Literature review and hypothesis

The topic of political careers is linked to the traditional political science theme of political professionalization (e.g. Borchert, Reference Borchert, Borcher and Zeiss2003) and it has primarily focused – both with a country case or a comparative perspective – on parliamentarians (Norris, Reference Norris1997; Best and Cotta, Reference Best and Cotta2000; Siavelis and Morgenstern, Reference Siavelis and Morgenstern2008; Allen, Reference Allen2018; Ohmura et al., Reference Ohmura, Bailer, Meiβner and Selb2018). Notwithstanding from the 2000s onwards, an increasing interest in exploring also executive careers emerged (e.g. Dowding and Dumond, Reference Dowding and Dumond2009; Martocchia Diodati and Verzichelli, Reference Martocchia Diodati and Verzichelli2017; Müller-Rommel et al., Reference Müller-Rommel, Vercesi, Berz, Müller-Rommel, Vercesi and Berz2022), even with a gender perspective (e.g. Jalalzai, Reference Jalalzai2008; Escobar-Lemmon and Taylor-Robinson, Reference Escobar-Lemmon and Taylor-Robinson2009; Krook and O'Brien, Reference Krook and O'Brien2012; Beckwith, Reference Beckwith2015; Beer and Camp, Reference Beer and Camp2020), most works display a prominent national level perspective or a sort of nationalization bias. The challenge of considering a multilevel view as proposed by the seminal work of Borchert (Reference Borchert, Borcher and Zeiss2003, Reference Borchert2011) has been fruitful and it has provided several country case studies and certain comparative works (Teruel, Reference Teruel2011; Borchert and Stolz, Reference Borchert and Stolz2011a; Dodeigne, Reference Dodeigne2014, Reference Dodeigne and Dodeigne2020; Pilet et al., Reference Pilet, Tronconi, Onate, Verzichelli, Deschouwer and Depauw2014; Di Capua et al., Reference Di Capua, Pilotti, Mach and Lasseb2022) so far. Furthermore, it has also included studies on European politicians (e.g. Scarrow, Reference Scarrow1997; Daniel, Reference Daniel2015; Salvati, Reference Salvati2016; Aldrich, Reference Aldrich2018) and it has also focused on the careers of national ministers with a multilevel perspective (see the recent special issue edited by Semenova and Dowding, Reference Semenova and Dowding2023) or with a specific focus on the territorial links of their previous (Tronconi and Verzichelli, Reference Tronconi and Verzichelli2023) or subsequent careers (Claveria and Verge, Reference Claveria and Verge2015). However, regional executive research remains a limited field so far (see for exception: Botella et al., Reference Botella, Teruel, Barbera and Barrio2010; Jäckle, Reference Jäckle2013, Reference Jäckle2023; Stolz and Fischer, Reference Stolz and Fischer2014; Grimaldi and Vercesi, Reference Grimaldi and Vercesi2018; Astudillo and Martínez-Cantó, Reference Astudillo and Martínez-Cantó2020; Semenova, Reference Semenova2022; Boldrini and Grimaldi, Reference Boldrini and Grimaldi2023a) and this paper may be placed in such niche branch of literature as it aims at understanding in which direction Italian RCMs (both presidents and ministers) move across different tiers of government once they conclude their experience at the regional level.

In Italy, the political role of regional executives has increased since the 1990s due to two relevant processes of institutional discontinuity: regionalization and presidentialization.

Despite some inconsistencies and limitations (Baldini and Baldi, Reference Baldini and Baldi2014), regionalization in Italy can be traced back to the mid-1990s when several reforms were put in place to empower the subnational level of government starting from the so-called Bassanini law on decentralization (l. 59/1997), the constitutional reform of 2001 which devolves new powers to the regions, and the subsequent law on fiscal federalism (l. 42/2009) which allows a certain degree of autonomous taxation (Massetti, Reference Massetti2012). As a consequence, the level of authority at the meso-level has increased according to the regional authority index (Hooghe et al., Reference Hooghe, Marks, Schakel, Niedzwiecki, Chapman-Osterkatz and Shair-Rosenfield2021) on average from 18.19 in 1995 to 25.56 in 2020.

