Voters and the IMF: Experimental Evidence From European Crisis Countries

IMF interventions are often associated with rising political discontent in countries where the Fund intervenes. Studies examining this relationship, however, face the challenge of disentangling the impact of the IMF from the impact of the crisis that triggered the intervention. To address this challenge, we conduct survey experiments in Greece, Ireland, Portugal, and Spain and directly assess how voters evaluate the costs and benefits of an IMF intervention. We find that voters believe that the crisis will more likely be solved when the IMF intervenes, but they are also critical of the corresponding loss of national sovereignty. Because the former consideration, on average, dominates their assessment, IMF interventions increase the support of voters for unpopular economic policies. Nonetheless, cross-country differences suggest that continued public support for intervention hinges on the IMF’s ability to deliver on its promise to help resolve the crisis.


Design of Survey Experiment
Respondents were presented with a fictional scenario that varied in four randomly varying attributes: crisis severity, proposed spending cuts, main party in government/opposition, and IMF involvement.The main text of the vignette was presented over a series of screens to make it easier to absorb the information.The text in square brackets ([]) varies in the treatment levels listed, except for the country.The IMF treatment consists of whether Screen 4 is shown or not, and in whether respondents are reminded of the IMF involvement in Screen 5.
All treatments in Screen 2 vary together and describe the public debt situation.In Screen 3, the two government/opposition treatments vary together, as do the two spending cuts treatments.
The text of the vignettes is as follows: Screen 1: We are going to describe a situation [Greece/Ireland/Portugal/Spain] could face in the future, in 2026.Some parts of the description may seem important to you; other parts may seem unimportant.
Screen 2: Here is some background: In 2026, [Greece/Ireland/Portugal/Spain] is experiencing a [moderate/massive] increase in the level of public debt.The debt level is [somewhat/extremely] troubling.The government is finding it [fairly/very] difficult to borrow money on financial markets to pay for public spending.This presents a [somewhat/highly] urgent threat to the security of your savings and to the broader economic situation in [Greece/Ireland/Portugal/Spain].
Screen 3: Here is some more background: To address the economic concerns, the [Fianna Fáil/Fine Gael], [PASOK/New Democracy], [Partido Social Democrata/Partido Socialista], [Partido Popular/PSOE] Prime Minister announces in a televised speech that the government will implement [moderate/massive] spending cuts to limit the increase in public debt.These cuts will [slightly/deeply] affect funding for pensions, health care, education and public transport, for instance.The main opposition party, [Fine Gael/Fianna Fáil], [New Democracy/PASOK], [Partido Socialista/Partido Social Democrata], [PSOE/Partido Popular], criticizes the measures and doubts whether they will be successful.
Screen 4: Here is some additional background: The Prime Minister says that these spending cuts are necessary.This is because the International Monetary Fund (IMF) has made these cuts a precondition for [Greece/Ireland/ Portugal/Spain] to get an emergency loan that could stabilise the financial situation.The IMF is an international organization that provides emergency loans to countries in crisis, but only to governments who commit to carry out certain reforms.[Greece/Ireland/ Portugal/Spain] would not receive the IMF loan without the cuts.]Screen 5: Here is a summary of the situation again, for your reference: • COUNTRY has experienced a [moderate/massive] increase in the level of public debt.
• The Prime Minister is a member of [Fianna Fáil/Fine Gael] • The government will implement [moderate/massive] spending cuts to limit the increase in public debt.
• ONLY SHOWN TO TREATMENT GROUP: The Prime Minister says that these spending cuts represent the precondition for an emergency loan from the International Monetary Fund (IMF).
Outcome Questions Before answering the mediating questions, the respondents were again reminded about the content of the vignette (same as Screen 5).

Mediating Questions
• Competence: Do you think that, in this debt crisis, the government proved to be a competent or an incompetent manager of the economy?(Response categories: 'Completely incompetent' (0) to 'Completely competent' (10)) • Assessment of Policy Success: Do you think the government's decision to cut spending will be successful or unsuccessful in resolving the debt crisis?(Response categories: 'Completely unsuccessful' (0) to 'Completely successful' (10)) • Sovereignty: To what extent do you think that the government was free to choose its own response to this debt crisis?8Response categories: 'Entirely unfree' (0) to 'entirely free' (10)).
• Accountability: To what extent do you think that the decision to cut spending matches the views of the Irish [Spanish, Portuguese, Greek] voters?(Response categories: 'Not at all' (0) to 'fully' (10)).

