Runoff Comebacks in Comparative Perspective: Two-Round Presidential Election Systems

Abstract

In many countries, the president of the republic is directly elected through popular vote. Most such nations have an electoral system that provides for a double-round of voting. Often termed a two-round system, such elections have a second turn if no candidate obtains an absolute majority in the first round. This article presents an original dataset covering all the presidential and the semi-presidential regimes (73 countries) that have adopted two-round system along with the results of 423 elections for both the first and second round and whether the result was determined in the first round, whether the plurality winner of the first round was victorious in the second round, or whether there was a second round ‘comeback’ – when the runner-up of the first round was victorious in the second. A variety of exploratory hypotheses are presented that try to predict the likelihood of a second round or a comeback. Few are validated. The presence of an incumbent and a qualified method of presidential selection increases the probability of a second round occurring while a greater number of candidates and greater electoral volatility make it less likely. We find that no variables related to political institutions, presidential resources, or the structure of the first round predictive of a second round comeback. We make the dataset public so that future scholars can continue this form of investigation.

Keywords

presidential elections two-round systems runoff comebacks presidentialism semi-presidentialism

Introduction

Most regimes select their president via direct popular vote. Presidential and semi-presidential regimes account for 106 of 194 regimes in the world. In plurality – that is, one-round electoral systems –a candidate can be elected receiving relatively few votes. Alternatively, the top two candidates could be very close in vote totals. Some countries require a ‘qualified’ simple majority as a mean of avoiding a second round of voting.¹ Yet, among the presidential regimes, 60.4% have adopted a two-round double-ballot electoral system (TRS) to select their president while 77.4% do so in semi-presidential instances (68.9% for the whole sample) (Table 1, Appendix 1).² Often a week or a fortnight later, a second ballot ‘runoff’ is held if no one candidate wins an absolute majority on the first round. The latter can be reserved for the top two candidates (majority runoff) or more than two candidates (majority-plurality) (Martinez Martinez, 2006).³ The TRS represents one way to avoid candidates being elected with only a small proportion of the popular vote.

Table 1.

The Two-Round Electoral System in Presidential Elections in Comparative Perspective (1945–2020).

Regime type	Total (N)	Of which two-round system (n)	Two-round system (%)
Presidential	53	32	60.4
Semi-Presidential	53	41	77.4
. . . of which
Presidential-parliamentary	21	10	47.6
Premier-presidential	32	31	96.9
N	106	73

Source: Authors’ elaboration on Nohlen et al. (1999, 2001), Nohlen and Stover (2010) and Nohlen (2005). Minister of Interior and Electoral Committees.

The TRS’s runoff is a majority-producing system that requires a candidate to be elected by an absolute majority, if not in the first, then, in the second round (Cox, 1997: 65; Farrell, 2001: 45).⁴ In the event that the first-round runner-up wins the runoff (and is thus elected); then, we have a situation known as a ‘runoff comeback.⁵’ The chance of a presidential electoral comeback is linked to voter behaviour, parties’ strategies and alliances, and the electoral supply of the given profile of candidates. The possibility of comeback would thus require one to analyse the dynamics between the first and the second round; such dynamics of TRS merits further exploration (Samuels, 2002: 469, fn 7). This article introduces a dataset of two-round presidential elections from 1945 to 2020 and investigates the factors affecting the likelihood of runoffs and comebacks. A few relevant cases, among others, employing a two-round presidential⁶ electoral system are Argentina, Brazil, Chile, Colombia, France, Portugal, Poland and Russia. In this article, we analyse all direct presidential democratic elections in countries with presidential and semi-presidential regimes that adopted a two-round electoral system (with a 50% threshold or lower). The sample includes 73 countries and 423 elections for all the countries adopting a TRS. The article presents data on the conditions – both electoral and political – under which the likelihood of becoming president increases for the runner-up. Although a variety of exploratory hypotheses are presented, few are validated. We make the dataset⁷ public so that future scholars can continue this form of investigation.

The Two-Round System in Direct Presidential Elections

Two-round systems are more common for the election of a president than for the legislature (Birch, 2003; Blais et al., 1997). The need for legitimisation (as is inherently attained when an overall majority supports the presidential office holder) of a single national chief executive is greater than that for one legislator out of hundreds. Data show that the double ballot has been adopted in 73 countries, of which 32 have a presidential regime and 41 a semi-presidential one. In the latter case, the subtype distinction between presidential and premier-presidential regimes (Elgie, 2011; Shugart and Carey, 1992) cases is unbalanced in favour of the premier-presidential one (31 vs 10).

From a geographical standpoint, the countries adopting a TRS are distributed as follows: 27 in Africa (36.9%), 17 in Europe (23.3%), 15 in Latin America (20.5%), 13 in Asia (17.8%), and 1 in Oceania.

