Bicameralism and Italy’s present system of government

Perhaps the leading feature of the Italian constitution that has led to legislative gridlock is its system of full and symmetric bicameralism. Because both chambers of the legislature must approve identical bills in order for them to become law, the legislature tends to be unsuccessful in passing consequential legislation. This is not, however, a feature that is unique to Italy’s system of full bicameralism. Rather, bicameralism is known to generally reduce policy change, as it adds an additional veto player to a system of governance. Figure 1 depicts this dynamic. In Figure 1a, bill passage requires only a simple majority of all legislators in the upper and lower chambers: four out of six of the depicted legislators. This arrangement produces a winset [W(SQ); the set of policies that the hypothetical unicameral legislature prefers to the status quo] that includes the three shaded petals in Figure 1a. In Figure 1b, however, the legislature is now bicameral: members of parliament are divided into upper (U) and lower chamber (L), and concurrent majorities of both chambers are required in order for a piece of legislation to pass. Under this arrangement, W(SQ) shrinks to just one of the three petals present in Figure 1a. In other words, by adding a second chamber as a fully autonomous veto player within the legislative branch, the set of bills that could conceivably beat the status quo shrinks, and policy change abates (Tsebelis and Money, Reference Tsebelis and Money1997). This reduction in the size of the winset has an additional, important consequence: the role of the agenda setter is reduced under bicameralism. Indeed, the government of a parliamentary system has less prerogative when confronted with a bicameral legislature than when confronted with a unicameral one.

Figure 1 Winsets under a unicameral and bicameral arrangement. (a) Winset under a unicameral arrangement, (b) winset under a bicameral arrangement.

As is the case with all other veto players, however, the distance between chambers in a bicameral system also restricts the level of policy change possible within a legislature. As Madison describes it in the Federalist Papers, ‘the improbability of sinister combinations will be in proportion to the dissimilarity in the genius of the two bodies’ (Hamilton et al., Reference Hamilton, Madison and Jay1961 [1787–1788], 379, no. 62). The argument is presented with contemporary terminology in Figure 2. Here, the lower chamber (L) and the upper chamber (U) combine to create a winset, W(SQ), represented by the lighted shaded region. However, when the upper chamber shifts further away from the lower chamber (U’), the winset shrinks to the darkly shaded region. Lower and upper chambers may drift apart for any number of reasons, such as differences in election cycles or the size of their districts. Regardless of the reason, however, a growing distance between the chambers will exacerbate bicameralism’s quality of reducing policy change.

Figure 2 Winset of veto players (VPs) L and U’ is contained within winset of VPs L and U.

Illustrated another way, bicameralism works in much the same way as a qualified majority in a unicameral legislature. In an exchange that has since been largely lost (possibly in World War II, see McGrath, Reference McGrath1983; McLean and Urken, Reference McLean and Urken1992), Madison and Italian doctor (and American apologist) Philip Mazzei debated the merits of their favored systems: the first arguing in favor of bicameralism, and the second (citing personal correspondence with Marquis de Condorcet) in favor of unicameralism with qualified majorities. Figure 3 depicts the essence of what might have been their arguments in modern political science language. In the first half (Figure 3a), I present the ‘core’ [i.e. the set of status quo policies that can never be overcome; see Tsebelis (Reference Tsebelis2002) for greater detail] of a seven-person legislature under a 5/7ths qualified majority requirement.Footnote ⁴ Each line represents a different five-person coalition that might support a particular bill. When one accounts for all possible five-person coalitions, the intersection of these coalitions (the darkened center in Figure 3a) represents the set of status quo policies that can never be successfully changed – the core of this qualified majority, unicameral legislature. Policies under this configuration will possess at least two features. First, policies will be concentrated at the center of the political spectrum, as depicted in Figure 3a. Second, policies will change less frequently in the qualified majority case than in the simple majority one, all else being equal.

Figure 3 Cores of unicameral legislature with qualified majority and bicameral legislature. (a) Winset under a unicameral arrangement with a 5/7ths majority requirement, (b) winset under a bicameral arrangement.

Bicameralism, depicted in Figure 3b, produces a different shaped core. In this scenario, two chambers (P and C) must vote in favor of a bill in order for it to pass. Much as in the case of the qualified majority scenario, legislators P1, C1 prefer policies above line segment P2C2, legislators P3, and C3 prefer policies below, while legislators C2 and P2 are indifferent. Thus, the core of this legislature – the set of status quo policies that can never be defeated – is the line segment PC. This result produces a core with a different shape (a straight line in two dimensions), but it is still centrally located between the preferences of the two chambers. In other words, the unicameral, qualified majority legislature and the bicameral legislature will both result in policies that lie in the center of the political spectrum. Furthermore, both arrangements decrease the amount of possible policy change, relative to the simple majority unicameral case.