IMF Conditionality Data
The figures presented in section 3.2 are based on the data collected by Kentikelenis et al. (?).For Spain, we coded the 'Memorandum of Understanding' (from July 20, 2012) using the same coding instructions as suggested by in their code book on p. 15.All the policy changes mandated in Spain were linked to reforms of the banking system, the financial infrastructure, and monetary policy issues.
For the ease of presentation, we cumulated some policy categories, specifically: the categories 'privatization' and 'state-owned enterprises' were put into one.The categories for 'fiscal issues' and revenue/taxation' were also combined.No policy reforms were categorized under the following dimensions: environment, external sector & trade, redistributive policies, and social policies.

Descriptives & Balance Test
These tables show that the treatment and control groups do not differ on key covariates (gender, age, ideology and education) and present the simple bivariate difference in the key outcome variable between treatment and control groups.

Vote choice
The effect on vote choice is more complicated to present as we need to examine treatments conditional on whether each respondent is generally a supporter of the government or not.
Recall that the government party varied randomly between the two largest parties in the system.Figure 1 shows the effect of IMF conditionality on whether respondents say they would vote for the government in the vignette.We can see that those who are generally supporters of the government party do not respond to the treatment.However, opposition supporters are more likely to vote for the government if the IMF demanded the policy programme.The treatment is statistically significant across the four countries and for Ireland, with substantively similar effects in Portugal and Spain.Smaller sample sizes mean statistical significance is harder to reach here.Interestingly, independent supporters (i.e.those who before the experiment do not voice support for either the government or the opposition) do not respond to the treatment.6 Additional Analyses

Effects of other treatments
Figure 2 shows that the other treatments have varying effects on policy approval.Recall that the effect of our main treatment of interest was about 0.3 overall.The size of the public debt -in other words, the severity of the crisis -has no impact on policy approval, though Portuguese respondents react somewhat negatively to this treatment.Spending cuts lead to lower policy approval, though this effect is not quite statistically significant at the 0.05 level.Interestingly, Portuguese respondents respond positively to cuts.Finally, the effects of which party is said to be in government depend on which party each respondent supports.Specifically, those who support the opposition are less likely to support the policy.

Additional Heterogeneous Treatment Effects
The following graphs shows the heterogeneous treatment effect by voters' perceived government competence in addressing economic issues.The index is based on the answers to the following questions: Please indicate how confident you are that the government is able to deliver the following policy outcomes in the near future: a) protect jobs, b) preserve economic wealth, c) foster economic prosperity, d) balance the public budget.
Respondents indicated their level of confidence on a scale from 1 (not at all confident) to 5 (very confident).The treatment effect is not clearly moderated by the perceived competence of the government in delivering positive economic outcomes.

Effect of other mediators on policy approval
As mentioned in section 3.1 (footnote no 6) we specified two additional mechanisms related to a) respondents' perception of government accountability, and b) respondents' perception of whether they expect the chosen austerity path to be successful.Figure 6 shows that the four mediators all have an impact on the outcome variable (approval).Respondents view policy packages more positively if they also believe that their government is competent and sovereign and that the policy package will be effective and represents voters well.
Figure 6: Effect of mediators on policy approval While all mediators have an effect on policy approval, the relevant mechanisms for two of them, competence and accountability, proved to be irrelevant.Figures 7 and 8 show the corresponding results.Possibly, the reference to a scenario in the future influences perceptions of government competence,   10 present the treatment effect of IMF intervention on the perceived effectiveness of the policy and the perceived sovereignty of the government, conditional on respondents' economic ideology.We can see that those with left-wing economic ideologies are more likely to think that IMF intervention will increase the effectiveness of a policy (though note that this interaction effect is not statistically significant), while the treatment effect on sovereignty is not (linearly) moderated by economic ideology.

Figure 1 :
Figure 1: Effect of IMF conditionality on vote choice

Figure 2 :Figure 3 :
Figure 2: Effect of other treatments on policy approval

Figure 4 :
Figure 4: Effect of IMF conditionality on approval, conditioned on size of adjustment

Figure 7 :
Figure 7: Treatment effects on additional mediators, by country

Figure 8 :
Figure 8: ACME for additional mechanisms, by country

Table 2 :
Descriptives of 'Approval'-Variable (key DV) The figures included in the main text of the paper are based on the following regression tables.Specifically, Figure ?? draws on table 3, and Figure ?? draws on table 4. The ACME visualized in Figure ?? are based on table 5.

Table 3 :
Regression Results: Approval of Policy