The End of Cornucopia: Party Financing after the Great Recession

Abstract This article investigates the dimension and evolution of the financing of political parties. It focuses on 28 parties in the five major European countries (Germany, France, UK, Italy, Spain), analysing the parties’ budgets from 2002 to 2016. The article's assessment shows that the availability of funds increased until the beginning of the Great Recession (2008), and then decreased, mainly due to a decline in public support for parties. Diminished state generosity has led parties to look for different sources of financing: the article shows the proportion of self-funding resources in terms of membership fees and private donations that has sustained the parties’ finances. Finally the article presents a model that helps to explain the shrinking of parties’ income by including parties’ ideological alignment, electoral outcome, presence in government and share of public financing, and countries’ public spending and GDP level, to investigate the plausible causes of the reduction of parties’ income.

The importance of party financing for intra-party dynamics and parties' relations has also been analysed following different lines of research, but all focused on how the abundant flow of money from the state has reduced recruitment and mobilization capacity by the parties themselves, thus enforcing a higher level of dependence on the state (see, among others, van Biezen and Rashkova 2014;Detterbeck 2008;Piccio and van Biezen 2018;Scarrow 2006;Scarrow et al. 2017).
A further line of research has investigated the relationship between party/ candidate funding and democracy. The specific focus of this area of research is the intrusive effect of money and the corresponding effort in guaranteeing the fairness and equity of electoral competition; in this vein, particular attention is devoted to the means of counteracting political corruption, and of controlling and monitoring donors' and vested interests' contributions to political parties (e.g. Cagé 2020; Casal Bertoa et al. 2014;Casas-Zamora 2005;Mendilow 2012;Scarrow 2018).
A third line of research deals with the legal framework of party financing (e.g. Austin and Tjernsrtöm 2003;Falguera et al. 2014;Nassmacher 2009;Ohman 2012;Piccio 2012;Pizzimenti 2017;Scarrow 2007). This approach provides detailed information regarding the country-specific regulations on the distribution of public funds, the constraints on fundraising and the account and control of expenses. A wealth of data is offered by international organizations such as the Institute for Democracy and Electoral Assistance (IDEA), which provides universal coverage of 'funding regimes' (Koss 2011) all over the world. 1 Further details on the appropriateness of party behaviour are also available from the GRECO institute, and from the Venice Commission of the Council of Europe. 2 In addition to these approaches, analyses of country-specific or regional cases have enlarged and deepened knowledge of party financing. With no pretensions to this being an exhaustive overview, it is worth mentioning the most important and recent studies dedicated to the five European countries analysed in this article: Koss (2011) and Blumenberg et al. (2018) for Germany;Phélippeau (2013) and François and Phelippeau (2015) for France; Pizzimenti and Ignazi (2011), Piccio (2018) and Ignazi and Fiorelli (2019) for Italy; Casal Bértoa (2016, 2018) and Jimenez and Villoria (2018) for Spain; and Clift and Fisher (2004), Keith (2007), Pinto-Duschinsky (2016, Johnston and Pattie (2014) and Fisher (2018) for the UK.
This article follows the theoretical corpus proposed by Mair (1994, 1995) and re-stated in Katz and Mair (2018). These two authors suggest a process of party detachment from society and a consequent deeper interpenetration with the state. This process has an effect on party finances: it reduces the capacity for self-financing through membership fees as the members decline in number, and it encourages parties to exploit public resources to counteract the shrinking of their self-funding capacity. Therefore, according to Katz and Mair, and the subsequent literature, parties have become richer thanks to growing state provisions and also more dependent on this source of income.
The hypotheses and outcomes of that previous body of research are updated and we hope refined here due to: (1) a unique collection of data on the finance of 28 'relevant' parties (more than 5% of the votes, on average, in the 2002-16 period) in five European countries (France, Germany, the UK, Italy and Spain); 3 (2) the extension of the time span up to 2016, the last year of available comparable financial data for the countries under consideration; (3) a detailed examination of the single parties' financial entries rather than an overall analysis by country only; (4) a careful harmonization and standardization of the different entries in the parties' budgets derived from official party documents. 4 Thus, the present article seeks to determine whether: (1) the finances of the parties have continued to increase even after the Great Recession; (2) parties have abandoned their long-run dependency on state funds; (3) the (eventual) shrinking of state support has been counterbalanced by other sources of income such as membership fees and donations; (4) the Great Recession had an impact on the parties' budget and its composition.
The article proceeds as follows: the next section presents the main characteristics of the five funding regimes analysed; then we analyse the trends of the main income sources for each country and political party, followed by an evaluation of the impact of the Great Recession on parties' financial wealth; and finally, we present some concluding remarks for future developments.

