Technocracy, trust and democracy

During the eurozone crisis, technocratic governments have been appointed in several countries.Footnote ¹ These episodes cast a new light on the long-lived EU democratic deficit debate, traditionally focused on the legitimacy of EU-level policy. The subject of technocracy has been hitherto investigated mainly for its theoretical and conceptual relevance. A sizeable literature looks at the conditions that lead to the emergence of this form of government – concluding that it tends to occur during political or economic crises (Domínguez Reference Domínguez1997; Hallerberg and Wehner Reference Hallerberg and Wehner2013, Reference Hallerberg and Wehner2018; Neto and Strøm Reference Neto and Strøm2006; Pasquino and Valbruzzi Reference Pasquino and Valbruzzi2012; Pastorella and Wratil Reference Pastorella and Wratil2017). Others have looked at the technocratic governments appointed during the eurozone crisis as a symptom that national representative democracy has been weakened for the sake of preserving the single currency and economic integration (Matthijs Reference Matthijs2014; Sánchez-Cuenca Reference Sánchez-Cuenca2017). In search of an appropriate description of what countries in the so-called eurozone ‘South’ went through, scholars have resorted to concepts such as ‘democracy without choice’ (Alonso Reference Alonso2014), ‘democratic void’ (Mair Reference Mair2013), or even ‘authoritarian liberalism’ (Streeck Reference Streeck2015). Some have been looking into the erosion of trust in EU countries’ domestic democratic institutions, showing that economic factors matter (Armingeon and Guthmann Reference Armingeon and Guthmann2014; Foster and Frieden Reference Foster and Frieden2017; Norris Reference Norris2011). Others have been addressing the relation between technocracy and populism as two organizing poles of politics in contemporary Western democracies (Bickerton and Invernizzi Accetti Reference Bickerton and Invernizzi Accetti2015; Caramani Reference Caramani2017; Leonard Reference Leonard2011; Schmidt Reference Schmidt2006).

Hypotheses

Following Duncan McDonnell and Marco Valbruzzi (Reference McDonnell and Valbruzzi2014), the ‘ideal type’ technocratic government is one where: (1) all major governmental decisions are not made by elected party officials; (2) policy is not decided within parties which then act cohesively to enact it; and (3) the highest officials (ministers, prime ministers) are not recruited through party. An ideal type unelected technocratic government thus operates in the context of a system that remains democratic overall, but it is largely immune to the constraints that typically act on elected officials within an environment of democratic accountability. In fact, technocracy and democracy tend to be seen as incompatible in the literature (Fischer Reference Fischer1990; Shapiro Reference Shapiro2005), and especially so in the context of the eurozone crisis (Matthijs Reference Matthijs2017). The most common lines of criticism to technocracy are well exposed by Giulia Pastorella (Reference Pastorella2016), who concludes that while technocratic governments are not per se a threat to democracy, they meet the standards of party democracy only because those standards have been eroded over time. Technocratic governments are often portrayed as illegitimate and unaccountable (Skelton Reference Skelton2011) due to their ability to bypass regular deliberation (Wolkenstein Reference Wolkenstein2015) and because they constitute a break in the chain of delegation on which representative democracy relies (Neto and Strøm Reference Neto and Strøm2006; Riker Reference Riker1982). However, a few scholars have argued otherwise. Michael Schudson (Reference Schudson2006), for example, posits that technocracy can create positive externalities for democracy. Philip Pettit (Reference Pettit2004) claims that depoliticization may actually be instrumental to democratic deliberation, and Pierre Rosanvallon (Reference Rosanvallon2011) suggests that the impartiality typical of technocratic governments makes them legitimate. Building on these alternative views, our first hypothesis (H1) states that:

H1 implies that we can conceive of scenarios where the appointment of unelected technocrats to executive office increases (counter-intuitively) citizens’ satisfaction with the functioning of democracy. Why would that be the case? We think the explanation is to be found in a trade-off that the technocratic form of government establishes between input and output legitimacy (Schmidt Reference Schmidt2011, Reference Schmidt2013). In principle, one would expect any political change that increases a government's output effectiveness to trigger an improvement in its perceived legitimacy, and changes that reduce citizens’ input participation to produce the opposite effect. The appointment of a technocratic government – which, as the literature abundantly shows, tends to occur in crises – is an especially interesting phenomenon because it triggers both types of change at the same time.

