comparative politics - ulisboa€¦ · comparative politics april 2015 316. situations. we show in...

The Politics of Minister Retention inPresidential Systems

Technocrats Partisans and Government Approval

Marcelo Camerlo and Aniacutebal Peacuterez-Lintildeaacuten

Presidents and prime ministers often introduce policy adjustments by reallocatingcabinet portfolios and shield their favorite policies by protecting key cabinet membersIndividual minister profiles are a crucial factor in guiding such decisions The retentionof politically savvy partisan ministers helps consolidate legislative support while theretention of technocrats who are less vulnerable to the pressures of political partiesand constituents guarantees policy continuity Different types of ministers are there-fore protected or sacked in different political contexts Given these alternatives whattypes of ministers enjoy longer tenures and under what conditions How do individualprofiles and electoral incentives affect minister turnover

We argue that the effect of ministersrsquo profiles on portfolio reallocation is conditionedby the popular approval of an administration When the government is unpopular theloyalty of cabinet members to the chief executive is fragile and external factors drivecabinet turnover Partisan ministers leave the cabinet as the government coalition crum-bles and technical ministers are replaced in response to public pressures In such contextsgovernment chiefs only command the loyalty of an inner circle of ministers who arewithout partisan careers or independent policy agendas When the government is verypopular by contrast the cabinet aligns with the chief executive and external pressuresfor cabinet turnover decline In the latter scenario partisan and technical ministers tendto be safe in their posts but technocrats in charge of the administrationrsquos successfulpolicies have more leverage and a lower risk of exit than most partisan ministers

As a result the conventional distinction between partisan and non-partisan minis-ters proves insufficient to understand retention strategies While weak heads of govern-ment are forced to engage in defensive reallocation strategies which result in a higherrate of turnover for most ministers strong executives can engage in unilateral portfolioreallocation in order to strengthen their control of the cabinet This pattern typically

315

involves the preservation of technocrats responsible for successful government policiesbut not necessarily of outsiders who become less necessary in a context in which allministers align with the head of the government

From Cabinet Appointments to Minister Survival

Portfolio reallocation has been increasingly depicted in the literature as a tool todeal with difficult situations such as the presence of ldquoundesirablesrdquo and ldquoopportunistsrdquowithin the cabinet the decline of government popularity low policy performance andeconomic or socio-political conflicts1 The emerging research agenda departs fromthe well-developed literature on cabinet formation In the classic approach portfolioallocation was conceptualized as a tool to obtain legislative support for the establish-ment performance and survival of governments This perspective focused on thepartisan composition of the cabinet underscoring the role of institutional conditions(eg legislative majorities and the nature of coalitions) as the main explanatory vari-able Initially developed to explain the politics of European parliamentary democraciesthe approach later travelled to the realm of Latin American presidentialism2

Alternatively portfolio reallocation studies focus on the appointment and removalof individual ministers within cabinets and privilege explanations that underscore therole of critical events and individual attributes above and beyond partisanship Theemphasis on individual traits has led to a renewed interest in gender technical skillsand extra-party affiliation of ministers as well as the development of more exhaustiveand sophisticated datasets on individual backgrounds3

A central theme in this emerging line of research has been the role of non-partisansin government The issue of ldquoindependentrdquo ministers has been tackled by the literatureson parliamentary and presidential regimes although from different angles Research onEuropean democracies has focused on the increasing presence of ldquooutsidersrdquo in thecabinet over the last decade demonstrating a deviation from traditional career pathsThe need to recruit based on technical skills to face the growing complexity of multi-level governance and the organizational decline of political parties has been posedas the main explanation for the new phenomenon in addition to factors such as theldquopresidentializationrdquo of politics and the institutional separation between executive andlegislative powers4 In presidential democracies where rulers are expected to nominatemore ministers without party backgrounds scholarly attention has focused on expertswithout political experience but with considerable political power Initially triggeredby the performance of successful ldquotechnopolsrdquo during the nineties this research haslinked technocratic politics to the increased impact of economists neoliberalism andstructural reforms5

Despite the relevant contributions of these research agendas three implicit assump-tions present in the literature hamper an adequate consideration of minister profiles andtheir role in contemporary governments First portfolio reallocation is conventionallydepicted as a reactive tool used by weak heads of government to confront negative

Comparative Politics April 2015

316

situations We show in this article however that portfolio reallocation can also be aproactive tool when used selectively by well-positioned chief executives to consolidatetheir control over cabinets

Second the presence of non-partisan ministers in government is conventionallylinked to the de-politicization of public policy The European literature has largelyviewed the surge of non-partisans as a reflection of major trends related to the complexityof modern governance such as modernization democratization and EuropeanizationThe Latin American literature on the other hand has linked strong non-partisan minis-ters to the resolution of acute economic crises and the introduction market-orientedpolicies In both cases non-partisan ministers are considered apolitical individuals(at least in traditional terms) who join the highest levels of the decision-making processwhen there is a suspension of regular politics We argue that this image of non-partisanscan be misleading for two reasons To begin with ldquonon-partisan ministersrdquo is a hetero-geneous category that includes technocrats with an independent policy agenda as wellas outsiders who are personally loyal to the chief executive Moreover the idea thattechnocrats represent the de-politicization of public policy mostly reflects the fact thatprofessional politicians allocate blame for unpopular policies to technical ministers andclaim credit for successful policies for themselves This suggests that the framing ofthe relationship between government leaders and non-partisans in the cabinet is mediatedby public approval rates