Funding regimes
State regulationsthe so-called 'funding regimes' (Koss 2011)define the rules of access to public resources and provide guidelines for a party's fundraising and proper accounting procedures. A brief overview of some features of the funding regimes of the five countries helps to put the data analysis into context. 5 Direct state funding has been present in every country, and is customarily based on the share of votes and/or seats obtained at elections. Beyond that common prerequisite some national peculiarities are worth mentioning. Germany has a unique mechanism of 'matching funds', in which the state subsidy received by the party should not outweigh the annual revenues collected through fundraising activities; moreover, the state provision cannot exceed a fixed threshold in any case (from 133 million euros in 2002 to 160 million euros in 2016). 6 France is the only country where not only parties but also candidates receive public support in terms of reimbursement of expenses, up to a pre-fixed maximum. 7 Only the UK allocates public money specifically to opposition parties (Short Money for House of Commons and Cranborne Money for House of Lords). Italy is the sole country where state financing is formally allotted exclusively for the reimbursement of electoral campaigns even if parties use the money for any purpose. Moreover, Italy shares with Spain the peculiarity of imposing no cap on state support.
Indirect state funding in the form of tax deductions is common to every country, with different degrees of deduction. Tax deduction is commonly seen as an incentive to promote private donations to political parties. However, as Julia Cagé (2020: 49) has highlighted, as the benefit in tax deduction applied to private donations increases with the increase in donation, this might attract wealthy donors rather than small contributors, reproducing and even reinforcing social inequalities within the party.
Membership fees represented the most important source of income during the mass-party model era. According to the cartel party thesis, and further theoretical and empirical considerations (Katz and Mair 1995, 2002, the parties' drive to the state has distanced citizens from direct involvement in political parties. The consequent shrinking of membership, and membership dues, has thus reduced the capacity to self-finance party activity. Moreover, state intervention in supporting parties' finances has further demotivated parties from attaining resources from the rank and file (and, to a certain extent, from private sources as well).
Private donations are a highly sensitive area and are regulated in a detailed manner in the five countries under examination. There are some common patterns such as: an ever stricter and lower cap limitation (as in France, Spain and Italy), an increasing amount of publicity and disclosure of donors (in every country, with some exceptions for Germany) and a growing restriction on funds from legal entities, state-owned companies and foreign donors (apart from the UK, where no limit exists). A peculiar form of private donation is the so-called 'party tax'; that is, the delivery of a quota of the party representatives' salary to the party headquarters (Bolleyer 2012;Bolleyer and Trumm 2014;Pinto-Duschinsky 2016). 8 These contributions have become quite important for party finances, especially in a period of declining state support.
A unique form of financing 'from other political organizations' is present in France. This regards the flow of money which the so-called 'micro-parties' redirect to the main parties. Micro-parties, which numbered 338 in 2016, can be understood as empty boxes with a double aim: first, they field candidates in order to receive reimbursement for the candidate's campaign and, more unlikely, for the party, if it gets more than 1% of votes; second, they collect 'extra money' from private donations as they can easily bypass the limit for the amount received given the relaxed accounting system. Most of the time this extra money is then conferred to the established parties because most of these micro-parties are created or solicited by the main parties themselves, or some kind of agreement between them has been established (François and Phélippeau 2018: 76ss).

Party-financing macrotrend
The most recent comparative empirical analysis by country confirms a positive trend in the parties' overall budgets, until the end of the first decade of the 2000s (Bardi et al. 2017; van Biezen and Kopecký 2017;Webb and Keith 2017). In this section, we examine whether the trend continued until 2016 or if a different trajectory emerges.

The ups and downs of party finances
In four out of five countries, political parties in 2016 display larger availability of finances than in 2002 (Table 1). In the 14 years between 2002 and 2016, only Italy presents a remarkable reduction of the parties' income because of the new legal provision which first reduced (in 2012) and finally eliminated (in 2014) direct public financial support. The rate of change between the initial and final year (t2 − t0) is in fact markedly negative (−74%) for the Italian parties. The other parties have instead increased their income, and particularly so the UK ones (+89%), whereas in Germany the parties' budgets moved upwards almost imperceptibly because of the 'rigidity' of this funding regime, as highlighted above.
This general overview assesses that, in most of the countries, money available to the parties, considered altogether, is now more abundant. However, if 2008 is taken as the year of the onset of the Great Recession, two considerations emerge from Table 1. First, in 2008 much more money was available to parties than in 2002. The average increase is 23% and, in four countries, the increase in parties' income was above 40% (and in two of them, France and Italy, even above 60%). Second, with the notable exception of the UK, such largesse of resources shrank in the years after 2008 (the average reduction was −17%). This trend raises the question of whether the shrinking of financial means is related to the Great Recession. The apparent relationship emerging from this data will be tested later in the article with an ad hoc regression model.

Analysis of single parties
In this section, the most significant data of the single party budgets are analysed.
The richest parties are, by far, the German ones ( Figure 1b). Due to generous state financing and incentives for self-funding, the two major German parties, the Social Democratic Party (SPD) and the Christian Democratic Union (CDU), each had some 150-160 million euros available throughout the period covered. If one adds to the CDU budget the funds of its sister party, the Christian Social Union (CSU), the 'Union' touches the threshold of 200 million euros.