The possibility for a second round is not only mechanical, but also has various political effects in terms of electoral and political repercussions and the overall openness of political competition. As argued (McClintock, 2018a: 3), TRS opens the political arena to newcomers and it lowers barriers to entry into effective competition in presidential elections (Carreras, 2012; Carreras, 2014; Carreras, 2017). In addition, through the TRS, the winner receives a majority of the electorate increasing the level of legitimisation and decreasing the risk of extremist presidents, as ‘the runoff system allows moderate candidates to pursue their own platform without being forced to merge and compromise with the neighbouring extreme candidates’ (Bordignon et al., 2016: 2349). A TRS subset electoral system, the so-called ‘double complement rule’ (Shugart and Taagepera, 1994), involves providing two distinct ‘ways for a candidate to be elected in the first round: either via surpassing a specific threshold (e.g., 50 percent of the valid vote) or by crossing a lower threshold (e.g., 40 percent of the valid vote) while simultaneously besting the first runner-up by a substantial margin (e.g., 10 percent of the valid vote)’ (Jones, 1994; Jones, 2018: 287).⁸

The TRS ‘underscores the problems created when the initial front runner is defeated in the second round’ (Pérez-Liñán, 2006: 130). With the adoption of rigid thresholds and the subsequent implementation of a runoff election when no candidate receives the required number of votes, there is also the ‘possibility of an inversion of the initial outcome, making a candidate who finished second (or third) in the first round the new president elect’ (Pérez-Liñán, 2006: 131). Presidential electing at the first round (with a plurality or majority electoral system), in the second round, or the possibility of a ‘comeback’ each has potential consequences as it relates to legitimisation, openness of the competition, and role of the incumbent and/or the challenger.

In terms of the independent variables, what is lacking in the literature is how variables (namely, the presidential resources) can affect the probability of a (1) second round or (2) comeback. From a theoretical point of view, analysing TRS provides evidence on the consequences of the potential electoral opportunity for new candidates, the level of electoral legitimisation of the winner, and possibilities for alternation in power.

The electoral results (runoff and comeback) should vary in fact depending on the different guidelines governing TRS as well as on the presidential features and the institutional context. This is the standing knowledge we have on the topic (Shugart, 2007). Below, we add another step: an analysis of the whole process, differentiating how and why/when we arrive to a runoff occurrence and the cases of a comeback. Both rounds need to be analysed to determine to what extent the second round is a new ball game, with its own theoretical explanations.

Second Rounds and Comebacks

As discussed in the introduction, the causes and the effects of a second round in presidential elections are important from a theoretical standpoint. Together with a substantial comparative gap in the literature, the significance of the study relies on the potential destabilizing effect of the plurality electoral system. Going to the second round would increase the legitimation and may also open up the race to a wider number of competitors. It is important to see whether the TRS is in name only, meaning that there is no real competition and almost always the first-round leader wins, with the system serving to anoint plurality winners with greater legitimization. As evidenced below, the typical outcome is indeed that the plurality winner from the first round wins the second. So, investigating the causes of a runoff or comeback is important to explain when the TRS operates different from a plurality election.

The number of comebacks in direct presidential elections, as may occur in presidential and semi-presidential regimes, constitute important data for understanding the political process in those countries. Nevertheless, the most intriguing aspect to be investigated concerns the possible factors producing the comebacks. In this sense, it is possible to present a few hypotheses on what exactly contributes to the likelihood of having a second round and on the likelihood of an electoral comeback in the presidential runoff.

First, the regime type might play a role in affecting the probability of having a second round. Presidential and semi-presidential regimes differ in terms of their institutional structure, the role conferred to the president and the perception of his or her capacities by the electorate and, in turn, his or her accountability (Cheibub, 2012; Hellwig and Samuels, 2007). To test the possibility of a second round, we can expect that the incumbent can be electorally awarded or sanctioned depending on the role he or she plays in the political arena. Therefore, we can assume differences between the presidential and semi-presidential regimes. Most of the former presidential political activities depend, in large part, on the powers conferred to the head of the state (Doyle and Elgie, 2016; Shugart and Carey, 1992):

H1: Presidential systems are less likely than semi-presidential systems to result in a second round, and if they do, are less likely to have a comeback.

Moreover, as discussed above, the TRS can have various structures, for example, a threshold fixed at 50% or a so-called ‘qualified’ second round, in which candidates can win in the first round if they obtain a certain percentage of votes. It follows that we can expect variations in the frequency of electoral comebacks, depending also on the electoral system, which exerts mechanical and psychological effects on the parties, voters and candidates alike. The public’s electoral behaviour might be conditioned accordingly knowing that a certain threshold can ensure a victory for a candidate in the first round and might vote strategically to prevent the risk of a second round occurring at all:

H2: The presence of a qualified second round should reduce the likelihood of a second round occurring, but have no impact on whether the front runner or runner-up is ultimately victorious.