On the other hand, this transformation coincided with the change of the institutional regional framework from a parliamentary to a presidential system (Musella, Reference Musella2009), despite even previously an electoral reform foreseeing the direct election of RPs (since 1999) by mimicking the direct election of majors introduced in 1993. As a result, regional (and even local) posts have become more attractive and appealing to political leaders because there is ‘more at stake’ than in the past and executive decisions are more autonomous and broader in scope. In such a context, not only the figure of the RPs was empowered (Grimaldi and Vercesi, Reference Grimaldi and Vercesi2018), but also that of the entire cabinet compared to that of the regional deputies (anonymized for review).

Such transformations made Italy an intriguing case study to explore if and to what extent the emergence of attractive subnational alternative offices has changed the traditional ambition toward an upward career path that has traditionally characterized the Italian political class (Tronconi and Verzichelli, Reference Tronconi, Verzichelli, Chiaramonte and De Sio2019). This topic is especially interesting also by considering two other political and institutional elements. On the one hand, the decentralization process has led to increasing stability of regional and local governments, especially in comparison to national cabinets (Tronconi and Verzichelli, Reference Tronconi and Verzichelli2023). On the other hand, recently the number of national posts in Parliament has decreased. The combination of such elements would likely lead an increasing number of politicians to prefer investing their efforts in a municipal or regional career rather than a national one in the future.

Our research questions can be synthesized as follows:

What factors are associated with upward movements (i.e. toward national offices) of regional cabinet ministers? Which ones to backward movements (i.e. toward local offices)? And which ones to exit from politics?

To address such questions, we combined context-oriented and actor-oriented approaches to political career research (Jahr and Edinger, Reference Jahr and Edinger2015; Vercesi, Reference Vercesi2018) taking into account both political–contextual factors (such as belonging to small or large parties; the party outsiderness, certain positions within the cabinet) and individual factors (such as gender, age, the length of previous political career at the local, regional, or national level).

Among political–contextual drivers, international literature (Semenova and Dowding, Reference Semenova and Dowding2023) suggested many factors that unfortunately cannot be used in single case studies because there is no (or not enough) variance, such as the electoral system or the national political system. Moreover, other drivers as party dominance and party system congruence are no longer applicable to the Italian case or are not fully convincing. In fact, in Italy, a dominant party across levels existed from 1970 to 1990 (DC) but later it disappeared, and its heirs got a handful of votes in comparison to new or refreshed parties. Furthermore, previous research pointed out that – contrary to international expectations – the unidirectional model prevailed before the 1990s (Grimaldi and Vercesi, Reference Grimaldi and Vercesi2018), and therefore movements upward the tiers of government were the most frequent. This is due to the postponed implementation of the ordinary statute regions that did not correspond to an actual decentralization, as Italy remained a unitary state until the mid-90s, making regional and local posts clearly less appealing than national ones.

Party system congruence is not fully convincing as a driver both for its definition and then as it partially overlaps with other drivers in political career research. Party system congruence, namely the situation when a sub-national party system reflects that of a national and federal one, is intended by Semenova and Dowding (Reference Semenova and Dowding2023) as the existence of the same parties across territorial levels. This means that congruent parties compete at multiple territorial levels otherwise they are defined as incongruent parties. However, differently to Semenova and Dowding (Reference Semenova and Dowding2023), we believe that party congruence implies not only the identification of the presence of the same parties across different territorial levels but – following Schakel (Reference Schakel2013) − the overall similarity in terms of votes between two party systems at different territorial levels. In particular, party system congruence measures how different electorates (e.g. national and regional) vote in different types of elections (e.g. national and regional) (see also Bolgherini et al., Reference Bolgherini, Grimaldi and Paparo2021). This aspect is important as we cannot use the findings of certain studies.Footnote ¹ As a consequence, we decide to disregard this misleading conceptualization of party congruence and focus on the idea of the presence of the same parties in different territorial competitions. In this regard, two points are crucial. On the one hand, the differentiation between parties that are present or not across territorial levels in regionalized countries resembles the most widespread distinction between national/federal and regional/local party lists (Massetti and Sandri, Reference Massetti and Sandri2013) as parties present in both types of elections are often – with only a few exceptions – national parties, whereas parties present only in one type of election are often regional/local parties which never enter national parliament.