The third deficit relates to the assumption that minister profiles can be properly cap-tured by certain rigid dichotomies insideroutsider generalistspecialist or politicianamateur The early European literature developed binary taxonomies that mostlyremained unchallenged in subsequent research6 In turn the Latin American literaturefocused on a very specific profilemdashsuccessful technocrats with significant politicalcloutmdashwithout considering other types of non-partisans and their relations with partisanministers Additionally these taxonomies overlooked portfolios not specifically relatedto economic policy and relied on a very basic operationalization of technocratic profilesthat measured whether ministers had PhDs in economics We address this problem bydeveloping a systematic typology of minister profiles and by operationalizing the con-cepts of technocrats partisans and outsiders based on our conceptual scheme

Portfolio Reallocation and Minister Profiles

To overcome the limitations discussed above we broadly define portfolio reallocation asa government tool for managing multiple political resources through either cooperativeor unilateral strategies We argue that the strategy chosen is conditioned by the politicalstrength of the chief executive Our argument therefore involves three claims

1PortfolioReallocation is aPolitical Instrument forManagingDifferentResources Theliterature suggests that heads of government allocate portfolios with the goal of optimizingthree important resources political support (governability) technical skills (competence)


317

and loyalty (preference alignment) Political support is maximized through the appoint-ment of ministers affiliated with external collective actors such as parties or mass orga-nizations The crafting of legislative coalitions by a formateur is the most common andthe best-studied mechanism of this kind but it is not the only strategy to ensure govern-ability7 Cabinet appointments also secure support from organized groups such as tradeunions churches and the military which can be crucial in certain historical contextsIn turn technical skills are secured through the appointment of expert ministers withformal instruction or administrative experience in relevant areas The literature hasfocused on the skills of those responsible for market-oriented reforms but concernsabout expertise also affect most cabinet appointments and are independent of ideologicalpreferences8 Furthermore proximity in policy preferences between the chief executiveand cabinet members has been a critical factor in studies underscoring adverse selectionand moral hazard problems9 We therefore consider the alignment of cabinet preferencesas the third goal pursued through portfolio reallocation strategies

Although the three goals are not mutually exclusive political support technicalskills and loyalty are hard to maximize simultaneously and they may become moreor less salient in different contexts We argue that different types of ministers mapdistinctively into these priorities Partisan ministers who command strong politicalaffiliations are uniquely positioned to deliver political alliances Technocrats whocommand expertise are renowned for their competence Lastly ministers without partyties or independent policy expertise are more prone to behave as loyal agents of thechief executive We refer to members of the third group as outsiders

Figure 1 summarizes our typology of minister profiles This typology underscoresan important conclusion commonly overlooked by dichotomous classifications Tothe extent that the search for technical skills and loyalty presents important tradeoffsthe appointment of non-partisan ministers by the head of the government may serveopposite purposes

2 Government Heads May Choose between Cooperative and Unilateral Realloca-tion Strategies It follows from previous paragraphs that the executive may allocatecabinet portfolios to pursue different strategies We define as cooperative any manage-ment of portfolios oriented towards the inclusion of ldquootherrdquo parties or social actors in thecabinet and as unilateral any management of portfolios oriented towards the inclusion

Figure 1 Minister Profiles


318

of the chief executiversquos ldquoownrdquo people Although the specific meaning of ldquoothersrdquo andldquoownrdquo varies across cases and historical circumstances we claim that the choice overthose strategies constitutes the main causal mechanism linking government approvaland minister retention Unilateral strategies are easy to sustain when the chief executiveis politically strong and cooperative strategies are necessary when the chief executiveis politically weak

Our distinction resonates with a broader literature that describes the contrastbetween consensual and majoritarian models of policymaking ldquostatutoryrdquo and ldquoexecu-tiverdquo policymaking strategies and pluralist and majoritarian styles of presidentialism10

Yet it allows us to conceptualize government styles beyond the creation of formalmulti-party coalitions For instance the head of a single-party cabinet recruiting par-tisan ministers who previously opposed the leaderrsquos nomination within the partywould be an instance of the cooperative modality while the head of a coalition cabinetusing his or her party quota to nominate outsiders who are personally loyal to theindividual leader would be an instance of the unilateral strategy11