Spanish major parties come second ( Figure 1e). 9 Apart from some oscillation over time, until the beginning of the 2010s, the People's Party (PP) and the Spanish Socialist Workers' Party (PSOE) had around 90 million euros and  The fragmentation and instability of the Italian party system, and the radical changes in the rules of state financing, necessitate a more detailed analysis ( Figure 1d). First, for the sake of parsimony the two major parties, the PD and the PDL, which were established in 2007 and 2009 respectively, are considered together with their respective founding partners: the Democrats of the Left (DS) and the Daisy (La Margherita) for the PD, the National Alliance (AN) and Forza Italia (FI) for the PDL. 10 Second, the PDL seems to undergo a dramatic financial decline after 2013, whenby then renamed Forza Italia because of a splitit declared only a handful of million euros. This figure actually hides the party's real financial condition because, in 2014, it benefited from 90 million in personal loans from the founding leader, the multimillionaire Silvio Berlusconi. 11 Third, the financial accounts of the newcomer, the 5 Star Movement (M5S), at present the largest party in the Italian parliament, are quite difficult to ascertain given its organizational peculiarities (Vittori 2019). 12 The official balance sheets, reported by all the components of the M5S, present astonishingly small figures: summing up all the components' budgets, the total income was 737,000 euros in 2013, 618,000 euros in 2014, 2,500 (sic.) euros in 2015, and 406,000 euros in 2016. These minimal revenues apparently do no harm to the party as it declares that no expenses are afforded for electoral campaigns or the ordinary management of the party. 13 When the M5S is in need of money for its special events, it sets up an ad hoc structure which is dissolved immediately afterwards. 14 Therefore, the figures relating to these events do not appear in any of the official party budget components.
As for the other Italian parties, the finances of the PD have been overstated by its direct competitor, the PDL, until 2013 when, after the elimination of the state direct fund, it reacted very effectively, becoming the richest Italian party but at a much lower level than in the past. 15 The other parties display less affluence; particularly dramatic is the Lega Nord's financial collapse after 2012, following its political crisis due to the eruption of cases of malpractice and corruption.
UK parties are the least wealthy ( Figure 1c): around 45 million euros on average for both Labour and Conservatives, with the Conservative Party even registering less than 30 million euros for a couple of years in the early 2000s, when it underwent a major political/electoral crisis. On the other hand, in the same years the Labour Party, under the premiership of Tony Blair, almost touched the ceiling of 60 million euros. This level has also been surpassed in recent years by both parties (although they underwent a reversal in 2016).
Looking at the minor parties included in this analysisthose receiving between 10% and 5% of votes on average in the periodwe can see a common pattern of stability over time. Beyond their ideological alignment they do not fluctuate as widely as the main parties and keep their total revenues quite constant throughout the period. Some parties are, however, richer than others. For example, the French Communist Party (PCF), notwithstanding its limited electoral success, displays an amazingly solid budget, always above 30-35 million euros; and the Greens' (EELV) income is not so distant from that of the National Front (FN) (both are around 8 million euros), although the electoral results of the two are not comparable during the period. In Italy, the radical left Communist Refoundation Party's (PRC) budget equalled that of the right populist Lega Nord until 2008, but the PRC offspring, Left Ecology Freedom (SEL), later displayed very limited resources, whereas the Lega Nord increased its budget until the early 2010s; then, the emergence of a series of scandals dramatically reduced the Lega Nord's income to few million euros. The German minor parties of the radical left (Party of Democratic Socialism (PDS), Work and Social Justice, the Electoral Alternative (WASG), the Left (Die Linke) and the ecological left (Alliance90/the Greens (Grüne)), present a somewhat similar income to that of the liberal-secular Free Democratic Party (FDP) and the confessional Bavarian CSU; the latter party is, however, slightly richer, with a peak in revenues in the mid-2010s. The British Liberal Democrat Party finances oscillate between 6 million and 10 million euros, far below the main UK parties.
Still dependent on state support?
Public financing represents the largest part of political parties' income in most of the countries where it was introduced ( van Biezen 2008;Falguera et al. 2014;Katz and Mair 2018;Weekers et al. 2009). State subsidies were constantly increased until the end of the first decade of the 2000s ( van Biezen and Kopecký 2017;Ignazi 2017). Table 2 displays the average share of public subsidies in the parties' budget. 16 In general, until 2008 parties were dependent on the state for around 40% of their income; later this quota dropped, arriving at 33% by 2016. Differences between countries and parties are notable. German and UK parties in 2002 were used to obtaining less from the state, respectively, 29.9% and 22.2% of their resources, a quota that is related to their peculiar funding regimes (see above); in the other countries, rather, parties were heavily dependent on public support, from a minimum of 54.5% (France) to a maximum of 76.7% (Spain). In the following years this dependence remains overall stable, save for a minimal increase, until 2008 (with French and Italian parties going in opposite directions: the former being less dependent, the latter more dependent), and only after it had dropped. In 2016 the balance is different. The share of public funds drops to one-third for French, German and Italian parties, and even less for the UK ones. Spain is an outlier in this context: although public support shrank, Spanish parties are even more reliant on it.
In sum, although the economic crisis has tightened the availability of state intervention (especially in Spain and Italy), 'continental' parties remain somewhat dependent on it, but not at the same level as in the first decade of the 2000s. Apart from some different criteria of data collection, the updating of data to 2016 allows us to reconsider previous evidence provided by the PPDB project (especially Bardi et al. 2017; van Biezen and Kopecký 2017): with the exception of the UK, where state funds are of relative importance, parties are now less dependent on public resources that, in turn, have been shrinking in the most recent years.