In conclusion, because bicameralism adds an institutional veto player to a political system, it is bound to decrease policy change – a result that is exacerbated when the chambers drift apart. It is also likely to concentrate a country’s policies into a small region within the center of a country’s ideological spectrum. These theoretical results explain Italy’s history of ‘inconclusive politics’, as Prime Minister Renzi describes them. For much of Italy’s history under the 1947 constitution, the Senate did not restrict the set of feasible outcomes (winset of the status quo) because of its ideological make-up per se: after all, it had the same composition as the Camera dei Deputati. Instead, lack of political alteration (dominance of the DC throughout the period) as well as multiple veto players (either in the form of compromesso storico or pentapartito governments) led to ‘immobilismo’. The electoral reforms in 1993 and, even more in 2005, created ambiguous results. On the one hand, alternation in the Italian political system reduced of the number of partisan veto players, since the winning coalitions were either center-left or center-right and did not include parties on both sides of the spectrum. On the other hand, in the 2005 electoral reform, the bonus distribution at the national level for the House of Deputies and at the regional level for the Senate, in combination with ‘perfect bicameralism’, increased the number of institutional veto players. The resulting complication was compounded by the intervention of the Constitutional Court, which declared the electoral reforms unconstitutional.

This ambiguity generated the present need for reform. The remainder of the paper examines the results, arguing that although the proposed reforms would likely achieve their goal of decreasing the difficulty of policy change, they would also render constitutional revisions more difficult and increase the role of the judiciary in constitutional interpretation.

Consequences for constitutional revision

While Italy’s reforms drastically decrease the powers of the Senate in most legislative situations, one area in which full bicameralism remains is constitutional reforms and constitutional laws. According to the Italian Constitution (Part II, Title VI, Section II, article 138), the Senate indeed plays an important role:

This set of provisions specifies two alternative procedures for constitutional revision: revisions may occur via two-third majorities in both chambers of the legislature, or by simple majorities plus a referendum.Footnote ⁵

Figure 4a demonstrates the core of the first procedure for revision, two-thirds majorities in both chambers. I represent each chamber with five members (L1-5 and U1-5) and look for agreement from four out of the five members in each chamber, as four out of five most closely approximates the two-thirds majority required to successfully pass constitutional reforms under the first procedure. Under this arrangement, the pentagon C_L represents the core of the lower chamber: any point outside this pentagon can be defeated by its projection on the pentagon, since four out of the five members of the lower chamber would prefer this solution. Similarly, the pentagon C_U depicts the core of the upper chamber. Similar to Figure 3b, then, the bicameral constitutional core in this case is comprised not only of both C_L and C_U, but also of the area between them (the shaded area in Figure 4a).Footnote ⁶

Figure 4 Constitutional core of Italy under article 138. (a) Constitutional core for revision by concurrent two-third majorities in each chamber, (b) constitutional core for revision by concurrent majorities and referendum, and (c) constitutional core (intersection of two alternative procedure cores).

Figure 4b captures the second possible procedure for constitutional revision: concurrent simple majorities in both chambers, plus a referendum. In this case, the bicameral core by simple majority is located along the line segment L2U2. Indeed, this line is a ‘bicameral median’; that is, it has a majority of members of both houses on either side of it, like P2C2 in Figure 3. It follows that there is a majority in both houses that would prefer, over any point outside this line, its projection on the line itself. The bicameral core, however, does not extend beyond point L2 and U2: anything outside the solid line in the figure does not command a majority in one of the chambers. So, the bicameral core is the solid line between L2 and U2. In order to calculate the whole constitutional core, including the referendum requirement, one must factor in the voters (point P) and expand the core. The shaded triangle depicts that addition.Footnote ⁷

The core of the Italian Constitution exists at the intersection of the cores of the two procedures delineated in Figure 4a and 4b. Indeed, any point that belongs in only one of the procedural cores can be changed by using the alternative procedure. This intersection is represented by the darkly shaded area in Figure 4c. It cannot be a priori assessed which one of the two procedures is easier to be used. Indeed, this depends not only on the institutional rules, but also on the actual preferences of the actors. However, the system does behave in predictable ways. For example, if the preferences of the people are moved much closer to both houses than Figure 4 indicates, then it is easier to make a constitutional revision with a referendum than with two-thirds of both chambers. This seems to be the case today, because it is difficult to imagine that it would be easier to get a two-third approval of the Senate to reduce its own powers than an approval by the people. Similarly, if the two houses drift apart, the segment LU in Figure 4 will become longer, and the core will expand. Indeed, all three cores in Figure 4a–4c will become larger, and constitutional revision will become more difficult. This will be the case under the electoral system for the Senate created by the proposed constitutional reforms.