The Power of the Chief Executive is Volatile The political strength of governmentshas been commonly assessed through indicators that present limited variation (if any)during the life of an administration such as the composition of the legislature and theconstitutional powers of the executive branch However actual leverage of governmentleaders may fluctuate significantly over time Argentine President Fernando de la Ruacuteaillustrates this fact in dramatic ways Sworn into office with considerable electorallegislative and party support in late 1999 he was forced to resign before completingthe first half of his term in 2001 Following an opposite trajectory Argentine PresidentNeacutestor Kirchner started his administration with limited electoral and legislative supportin 2003 but managed to finish his administration in 2007 as one of the most successfulpresidents in the countryrsquos history Thus the need to secure political support technicalcompetence or personal loyalty may vary during the lifetime of an administrationRecent studies underscore the relevance of maintaining legislative support after theinitial moment of cabinet formation and show that the use of executive prerogativesis mainly conditioned by the political context12

To the extent that government strength can be a volatile factor public approval forthe chief executive is the most sensitive indicator reflecting its fluctuation over timeApproval rates are a real-time indicator of possible electoral outcomes and as suchfacilitate (or imperil) the relations of the government with the legislature the rulingparty and the main interest groups As a marker of specific support popularity createsincentives for the alignment of partisan ministers with the goals of the chief executiveand strengthens the credibility of technocrats blessed by the administration

Our main argument is that approval rates condition the ways in which governmentchiefs can employ portfolio reallocation strategies to manage political support technicalexpertise and internal alignments in the cabinet When government leaders are popularthey can avoid the more complex time-consuming and conceding cooperative modali-ties Strong leaders are able to implement unilateral strategies in the management of


319

cabinet portfolios protecting close party members but replacing more distant partisanministers to the extent that coalition agreements allow them to do so With governabilityassured articulators of political support will be kept in the cabinet as long as they donot challenge the chief executive At the same time the protection of ministers withtechnical skills becomes a priority as they are critical to sustain the successful policiesthat generate public support Heads of government may also preserve loyal outsidersin office but they have few reasons to do so Given the incentives of all ministersto align with the chief executive mere loyalty is not a wholly advantageous qualitywhen securing onersquos position in the cabinet

By contrast the leaders of unpopular governments are forced to adopt morecooperative strategies in order to secure governability In bad times coalition memberswho represent other parties or even alternative factions within the ruling party may aban-don the cabinet to distance themselves from the government and the chief executive oftenneeds to renegotiate the composition of the government Even though technocrats maypromise successful outcomes over the long run public pressures against unpopular poli-cies may force their removal In this context incentives for shirking among cabinet mem-bers are great and outsiders may be the only source of loyalty in the cabinet Yet because oftheir limited weight outsiders are also easier to replace as part of reallocation bargains13

The three claims developed in this section suggest two sets of hypotheses Thefirst one refers to the effects of public support on cabinet stability indicating that

H1 The higher government approval the lower the risk of exit (ie the longerthe survival) for partisan as well as technical ministers yet

H2 Government approval does not affect the survival of outsiders

The second set of hypotheses underscores the heterogeneous nature of non-partisanministers indicating that

H3 Technocratic ministers will confront a lower risk of exit than partisanministers when governments are popular but a similar (or higher) risk whengovernments are unpopular while

H4 Outsiders will confront a higher risk of exit than partisan ministers whengovernments are popular but a similar (or lower) risk when they are unpopular

The two sets of hypotheses are logically related Hypotheses 1 and 3 suggest thatpublic approval will benefit partisan and technical ministers but technocrats shouldbenefit considerably more when policies are successful Hypotheses 2 and 4 suggest thatthe effect of approval on the fate of outsiders is uncertain and thus partisan ministersand by implication technocrats will gain a relative advantage during periods of greaterpublic support Together these hypotheses underscore that non-partisan ministers arefar from being a uniform group


320

Minister Profiles An Operational Definition

The previous section has emphasized that different minister profiles are functional to securedifferent political resources For instance while partisan ministers are more able to mobilizesupport from their parties in parliament experts provide more technical skills Howevertwo issues complicate the operationalization of minister profiles Our discussion abovesuggests that non-partisan ministers are a heterogeneous group Additionally ministerialtraits may be present to different extents Ministers may have stronger or weaker partyaffiliations or they may display more or less expertise in a policy area Here we proposean operational definition of minister profiles that deals with those challenges

As reflected in Figure 1 the main distinction between minister profiles introducedby the literature is driven by party affiliation which results in the identification of par-tisan and non-partisan cabinet members In turn the most distinguishing feature amongthe latter is expertise which allows for the distinction between experts and outsiders LetA refer to the set of ministers with strong party affiliations and E to the set of individ-uals with technical expertise Because the ideal type of a technocrat refers to a ministerwho is an expert and lacks strong political affiliations we define this set as T equiv (E cap Aprime)where E denotes membership in the set of technical experts and the superscript forA denotes the complement (ie negation) of the set of ministers with an affiliationIn turn because the ideal type of an outsider refers to individuals with no politicalaffiliation and no expertise we define this set as O equiv (Eprime cap Aprime)

Given this conceptual scheme any operational definition of minister profiles mustrely on systematic criteria to determine a ministerrsquos membership in the sets of technicalexperts and of politically affiliated officials Membership in those groups however isnot clear-cut For instance an outsider who remains in office through several adminis-trations of the same party may be increasingly recognized as a partisan This transitionillustrates a progressive passage from not being a member of set A to becoming a fullmember To the extent that affiliation and expertise are variable attributes the ministertypes identified in Figure 1 should be considered poles in a continuous property spacerather than discrete categories