Proportion of state funding for single parties
Among to the generally high dependence of all Spanish parties on state money (Figure 2e), the PP is the most reliant of all the parties surveyed: around 85% of its budget comes from public sources; and the PSOE ranges around 70-75%. In addition, a new party such as Podemos, which emphasized the virtues of crowdfunding, seems quite dependent on state support. It is worth noting that when Spanish state funds decreased by 20% in 2014, both major parties -PP and PSOEeven increased their dependence up to 80-90%. Some Italian parties have followed suit (Figure 2d): the Daisy and AN before 2008, and the PD after 2008, are around 80%. 17 Contrary to the evolution in Spain, when in Italy direct funding was first reduced by 20 million and then cut by 50% for the 2013-18 legislature, the dependence of parties on state support collapsed. However, with the cuts in direct funding, indirect state support in the form of tax deductions was introduced; this new provision was particularly effective for the PD, which attracts more than 60% of total indirect funds, totalling some 7 million euros. State funds still constitute an important source of income for the three main French parties (UMP/LR, PS and FN) (Figure 2a): in 2016 they depended on public funds for more than 40% of their incomes, and the Greens are somewhat below that threshold. The UMP/LR used to be highly dependent for around 60% of its income during its time in power (2002-12), but afterwardsquite significantlythe proportion of its income from state funds declined by 20 percentage points. 18 The PS oscillated around 40% during its opposition, but after its success in 2012 it slightly increased. The FN's dependence on state funding is strictly linked to its electoral performance. After its success in 2002, when state funds represented 60% of its income, it started a traversé dans le desert, particularly after the electoral downturn of 2007 and Jean-Marie Le Pen's leadership crisis. Only with the change of leadership and the electoral recovery in the early 2010s has the party attracted more state funds and increased their share. The Greens exhibit a pattern similar to the FNa fall and later a recoverywhile at a much lower level. The PCF stands apart as state funds formand have constantly beenaround 10% of party income, demonstrating a unique capacity for self-financing.
In Germany (Figure 2b), public money consistently covers around one-third and one-quarter of the CDU's and SPD's total income, respectively. The proportion is moderately higher for the Grünen, FDP and Linke. On the other hand, the CDU is the least dependent party due to its efficacious self-funding capacity, especially in terms of private donors. The rigidity of the funding regimes inhibits the ups and downs that have been seen in other countries.
At the bottom are the UK parties. These parties' peculiar funding regimead hoc state support for opposition parties, plus a small grant for all partiesmakes this entry in the parties' budget quite volatile. This is why in the period considered, when Labour and Conservative stayed in opposition for more or less the same number of years (eight and six, respectively), they received a similar amount of state support; and this is why their lines, as shown in Figure 2c, cross each other in 2010, when the Conservatives came back into power. Instead, the Liberal Democrats (who stayed in opposition for a longer period) not only availed themselves of Short Money for many years, but 'needed' more of that source of income as it represented around one-fifth of the party's total budget. Even if the amount of public funds is modest compared with other countries, the limited income of UK parties, again in comparative terms, 'pulls up' the share of public funding to some 40% when a party is in opposition. This quota is perfectly comparable with the other countries.

The return of liberality and members' contributions?
All sources of financing other than the public ones present an inverse trend with respect to the traditional dependence on public subsidies (Figure 3). In Germany, the UK and France they cover more than half of the income of political parties. The German parties show a stable and constant average share (around 65%), and the UK and French ones exhibit an increasing rate of self-funding, with the former attaining more than 80% of their income from that source in the 2010s. In Spain, where public subsidies have been very generous until 2012, when public funds were curbed, parties rely on self-financing much less than the other parties: only around 20-25% of their total income came from that source. Italian parties did not obtain more than one-third (and sometimes even less) of their revenue from private money for a large part of the period. After the cut in state provisions in 2012, private sources increased dramatically, up to a high point of 80% in 2015.
This general picture presents some variations when the general heading of autonomous sources is divided into three main components: the traditional source of self-funding represented by membership fees, private contributions (donors), and that peculiar and quite disguised form of self-financing provided by the party representatives, the so-called 'party-tax'.

Membership fees
According to the cartel party thesis, and further theoretical and empirical considerations (Katz and Mair 1995, 2002, the parties' drive towards the state has distanced citizens from direct involvement in political parties. The consequent shrinking of membership, and membership dues, has thus reduced the capacity of a party to self-finance. On average, as Table 3 demonstrates, around 15% of a party's total income stems from membership dues in all years between 2002 and 2016. This value, in line with what was reported by van Biezen and Kopecký (2017: 87) for the early 2010s, is increased by the German parties' figures because more than 25% of their budget comes from this source; this high value reflects the incentive to enrol members and thus increase self-financing, to meet the legal requirement for the maximum allotment of public subsidies. The SPD and, even more, PDS-WASG-Linke profit more than others (see Figure 4b). Beyond the German parties, revenues from membership are everywhere below 20%. In France parties are close to that level, as for the PS in the first decade and the FN in the last years ( Figure 4a); in Spain ( Figure 4e) and Italy (Figure 4d) they are consistently below this, at around 10-15% maximum, as occurred for the PP in Spain. 19 This is also true for most of the Italian parties, but we should take into consideration that the PDalthough having the largest membershipdoes not report membership fees in it national accounts. 20 The British Conservatives are the least reliant of all the parties in every country, as they do not receive more than 3% of their income from membership fees. UK Labour, on the other hand, in very recent years increased the contributions of the membership spectacularly, up to 30% of its budget thanks to skyrocketing numbers propelled by a tough competition for leadership and a reduced fee (Figure 4c).