In a crisis, the appointment of unelected technocrats trades off a promise of increased output effectiveness (through the signalling of non-partisan expertise) against an immediately reduced input participation. Therefore, we would expect perceived input legitimacy to decrease, and perceived output legitimacy to increase. The net effect is uncertain a priori, and it will depend on which of the two opposite effects dominates. If the output effect dominates, then the appointment of an unelected technocratic government may well occur without any deterioration in citizens’ satisfaction with the functioning of democracy. If perceived input legitimacy is low to start with – such as in the case of Italy – the output effect may more easily dominate, and satisfaction with democracy may increase. Let us formulate these claims in a way that we can test empirically. On the input side, we expect the appointment of a technocratic government to have a negative effect on perceived input legitimacy, because by sidestepping elections it reduces the say that voters have within the process of democratic accountability:

On the output side, we expect people to see a technocratic government of impartial experts as better placed to deliver a solution to the crisis that triggered its appointment, thus improving perceived output legitimacy:

If the crisis is of an economic nature, we therefore expect to observe an improvement in citizens’ expectations regarding the future economic situation, following the appointment.

A second aspect that we want to investigate is the effect of appointing a technocratic government on citizens’ trust in the government on one hand, and in representative institutions on the other. Cultural and political factors are obvious candidates to explain citizens’ trust in government (Inglehart and Norris Reference Inglehart and Norris2016; Levi and Stoker Reference Levi and Stoker2000; Norris Reference Norris and Norris1999), but research has also highlighted how trust is affected by objective economic factors (Armingeon and Ceka Reference Armingeon and Ceka2013; Armingeon and Guthmann Reference Armingeon and Guthmann2014; Foster and Frieden Reference Foster and Frieden2017; Guiso et al. Reference Guiso, Sapienza and Zingales2016) and by perceived economic or political performance (Williams Reference Williams1985). This points to an ‘output’ dimension of trust, which is especially relevant for the technocratic mode of governance. Eri Bertsou and Giulia Pastorella (Reference Bertsou and Pastorella2016) find that citizens are more inclined to support the technocratic mode of governance when they have weaker democratic attitudes and are distrustful of representative institutions. Their work is limited to the pre-crisis period and looks at trust as an independent variable. We believe further insights can be gained by investigating this question from the opposite angle. Consistent with our discussion in the previous section, we would expect the appointment of technocrats to be associated with an increase in people's trust in the government – which would provide robustness to our claim in H2.2.

The appointment of expert technocrats to executive office during economic crises could serve as a signal to voters that the normal political game is temporarily suspended in favour of a government of experts who – by means of their technical skills and political independence – can be in a better position to resolve the crisis. If this signal works, we would expect trust to increase:

Empirical validation of these hypotheses would suggest that the appointment of unelected technocratic executives could have counter-intuitive effects on citizens’ attitudes towards domestic institutions as well as on their satisfaction with the functioning of democracy. This does not imply that the effects will be permanent. Output legitimacy ultimately derives from the legitimacy of the outcomes – that is, if the policy measures enacted are effective in addressing the issues facing citizens. Because technocratic governments rely on the signalling effect of their expertise, their perceived legitimacy is linked to the actual output they are able to deliver. Failure – real or perceived – to resolve the crisis that triggered their appointment could cost them dearly in the longer term.