In order to capture the idea of partial membership in any given set we employfuzzy set notation14 Rather than being dichotomous and exclusive membership ina fuzzy set is determined by a continuous function ranging between 0 and 1 with 0indicating the complete absence of the attribute 1 the complete presence of the attri-bute and 05 being the cross-over pointmdashvalues below the cross-over point indicatethat an element is ldquorather outsiderdquo the set and values above indicate that the ele-ment is ldquorather insiderdquo the set The degree of membership in the complement (nega-tion) of a fuzzy set is given by 1 ndash fi where fi is the membership score for element iin the category that is negated In turn the degree of membership at intersection oftwo fuzzy sets is defined by the minimum score for the two membership functionsThis approach provides greater flexibility to code the nuances of minister profileswhile preserving a logically consistent conceptual framework We operationalize thetwo constitutive sets using empirically observable attributes as follows


321

Technical Expertise (E) Following the literature we identify technical expertise withthe presence of (E1) advanced academic training in the policy field or (E2) extensiveon-the-job training in the specific bureaucracy The first item is coded trichotomouslywith values of 0 if the minister has a degree unrelated to the portfolio area 05 if theminister has a related BA or equivalent (eg BA in economics for ministers of Eco-nomics and Finance in medicine for ministers of Public Health) and 1 if he or she has arelated graduate degree (MA MS MBA PhD or equivalents) The second itemwas defined as an ordinal scale reflecting whether the minister had administrative expe-rience in the specific policy area or professional experience in a related field (egschool administrators for ministers of Education union leaders for ministers of Labor)and the amount of that experience We assign values of 0 if the minister has neverworked in a related field before 025 if he or she served in a related field during oneadministration 050 if he or she previously served in the same policy area in one admin-istration 075 if he or she served in a related field for more than one administration and 1 ifhe or she acquired experience in the specific policy area for more than one administrationBecause the two attributes E1 and E2 are interchangeable membership in set E is definedas the maximum score for E1 and E2 (ie as the union of the two conditions)

Political Affiliation (A) The idea of affiliation reflects the extent to which a givenminister serves as the agent of a party (or any other group) in the cabinet or simplyrepresents his or her personal views We coded five dichotomous indicators reflectingwhether the press identified the minister as (A1) a well-known member of a politicalparty (A2) a leader of the trade unions (A3) a representative of business associations(A4) a spokesperson for powerful groups (the military the church the bureaucracyetc) and (A5) a sympathizer of a political party or an individual with personal ties tothe president The membership function is an ordinal scale 100 for ministers affiliatedwith a political party 050 for ministers affiliated with other political organizations025 for party sympathizers or individuals related to the president and 000 for min-isters with no affiliation If an individual met more than one condition we used thehighest score to capture the strongest affiliation

Based on the operationalization of the two components membership in the set oftechnocrats and outsiders was measured as

[1] Technocrat 5 min (E (1 ndash A))

[2] Outsider 5 min ((1 ndash E) (1 ndash A))

where min denotes the minimum membership score (ie the intersection of the twodefining conditions) E equiv max(E1 E2) and A equiv max(A1 05A2 05A3 05A4025A5) The resulting variables range between 0 (eg for non-technocrats of anykind) to 1 (eg full members of the technocratic ldquoclubrdquo) The degree of membershipin the set of partisan ministers is given by A full members of this group are by impli-cation individuals for whom the variables ldquoTechnocratrdquo and ldquoOutsiderrdquo have a score of 0


322

Data and Method

The Argentine political system is a crucial case for testing our hypotheses It is amongthe democracies with the highest levels of minister turnover and it presents considerablevariance for our key explanatory variables as well as for most alternative explanationsAdditionally it is a presidential system which assures that decisions about portfolioreallocation are mainly handled by the head of the government This tendency towardsa ldquopresidentialrdquo control of portfolios is also expanding in parliamentary democracies15

In order to test our hypotheses we observed all 159 ministers in office in sevenArgentine administrations from 1983 to 2011 The boundaries of administrations aredefined by presidentsrsquo inauguration dates The dependent variable is the duration ofeach minister in a specific portfolio Because we are interested in the use of portfolioreallocation during the administrationrsquos lifecycle we treated all ministers leaving theportfolio at the end of the administration as censored cases Exits caused by healthreasons were censored as well We identified 94 ministers leaving their portfolios inthe midst of an administration

The key explanatory factors in our argument are public approval for the govern-ment and minister profiles Our independent variable approval reflects the proportionof respondents who approved the work of the incumbent administration in national pollsbetween 1983 and 2011 Aggregate monthly figures were compiled from surveys con-ducted by IPSOS-Mora y Araujo and the data were interpolated to impute missingvalues16 The evolution of this predictor includes peaks of popularity for Rauacutel Alfonsiacutenin 1984 Carlos Menem in 1989 Fernando de la Ruacutea in 1999 and Neacutestor Kirchner in2003 The average approval rate was 45 percent with a minimum of 7 percent to amaximum of 84 percent