The amplitude of the presidential resources can help the incumbent obviate the need for a second round. Presidents who are more powerful have been shown to install a greater number of non-partisan ministers, affect the level of turnout in presidential and legislative elections, create more intra-executive conflict, issue more decrees and vetoes and produce more unpredictable and unstable cabinets (Elgie, 2016). From a theoretical point of view, then, there is reason to suspect that a president could use her powers to reduce the likelihood of a second round or a comeback for either herself or a designated successor. She or he can use the mass media, dismiss the assembly (in semi-presidential regimes), reshuffle the government and veto legislation. The president, possessing stronger political power, should benefit from greater resources (Elgie et al., 2014); as a result, she or he so being able to avoid the second round and the comeback be re-elected:

H3: The stronger the presidential power, the less the likelihood of a second-round or a comeback.

Although many political scientists are opposed to the use of term limits, they do increase the ability to remove corrupt leaders who have abused democratic institutions to keep themselves in power (or even well-trusted leaders who use that support to keep themselves in power unduly). This might produce greater levels of executive instability (Shapira, 2022) and lead to fewer outright first-round victories. On the contrary, the presence of term limits might also produce a set of oligarchic, or immoral, powerholders or incentivize leaders to groom potential successors (Shapira, 2022). Thus, one might then expect that term limits would induce a current president to use his or her powers to assist a designated successor:

H4: The presence of term limits increases the likelihood of a second round, but decreases the likelihood of a comeback.

Relatedly, longer terms would also allow presidents more time to use the resources afforded to them in either to support their own candidacy or to set things up for a successor:

H5: Longer Presidential term lengths would decrease the likelihood of both a second round and a comeback.

In addition, the period in which the elections take place can have differential effects on the campaign and the result as well (Edwards, 1979; Jones, 2018: 294; Jones West and Spoon, 2013; Smith, 2004). In this sense, the fact that the candidate is an incumbent at the end of his or her first term should be different from the circumstances of a challenger running for the first time. Therefore, considering the presidential term has theoretical importance: the presidential term duration, and the presence and the nature of the presidential term limits (Baturo and Elgie, 2019). There are countries in which the president can serve only one term (without any possibility of re-election, either immediately or in the future). Conversely, countries exist where no limits are placed on consecutive presidential re-election. An incumbent president may have significant advantages; therefore, ‘whether or not a president is eligible to run for re-election can have profound impact on the pattern of competition in a presidential race’ (Jones, 2018: 288). All these data represent the institutional and political structure of possibilities that the candidates – especially the incumbent– encounter when facing the electoral race. Incumbency furnishes many resources (visibility, use of governmental resources, etc.) to the candidate both in the first round and in the case of runoff to avoid a defeat, a comeback:

H6: The presence of the incumbent running for re-election would decrease the chance of a runoff and comeback.

The level of electoral volatility – although measured for the legislative elections – might serve as an indicator of the likelihood voters will switch their choice of candidate, thus changing opinion between the first and second rounds. The openness of the electoral behaviour ‘market’ and the voters’ propensity to change their minds can increase the likelihood of a second round, as no one candidate would consistently have a steady share of voter support. Along the same lines, the runoff might generate a comeback if even a small portion of the electorate decides to switch their vote. We need to consider the electoral data and the ratio of the two-round electoral system in terms of voters’ ‘rational’ choice:

H7: The greater the electoral volatility, the higher the likelihood of a second round and with that a comeback.

As argued by Cox (1997: 123), the nature of the electoral system affects the number of potential winners: in TRS, this number is equal to 3 = (M + 1), as only the top two candidates pass to a runoff. According to the strategic voting theory (Cox, 1997: 137), in TRS voters should be psychologically conditioned to concentrate their votes on three candidates in the first round, as two candidates are able to reach the second ballot. If voters concentrate their votes on only two candidates, however, a second round might not be needed:

H8: The greater the number of candidates, the higher the likelihood of a second round and with that a comeback.

Method

The present study is based on an original dataset that covers all popularly elected presidential offices in both presidential and semi-presidential regimes since 1945 till March 2020. We have 423 elections and 73 countries in all regions of the world.⁹ In TRS electoral systems, it is possible to have a second round; in the latter case, the winner can be the candidate who placed second in the first round. Voters can change their minds between the two rounds (either by supporting another candidate or by voting for a candidate whose is ideologically similar to their first choice), thereby generating the so-called ‘runoff comeback’. The dependent variable is categorical, measuring whether an election is won in the first round, whether the election is won in the second round by the first-round winner or whether there is a second round runoff comeback.