A natural experiment: the 2011 Italian crisis

We investigate the validity of our hypotheses in the context of the Italian economic and political crisis of November 2011, which resulted in the collapse of the fourth Berlusconi government and its replacement with a fully technocratic cabinet led by Mario Monti, economics professor and former European Commissioner.Footnote ² The crisis was precipitated by the intensification of market pressure on Italy during the summer of 2011 and preceded by a period of boiling political tensions. Disagreement between Prime Minister Berlusconi and his long-time ally Gianfranco Fini degenerated into an outright split within the majority. The judicial investigations involving the prime minister uncovered a sex scandal, which inflicted a serious blow to the credibility of the Italian government both at home and abroad. In the 2011 local elections, the party headed by Berlusconi (People of Freedom, POF) underwent a historic defeat in the stronghold of Milan, where the candidate of the centre left prevailed. The defeat made the Northern League – historically an ally of the POF – uncomfortable with the alliance.

These political events formed the background to the negotiation of the annual budget law, which became a cause of disagreement within the majority. Finance Minister Giulio Tremonti was in favour of austerity, but the dismal result in the regional elections made this stance unappealing to the rest of the coalition, and the budget presented in June 2011 showed signs of a political compromise in favour of delaying fiscal consolidation. The budget failed to reassure the markets, and during July the spread on Italian 10-year government bonds reached levels unseen since 1999. On 15 July both houses of parliament approved an austerity package, but discontent was growing within the POF and market pressure did not abate.Footnote ³ In August, after the spread had reached 390 basis points, the Italian government received a letter from the European Central Bank (ECB), asking Italy to ‘urgently underpin the standing of its sovereign signature and its commitment to fiscal sustainability and structural reforms’.Footnote ⁴ The government approved an additional austerity package worth EUR 45.5 bn (see Sensini Reference Sensini2011), and the ECB started buying Spanish and Italian government bonds on the secondary market. By the end of August, as the ECB's purchases had succeeded in temporarily lowering bond yields, Berlusconi announced a softening revision to the emergency austerity package (see Atkins et al. Reference Atkins, Sanderson and Segreti2011).

This backtrack on fiscal consolidation created confusion in the markets and, at the urging of the president of the republic, the emergency package was again reinforced in September. This notwithstanding, the rating agency Standard & Poor's downgraded Italian debt, stating that the ‘Italian government's tentative policy response to recent market pressures suggest[ed] continuing future political uncertainty about the means of addressing Italy's economic challenges’.Footnote ⁵ In November, the crisis precipitated after the Greek government announced its intention to call a referendum on the austerity measures requested by its creditors. This triggered a generalized increase in yields on the government debt of weaker countries, and the spread on Italian debt reached 441 basis points. On 4 November, at the G-20 Summit in Cannes, Italy came under pressure to accept EU/International Monetary Fund (IMF) surveillance (Wintour and Elliott Reference Wintour and Elliott2011; see Benvenuti Reference Benvenuti2017). Despite claiming that he would remain in power due to the absence of eminent public figures ‘able to represent the interests of the country internationally’, Berlusconi was under pressure to resign (Benvenuti Reference Benvenuti2017).

The political crisis reached its climax on 8 November, when the government was outvoted in parliament.Footnote ⁶ On the same evening, the prime minister announced his intention to resign once parliament approved the budget law. On 12 November Berlusconi resigned, and one day later the president of the republic conferred the task of forming a new government on Mario Monti. On 19 November Monti announced the composition of his fully technocratic cabinet (see Marangoni Reference Marangoni2012), which remained in charge until December 2012 – when the government lost support in parliament. New elections were held in February 2013. Monti, after lengthy hesitation, decided to run and was heavily defeated.