To capture minister profiles two additional independent variables follow theoperational definition of ldquoTechnocratrdquo and ldquoOutsiderrdquo proposed in the previous sec-tion The average membership score in the set of technocrats is 030 while the averagescore for outsiders is 019 About 43 percent of the ministers in our sample are mem-bers of the set of ldquotechnocratsrdquo to some extent 17 percent of the ministers have amembership score of 025 143 percent of 050 166 percent of 075 and 103 percentmatch the ideal type of technocrat with a membership score of 10 In turn about34 percent of the ministers have some membership in the set of ldquooutsidersrdquo 8 percentat 025 103 percent at 050 137 percent at 075 and 17 percent at 10

Following the previous literature we introduce three sets of control variables Thefirst set reflects the institutional context ldquoMinorityrdquo captures whether the presidentrsquosparty or coalition has a minority in any of the legislative chambers We observedminority governments for about half of the period in our study ldquoCoalitionrdquo registerswhether the cabinet includes members of parties other than the presidentrsquos Onlytwo out of seven administrations had coalition cabinets These attributes are commonlyused by the literature to measure institutional constraints and government strengthIn addition all administrations in our sample were controlled by either the PJ (PartidoJusticialista) or the UCR (Unioacuten Ciacutevica Radical) Because those parties have different


323

levels of institutionalization and styles of leadership ldquoUCRrdquo is an indicator that captureswhether ministers belong to an administration of the Radical party The possibility ofreelection and the time left to the end of the term have been identified as critical factorsaffecting the strategic use of portfolio reallocation17 ldquoRe-eligiblerdquo indicates whetherthe incumbent president can run for immediate reelection (we observed five re-eligiblepresidents out of the seven) while ldquocalendarrdquo measures the number of days left to thedate of the next presidential inauguration (as provided by the constitution) As indicatedpreviously research on ministerial expertise has paid special attention to officials incharge of the economy

As far as we know there is no evidence indicating that ministerial tenures areshorter or more fragile in some portfolios than others However because economicmanagement has been the main concern for most Argentine presidents since 1983we include a dichotomous variable that identifies ministers of ldquoFinancerdquo Adding con-trols for other ministries did not alter the results18

The second set of control variables registers the state of the economy a conven-tional explanation for portfolio reallocation ldquoGrowthrdquo records quarterly percentchange in the gross domestic product and ldquoinflationrdquo registers the monthly percentchange in consumer prices Both indicators were taken from the Argentine NationalInstitute of Statistics and the Census (INDEC) Average economic growth was 33 per-cent ranging from -163 percent to 126 percent while average inflation was 57 percentranging from minus075 to 1966 percent19 Since market-oriented policies were presum-ably related to the presence of technocrats in the cabinet we introduce a dichoto-mous variable to capture the period when the ldquoneoliberalrdquo economic plan (ldquoPlan deConvertibilidadrdquo) was in force (April 1991ndashDecember 2001)

The last set of control variables registers the individual background of ministersincluding their age and gender In models not including the measure of technocracywe also control for whether a minister was an economist and his or her level of formaleducation (basic college graduate) The average ldquoagerdquo of the observed ministers wasfifty-two years ranging from thirty to seventy-four only 94 percent of them wereldquowomenrdquo Less than 9 percent had only basic education while about 25 percent had agraduate degree ldquoEconomistsrdquo represent 23 percent of the observations

We model the duration of ministers in office using a Cox proportional hazardsmodel which estimates the probability that a minister will exit the portfolio attime t The use of a semi-parametric model allows us to analyze this phenomenonwithout assuming a specific shape for the hazard function

Results

Table 1 presents the results of the hazard analysis of minister turnover The effect ofeach covariate is shown in the table as a hazard ratio Model I estimates the uncondi-tional effect of presidential approval on minister turnover controlling for the covariatesrelated to institutional features (ldquominorityrdquo ldquocoalitionrdquo ldquoUCRrdquo ldquore-eligiblerdquo ldquocalendarrdquo


324

Table 1 Proportional Hazard Models of Minister Duration

I II III IVApproval 976

(009)976(009)

973(011)

972deg(017)

Technocrat - 422(146)

806(574)

810(578)

Outsider - 1583(567)

378(325)

373(256)

TechnocratApproval - - 984(017)

983(017)

OutsiderApproval - - 1036deg(020)

1037deg(020)

Institutional Attributes (Strength)Minority 1300

(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)

Institutional Attributes (Other)UCR party 1607

(699)1776(771)

2014(878)

2095(1122)

Re-eligible 1813(686)

1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)

Finance Portfolio 1990(799)

2525(866)

2451(847)

2455(849)

Economic ConditionsInflation 1102

(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)

Neoliberalism 942(273)

867(257)

865(256)

861(258)

Individual FeaturesAge 982

(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -

Basic Education 1007(379)