Given the nature of our dependent and independent variables, we have adopted a sequential logistic regression (SLR).¹⁰ SLR takes the temporal aspect into account as voters select a presidential candidate conditional upon the first-round results. This technique assumes that one can be ‘at risk’ of passing a ‘transition’ only if one has passed, or ‘survived’, previous transitions (Buis, 2017; Fullerton, 2009). In our case, this means that a voter can only be ‘at risk’ of voting for the runner-up once she or he has passed the ‘transition’ of not attaining a presidential victory in the first round.¹¹ Results can be interpreted as estimating the likelihood of each category of the dependent variable.

Of course, similar statistical results can be attained by running separate logistic regressions on each choice set on the appropriate subsample (Mare, 1981). However, the sequential logistic model has the added benefit that it can estimate varying effects of variables across multiple stages (Albert and Chib, 2001; Fullerton, 2009),¹² and thus allows us to analyse both rounds at once. In sum, SLR allows for an analysis of both the effect of a variable on each transition, first round and second round, and the effect on the final outcome: the president elected (Buis, 2017). The SLR makes it easier to test hypotheses across transitions, since the entire model is estimated simultaneously.

We operationalize the variables in the following way:

H1 (Presidentialism)	0 = Semi-Presidentialism; 1 = Presidentialism
H2 (Electoral System)	1 = TRS; 2 = Qualified TRS
H3 (Presidential Power)	0 – 1 standardized scale (Doyle and Elgie, 2014¹³)
H4 (Term limit)	0 = none; 1 = term limit present
H5 (Term length)	Length of presidential term (measured in years)
H6 (Incumbent re-candidate)	0 = not incumbent; 1 = Incumbent re-candidate
H7 (Electoral volatility)	(Pedersen, 1979)
H8 (Number of Presidential Candidates)	effective number of presidential candidates (Jones, 2018)¹⁴

TRS: two-round system.

Data Analysis

In most of the cases included in this research (57.2%), the president was elected in the first round; in the remaining 42.8%, the election was decided in the runoff (Table 2).

Table 2.

Presidential Elections by Electoral Round in Comparative Perspective (%) (1945–2020).

Presidential election	Europe	Latin America	Africa	Asia	Oceania	Total
First round	31.3	52.8	69.4	84.7	55.6	57.2
Second round	68.7	47.2	30.6	15.3	44.4	42.8
Total (N)	100 (115)	100 (106)	100 (121)	100 (72)	100 (9)	100 (423)

Source: Authors’ elaboration on Nohlen et al. (1999, 2001), Nohlen (2005) and Nohlen and Stover (2010). Minister of Interior and Electoral Committees.

The data indicate the presence of important geographical differences across countries adopting a TRS for their presidential elections. All regions except Europe show values well above 50% of cases in which the presidential race was decided in the first round. Overwhelming majorities of cases are present in Asia (84.7%) and in Africa (69.4%). In Latin America, the number of presidents elected at the first round exceeds half the total elections (52.8%); the same is found in Oceania (55.6%). In contrast, in Europe, the number of heads of state elected in the first round is significantly less than 50% (31.3%).

If we consider the regime type, the differences are less salient: both presidentialism and semi-presidentialism register about 40% (44.4% and 41.5%, respectively) of elections decided by runoff.¹⁵

These considerations become crucial when analysing runoff elections. In the latter cases, that is, when the president was decided in the second round, the vast majority comprised candidates who placed at the top in the first round. Nonetheless, the analysis of all presidential direct elections using a two-round electoral system suggests that electoral comebacks, overall, are quite relevant. About 30% of all second-round competitions in presidential and semi-presidential regimes result in the runner-up winning. Fifty-three of the 181 second-round elections saw the ‘surprising’ electoral comeback of the candidates who placed second in the first round. It is worth noting that a significant percentage of all the runoffs in semi-presidential and presidential regimes are won by the candidate who finished second in the first round. Comebacks at the presidential elections were reported in 34 of 73 countries included in the research. Despite representing a significant percentage (29.3%)¹⁶ of total runoffs, one finds that amounts vary across regions.¹⁷

It is important to note differences and similarities within the sample of total comebacks. From a regime type perspective, the data show that there is no important difference between presidential and semi-presidential cases. In fact, the former register 50.9% of all comebacks while the semi-presidential countries represent the remaining 49.1%¹⁸. In synthesis, data show that in all the presidential elections analysed, the majority has been decided in the first round. In almost one-third of the total, the front runner won in the runoff, and in about 12% of all two-round presidential elections, a comeback has decided the electoral outcome (Figure 1(a) and (b)).

Figure 1.

(a) The Comebacks in Presidential Elections (% of Elections That Have a Runoff) (1945–2020). (b) Distribution of the Electoral Comebacks in Presidential and Semi-Presidential Regimes (% All Runoff) (1945–2020).