On the other hand, the divide between national and regional parties partially overlaps with another contextual–political aspect that pertains to the organizational nature of parties, particularly their size. Indeed often, national parties are also the largest ones. In this respect, a breach of literature emphasizes how the size of the party may affect career trajectories. Large parties are generally those with better resources in terms of organization, money and especially available positions at the national level in comparison to small parties that have few resources, they struggle to be present across different tiers of government and have definitely fewer available positions to fill at the national level. In particular, small parties that fail to enter the national parliament show scarce organizational capacity and candidate appeal and favor the exiting of skilled political personnel (Dinas et al., Reference Dinas, Riera and Roussias2015).

Consequently, it is more likely that political personnel belonging to small parties exit from politics rather than move upwards toward a national position in comparison to politicians belonging to large parties.

However, it is also likely that politicians belonging to big parties (Jäckle, Reference Jäckle2023) may move backward to fill local positions once they have terminated their regional career rather than exiting from politics.

Consequently, we prefer to focus on party size and hence our first hypothesis is as follows:

Another interesting aspect pointed out by international literature is related to non-partisan ministers both as experts and outsiders. Certain national-level studies pointed out that non-partisan ministers have fewer chances to develop a relevant ministerial career, staying in the cabinet shorter than political ministers (Teruel and Mir, Reference Teruel, Mir, Costa Pinto, Cotta and Tavares de Almeida2018). In particular, they seem not able to maintain a position at the national level and also, they cannot move backward toward subnational positions. However, not all ‘non-partisan’ are ‘non-political’ ministers as both in France and Italy some of them used access to government as the first step in a political career (Bruerè and Gaxie, Reference Bruère, Gaxie, Costa Pinto, Cotta and Tavares de Almeida2018; Verzichelli and Cotta, Reference Verzichelli, Cotta, Costa Pinto, Cotta and Tavares de Almeida2018). At the subnational level, the picture seems clearer as case-studies reveal (Jäckle, Reference Jäckle2023) that non-partisan regional ministers are more likely to exit politics and return to their previous professions than partisan ministers. As a consequence, we also assume that:

Finally, when it comes to understanding the role of prestigious or less prestigious portfolios on minister careers, there is certainly a gap in literature with few exceptions (e.g. Camerlo and Gallardo, Reference Camerlo and Gallardo2022), as most of the interest revolves around how the distribution of portfolios is gender-biased (Reynolds, Reference Reynolds1999; Studlar and Moncrief, Reference Studlar and Moncrief1999; Escobar-Lemmon and Taylor-Robinson, Reference Escobar-Lemmon and Taylor-Robinson2009; Krook and O'Brien, Reference Krook and O'Brien2012). Such a gap is clearly larger at the subnational level where there is almost no debate on what portfolios are high or low prestige in regional executives and there is scarce literature also on specific cases (see as exception Jäckle, Reference Jäckle2023).

Despite Jäckle (Reference Jäckle2023) not distinguishing between high- and low-prestige portfolios, he found that in Germany the regional prime minister, the minister of federal affairs, and the head of state chancelleries are those with the better chance to move upwards because they already work with the federal government, whereas the regional ministers of economy, finance, transportation, and construction and housing are those who have the better chance to enter the private sector. In Italy, the only cabinet members who can work together with national politicians regularly are RPs in the so-called State-Regions Conference (Conferenza Stato-Regioni). Consequently, we can hypothesize that:

Among individual factors, gender and the length of previous political careers have been considered.