These data support the interpretation that organizational traditions still play a part as leftist parties seem to emphasize the traditional role of membership for party financing whereas moderate conservatives appear less dependent on this source of income. Moreover, smaller, and radical, parties -FN on the right, PDS-WASG-Linke and PRC-SEL on the lefttend to be most dependent on members as they can profit less from voluntary donations coming from wealthy donors or vested interests.
In sum, membership fees represent a residual component of the parties' income, even if a modest increase of their share emerges in the most recent years. Considering long-term analyses which have shown the growing marginalization of membership fees in sustaining parties' finances (Pizzimenti and Ignazi 2011 for Italy; Koss 2011 for Germany; Kelly 2018 for the UK; Rodríguez-Teruel and Casal Bértoa 2018 for Spain; Phélippeau 2013 for France), this relative comeback of the financial role of membership might suggest that the descent has reached its lowest level, and there are even some signs of some revamping in the face of reduced support from public sources.

'Party tax'
A peculiar kind of voluntary contribution comes from party representatives in subnational, national and European assemblies (Bolleyer 2012; Bolleyer and Trumm 2014; Pinto-Duschinsky 2016). These contributionsthe 'party tax'are more or less directly defined in party statutes or by the parliamentary party groups. In fact, this kind of financial flow manifests in parties' budgets only in Germany and France; other countries' regulations do not require this distinction and thus this peculiar financing could not be identified as such.
Although data are limited for France and Germany, a closer investigation of the role of the party tax reveals that only leftist parties collect a large share of the private donations from their elected officials. 21 The French PS and the Greens officeholders' contributions represent 85-90% of donations, and their German counterparts -SPD and Grünearound 65%. On the other side of the political spectrum, large moderate parties such as the CDU or the UMP do not raise much money from their representatives, as this entry provides 35% and 22%, respectively, of the share of private money.
Party tax distribution reveals how ideological alignments still play a role in intraparty dynamics. The tradition of attachment to the party, much stronger in the socialist tradition than in the conservative one, is revealed quite clearly in the different importance of representative contributions to the respective parties. Furthermore, in the two countries where data are available, the party tax has become a substantial factor after the Great Recession: almost every party has increased contributions from officeholders in relation to other private financial support starting from the 2010s. A further, albeit partial, confirmation of this tendency is provided by the Italian parties: in 2016, Forza Italia obtained 75% of private money from the party tax, Lega Nord 92%, and PD 99%. 22

Donors
Even including the 'party tax', private donations have been less generous in more recent years as they generally shrank (apart from the UK, where they have increased in absolute terms from 12.8 million euros to 15.9 million euros) (see Table 4). Notwithstanding the general abatement of private donations, their share in the parties' budget increased, especially with respect to 2008 (see Figure 5). Italy and the UK exhibit the largest growth in the share of donations (above 50%). However, they reflect two different patterns: a tradition of reliance on this source for the UK parties (the 'big donor culture'; Kelly 2011), and a vital necessity for Italian parties after the shut-down of public financing. Of the UK parties, not surprisingly the Conservatives depend on donors for as much as 70% of their total budget: the highest share among all parties considered. The Labour Party follows suit with 50%. In fact, the two parties attract different kinds of private money: corporations and big business for the Conservatives, trade unions (plus some wealthy donors in the Blair years: see the 'cash for honours' case) for Labour.
The other parties are less capable of raising money from donors: the PP seems almost totally uninterested, or finds the idea unappealing, and the UMP and CDU rely on that source for 25-30% only. Italian parties were quite uninterested in fundraising activities until the reforms of 2012-14: the elimination of public funds has forced Italian parties to seek private contributions, and, in fact, after that date, every party has exceeded the 40% threshold (with Forza Italia even above this).
In conclusion, membership fees and private donations have increased in importance for the financial sustainability of political parties. The UK and German parties have traditionally based their financial equilibrium on these sources although for different reasons: in the former case, following a long tradition of self-financing, with all the problems related to it, and the latter for the matching mechanism of the German financing system. Italian parties have dramatically increased their reliance on private funds in recent years: until the implementation of the new law in 2012-14, the share of private support had not reached 20%, but afterwards jumped up to 54%.
Party financing and the Great Recession As described above, Table 1 reveals a curvilinear trend in the total income of political parties, with a growing availability of money until the end of the 2000s and a shrinking of resources after 2008, taken as the turning point of the economic crisis. This trend leads us to conclude that the Great Recession had an impact on the financial wealth of political parties. To test this hypothesis, a regression model which takes into account country and party variation for different covariates has been introduced. The model uses one dependent variable, two independent variables at the country level and five control variables at the party level. The dependent variable is the total income of political party. The variable has been standardized to account for the intra-country variation: thereby the differences in parties' income and the peculiarities related to the different funding regimes are controlled. Therefore, in the model the natural logarithm of the total income of a party (log_income) is used.