- - -

Graduate Education 740(197)

- - -

Control function 1002(020)

LR χ2 (15) 4772 (14) 5355 (16) 5826 (17) 5827Log likelihood minus37343 minus37051 minus36816 minus36815Prob gt chi2 0000 0000 0000 0000Observations 6727 6727 6727 6727Subjects 159 159 159 159Failures 94 94 94 94

Entries are hazard ratios (standard errors in parentheses) deg p lt 10 p lt05 p lt01


325

and ldquofinancerdquo) economic factors (ldquoeconomic growthrdquo ldquoinflationrdquo and ldquoneoliberalismrdquo)and individual backgrounds (ldquoagerdquo ldquowomanrdquo ldquoeconomistrdquo ldquobasic educationrdquo andldquograduate educationrdquo) Model II introduces the two non-partisan profiles (ldquotechnocratrdquoand ldquooutsiderrdquo) as predictors keeping partisan ministers as the category of reference Pre-senting our main results Model III assesses the conditional effects of popular approvaland minister profiles Model IV addresses issues of endogeneity discussed in the nextsection Equations II through IV exclude three individual background variables thatcontain information already incorporated by the minister profile scores (ldquoeconomistrdquoldquobasic educationrdquo ldquograduate educationrdquo)

Model I shows a significant and negative effect of presidential approval onthe risk of exit Each additional percentage point of popular support for thepresident reduces the relative risk of a minister leaving his or her portfolio byabout 2 percent This result confirms previous findings as well as general presump-tion that weak presidents are inclined to engage in portfolio reallocation to manageadverse conditions

Regarding institutional controls both ldquocoalitionrdquo and ldquominorityrdquo have positive butinsignificant effects ldquoCalendarrdquo shows a significant and positive effect suggesting thatthe longer the time to the end of the term the higher the risk of ministers leaving officeAll other institutional attributes have insignificant effects indicating that ministerswere equally exposed to anticipated exits irrespective of whether presidents belonged tothe ldquoUCRrdquo or PJ or whether they could be re-elected Ministers occupying the ldquofinancerdquoportfolio appeared to be at greater risk this effect is not significant in Model I but itbecomes statistically significant in Models II and III Results remain similar in modelsalso controlling for portfolios in Interior Defense Foreign Affairs and the Chief of Staffand in models with random frailties

Among the set of economic covariates only ldquoinflationrdquo encouraged portfolioreallocations while ldquoeconomic growthrdquo and ldquoneoliberalismrdquo present insignificanteffects Finally none of the variables capturing ministersrsquo individual backgroundspresent significant effects

Model II estimates the risk for technocrats and outsiders when compared to parti-sans (the reference category when both fuzzy membership scores are zero) This modelconfirms that non-partisan ministers are a heterogeneous group with different careerdynamics While ldquotechnocratsrdquo show a lower (and statistically significant) risk of exitldquooutsidersrdquo present a greater (but statistically indistinguishable) risk than partisan min-isters The effect of ldquoapprovalrdquo remains negative and significant

Our argument however is that presidents protect or remove cabinet membersdepending on popular approval and the ministerrsquos profile Model III evaluates ourhypotheses about conditional effects by including two interaction terms ldquotechnocratapprovalrdquo and ldquooutsiderapprovalrdquo In this model the coefficient for approval cap-tures the effect of government popularity for partisan ministers (ie when indi-viduals score 0 for both ldquotechnocratrdquo and ldquooutsiderrdquo) This effect remains negative andsignificant consistent with Hypothesis 1 partisan ministers are more secure in their postswhen the government is politically stronger


326

Hypothesis 1 also anticipates that technocrats should be safer in contexts of higherpopular support By contrast our second hypothesis suggests that the survival of out-siders should be independent from government approval To assess these empiricalexpectations Figure 2 compares the marginal effects of approval for different degreesof membership in the sets of technocrats (21) and outsiders (22) The figure sup-ports both predictions the effect of approval is negative (ie reduces the risk ofexit) for partisan as well as technocratic ministers but it is statistically indistinguish-able from zero for ministers with any meaningful degree of membership in the setof outsiders

Hypotheses 3 and 4 focus instead on the consequences of minister profiles forminister tenure They predict that when governments are popular technocrats willconfront a lower risk of exit but outsiders will confront a greater risk of exit thanpartisan ministers Testing these hypotheses requires reversing the analysis of con-ditional effects for Model III Rather than comparing the effect of approval for differentminister profiles in Figure 3 we contrast the marginal effect of profiles at different levelsof presidential approval

Figure 2 Marginal Effect of Presidential Approval for Technocrats and Outsiders

21 Effect for Technocrats 22 Effect for Outsiders

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat

Dashed lines give 95 confidence interval

OutsiderDashed lines give 95 confidence interval

Note Values in the vertical axis represent conditional Cox regression coefficients (95 confi-dence intervals) for Model III The value when membership in the fuzzy sets is zero (minus0026)reflects the unconditional effect for partisan ministers The second mediating variable is set to zeroin each graph