As pointed out elsewhere (Boldrini and Grimaldi, Reference Boldrini and Grimaldi2023b) gender likely affects the length and the type of political career. In particular, looking at comparative national literature, we see that female national ministers are less likely to return to politics in comparison to male counterparts and they are more likely to return to their previous occupation (Claveria and Verge, Reference Claveria and Verge2015), hence being a minister is a one-shot experience for women. This evidence has been found also at the regional level as Jäckle (Reference Jäckle2023) found that female regional ministers in Germany are less likely to jump to the federal level than male ministers. Therefore, we can assume that:

Another important individual aspect to consider is the degree of political professionalism of RCMs which can be evaluated through the length and type of their previous career as a fundamental resource. In fact, recent studies have pointed out that previous political experience at local and regional levels is key to maintaining offices (Boldrini and Grimaldi, Reference Boldrini and Grimaldi2023a) for regional ministers and it is likely that the length of a previous career is also crucial for moving upwards toward national positions. However, to assess the role of political professionalism we think it is useful to take into account not only the length of the career per se but also the type of previous positions covered across territorial levels. In other words, political professionalism should increase not only as time spent in politics increases, but also as different positions and roles held at various territorial levels increase. Consequently, we can assume that politicians with longer political careers in multiple positions across different territorial levels are more likely to move upwards in comparison to those with shorter careers or with careers concentrated on local positions only. In particular, politicians with long careers and who previously held national, regional, and local posts are more likely to be considered political professionals and hence have more chance to move or return to national offices. On the other hand, politicians whose previous career is long but only focused on local positions are less likely to move upward, but they can at least hope to return to the local level once their regional cabinet experience ends.

Consequently, our last hypothesis is as follows:

Data and method

As previously mentioned, the research investigates what factors influence the movements toward positions at the national or local level, or the absence of any of such movements and therefore the exit from politics of RCMs once their executive mandate has fully ended. To this end, the research focused on RCMs at the end of their last term (i.e. those who did not receive subsequent appointments as RCMs) in the 15 Italian ordinary statute regions starting from the regional elections of 1995 to those of 2018. Data were extracted from an original dataset on Italian RMs (Grimaldi and Boldrini, Reference Grimaldi and Boldrini2022) and integrated, based on the research needs, with data from the National Registry of Local and Regional Administrators and the National Archive of Elections of the Italian Ministry of the Interior.

We decided to focus on RCMs who fully ended their regional executive career to assess what are the drivers that favor jumps upwards and backward within the subsequent 5 years or the exit from politics.

The selection of such a time frame was dictated by the need to provide the most comprehensive analysis as possible. In particular, the decision to focus on RCMs starting from the regional elections of 1995 was due to the possibility of analyzing reliable and comparable data. As observed by other studies (Boldrini and Grimaldi, Reference Boldrini and Grimaldi2023a), information on local and regional politicians during the so-called First Republic (1948–1990) is often unreliable, and the inclusion of RMCs before 1995 would have posed problems of data comparability due to the significant political discontinuity that occurred after the Clean Hands scandal and the 1994 general election.Footnote ²

In contrast, the choice of 2018 as the endpoint of our study (hence, the inclusion of all RCMs who finished their last mandate in that year) was dictated by specific research needs. For the purposes of the analysis, it was chosen to consider the career movements within the 5 years after the end of RCMs final term. The decision to consider political positions in this time span, although it may appear discretionary, was driven by both theoretical and empirical reasons. Firstly, 5 years is the maximum length of each legislature at the local, national, as well as European level. This time span, therefore, ensures coverage of at least one election for each level, and it also allows the inclusion of any assignments at various levels obtained following these elections. Moreover, from an empirical standpoint, almost all RCMs who obtain a new position after the end of their term got it precisely within 5 years.Footnote ³

Consequently, we have a population of 721 RCMs.

Given the nature of our investigation, a quantitative method was deemed the most appropriate and a multinomial logistic regression was used to test which variables significantly influence career movements once their regional executive experience ends. Considering our interest in the movements upwards toward national offices or backward toward local offices or on the exit from politics, we also opt for multiple logistic regression as a robustness test, by testing different models with various operationalizations of the dependent variable corresponding to the different career outcomes of RCMs (see Annex 1).