Figure 2 displays all the 53 comebacks. The Y axis reports the front runner percentages of votes at the first round, while the X axis indicates the value for the runner-up. The most striking cases is France 1995 (bottom left), when the top two candidates collected the lowest percentage of votes (44.1%). At the other corner (top right), we find Zimbabwe (2008), where the top two candidates received 91.1%, and Ghana (2008) (97%).

Figure 2.

The Electoral Structure of First Round for Presidential Comebacks in Presidential and Semi-Presidential Regimes (1945–2020) (%).

In all but one case, the difference between the top two candidates in the first round was less than 20 points, meaning that is far from the diagonal that indicates a closer pattern of the two top candidates.¹⁹ There also were candidates who lost the runoff, having failed to win in the first round by less than 1%.²⁰ Furthermore, the sum of the votes for the top two candidates at the first round, on average, was equal to 64%. In few striking cases (13 of 53) of runoff comebacks, the front runner in the first round was relatively close to 50%. Also highlighted is the Portuguese election in 1986. This exemplifies a ‘runoff comeback’: the first round runner-up, Mário Soares, received just less than 26% of the votes in the first round, but went on to win the second round (and become President) with 51% of the vote in the second round.

We decided to run separate models for institutional and electoral variables because the first refers to ‘fixed’ –constitutional – factors, while the second elements relate to the dynamics of the political system. All model results are presented in Table 3.

Table 3.

Sequential Logit Estimation of Presidential Election Results in Presidential and Semi-Presidential. Direct Presidential Elections 1945–2020.

	Model 1		Model 2		Model 3		Model 4
	Second round	Comeback	Second round	Comeback	Second round	Comeback	Second round	Comeback
Institutional regime
Presidential	0.33	0.14	0.36	0.10	–1.36	0.36	–4.71	–1.78
	(0.32)	(0.43)	(0.48)	(0.86)	(0.88)	(0.54)	(0.07)	(0.41)
Electoral System: Qualified	–1.33** (0.01)	–0.10(0.92)	–1.79** (0.01)	0.42(0.65)
Presidential resources
Presidential power			–1.08	0.10			12.87	5.53
			(0.38)	(0.96)			(0.09)	(0.22)
Presidential term limit			0.03	0.37
			(0.96)	(0.44)
Presidential term length			–0.20	0.58			1.94	0.24
			(0.49)	(0.17)			(0.42)	(0.63)
Incumbent re-candidate			–1.37***	–0.46			–0.05	–0.36
			(0.00)	(0.39)			(0.96)	(0.74)
Electoral structure: First round
Electoral volatility					0.14***	–0.04	0.21	–0.02
					(0.00)	(0.08)	(0.07)	(0.52)
Effective number of presidential candidates					4.74* (0.04)	0.06(0.32)	4.55* (0.04)	0.07(0.48)
Constant	–.99	–1.06	3.32	–4.33	–21.33*	–0.45	–34.26*	–3.00
	(0.07)	(0.22)	(0.07)	(0.14)	(0.03)	(0.55)	(0.02)	(0.38)
Wald χ²	7.79		42.47		51.78		76.48
No. of observations	423		318		85		73

Source: Authors’ elaboration.

Note: In parentheses is presented the p < |z|.

p < 0.05*. p < 0.01**. p < 0.001***.

The first model includes the variables related to the institutional regime. As compared with semi-presidential regimes, presidential regimes are more likely to head to a second round and when they do, are more likely to result in a comeback. However, the p-values are not significant in these models. The second variable to analyse would be the electoral system. Here, we find that those having a ‘qualified’ second-round indeed are less likely to result in the need for a second round. Also as expected in H2, there is no effect of the electoral system of which candidate is the ultimate winner.

Moving to presidential resources, model 2 includes the presidential term (length and limits), the presidential power as a factor potentially able to affect the likelihood of having a runoff and the presence of an incumbent running for re-election. Along the same lines, the role of the incumbent is very important. We have measured the impact of the incumbent running for re-election and, if so, it is less likely that a second round will be needed. The coefficient for presidential power is negative, meaning that an increase in this dimension reduces the probability of a second round, although not at standard levels of significance. Term limits have no notable role in impacting the probability of a presidential second round or reducing the likelihood of a comeback. The term length variable also does not have a notable role in impacting the probability of a presidential second round or reducing the likelihood of a comeback. None of the presidential resources impacts which candidate is more likely to be victorious in the second round.