The case of the Monti government constitutes an especially good testbed to study the effects of technocratic appointments. Not only it was a full technocratic government, but in comparison to the other such cases identified by McDonnell and Valbruzzi (Reference McDonnell and Valbruzzi2014) it is the one that most closely matches the ‘ideal type’ definition given above. No major governmental decisions were made by elected party officials, policy was not decided within parties, and the highest officials were not recruited through parties. This differentiates the Monti government from, for example, the Papademos government which came to power in Greece at about the same time. While Monti's cabinet was entirely composed of non-partisan technocrats with no previous experience in politics, Lucas Papademos's cabinet had only three such members and the others were politically affiliated. The technocratic character of Papademos's cabinet was diluted by the fact that most members had served in the previous Papandreou government. From a conceptual standpoint, the Italian case allows us to investigate citizens’ attitudes towards technocracy in a paradigmatic setting.

Research design

We exploit the 2011 Italian crisis as a natural experiment. Experimental methods have increased in popularity in both political science and economics, due to their capacity to better isolate causal effects.Footnote ⁷ Our research design belongs specifically to the class that Jordi Muñoz et al. (Reference Muñoz, Falco-Gimeno and Hernandez2019) label unexpected events during survey designs (UESDs). UESDs exploit the occurrence of unforeseen events during the fieldwork of opinion surveys, to gauge causal effects. In a UESD, the sample is divided into a treatment and a control group by an event occurring at time t _e. Respondents interviewed before are the control group (T _i = 0), while respondents interviewed after are the treatment group (T _i = 1). Operationally, we can construct a treatment dummy as follows:

$$T_i = \left\{ {\matrix{ {0\; \; \; \; \; \; \; \; \; if\; \; \; \; \; t_i \lt t_e} \cr {} \cr {1\; \; \; \; \; \; \; \; \; if\; \; \; \; \; t_i \gt t_e} \cr}} \right.$$

where t _i is the time at which individual i is interviewed. The quantity of interest is the average treatment effect (ATE):

$$ATE = [E\lpar {Y_i\lpar 1 \rpar {\rm \vert} T_i = 1} \rpar -E(Y_i\lpar 0 \rpar \vert T_i = 1)]$$

This cannot be directly computed, because we never observe the same individuals in both states of the world. Under specific assumptions, however, we can estimate the ATE by comparing mean outcomes of treatment and control groups:

$$\lsqb {\; \overline {Y\lpar 1 \rpar } -\overline {Y\lpar 0 \rpar } \;} \rsqb \; \mathop \to \limits^p \; [E\lpar {Y_i\lpar 1 \rpar {\rm \vert} T_i = 1} \rpar -E(Y_i\lpar 0 \rpar \vert T_i = 1)] = ATE$$

Following Thad Dunning (Reference Dunning2012) and Muñoz et al. (Reference Muñoz, Falco-Gimeno and Hernandez2019), the validity of UESDs for estimating ATEs hinges on two assumptions, which we will discuss in the next sections.Footnote ⁸

Excludability

We use a cross-section wave of the Eurobarometer survey data that were collected in Italy during the days of the November 2011 crisis.Footnote ⁹ The survey includes a variable associating to each individual the exact date when their interview took place. The interview window was 6–20 November 2011, thus including the day when Monti was assigned the task of forming a government (13 November). We exploit the occurrence of this unexpected event during the fieldwork of the survey, to estimate the causal effect that the appointment of a technocratic government had on trust and satisfaction with democracy in Italy. In our design, the ‘treatment’ is the conferral of the task to form a government to Mario Monti. This event exogenously splits the sample into a control group (respondents interviewed before 13 November) and a treatment group (interviewed after). The daily distribution of people interviewed each day is shown in Table A1 in the Appendix: 52% of our sample was interviewed before, 48% after the event.Footnote ¹⁰

Excludability requires that any difference between respondents interviewed before and after the event is the sole consequence of the event. From an operational standpoint, this means that assignment to the treatment and control groups should be ‘as-if’ random. To test for the validity of this assumption, we ran balance tests for a variety of respondents’ features, across the two groups. The variables considered include location, demographic factors, socioeconomic features, as well as attitudinal aspects and proxy of economic status such as ownership of various durable goods. All these variables are coded numerically so we can statistically test whether the mean value of these characteristics is the same for respondents in the treatment and control groups.Footnote ¹¹ Our results (Table 1) show that we cannot reject the null hypothesis of equal means in any of the tests we ran except one (ownership of a DVD player, which is irrelevant in the context of our research question). This strongly suggests that the treatment and control groups are statistically equivalent, except for the dates of interview, and respondents can indeed be considered ‘as if’ they had been randomly assigned to the group to which they belong.