Under the proposed reforms, the Senate will be elected in a method far different from the lower chamber of the legislature. That is, Senators will be indirectly elected through regional councils. While these changes are likely to have a wide variety of ramifications, one likely consequence will be that the difference in electoral make-up of the upper and lower chambers will increase the distance between these two constitutional veto players. As a result, constitutional change will become more difficult in Italy, increasing the rigidity of an already rigid constitution. I remind the reader that a previous attempt to change the Constitution (under the Berlusconi government) failed. The simple reason is that the required procedures discussed above make constitutional revisions difficult in Italy. For comparison, Table 1 shows measures of constitutional rigidity in different countries, indicating that Italy has one of the most difficult constitutional revision procedures (Tsebelis, Reference Tsebelis2016). An affirmative answer on the referendum regarding constitutional reform will make future constitutional revisions even more difficult to achieve, because the chambers are likely to drift apart.

Table 1 Constitutional rigidity in some democratic countries

Bolded value underscores where Italy lay in comparison to other countries’ constitutional rigidity. Italy already ranks quite highly.

Consequences for policymaking

In addition to this change to the process of constitutional revision, probably the most consequential change initiated by the proposed reforms occurs within the policy-making powers of the Italian Senate. Rather than maintaining a perfectly symmetric bicameral system, the reforms strip the Senate of most of its legislative veto powers, reducing the number of institutional veto players in the Italian system of governance. As a result, Italy is likely to experience far more policy change than it has in years past. In addition, the power of the government will increase (given that it controls the agenda, in a system where the number of veto players decreases). Whether or not more policy change is a positive or normative good, however, depends on one’s perspective. Economists, for instance, tend to argue in favor of less policy change, which they argue allows for the stability that businesses need in order to make long-term investments. In their Nobel Prize-winning article Kydland and Prescott (Reference Kydland and Prescott1977) argue that rules are better than discretion, and that the multiplicity of veto players makes discretion impossible. Similarly, political conservatives prefer less policy change in general. Political scientists, however, argue that too much policy stasis disallows governments to adapt to an ever-changing world. Progressives adopt a similar logic in their stance in favor of more policy change. The previous analysis does not lead to a confirmation of either of these unconditional judgments. A strong Senate leads to less policy change, but the normative assessment of policy stability depends on one’s assessment of the status quo.

Many empirical studies confirm the straightforward relationship between multiple veto players and policy inertia (or the absence of policy change). Kastner and Rector (Reference Kastner and Rector2003), for example, find that democracies with more partisan veto players enact fewer changes in capital controls. Hallerberg and Basinger (Reference Hallerberg and Basinger1998) show that OECD countries with more veto players tend to change their corporate and personal income tax rates less (in response to the US tax reform of 1986). O’Reilly (Reference O’Reilly2005) finds similar results for tariffs, as do Tsebelis (Reference Tsebelis2002), Tsebelis and Chang (Reference Tsebelis and Chang2004), and Giuriato et al. (Reference Giuriato, Cepparulo and Barberi2016) for labor laws, budget distribution, and fiscal forecast manipulation, respectively. All these studies indicate that increasing the number of veto players increases policy stability; however, they do not comment on the positive or negative effect (valence) of this stability.