327

Figure 31 reflects the consequences of being a technocrat for tenure stability Thecoefficient for ldquotechnocratrdquo in Model III captures the expected effect of individualexpertise (ie of becoming a full member of the technocratic group) when presidentialapproval is 0 In this hypothetical scenario the effect is negative but non-significantThe interaction ldquotechnocratapprovalrdquo reflects the marginal change in the effect of tech-nocracy for each additional percentage point in presidential popularity The first panelof Figure 3 shows that the conditional coefficient for ldquotechnocratrdquo (that is the effectof technocracy when approval is greater than zero) is negative and significant (atplt05) for the most relevant range of approval levels The effect becomes significantwhen ldquoapprovalrdquo is above 30 percent with a hazard ratio ranging from 049 (whenapproval is at 31 points) to 026 (at 70 points) That is when presidential popularityis high the relative risk of technocrats leaving their portfolios is 74 percent lower thanthe risk for non-technocratic ministers

Similarly Figure 32 assesses the conditional effect of being an outsider at differentlevels of popular approval The coefficient for ldquooutsiderrdquo in Model III captures theexpected consequences of being an individual with no expertise and no party affiliationwhen presidential approval is 0 As in the previous case the effect is negative and not-significant The interaction ldquooutsiderapprovalrdquo reflects the marginal change in theeffect of ldquooutsiderrdquo for each additional percentage point in the presidential popularityThis estimate is positive and significant The second panel of Figure 3 shows that the

Figure 3 Effect of Minister Profiles at Different Levels of Government Approval


-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval

Dashed lines give 95 confidence interval Dashed lines give 95 confidence interval

Note Values along the vertical axis represent conditional coefficients for Cox regression model III(dashed lines are 95 confidence intervals)


328

conditional coefficient for ldquooutsiderrdquo becomes positive and significant (at plt05) whenapproval is above 49 percent with a hazard ratio ranging from 224 (when approval isat 50 points) to 652 (at 80 points) That is when presidential popularity is very highthe risk of outsiders leaving their portfolios is more than six times greater than the riskconfronted by other ministers

The comparatively fragile position of outsiders when the government is popu-lar does not reflect an intrinsic change in the position of these ministers in thecabinet Figure 22 showed that public support does not affect the tenure of out-siders However Figure 21 offers strong support for our first hypothesis anticipatinga longer tenure for technocrats and partisans Therefore it follows that the relativeincrease in the proportional hazard for outsiders when government approval is highsimply reflects the safer position achieved by the other two groups under suchfavorable political conditions

The systematic analysis of conditional effects of popular approval and minis-ter profiles shows that the fate of different ministers is hard to distinguish whenadverse political conditions force governments to engage in reactive portfolio reallo-cation strategies However minister types matter above some threshold of popularapproval Differences in the length of minister tenure become more pronounced aspopularity increases and the government is able to employ reallocation strategies moreproactively Our findings clearly support the theoretical expectations articulated in thesecond section

Endogeneity Concerns

A potential concern with the findings presented in Table 1 is that presidential approvalrates may be correlated with the residual of equations I through III Reverse causality or(more plausibly) omitted variables could cause endogeneity bias For example presiden-tial approval would be endogenous if charismatic presidents were more likely to mobi-lize mass popular support and also to retain faithful ministers for longer periods Anincrease in presidential charisma would simultaneously make the administration morepopular and its ministers more durable But because presidential charisma cannot bemeasured directly and is omitted from the equation residual variance in minister dura-tion would be correlated with approval As a result empirical estimates for the effectof approval on minister survival would be biased (upward)

Conventional treatments of endogeneity based on instrumental variables are com-plicated in this context by two features of our study First the effect of the potentiallyendogenous variable is mediated by minister profiles (ldquotechnocratrdquo and ldquooutsiderrdquo)meaning that estimates for the two interaction terms may also be biased Secondbecause the main equation is a proportional hazards model most available implementa-tions of 2SLS or GMM estimators are of little use

To overcome these limitations we introduce a control function in Model IVThe control-function approach relies on the same foundations of instrumental variable


329

models but it allows for a more flexible implementation in non-linear settings20 Theestimator is implemented in two stages in the first stage we predict presidentialapproval using the exogenous variables and four additional instruments discussedbelow We retrieve the residual of this equation as the control function In the secondstage we re-estimate the proportional hazards model adding the control function asan independent variable In a linear model this procedure yields the same estimatesas the 2SLS estimator but in a model with interaction terms it produces more reliableestimates of the parameters of interest

In order to instrument presidential approval we added four exogenous predictorsto the first-stage equation21 The first two items reflect the performance of the Argentinesoccer team in the World Cup We do not claim that soccer dominates the Argentinepolitical mood presidential approval declined consistently after June 1986 whenArgentina won the World Cup and improved consistently after 1990 when Argentinaended as the runner-up But a strong performance in the World Cup presumably boostspublic sentiment while poor results negatively impact public opinion