Table 1.

Balance Tests

A potential threat to excludability may come from violation of the non-interference condition, which requires treatment status for one unit to be uncorrelated with the outcome for other units. This assumption is very plausible in UESDs (Muñoz et al. Reference Muñoz, Falco-Gimeno and Hernandez2019). In our framework more specifically, all people interviewed after the announcement that Mario Monti had been tasked with forming a government had very likely been exposed to this news – which received immediate and major media coverage – whereas all people interviewed before the announcement had obviously not. One possible counter-argument is that respondents interviewed before the announcement could perhaps already foresee that a technocratic government would be appointed in the near future. We believe that this argument does not invalidate our design, for three reasons. First, while there had been rumours in the summer of a possible changeover involving Monti, technocratic governments are a rare occurrence – even in Italy, which has had more technocratic governments than most countries in Western Europe. The Italian Republic has been ruled by 65 governments in its 73 years of life, but political crises typically result in cabinet reshuffles, and before 2011 there had been only one case of a fully technocratic government, in 1994.Footnote ¹² Thus there is no reason to believe that before the announcement on 13 November, people would necessarily anticipate a technocratic government as the natural outcome of one of many political crises.

Second, while parallels were drawn in 2011 with the 1994 experience, there was a great deal of uncertainty as to whether a technocratic government was feasible in 2011. Until the very end, Prime Minister Berlusconi maintained that he could still command a parliamentary majority enabling him to govern (Benvenuti Reference Benvenuti2017). The former centre-left finance minister, Visco, reportedly said to the Financial Times that ‘the Monti solution would be excellent’, but ‘to have a Monti government, this one has to fail – and there's no one to make it fail’. Unlike 1994, when it had pulled the plug on the government, in 2011 the Northern League was weakened by internal power struggles and seemed to have little interest in bringing Berlusconi down. Despite tensions, the Northern League eventually supported the government during a confidence vote as late as the end of October 2011 (see Corriere della Sera 2011b). Most importantly, Monti himself had expressed scepticism during the summer about the idea that he could head a technocratic government.Footnote ¹³ Overall, we believe that the risk of respondents in the control group anticipating a technocratic government before 13 November is low. Even if there had been some anticipation, however, this would make it theoretically less likely that we would detect a difference in outcomes after the announcement, so it would actually strengthen – rather than weaken – our findings, if we found any statistically significant effect.

Empirical results

We estimate ATE via a regression adjustment where we specify the outcome as a probit model. To provide evidence in support of our H1, we need to show that the appointment of the Monti government is associated with an increase in satisfaction with democracy. Our results show that in fact there is a positive and statistically significant effect of the appointment on Italians’ satisfaction with the functioning of democracyFootnote ¹⁵ (Table 2, bottom).

Table 2.

ATEs: Expectations and Satisfaction

Table 2 also reports the estimated ATEs for respondents’ assessment of the current economic situation and for their economic expectations. Respondent in the treatment group appear to be slightly more pessimistic about the country's economic situation – possibly because witnessing a rare and extreme event such as the appointment of a fully technocratic government made them realize the gravity of the situation – but the effect is small and weak. Respondents in the treatment group are, however, significantly more optimistic in their expectations for future domestic economic conditions. This effect is very large – almost 10 percentage points – and strongly significant, lending support to the channel identified in H2.2.