Nevertheless, some studies do appear to underscore clear benefits or drawbacks of policy stability from veto players, but these studies typically require additional arguments above and beyond the argument of ‘inertia’, to substantiate their claims. Henisz (Reference Henisz2000), Acemoglu and Johnson (Reference Acemoglu and Johnson2005), and Fatás and Mihov (Reference Fatás and Mihov2013), for example, all argue that policy stability leads to security and investment, ultimately increasing economic growth. Henisz and Mansfield (Reference Henisz and Mansfield2006) find that increases in policy stability lead to more trade openness, because protectionism occurs in periods of recession, and many veto players make policy change (including increases in protectionism) more difficult. Roubini and Sachs (Reference Roubini and Sachs1989), Alesina and Drazen (Reference Alesina and Drazen1991), and Alesina and Perotti (Reference Alesina and Perotti1995) find that large numbers of veto players lead to increases in public debt. A more recent study by Franzese (Reference Franzese2002), however, presents a more skeptical view of the impact of veto players on the quality of outcomes. By increasing the number of countries (in particular including Switzerland with many veto players and low budget deficits) in his sample and examining longer periods of time, Franzese finds that large numbers of veto players simply create inertia behind whatever level of debt a country already has, rather than uniformly increasing public debt. In this respect, Franzese’s analysis is more similar to studies on inertia presented in the previous paragraph. Similarly, while some other studies do find that veto players may have an impact on the quality or content of outcomes, the relationship is rather nuanced. In responding to financial crises, for example, MacIntyre (Reference MacIntyre2001) and Angkinand and Willett (Reference Angkinand and Willett2008) find that both too many and too few veto players may intensify the effects of banking crises. Indeed, while many veto players disallow a government from adequately responding to a crisis, too few veto players may create instability that undermines consumer/business confidence.

In his 2007 book on decentralization, Treisman analyses all existing outcomes-based arguments concerning decentralization, whether in favor of (including administrative efficiency, competition among local governments, support to local economic, citizen participation and development of civic virtue, checks to the power of central government, policy innovation, diffusion of ethnic conflict) or against (pressures from central government for aid, inefficiency, and failures of fiscal coordination). His analysis is not focused on veto players, and extends both to the theoretical and the empirical level. Here are Treisman’s conclusions: ‘it is hard to reach any general conclusions about whether political—or administrative, or fiscal—decentralization will improve or impair the quality of government and economic performance. They will have many effects, driving in different directions on different dimensions’ (Reference Treisman2007: 274). In fact, according to Treisman, only one general conclusion may be made about decentralization: ‘If political decentralization increases the number of actors whose acquiescence is needed to change policies, this will—other things being equal—tend to entrench the status quo’ (Reference Treisman2007: 274). This, of course, was the reality in Italy for several decades, and the present reforms of the electoral system and of the Senate move in the opposite direction. By removing the Senate as an institutional veto player in the system of governance, Italy will experience more policy change in the future, should it follow through with its reforms. The effect on the economy or growth, however, remains indeterminate.

Consequences for institutional balance of power

In addition to constitutional revision and policymaking, Italy’s reforms would have important consequences for balance of power between branches of government if adopted, particularly between the legislature and judiciary. More specifically, this analysis argues that while the reforms would decrease the statutory interpretation power of the judiciary, they would amplify the courts’ power in constitutional interpretation.

Figure 5a illustrates the intuition for the statutory component of this argument. Assume for a moment that there are three legislative veto players in Italy, such that the triangle 123 connecting the players forms the legislature’s core (again, the set of points that they cannot possibly agree to change). Assume also that J and K represent statutory interpreters (namely, judges). Under this arrangement, if a judge selects one of the points inside of the core, a legislative override is not possible. Thus, Figure 5a illustrates three different possibilities. In the first two cases, the judge’s ideal points J and K are outside the legislative core, and they select the closest core point to them (J’ and K’, respectively). Despite the fact that these two choices are significantly different from each other, the legislative veto players are incapable of changing either of them. In the third case, the judge is located inside the legislative core, but changes her mind and moves from point L1 to point L2. Since her ideal point is inside the core, she can select it. Here again, the legislative veto players can do nothing to stop her.

Figure 5 Policy selection by judiciary within legislative and constitutional cores. (a) Policy selection by first movers (judges) within legislative core, (b) policy selection by judges within legislative and constitutional core.

In each of these cases, the power of a judge to move policy increases with the size of the legislative core. In other words, if judges hold the power to interpret existing statutes, they can move the statutes unilaterally, and then wait for the legislature to respond. The larger the core, the larger the number of status quo relocations to which the legislature will be unable to respond. This basic point explains why the judiciary’s statutory interpretation powers should decrease in response to Italy’s proposed reforms. As noted above, the removal of the Senate as a legislative veto player would decrease the size of the policy-making core within the legislature. As the core shrinks, the judiciary will find fewer opportunities to relocate the status quo without legislative overrule: a small core means that the legislature can retaliate on a larger number of possible statutory interpretations. Empirical evidence in favor of this argument has been presented by Tsebelis (Reference Tsebelis2002) for advanced industrialized countries and from Andrews and Montinola (Reference Andrews and Montinola2004) for developing ones. Thus, the proposed reforms would decrease the courts’ power in statutory interpretation in Italy.