The two independent variables at country level are related to economic indicators; they are: (1) the level of the GDP per capita between 2002 and 2016 and (2) the amount of public expenses. The hypotheses underlying the choice of these two variables are: (1) the higher the country's GDP, the higher the party's income; and (2) the higher the public expenses, the more parties receive public funds. Given the different levels of GDP and public spending between the five countries, both variables have been standardized; this was particularly important for the public spending because of the different funding regimes.
Finally, five control variables at party level are introduced. The first one distinguishes election years from non-election ones, as during election years the state usually provides parties with high levels of subsidies; this variable is treated as a dummy variable, where the election year has been coded as 1, and other years as 0. The second variable is the votes in general elections as the election outcome shapes the distribution of state funding: the more votes a party gets, the higher is the state financial support. The third variable is the party's presence in government or in opposition, as holding office might boost the party's income through generous donations; this variable is treated as a dummy variable, where 0 indicates a party in opposition and 1 party in government. The fourth variable takes into consideration the level of public financingthat is, the party's dependence on state fundingfor each party; given the different regulations in allowing public funds, the amount of public financing by country has been standardized. The last control variable differentiates political parties according to their ideological alignment. This variable is treated as a dummy where left-wing parties (DS, PD, PRC, SEL in Italy; PS and PCF in France; SPD and Linke in Germany; Labour in the UK; and PSOE and Podemos in Spain) are coded as 1, and all the others (moderate, conservative and right extremist) as 0. As we have seen above, political parties on the right of the political spectrum are usually more dependent on private financing than their leftist counterparts, and parties from the left side rely more on membership and public financing. The model thus seeks to test this relationship. Table 5 presents the results of the linear regression model at party level. 23 The data set can be considered as a panel set because it analyses the financial budgets of political parties over time. However, the country-level variation of the number of political parties during the time span considered has led to the adoption of a linear regression model rather than a fixed effects model. The overall variance explained by the model (68%) indicates that the covariates identify the factors shaping the income of political parties. 24 The level of public expense, the share of votes collected by a party at a national election and its presence in government are significant and, as expected, positively related to the income of the political party. The same positive relationship emerges from the public funding dependence: the higher the level of electoral support for a party, the higher its income. Finally, the ideological spectrum plays a role as well: left-wing parties rely more on membership fees and state supportcombined together, these represent the larger share of income rather than private fundingcontrary to what occurs in the right-wing parties. The model assesses whether the level of public expense (beta = 0.310), regardless of the general economic situation (GDP per capita) of each country, contributes to explain the variation among parties. On the contrary, the GDP per capita, although it is positively related to the increase in income of political parties, is not significant. The same is true for the election year. Even if the flow of money in a party's coffers is directly related to the electoral cycle (as the positive coefficient indicates), the variation in the amount of total resources available to parties seems not to be dependent on it.
In conclusion, the regression model indicates that the most significant covariates which are positively related to the parties' income are some party-related variables, such as, in ascending order, political alignment (beta = 0.103), electoral outcome (beta = 0.219) and amount of state support received (beta = 0.676). The quite high coefficient of the state support suggests that, beyond the actual (decreasing) amount of public contributions, this entry represents by far the most important explanatory factor of parties' level of income. In other terms, parties still rely on state support.

Conclusion
This article partly confirms and partly revises some of the information the literature on political financing has offered in recent years. The novel outcomes of the present research were determined by an extension in time and an in-depth analysis of the major parties' official budgets, in five European democracies.
The first primary research question of this article dealt with the trend of parties' income. Our data challenge the hypothesis of continuous financial growth. Having expanded the time span to 2016, we have demonstrated that the main parties' income, on average, has faced a retrenchment in more recent years. While parties did increase their income from 2002 to 2008 (with the exception of the UK), their revenues were curbed in 2016 by almost one-quarter compared with 2008. The connection with the Great Recession is thus likely. This outcome has raised a second question: what is the source of this reduction in income? Which of the three (comparable) main entries in the party's budgetdirect and indirect state financing, membership fees and private donationsshrank the most?
Before answering this question, let us recall the relative weight of these entries. Overall, at the beginning of the 2000s the single most important source of money was provided by the state, although with some remarkable variations among parties. In general, the state covered around 40% of the budget, a quota which rises to 50% if one excludes the UK, where state support is very limited. Private donations covered around 17% of the total income (but more than 20% if we project on 2002 the level declared later by UK parties, when data on this entry were finally included in the official budgets). Membership fees ranged around 15% (but less than 10% if we exclude the German outlier, where parties collect around one-third to one-quarter of their money from their members). The remaining source of income comes from a series of entries such as occasional fundraising, promotional/commercial activities, property/legacies/leases, services and financial activities, and extraordinary sales.
The parties' financial opulence ended by the turn of the decade, in connection with the Great Recession: in 2016 the income of all parties decreased by 17% overall, and went back to 2002 levels, save for a modest +2%. Much of this retrenchment is due to the radical cut in state provisions for Italian parties since 2012, and is not counterbalanced by the (sole) increase of UK parties' income.