The dependent variable in our analysis is related to the office gained by each RCM within 5 years of the end of his/her term. Hence, there are four modalities: (1) whether as an RCM, an individual obtained another office at the regional level (no movement) within 5 years of the end of their mandate; (2) whether the RCM obtained as first office within 5 years of the end of their mandate a (both legislative or executive) role at the national/EU level; (3) whether the RCM obtained as first office within 5 years of the end of their mandate an office at the local level; or (4) whether the RCM did not obtain any institutional role (neither at the local, regional, national, nor European level) within 5 years of the end of his/her term. Table 1 illustrates the frequency distribution of different operationalizations of the dependent variable across all RCMs.

Table 1. Distribution of the dependent variables

As shown in Table 1 around 23% of our sample succeeded in exploiting the multilevel structure of Italian institutional arrangement with basically the same quota of RCMs who decide to move upwards such as the president of Tuscany (1992–2000) Vannino Chiti and the president of Apulia (2000–2005) Raffaele Fitto who became junior ministers, and those who moved downwards such as Carlo Masci former Budget minister of Abruzzi (2009–2014) who became Mayor of Pescara in 2019.

Furthermore, 57.3% of the sample exits from politics once the executive commitment ends. In particular, this is especially true for female executives such as Maria Rita Lorenzetti, the first female directly elected RP in Italy, who after the conclusion of her term in the Umbria cabinet, stepped down from any other institutional offices, and even for her successor, President Catiuscia Marini (2010–2019). Finally, 19.7% of the sample remains at the regional level, thus we cannot observe any kind of movement.

As far as the independent variables are concerned and in line with the theoretical perspective outlined above, both personal and political factors were considered. These include gender, non-partisanship, being a RP among RCMs, belonging to a small party, and the length of the previous career at the local, regional, and national levels.

Gender, non-partisanship, being an RP, and belonging to a small party were all operationalized as a dichotomous variable with a value of one if the RCM is respectively female, if appointed externally to the members of the regional council or indicated as ‘independent’ or ‘without a political party’, if holding the position of RP and if elected with a list that in the previous regional elections received less than 3% of the votes.Footnote ⁴ Table 2 illustrates the frequency distribution of such independent variables.

Table 2. Distribution of the independent variables

Turning to the previous career of RCMs, the three variables were all operationalized in the same way as a continuous variable, calculated according to the number of days in office. In particular, the local career, which includes institutional posts such as municipal councilor, municipal executive, mayor, provincial councilor, provincial executive, and provincial president, was calculated as the number of days in office before the conclusion of their last term as RCM. The regional career includes roles such as regional councilor, regional minister, and RP and it was calculated as the number of days in office before the conclusion of their last term as RCM. The national career includes roles such as deputy, senator, minister, junior minister, or member of the European Parliament and it was calculated as the number of days in office before the conclusion of their last mandate as an RCM. For comparability purposes, in the logistic regression, all these three variables were standardized. The previous career at the local level is highly prevalent among RCMs being held by 62.4%, while predictably, the national one is much less frequent, covered by only 8.6% of RCMs.Footnote ⁵ Table 3 presents the descriptive statistics related to the previous career at the local, regional, and national levels of RCMs.

Table 3. Descriptive statistics of previous local, regional, and national careers (in days)

Furthermore, we also added the following control variables to the analysis: age, geographical origin, and early termination. First, we controlled for RCM's age as it can deeply influence the subsequent career paths of each RCM. The basic idea is that as regional ministers get older, they have fewer opportunities to serve in other political roles or positions once their last term is over. To this end, age was calculated in years by subtracting the RCM's date of birth from the end of their last term, and it was operationalized as an ordinal variable with four different categories: under 40 years old, between 40 and 49 years old, between 50 and 59 years old, and 60 years old or older. We decided to include only four categories due to the low number of RCMs under 30 and those over 70 (only 12 in the entire sample).

Secondly, given the relevance of political geography in the Italian context, we control for the geographical origin of the RCMs. In other words, to verify if any specificities related to the traditional strongholds exist, that is where the two different political alignments prevail. Such a variable was operationalized according to the classification used by Valbruzzi and Vignati (Reference Valbruzzi and Vignati2021) into center-left stronghold (C-L) when only center-left regional cabinets were in power from 1995 up to 2018, center-right (C-R) stronghold when only center-right regional cabinets were in power up to 2018, and swing and competitive regions in all other cases.

