Investing in the roots of your political ancestors

Public infrastructure is best described as a centrally provided local public good (i.e. public good that generates localised benefits) or as a geographically targetable private good. See Knight (2004) for more details on this.

A parallel strand of the literature investigates the possibility of party favouritism in the allocation of intergovernmental transfers (and public capital) to politically aligned local authorities. To address the possibility that alignment and public funding can be entirely driven by socio-economic factors influencing both dimensions, several of these studies employ a Regression Discontinuity Design in closely contested electoral races, as pioneered by Lee (2008) (see, e.g., Brolo and Nannicini, 2012; Bracco et al., 2015; Baskaran and Hessami, 2017; Curto-Grau et al., 2018; Kammas et al., 2023).

Actually, Greece was part of the European ‘democratisation wave’, along with Portugal and Spain, which struggled with dictatorships in the second half of the twentieth century (see Huntington, 1993; Torregrosa-Het al.,and, 2015).

See Appendix A for a detailed description of the electoral system.

Proportional Representation (PR) electoral systems allocate seats to parties based on their electoral strength and then distribute these seats to individual candidates using various formulas. More specifically, in Closed-List Proportional Representation (CLPR), it is the party that ranks the candidates (prior to the election) and after the election the seats are allocated to individual candidates according to their rank in the party list. In OLPR, parties present a set of candidates running under their label but they do not rank them prior to the election. So, voters cast a ballot for an individual candidate and the parties receive seats in proportion to the sum of votes received by all the candidates running under their label. After the election, seats are distributed to individual candidates according to the number of personal votes they received. For example, suppose that a party is allocated n seats. The top n candidates, as determined by the number of votes they personally received, are those who win the legislative seats (for more details on this, see Cheibub and Sin, 2020).

The Greek political landscape has historically been characterised by interpersonal patron–client networks, which were personally linked to specific MPs, primarily benefiting them and only indirectly benefiting their political parties (see, e.g., Meynaud, 2002).

Notably, ‘long-term partisan loyalty’ refers to the repeated voting in favour of a party over extended periods and should not be confused with the notion of core supporters, which typically denotes voting in favour of a party in the previous election. Previous studies on pork-barrel politics primarily examine whether national governments allocate public resources to regions that supported the incumbent party in the most recent election (see, e.g., Cadot et al., 2006; Kauder et al., 2016; Luca and Rodríguez-Pose, 2019). To our knowledge, Joanis (2011) is the only other study that focuses on the concept of ‘long-run partisan loyalty’—that is, repeated voting in favour of a party over longer periods of time—rather than solely in the previous election.

In fact, these 51 prefectures correspond to a total of 56 electoral constituencies. This discrepancy arises from the fact that the two major metropolitan areas of the country, Attiki and Thessaloniki, contain five and two electoral constituencies, respectively. To align our analysis with the available data on public investment flows at the NUTS-3 level, we have merged the electoral constituencies of Attiki and Thessaloniki to correspond to one prefecture each (see Psycharis, 2008).

Notably, similar electoral systems ensured total control of the parliament by all subsequent one-party governments between 1974 and 1989. In contrast, the electoral law of proportional representation passed by PASOK before the election of 1989 prevented ND from forming a government despite its 5% lead in the popular vote over PASOK (see Verney, 1990).

PASOK’s success led to a decline in the vote share obtained by Georgios Mavros’ centrist party, which slumped to 11.95%, eventually resulting in its gradual disintegration from the political system within a few years (Mavrogordatos, 1984).

Specifically, we forward prefecture-level electoral results up to (and including) the year of the next general election (see, e.g., Jablonski, 2014).

The opposition is measured as the combined share of votes received by the two leading opposition parties, EKND and PASOK, between 1974 and 1981. For the period between 1982 and 1989, the opposition share is represented solely by the dominant opposition party, ND. This distinction arises from the presence of a dominant opposition party (ND) during 1982–1989. By contrast, between 1974 and 1981, the centrist party EKND and PASOK alternated in second and third place, with their combined strength reaching approximately 35%. Additionally, we combine the vote shares of the two parties since PASOK absorbed the majority of EKND supporters during its growing influence.

NUTS, which stands for Nomenclature of Territorial Units for Statistics, is a geocode standard established by EUROSTAT, the Statistical Office of the European Union. Within this system, LAU, denoting local administrative units, functions as the fundamental building block of NUTS and encompasses the municipalities of the European Union.

In 1986, Greece introduced regions as second-level administrative units, dividing the country into 13 regions. These regions operate as an intermediary administrative level between the central government and the individual prefectures.