Some Eurobarometer waves include a question on whether respondents feel their voice counts in their own country. This would have been the perfect variable to use as a direct measure of perceived input legitimacy at the domestic level. Unfortunately, this question is not asked consistently across the different Eurobarometer editions, and it was not asked in the November 2011 wave of the survey that we use. The closest substitute is the answers to a different question, which asks respondents whether they think their own voice counts in the EU. No statistically significant change is detected in respondents’ perception of this (Table 2). We are aware that this is an imperfect measure of perceived input legitimacy as we frame it above, because it does not relate to the domestic environment and it does not allow us to clearly validate or invalidate H2.1. We think it is nonetheless interesting to look at this indicator, in light of the prominent role that Europe played in the Italian crisis. The Italian crisis situates itself within the broader context of the eurozone crisis, and EU institutions such as the ECB played a direct and very visible role in the unfolding of the crisis at home. Relatedly, we find that the appointment of the technocratic government is associated with a large increase in respondents’ perception that Italy's interests are respected in the EU (Table 2). This is a strong signal that the switch from the Berlusconi to the Monti government – headed by a former EU Commissioner – led to an improvement in Italians’ perception of Italy's assertiveness at the EU level – likely through the positive effect the event had on the international standing and credibility of Italy. This also helps rationalize the increase in satisfaction with the functioning of democracy in the EU, which we detect among respondents in the treatment group. Overall, our estimates show that the appointment of the Monti government was associated with a strong improvement in perceived output effectiveness, without a deterioration in perceived input legitimacy (Figure 1).

Figure 1.

Effect of Monti's Appointment on Output and Input Factors

Table 3 reports estimated ATEs for trust in various domestic and EU institutions, and confirms that the appointment of the Monti cabinet had a positive effect on people's trust in the national government and – more weakly – parliament. The strongly significant increase in trust that we detect for political parties may appear prima facie puzzling, but it may be rationalized in light of the fact that political parties played a key role both in triggering the changeover through the vote of no-confidence to the Berlusconi government, and in making the technocratic solution possible through the support granted in the confirmatory confidence vote. Both votes occurred in parliament, but respondents seem to attribute (correctly) the outcome more to the independent choices of political parties than to the institution of parliament itself. We do not detect any change in trust for regional/local authorities, which is what we would expect given that these did not play any role in the 2011 political crises. Similarly, the third section of Table 3 reports ATEs estimated for placebos, i.e. institutions that we would not expect to undergo any change in trust due to the appointment of a technocratic government. These include the UN, the press, the radio, the TV and the internet.Footnote ¹⁶ As expected, no significant change in trust is found for these placebos, which lends robustness to the causal interpretation of the other effects we find.

Table 3.

ATEs: Domestic and EU Institutions, Placebos

The middle section of Table 3 reports estimated ATEs for trust in the EU and EU institutions. No statistically significant difference is detected for trust in the EU, but the increase in trust that we find for national institutions appears to ‘spill over’ to EU institutions as well. Why would trust in (supranational) EU institutions be affected by a domestic change in government? This may appear counter-intuitive at first, but it can be rationalized in light of two facts. The first is the already-mentioned prominent role played by EU institutions in the crisis. After bailouts were engineered for Greece, Ireland and Portugal, the prospect of Italy losing market access was associated with the collapse of the eurozone, because Italy was ‘too big to fail and too big to bail out’ (Jones Reference Jones2014; Summers Reference Summers2011). The strong involvement and concern of EU institutions in the Italian crisis became very clear at various points during the summer and autumn of 2011, in particular when the existence of the ECB letter mentioned above was made public (Benvenuti Reference Benvenuti2017). Because of the active and visible role that EU institutions played throughout the unfolding of the Italian crisis, it is thus not surprising that trust in these institutions was affected by the change in government that brought (at least temporary) relief. The second – and related – fact is the positive effect that the replacement of the Berlusconi government with a technocratic cabinet headed by a former European Commissioner had on people's perception of Italy's standing in the EU, as identified in Table 2.