This reduction in four out of five countries affects the source of income of the parties under analysis in various ways. As far as state funding is concerned, even after the 20% cuts in state funds, the two main Spanish parties (PP and PSOE) prove to be even more dependent on the state: the share of public money in their accounts reaches the highest level ever: 83.8%. UK and French parties moved slightly downwards (−2%). Italian parties, on the other hand, saw the share of their income from public funds shrink dramatically from 78.5% in 2008 to 33.1% in 2016: the state is no longer as generous as it used to be.
Membership fees continue to be marginal as a financial resource. Not even in the face of a state retreat have members' contributions increased significantly. The figures, however, suggest a possible change of attention towards the capacity of extraction of more money from the members: the increase, as modest as it is, should be considered in relation to the general decline in the number of members experienced by every party (van Haute and Gauja 2015). Apparently, the fewer the members, the more generous they are in contributing to the party (or the fees introduced by the party have been higher and higher). 25 The only exception to this rule relates to the British Labour Party. The internal harsh competition for the leadership under Jeremy Corbyn, plus the highly publicized reduced cost of the party card boosted individual party enrolment, which even exceeded (with more than 500,000 individual members) the previous record during the early Blair years (Audickas et al. 2019). These intra-party dynamics in the Labour Party could explain the general 6% increase in income from UK parties' membership fees.
As state funds diminish and membership fees stagnate in the parties' budgets, the only financial provision to compensate for this decline should come from private donations. Indeed, in 2016 this source represents one-third of parties' income, with a constant growth over time. Even after the Great Recession, private donations increased their share of the total budgets by 9%. This general trend, however, hides large differences among countries. While in France, Germany and Spain, the share of private donations moved up only imperceptibly, in the UK and Italy it changed in scale. UK parties, which traditionally rely on donors, and eventually on big ones (and on trade unions, in the case of Labour), because of the paucity of public funds, get 57.8% of their money from that sourcedouble that of the French and German shareand in 2016 private donations increased by 15% compared with 2008. Italian parties, entirely deprived of direct state financing in the past few years, have had to resort to private donations; they were effective, as they jumped from 17.4% in 2008 to 54.2% in 2016, although with reduced budgets.
Inside the basket of private donations, there is also a peculiar form of financial support, the 'party tax', that is, the contributions from elected officials. Unfortunately, it is not possible to disentangle this peculiar kind of contribution from the main entry of donations in every country. In Germany and France, where the official budgets provide separated figures for this entry, it represents, respectively, 62% and 42% of the private donations: a very important source of financing. On the basis of these figures, it could be guessed that in the other countries contributions from elected official also sustain the party budgets in a substantial way.
This consideration leads one to believe that parties have become more dependent on their representatives. This would suggest a further dominance of the party in public office over the party in central office, as suggested by Katz and Mair (2018). However, one can also view it in a different way. Indeed, as the current name of this practice -'party tax'suggests, it instead seems to be a kind of levy imposed by the party in central office on the representatives. Therefore, the internal balance of power seems to have shifted to the hands of the party headquarters that still employ symbolic (loyalty and attachment to the party) or more mundane (reselection) incentives to convince representatives to pay their, often substantial, contributions. On the other hand, financial state support, as well as MPs' salaries, are determined by the rules approved by national parliaments; these rules are usually approved by a large, sometimes unanimous, consent, thus reinforcing that image of a cartel of parties.
In conclusion, political parties no longer profit from the financial largesse they enjoyed before the Great Recession. Their budgets have shrunk to some extent, with large variations between parties and countries. The main cause of this change, as revealed by the regression model, is due to a reduction in state funds: the public cornucopia has lapsed. This downsizing has not been counteracted by a more active party self-financing regime. The money coming from membership fees has not moved upwards significantly, and private contributions have increased only in a few cases, concentrated mostly in UK and Italian parties (in the last years). In addition, large amounts of private money, where data are available, derive from the party tax on elected officials. Parties still need state support to keep their balance secure. But that source is apparently less and less attainable.
Notes 1 See www.idea.int. 2 See www.coe.int/en/web/greco and www.venice.coe.int/webforms/events/. 3 The term 'relevant' is not used here in a Sartorian sense (Sartori 1976: 121-125), but in a more colloquial way to indicate parties with more than a certain percentage of votes. See the Appendix for the list of political parties included. 4 The present data set -Party Financing Comparative Datasetis based on the official national budget records of the national party's head office, usually under the heading of 'Consolidated Income and Expenditures Account'. In addition to the budget analysis, party documents from the accounts department/treasurer have been taken into consideration to validate some unclear points.