Still, while the proposed constitutional reforms would decrease the judiciary’s statutory interpretation power, they would increase the courts’ power in another realm: constitutional interpretation. Consider now Figure 5b. Like Figure 5a, the triangle 123 represents the legislative or statutory core, and actors J and K are the ‘first movers’ (judges in our case, though they could also be bureaucrats). Unlike Figure 5a, however, another actor, actor 4, is present in Figure 5b. Actor 4 represents the additional approval needed for constitutional revisions – either additional approval in the legislature (up to two-thirds of both chambers), or the approval of the populace in a referendum. With the addition of 4, the core grows – now encompassing the quadrilateral 1234. Under these circumstances, J and K’s power of interpretation increases: instead of having to project her ideal point onto the perimeter of 123, K (for example) may now select her ideal point. And on constitutional issues the preferences of the Court may shift from C1 to C2 (or conversely) without the possibility of reaction by the political system (constitutional revision). As before, when the size of the core increases, the ability of the legislature (and/or other veto players) to overrule a court’s decision decreases – ultimately empowering the courts.

This theory of the judiciary’s power in constitutional interpretation has already received empirical support in the Italian context. Indeed, in examining unconstitutional sentence reversals by Italian courts, Santoni and Zucchini (Reference Santoni and Zucchini2004; from now on S-Z) find that the number of reversals closely tracts the ‘constitutional range’ in Italy (i.e. the size of the constitutional core). It is interesting to note that S-Z calculate the constitutional core in two dimensions, along the argument presented in Tsebelis and Chang (Reference Tsebelis and Chang2004), using the first two dimensions from Laver and Hunt’s (Reference Laver and Hunt1992) book. Figure 6 depicts the relationship between sentence reversals and constitutional range, derived from S-Z’s data. Indeed, as constitutional range increases, so too does the independence of the judiciary.

Figure 6 Core size and judicial independence in Italy.

S-Z’s analysis has been criticized by Fiorino et al. (Reference Fiorino, Padovano and Sgarra2007). They consider the appointment process of the judges, and when they include their own set of variables, they claim that S-Z’s findings (on which the analysis in this article relies) are reversed. Yet, as Fiorino et al. acknowledge, they do not consider constitutional decisions alone. They claim that ‘The consideration of all types of decisions that the Court may undertake in reviewing legislation makes our analysis more adherent to reality’ (Fiorino et al., Reference Fiorino, Padovano and Sgarra2007, 695). As a result, their dependent variable (unlike S-Z and this article) does not address constitutional decisions. In addition, in terms of their independent variable, they consider the ‘Herfindahl index of the fragmentation of parties in the government coalition’ (Fiorino et al., Reference Fiorino, Padovano and Sgarra2007, 692) as a proxy. However, this variable has nothing to do with the constitutional core used in this analysis (as well in the S-Z argument) for at least two reasons: first, it covers the government alone, and not the required two-third majority required by the constitution, and second, it is a function of the number of seats in Parliament that each governing party has, and not of policy positions. To make things clear, the Fiorino et al. variable depends on the distribution of seats in the government coalition: fragmentation increases the weak main coalition partner. On the other hand, the S-Z variable of core size depends on the positions of the parties of the constitutional core (including the Communists), and increases when the distance between Communists and the most conservative government coalition partner increases. Consequently, the Fiorino et al. arguments (with different independent and dependent variables) not only do not invalidate but cannot be even considered relevant to the S-Z arguments or the arguments in this article.Footnote ⁸

With regard to Italy’s proposed constitutional reforms, then, both empirical and theoretical research suggests that judges’ power of constitutional interpretation will increase. Indeed, because the reforms will render constitutional revision more difficult (growing the constitutional core), the judiciary will face fewer possibilities of constitutional revision for which the legislature could agree to overrule their decision.Footnote ⁹ Figure 6 uses S-Z data to show the historical evolution of Constitutional Court decisions to invalidate government decisions on grounds of constitutionality, along with the size of the constitutional core. The two time series seem correlated, and the next pictures shows that the visual impression is correct: the independence of the court is correlated with the size of the constitutional core.

As a result of the proposed reforms, then, Italy would see two important changes that are relevant to the power of the judiciary, depicted in Figure 7. First, it would see a decrease in the size of its legislative core, shrinking from triangle 123 to 12’3. This would limit the courts’ power in statutory interpretation. However, the reforms would also increase the size of the constitutional core, growing from quadrilateral 1234 to 1234’. This would empower the courts, in terms of constitutional interpretation.

Figure 7 Legislative and constitutional cores, old and new.