Finally, we also control for early termination, that is, when RCMs' term ends prematurely compared to the natural expiration of the legislature. In fact, early resignation – both as the result of an individual resignation or as the resignation of the entire cabinet – can influence the career path of RCMs by favoring or hindering subsequent movements across territorial levels. In particular, it has been operationalized as a dichotomous variable with a value of 1 if early resignation occurred and 0 otherwise.

Explicative analysis

Table 4 shows the results of the multinomial logistic regression. In the table the relative risk ratios and in parentheses the standard errors clustered at the regional level are displayed. In our model, RCMs remaining at the regional level represent the baseline, so the model tests the influence of the independent variables, respectively, in exiting from politics, moving downwards, and moving upwards against remaining at the regional level.

Table 4. Results of the multinomial regression analysis

Note: Results of multinomial regression, relative risk ratio displayed, standard errors in brackets; P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001.

First, our analysis shows that RPs are less likely to exit from politics or move downwards toward local offices, but they are more likely to stay at the regional level. On the contrary, there is no statistical significance for moving upwards. These results may seem somewhat counter-intuitive, as they seem to highlight a greater ease of holding office at the regional level for RPs. However, it should be specified that this result is heavily influenced by those RPs who decide to run for a second term and lose and consequently they obtain as first position that of regional councilor. Furthermore, the result is also due to the overall number of RPs that is relatively small. However, our third hypothesis (H3) is not confirmed.

Non-partisanship shows a positive statistically significant effect both for exiting from politics and for a movement upwards. In other words, as literature pointed out RCMs who are not affiliated with a political party are more likely to exit from politics after the end of their mandate than remain at the regional level. Therefore, our second hypothesis (H2) is confirmed. Furthermore, the marginal effect of non-partisanship in relation to an upward movement is neglectable (see marginal effects in Figure 1).

Figure 1. Predictive margins of statistically significant variables with 95% CIs.

Note: The dependent variable is standardized from 0 to 1.

As far as the previous career is concerned, our analysis shows that only a national one does have a clear impact on future movements. In particular, those RCMs who previously had a national career are more likely to move upwards toward a national or EU office or at least to exit from politics than remaining at the regional level. As a consequence, our H5a hypothesis is confirmed.

However, looking at the marginal plot in Figure 1, there is a clear non-linear effect of the previous national career especially in exiting from politics. So, it can be argued that certain RCMs with a relatively shorter previous national career are likely to exit from politics but, as far as the previous national career became longer and more relevant, they likely survive in politics. However, further specific analysis is needed to explore this aspect.

Turning to previous local career, our model shows that there is not a significant effect concerning a downward movement toward local positions, despite the association going in the correct direction and the effect approaching the statistical significance. In fact, when opting for multiple logistic regression as a robustness test, we found that previous local career is statistically significant for RCMs who move backward (see Annex 1). Nevertheless, our H5b hypothesis is not confirmed.

Alternatively, there is no statistically significant effect on being a woman in relation to an upward movement or to the exit from politics, even though the direction of the correlation goes in the expected direction in both cases. However, hypotheses H4a and H4b are not confirmed.

Finally, we did not find any evidence that belonging to a large party influences moving upwards or decreasing the possibility of exiting from politics. Thus, our H1a and H1b hypotheses are not confirmed. Nevertheless, when running a logit analysis, the size of the party size matters. In particular, we found RCMs belonging to small parties have less probability to jump at the national or EU level and more probability to exit from politics (see Annex 1).

As far as control variables are concerned, age is positively correlated with RCM who exit from politics and the effect is stronger when the age increases. This makes sense as aging increases the possibility of exiting from politics. Finally, being in a CR stronghold is negatively associated with a movement downward. This is probably because the large mainstream party of the right – Forza Italia (from 2000 to 2010 in most of the regions) – had a weaker control of the territorial structure in comparison to the main left-wing party and therefore its political personnel was less likely to move downwards than remaining at the regional level.