From a descriptive point of view, it is important to underscore that the number of incumbent presidents running for re-election was 156 (36.9% of the total), of which 102 candidates won. The more interesting data are that, among these latter politicians, the vast majority (88) were re-elected in the first-round and only 14 after a runoff. Finally, only two of these 14 were confirmed as heads of the state who got their second term in a runoff comeback (Austria 1951 and Cyprus 1998). Conversely, among the remaining 54 incumbent presidents who ran for re-election, 33 failed in the first round while 21 were defeated in the runoff. Of the latter, seven faced electoral defeat in a runoff comeback (Argentina 2015; Palau 1992; Poland 2005; Senegal 2000 and 2012, Serbia 2004, and Zimbabwe 2008). Therefore, these data indirectly confirm that the comeback is mainly a ‘prerogative’ – or better – an opportunity for challengers and newcomers rather than for incumbents.

Model 3 includes political variables linked to the electoral structure in the first round. The variables are the effective number of candidates and the electoral volatility. The volatility variable was only available for 85 cases. In these cases, the electoral regime had perfect prediction of not having a second round and so the coefficient is removed from the analysis. The data clearly indicate that, when a presidential race has many candidates in the first round, the likelihood of electing the president in the runoff increases considerably. Related to the electoral supply in the first round, we have measured the impact of electoral volatility of legislative elections²¹ (Casal Bértoa, 2020; Emanuele et al., 2020). This variable indicates the voters’ tendency to change their electoral behaviour. The data indicate that volatility has a positive impact on the probability of a second round, signalling the tendency that when voters are more likely to switch their votes, so do is it likely that no candidate will receive the necessary number of votes in the first round of a presidential election to prevent the necessity of a second round.

Model 4 is a pooled model, including all the variables and dimensions taken under investigation. Due to the reduced number of observations, term limit now also is a perfect predictor and is removed from analysis. According to the statistical significance, the effective number of presidential candidates is a positive predictor associated with the likelihood of a second round. In this pooled model, like in the base models, we find that nothing is a predictor of a second-round comeback.

Discussion

In many direct presidential elections, the electoral system states that in cases where no candidates reach a given quota of votes (usually 50%), a second round is scheduled. The data presented in this research indicate that a significant number of all the presidential elections in presidential and semi-presidential regimes end with a runoff comeback. The 29% represents important information never collected and analysed in the past in a comparative perspective. Table 4 summarizes our empirical findings.

Table 4.

Summary of Findings.

Hypothesis	Finding
H1 (Presidentialism)	No difference between Presidentialism and Semi-Presidentialism
H2 (Electoral System)	Qualified systems less likely to have a run-off; no impact on comeback
H3 (Presidential Power)	No impact of standardized presidential power scale (Doyle and Elgie, 2014)
H4 (Term limit)	No impact of term limit
H5 (Term length)	No impact of term length
H6 (Incumbent re-candidate)	Decreases likelihood of a run-off; no impact on comeback
H7 (Electoral volatility)	Increases likelihood of a run-off; no impact on comeback
H8 (Number of Presidential Candidates)	Increases likelihood of a run-off; no impact on comeback (Jones, 2018)

Although there are no relevant differences between presidential and semi-presidential regimes, relevant variations arise when considering region. In fact, even if Europe represents the only case where the regime type matters in terms of comebacks, the data must be analysed in light of the overwhelming spread of semi-presidential regimes among European countries – despite the unique case of the presidential regime. The data confirm the relevance of the electoral formula (Shugart, 2005) but not the impact of the presidential power (Doyle and Elgie, 2016). Besides the electoral system, the variables that most influence the likelihood of a runoff are the effective number of candidates in the first round and the incumbency status. In the case of the runoff comeback, we have no evidence that supports any of our presented hypotheses.

We suggest the addition of contextual factors as a fruitful avenue for future research. Polarization of political competition and voting behaviour trends at country/region level might also affect the likelihood of comebacks. Furthermore, the candidates who drop out after the first round should also play a role in whether a comeback is possible. Should an ideological neighbour to a challenger drop out, for example, that should increase the likelihood of a comeback. While we did find incumbency to be an advantage, this study did not account for the performance of the incumbent while in office. Economic indicators may also be found to affect the likelihood of second round or comebacks. Slowing gross domestic product (GDP) growth may, for example, increase the likelihood of an incumbent’s defeat at a comeback.

Another possible factor influencing the electoral outcome of the second round is the presence of cohabitation/divided government (Elgie, 2001; Shugart, 1995), that is, the incumbent (president or his or her party) (Passarelli, 2020) is facing a situation in which the legislature is controlled by the opposition (one or both chambers, depending on the structure of the parliament). The analysis above did not include any partisanship-related variables. The different political support for the president and the legislature may influence the role of the president, so we can hypothesize that the president under divided government rule can be negatively affected in electoral terms (Shugart, 1995: 328–330).