The source of income of different parties' official budgets, expressed in euros and indexed at GDP 2010 are standardized in the following way: total income; public financing (for ordinary activities, for electoral expenses, subventions from taxes, from coalitions or parliamentary groups); private financing (corporate donations, individual donations); internal resources (membership fees, donations from elected officials); donations from other political organizations; fundraising and promotional or commercial activities; property/legacies/leases; services and financial activities; extraordinary sales; other sources of income. 5 When not otherwise specified, all information is retrieved from the International IDEA Political Finance Database, accessed August 2018: www.idea.int/data-tools/data/political-finance-database. 6 The 1994 reform of public political financing (Gesetz über die politischen Parteien -Parteiengesetz) set a maximum annual amount (around 133 million euros) to be distributed among political parties on the basis of their fundraising performances. Therefore, the total income of the main political parties tends to be stable (at the highest level) and not amplified by public intervention. Only a change in the party system with the entry of new parties could have modified the distribution of state funds. This is precisely what occurred in 2016, with a shrinking of the total income for all (relevant) parties. 7 Each candidate reaching 5% of the votes receives a reimbursement for 47.5% of their expenses, provided that expenditure does not exceed the limit of 38,000 euros; that can be augmented at a rate of 0.15 euro for each inhabitant of the constituency. 8 In the German and French parties' budgets, the amount of money from party representatives is reported under a specific entry; in Italy, Spain and the UK, contributions from officeholders are considered private donations and thus are recorded without any specific entry in the budget. 9 The Spanish parties also receive generous public support at regional level, but these instalments are not included in the national budgets. See Rodríguez-Teruel and Bértoa (2018). 10 The PDL used to incorporate state subsides allotted for the entire legislature in the budget of the first year of elections; therefore, the budget skyrockets in the two electoral years (2006 and 2008) to 130 million euros and 160 million euros, respectively, only to plummet afterwards to much more limited amounts, as the party then receives resources from its membership and private donations only. This deviant accounting decision would have distorted any analysis, and thus the state subsidy is divided by the number of legislative years and distributed accordingly. 11 This is a case where the reading of the documents attached to the budgets clarify otherwise 'mysterious' figures. 12 In detail: (1) the M5S is not registered as a party on the national official roster and therefore it cannot raise any money from the 0.2% tax deduction, accessible to registered parties only; (2) no fee is required for membership; (3) it is articulated in several, distinct, national structures ( the 'official' M5S movement, the electoral committees for national, subnational and European elections, and, since 2016, the Rousseau Association, offspring of the internet private company Rousseau, owned by one of the party's founders), each one of which presents separate financial reports; (4) a large portion of the representatives' salary goes into a special fund which redistributes the money to non-partisan and non-political initiatives. (The total amount of money delivered by representatives to this fund from 2003 to 2019 reached 107 million euros; see www.tirendiconto.it.) 13 See www.movimento5stelle.it/rendiconti/Associazione_Movimento_5_Stelle_rendiconto_2015.pdf. 14 For example, the 'Committee Italy 5 Stars' was created to run a budget of 892,835 euros for a massive rally in Rome in 2018, and closed down just after the rally. See https://associazionerousseau.s3-eu-west-1. amazonaws.com/documenti/Relazione_sulla_gestione.pdf. 15 The PD budget is always underestimated because, unlike all other parties considered in this analysis, it does not include in the national budget the money coming from membership fees as they are collected and retained at the local level, save for those (few) directly enrolled via the central headquarters. This is why the membership fee share is so low in spite of the large number of members. 16 Public subsidies to political parties in the present Party Financing Comparative Dataset cover different forms of public intervention: subventions for ordinary activities of parties (such as in Spain, France and Germany), reimbursement for electoral expenses (in all the countries investigated), subventions from taxes (in Italy since 2012) and state funding received from the electoral coalition as reimbursement (as in Italy). The UK presents the sum of Short and Cranbourne Money for opposition parties and the Policy Grant divided among all parties. 17 As explained in note 15 above, the PD national budget does not include the membership fees that remain in the local structure: almost all members enrol in a local branch. Therefore, the PD dependence on the state is grossly skewed by this peculiar accounting method. Moreover, as specified below in the text, the distinction between direct and indirect funding is essential to gauge the PD's financial profile. 18 In 2003, the UMP's proportion of state funds peaked at almost 80% of its revenue because of its exceptional success in the 2002 presidential elections, when Jacques Chirac received 82.4% of the votes against FN leader Jean-Marie Le Pen, with consequent good performance in the following general elections. 19 The PSOE did not declare revenues from members until 2012; after that period their share is around 20% of total income. 20 Figure 4d presents a peculiar trend of Forza Italia membership revenues. Forza Italia and later the PDL had a rather uncommon recruitment style. In some years recruitment was not even implemented; at other times it was highly emphasized, even with plurennial membership. This procedure, and the scant emphasis on membership recruitment, do not allow for an assessment of a precise trend, even if the jump in 2014 might reflect the call to former members when Forza Italia was reinstated after the disruption of the PDL. 21 This is possible only for the German and French parties, as the respective financial regimes provide a specific entry in the parties' budget of the contributions from elected representatives, i.e. the party tax. 22 As financial regulation in Italy does not require any specific entry for representatives' contributions, data are calculated by the authors on the basis of the single donors' list. 23 Political parties in Italy have been considered until 2013 because the subsequent cut in public financing of political parties had such an impact on their income and cannot be compared with other political systems. 24 Correlations among covariates have been checked and a collinearity test was run to avoid redundant explaining factors. 25 This aspect would need to be tested by controlling the regulations over time, party by party. Unfortunately, no comparative data are available, to the best of our knowledge.