It is relevant to highlight that, when employing alternative specifications in our estimates of column (6), including the instrumental variable (IV) approach described in the subsequent section, the polynomial term of the variable victory margin is positive and statistically insignificant. This result is not consistent with the ‘swing voter’ hypothesis.

Note that the elections of 1961 and 1964 were the last before the military junta (1967–1974). In the 1961 election, ERE was the first party (and formed a government), whereas in the 1964 election, EK was the first party (and formed a government). We select to relate the electoral vote share of ND’s wins in the Metapolitefsi period (1974 and 1977) to its ancestor’s last electoral win (ERE in 1961) and, similarly, to relate PASOK’s first electoral wins in the Metapolitefsi period (1981 and 1985) to its ancestor’s last electoral win (EK in 1964) for consistency. In Sect. 3.2.3, we perform a robustness check by calculating the average vote shares of the ancestors from both elections before the junta period.

Building on the seminal study by Altonji et al. (2005), Oster (2019) assessed how large the bias due to unobservables should be in comparison to that of observables, in order to explain away the estimated effect. The ratio between the two components of the bias is denoted as δ. This exercise returns a ratio that is (above 1 and) negative, which indicates a negative correlation between observables and unobservables and implies that the coefficient increases when controls are added to the model. The exercise also indicates that for the results to become economically insignificant, the selection of unobservable factors would have to dominate the selection of the included observables and work in the opposite direction.

Given that the outcome is expressed in logged per capita Drachmas, the percentage change effect is calculated by e^λ − 1, with λ being the estimated coefficient on the incumbent share (α₄) multiplied by the standard deviation of the incumbent share.

Notably, according to both definitions, the strongholds include prefectures in the top quartile of the distribution of electoral strength.

The reduced-form estimates are not reported as they correspond to those already reported in column (2) of Table 2.

Moreover, funding from IMPs is added to this category as it is titled devolved IMPs.

We have fewer observations for investment flows by type because we could not track the regional fiscal account of 1979 in the Ministry of Economy and Development. Additionally, to prevent taking the logarithm of 0 for certain investment variables, we added the value of 1 before performing the logarithmic transformation.

In a precise manner, these patron–client linkages form pyramids, with powerful MPs at the apex, local party bosses in the middle and individual voters at the base (see, e.g., Mavrogordatos, 1983a). Traditionally, these networks of local bosses and middlemen were personally affiliated with the powerful MPs, and it was common practice for them to be transmitted as inheritance—or even as dowry—within the same family from one generation to the next. In such a political context, it is evident that powerful MPs were the ultimate centres of political power, and consequently, political parties were structurally built around these networks of local notables (see Meynaud, 2002, for more details on this). The absence of effective party organisation and mass membership led to the party’s parliamentary group becoming extremely powerful; this situation is often described as vouleftokratia (‘rule of the MPs’) in the relevant literature.

According to several scholars, observed electoral personalism can also be understood as stemming from a ‘political culture’—a set of attitudes and social values regarding power, law, family and the state—that traces its roots back to the era of Ottoman dominance and continues to influence political behaviour in contemporary Greece. Scholars such as Legg (1969), Campbell (1983) and Diamandouros (1983) have explored this aspect in detail. Based on this theoretical framework, efforts to introduce Western democratic institutions since 1864—such as extending the voting franchise to the entire male population—encountered challenges in a society where previous experiences had fostered deep suspicion of authority outside the family and a tendency to exploit authority for familial advantage. Thus, imported political institutions were often corrupted to serve traditional patron–client relationships (see, e.g., Diamandouros, 1983).

Notably, our classification considers ministers and deputy ministers that have been elected in the prefecture and also plan to be candidates in the next election.

It has been argued that since 1974, the pre-junta party system, which was based on traditional interpersonal patron–client relationships, has started to undergo transformation. The emergence of new parties during Metapolitefsi saw the development of stronger organisational structures, mass memberships and another type of clientelistic network described as bureaucratic clientelism or machine politics, rather than traditional patronage (see, e.g., Lyrintzis, 1984; Mavrogordatos, 1983b, 1997; Kammas et al., 2023). This transformation of the political system is a well-established fact. However, it is relevant to note that the interpersonal, clientelistic relationships of the past did not disappear entirely. On the contrary, during the early years of Metapolitefsi, newly established political parties sought to maintain the political networks of their pre-junta ancestors, which were structured around powerful MPs, while simultaneously developing strong organisational structures (see Kammas et al., 2023, for more details).