This study used an aggregate measure of the presidential power. There might be specific aspects of presidential power worth considering, for example, the power to veto legislation or issue decrees. Presidential power might also be best examined in a more qualitative framework, as de jure power might not be the same as how it is used in practice. For example, although institutionally weaker, Hungarian presidents are stronger than Irish ones (Tavits, 2009). Although the Doyle and Elgie’s (2016) measure is strong on reliability, it might be weak on other aspects of measurement validity that qualitative studies draw their strength from.

The findings herein have an important theoretical impact on the effects of institutions, electoral dynamics, and candidates’ behaviour: in the presence of a TRS electoral system, the chance of having a runoff decreases when the electoral competition is clear and reduced and the incumbent is seeking another term. Counter to our motivation, none of regime characteristics, presidential resources, nor the structure of the first round could predict the incidence of a comeback.

Footnotes

Appendix

Appendix 1.

Summary Statistics.

Front runner electoral increase between first and second round (Delta)
Variable	N	Mean	p50	SD	Min	Max
First round	128	19.1	17.2	12.3	0.0	66.5
Comeback	53	7.5	8.5	9.2	–38.0	24.0
N	181
Runner up electoral increase between first and second round (Delta)
Variable	N	Mean	p50	SD	Min	Max
First round	128	11.1	12.9	10.6	–33.2	29.7
Comeback	53	25.4	25.7	11.0	0.0	47.3
N	181

SD: standard deviation.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Gianluca Passarelli

Notes

Author Biographies

Gianluca Passarelli, Full Professor of Political Science. Department of Political Sciences, Sapienza University, Rome.

Matthew Bergman, Department of Government University of Vienna.

References

Albert

Chib

(2001) Sequential Ordinal Modeling with Applications to Survival Data. Biometrics 57 (3): 829–836.

Baturo

Elgie

(eds) (2019) The Politics of Presidential Term Limits. Oxford: Oxford University Press.

Bergman

(2021) Sorting Between and Within Coalitions: The Italian Case (2001–2008). Italian Review of Political Science 51 (1): 42–66.

Birch

(2003) Two-Round Electoral Systems and Democracy. Comparative Political Studies 36 (3): 319–344.

Blais

Massicotte

Dobrzynska

(1997) Direct Presidential Elections: A World Summary. Electoral Studies 16 (4): 441–455.

Bordignon

Nannicini

Tabellini

(2016) Moderating Political Extremism: Single Round Versus Runoff Elections Under Plurality Rule. American Economic Review 106 (8): 2349–2370.

Buis

(2017) Not All Transitions Are Equal: The Relationship Between Effects on Passing Steps in a Sequential Process and Effects on the Final Outcome. Sociological Methods & Research 46 (2): 649–680.

Carreras

(2012) The Rise of Outsiders in Latin America, 1980–2010: An Institutionalist Perspective. Comparative Political Studies 45 (12): 1451–1482.

Carreras

(2014) Outsiders and Executive-Legislative Conflict in Latin America. Latin American Politics and Society 56 (3): 70–92.

10.

Carreras

(2017) Institutions, Governmental Performance, and the Rise of Political Newcomers. European Journal of Political Research 56 (2): 364–380.

11.

Casal Bértoa

(2020) Database on WHO GOVERNS in Europe and Beyond, PSGo. Available at: whogoverns.eu (accessed 15 May 2022).

12.

Cheibub

(2012) Presidentialism, Parliamentarism, and Democracy. Oxford: Oxford University Press.

13.

Cox

(1997) Making Votes Count: Strategic Coordination in the World’s Electoral Systems. New York: Cambridge University Press.

14.

Doyle

Elgie

(2016) Maximizing the Reliability of Cross-National Measures of Presidential Power. British Journal of Political Science 46 (4): 731–741.

15.

Edwards

GCIII

(1979) The Impact of Presidential Coattails on Outcomes of Congressional Elections. American Politics Research 7 (1): 94–108.

16.

Elgie

(ed.) (2001) Divided Government in Comparative Perspective. Oxford: Oxford University Press.

17.

Elgie

(2011) Semi-Presidentialism. Sub-Types and Democratic Performance. Oxford: Oxford University Press.

18.

Elgie

(2016) Three Waves of Semi-Presidential Studies. Democratization 23 (1): 49–70.

19.

Elgie

Bucur

Dolez

(2014) Proximity, Candidates, and Presidential Power: How Directly Elected Presidents Shape the Legislative Party System. Political Research Quarterly 67 (3): 467–477.

20.

Emanuele

Chiaramonte

Soare

(2020) Does the Iron Curtain Still Exist? The Convergence in Electoral Volatility between Eastern and Western Europe. Government and Opposition 55 (2): 308–326.

21.

Farrell

(2001) Electoral Systems. A Comparative Introduction. London: Palgrave Macmillan.

22.

