonization (SPECPOL), (5) Administrative and Budgetary, and (6) Legal; See Voeten (2000), Appendix). Because of the contentious nature of the Israel/Palestine issue, these votes are separated from the more general Security measure in order to prevent excessive influence from just one issue specific area.

24. I drop abstentions. Deducing the motivation of an abstention is subjective and there is no preferred method or precedent in the literature. Kegley and McGowan (1981) and Russett (1966, 1967) measure UNGA abstentions and absences as half the weighting of a congruent vote, while others such as Wittkopf (1973) and Rai (1980) measure UNGA abstentions as non congruence or a coding of zero. In any event, a robustness check, calculated with abstentions coded as zero/no votes, was conducted. The two variables were highly correlated (r=0.83) and regression results using the inclusive variable produced no substantive differences in the results.

25. I take the yes–yes/no–no for congruency and the yes–no/no–yes corresponding noncongruent cases of the total votes and the issue specific votes for the US and Germany and coded them 1 and 0 respectively.

26. Since the data for my IV begins with 1973, I am limited to this period for the dependent variable.

27. While the Soviet Union provides the alternate pole during the cold war, according to Voeten, a ‘counter hegemonic bloc’ made up of countries such as China, India, Iraq, Iran, Libya, and North Korea provides the alternate pole in the period after the cold war. See Voeten (2000, p. 186–187).

28. Due to the nature of the data, the Public Opinion variable is scaled so that higher values correspond to greater Left opinion. I multiply the data by negative one to invert the data. This creates a measure that corresponds better to the dependent variable (higher values = rightist policy) scaled so that higher values correspond to less pacifism or (more security policy preference) Right opinion. See Tables 1 and 2.

29. Brooks also makes similar assumptions (1990, p. 508–529); Stimson also uses a type of L–R ‘mood’ indicator by approximating public opinion by using an aggregate measure of the public's preference for liberal or conservative policy (1998). Stimson's w-calc methodology would be extremely difficult to adapt to German data so I use the readily available MRG/CMP data.

30. For example, SPD Chancellor Willi Brandt's Oestpolitik.

31. Recently, studies have addressed the ideological differences, particularly with regard to the positioning of radical parties on a L–R scale, between East and West Germany (Shikano and Pappi 2004).

32. Not all issues will have a specific contrasting opposite issue and parties will rarely oppose each other directly on an issue by issue basis. For example, the corresponding issue to ‘peace’ is not ‘war’ and a left-of-center ‘peace’ party will not likely be competing with a right-of-center ‘war’ party for voters, at least not effectively. However, a ‘liberty’ party might compete with the ‘peace’ party. The assumption that all other issues are subordinate to the primary issue of liberty makes war an unstated possibility but not the directly corresponding unit of measurement to peace (Budge et al. 2001, p. 81). This approach provides an alternate to Down's (1957) ‘vote maximizer’ model.

33. See Erikson et al. (2002), Stimson (1998), and Stimson et al. (1995).

34. Impute is a Stata 8 command that fills in missing values by organizing patterns from a list of variables on which the imputations are based, using a best-subset regression method.

35. Tests suggest the extrapolated data point measure is valid and reliable, that is, the original measure and the extrapolated measure are nearly perfectly correlated. Using an issue specific measure of military spending support (from Eichenberg and Stoll 2003) show a high correlation to Left–Right partisanship, suggesting the validity of issue salience.

36. Kosimo uses a qualitative definition of conflict and war. Wars and conflicts of lesser intensity are classified according to the actual amount of violence observed and not according to the number of fatalities. See www.kosimo.de for a detailed account of the data.

37. See Pfetsch and Rohloff (2000).

38. For a detailed account of the methodology please see http://www.hiik.de/methodik/index.html.en.

39. Militarized interstate disputes variable of limited conflict taken from the Correlates of War project, events data variable (using a weighted index created from COPDAB, Weiss, and Panda data.

40. See Hibbs (1977). This argument has proponents at least as far back as 1977 and Hibbs’ Political Parties and Macroeconomic Policy. Hibbs showed that partisan electoral cycles have effects on policy.

41. Prais–Winsten fits a linear regression of a dependent variable on IVs that are corrected for first-order serially correlated residuals using the Prais–Winsten (1954) transformed regression estimator using multiple iterations.

42. Gujarati (2003) states that this method allows for the analysis without adding lagged difference terms. Achen (2000) suggests lagged dependent variables (LDV's) can suppress the explanatory power of other IVs. Regression models that compensate for potential serial correlation with lagged variables were also tested; however, the lagged variables are not significant at the 95 per cent level so I believe there is no serial correlation problem in the models.

43. Preliminary tests indicate that Germany is not an outlier, and that Page and Shapiro should be generalizable cross-nationally. See Davis (2008).

44. Alternate ‘threat’ variables were tested using ICB, MIDs, Kosimo, and events data. Using a MIDs variable of limited conflict taken from the Correlates of War project resulted in –0.03 (0.019), while the events data variable, using a weighted index created from COPDAB, Weiss, and Panda data, resulted in –0.06 (0.02).

45. Results of a model using a variable of UNGA votes on Israel/Palestine showed these results to be robust.

46. For a very good discussion of the effect of German floods of 2002 on the election of that year, see Roberts (2003).

47. See Hibbs (1977).

48. The so called ‘Peace movement’ (Friedens Bewegung) is routinely considered to be a Left and far Leftist phenomenon. See Kamp (2003).

49. Oskar Lafontaine, a former SPD politician/candidate, headed the splinter ‘Left-party’ until January 2010.

50. See Schwarz (2000) for a left-partisan analysis of these events.

51. Examples include Brandt's Oestpolitik disagreement with the US and the ‘Guillaume Affaire’, Brandt's assistant discovered to be a paid informant to the East German [Ministerium für Staatssicherheit (Sicherheits Dienst; STASI)] secret police.

52. An analysis of the content of the 1984 UNGA voting in that year shows an unusually high concentration of restriction on nuclear testing and nuclear proliferation issues in which Germany consistently voted against the US, as did most UNGA members that year. Yet interestingly, the policy of Pershing II installation remained unaltered. Other studies have noted this phenomenon as well and find that the more aggressive US policy becomes, as with Reagan administration's perusing the installation of Pershing II, the more fearful European populations grew of potential conflict with the Soviet Union and consequently began to move leftward ideologically.

53. See Kamp (2003).

54. Public opinion data provides support for this public sentiment. See Budge et al. (2001).

55. The NATO Kosovo intervention in 1999 corresponds with an increase in the Opinion variable for that year, so the fit is not perfect.

56. The SPD Schmidt government suffered a constructive vote of no confidence that resulted in its loss to the CDU in 1982.

57. See footnote 47. As stated earlier preliminary tests indicate generalizability cross-nationally.