Fullerton

(2009) A Conceptual Framework for Ordered Logistic Regression Models. Sociological Methods & Research 38 (2): 306–347.

23.

Hellwig

Samuels

(2007) Electoral Accountability and the Variety of Democratic Regimes. British Journal of Political Science 38 (1): 65–90.

24.

Jones

(1994) Electoral Laws and the Survival of Presidential Democracies. Notre Dame, IN: University of Notre Dame Press.

25.

Jones

(1995) A Guide to the Electoral Systems of the Americas. Electoral Studies 14 (1): 5–21.

26.

Jones

(1999) Electoral Laws and the Effective Number of Candidates in Presidential Elections. Journal of Politics 61: 171–184.

27.

Jones

(2018) Presidential and Legislative Elections. In: Herron

Pekkanen

Shugart

(eds) The Oxford Handbook of Electoral Systems. Oxford: Oxford University Press, pp.283–301.

28.

Jones West

Spoon

J-J

(2013) Credibility Versus Competition: The Impact of Party Size on Decisions to Enter Presidential Elections in South America and Europe. Comparative Political Studies 46: 513–539.

29.

Laakso

Taagepera

(1979) Effective Number of Parties: A Measure with Application to West Europe. Comparative Political Studies 12 (1): 3–27.

30.

McClintock

(2018a) Electoral Rules and Democracy in Latin America. Oxford: Oxford University Press.

31.

McClintock

(2018b) Reevaluating Runoffs in Latin America. Journal of Democracy 29 (1): 96–110.

32.

Mainwaring

Shugart

(eds) (1997) Presidentialism and Democracy in Latin America. Cambridge: Cambridge University Press.

33.

Mare

(1981) Change and Stability in Educational Stratification. American Sociological Review 46 (2): 72–87.

34.

Martinez Martinez

(2006) Ventajas y Desventajas de la Fórmula Electoral de Doble Vuelta. Barcelona: Cidob.

35.

Negretto

(2006) Minority Presidents and Democratic Performance in Latin America. Latin America Politics and Society 48 (3): 68–92.

36.

Nohlen

(ed.) (2005) Elections in the Americas: A Data Handbook. New York: Oxford University Press.

37.

Nohlen

Stover

(eds) (2010) Elections in Europe: A Data Handbook. Baden-Baden: Nomos.

38.

Nohlen

Grotz

Hartmann

(eds) (2001) Elections in Asia and the Pacific: A Data Handbook. Oxford: Oxford University Press.

39.

Nohlen

Krennerich

Thibaut

(eds) (1999) Elections in Africa: A Data Handbook. Oxford: Oxford University Press.

40.

Passarelli

(2020) The Presidential Party. A Theoretical Framework for Comparative Analysis. Political Studies Review 1: 87–107.

41.

Pedersen

(1979) The Dynamics of European Party Systems: Changing Patterns of Electoral Volatility. European Journal of Political Research 7: 1–26.

42.

Pérez-Liñán

(2006) Evaluating Presidential Runoff Elections. Electoral Studies 25 (1): 129–146.

43.

Remmer

(2008) The Politics of Institutional Change: Electoral Reform in Latin America, 1978–2002. Party Politics 14: 5–30.

44.

Reynolds

Reilly

Ellis

(2005) Electoral system design: the new international IDEA handbook. Stockholm, Sweden: International Institute for Democracy and Electoral Assistance (IDEA).

45.

Samuels

(2002) Presidentialized Parties: The Separation of Powers and Party Organization and Behavior. Comparative Political Studies 35 (4): 461–483.

46.

Shapira

(2022) A Majority Escalator Rather Than Term Limit Can Mitigate Leaders’ Toxic Oligarchic Entrenchment. Academia Letters 4715. https://wgalil.academia.edu/RShapira

47.

Shugart

(1995) The electoral cycle and institutional sources of divided Presidential Government. The American Political Science Review 89 (2): 327–343.

48.

Shugart

(2005) Comparative Electoral Systems Research: The Maturation of a Field and New Challenges Ahead. In: Gallagher

Mitchell

(eds) The Politics of Electoral Systems. London: Oxford University Press, pp. 25–56.

49.

Shugart

(2007) Mayoría relativa vs. segunda vuelta: la elección presidencial mexicana de 2006 en perspectiva comparada. Política y Gobierno 14 (1): 175–202.

50.

Shugart

Carey

(1992) Presidents and Assemblies: Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press.

51.

Shugart

Taagepera

(1994) Plurality versus majority election of presidents: A proposal for a “double complement rule”. Comparative Political Studies 27 (3): 323–348.

52.

Smith

(2004) Election Timing. Cambridge: Cambridge University Press.

53.

Tavits

(2009) Presidents in Parliamentary Systems: Do Direct Elections Matter? Oxford: Oxford University Press.