To capture those effects we included a dummy variable reflecting the timing of theWorld Cup in June and July every four years and another dummy capturing whenthe team made it to the final game The effect of both variables is significant at p lt 01The first coefficient is negative (minus21) and the second one is positive (74) indicatingthat bad soccer outcomes slightly hurt presidential approval while good results liftapproval by about 5 percentage points

The third instrument reflects the growth rate of Chinarsquos GDP (World DevelopmentIndicators) An expansion of Chinarsquos economy produces a favorable environment forArgentinarsquos agricultural exports indirectly boosting public revenue and thus the imageof the incumbent administration The coefficient for this variable is positive and sig-nificant (p lt 01) suggesting that a point of economic growth in China represents animprovement of 19 percent for the image of the president in Argentina

Our last instrument takes advantage of the institutional design of presidentialregimes Comparative studies indicate that presidential approval declines considerablyby the third quarter in office22 Unlike heads of government in parliamentary regimespresidents cannot control the election calendar so the timing for this effect in the seriesis determined exogenously by constitutional design We included a dichotomous indi-cator capturing the first six months in office of elected presidents The coefficient waspositive and significant reflecting on average a honeymoon advantage of twenty-ninepoints (p lt 01)23

The results of the equation including the control function presented as Model IVindicate that our initial estimates are reliable even after accounting for potentialendogeneity Estimates in Models III and IV are almost equivalent in size Statisticalsignificance for the effect of presidential approval declines slightly for partisan politi-cians (p lt 10) but it remains at conventional levels for technocrats The conditionalhazard ratio for a unit increase in presidential approval when the minister is a technocratis 096 (p lt 05) Consistent with our previous findings the effect remains insignificantfor outsiders (100 p 5 72)24


330

Conclusion

Our results underscore four important conclusions for the emerging literature oncabinet portfolio reallocation First although cabinet reshuffling has been conven-tionally presented as a reactive strategy employed to confront adverse situations wehave shown that portfolio reallocation can be employed as a proactive tool Whengovernments are weak the chief executive may be forced to pursue cooperative strate-gies to appease the opposition but when governments are strong leaders have greatleeway to consolidate their control over the cabinet The empirical evidence indeed con-firms that minister survival is considerably more fragile during hard times but it alsoshows that leaders are able to engage in selective choices about minister retentionwhen they secure popular support

Second the analysis of the Argentine case questions static definitions of govern-ment strength focusing exclusively on institutional conditions such as the level ofpartisan support in the legislature or the constitutional prerogatives of the executivebranch Because government approval is a real-time proxy for electoral outcomes itdrives the alignment of most political actors in the cabinet and the legislature evenwhile institutional factors remain stable over the short run Our analysis documenteda consistent effect of public support on minister survival even after we accounted forpossible endogeneity of approval rates

Third our operationalization of minister types has challenged the assumption thatcareer profiles can be properly captured by a clear-cut dichotomy We started from theconventional distinction between partisan and non-partisan ministers but augmented thetypology by identifying two distinct non-partisan profiles technocrats who commandample competence in a given policy area and outsiders who lack partisan affiliationas much as expertise Decisions to retain or replace ministers of each type reflect diver-gent incentives as government leaders seek to maximize different political resourcesgovernability competence and loyalty Moreover we have argued that minister pro-files should be fruitfully conceptualized as continuous rather than discrete personalattributes and advanced an operationalization based on fuzzy-set theory

A more nuanced conceptualization of government strength and minister profilesleads to a reconsideration of the claim that conventionally links non-partisan ministersand the de-politicization of public policy Our findings indicate that technocrats enjoylonger tenures when the government is popular and the chief executive can claim creditfor successful policies At the same time loyal outsiders remain unscathed by shiftsin public approval The interaction of credit-claiming and loyalty-procuring incentivesfor government leaders creates a pattern by which non-partisan ministers are hard todistinguish (among themselves and vis-agrave-vis partisans) in bad times but technocratshave safer jobs than outsiders in good times

Each of the four lessons opens new venues for theory development and empiricalresearch in the field of cabinet reallocation strategies Because data limitations con-strained our analysis to a single country further studies will be needed to explore theseinsights in other geographic or institutional contexts It is undeniable however that


331

the management of cabinet portfolios has great political significance beyond the initialmoment of government formation

NOTES

We are indebted to Luis Costa (IPSOS-Mora y Araujo) Carlos Gervasoni and Ryan Carlin for sharing data onpresidential approval and to Despina Alexiadou Sebastian Jaumlckle Matthew Kerby Cecilia Martiacutenez-Gallardoand Sebastian Saiegh for helpful comments This project was supported by the Global Studies Center and bythe Center for Latin American Studies at the University of Pittsburgh Replication files for this articleare available at httpthedataharvardedudvndvanibal

1 See Samuel Berlinski Torun Dewan and Keith Dowding Accounting for Ministers Scandals andSurvival in British Government 1945ndash2007 (Cambridge Cambridge University Press 2012)

2 See Cecilia Martiacutenez Gallardo ldquoOut of the Cabinet What Drives Defections from the Government inPresidential Systemsrdquo Comparative Political Studies 45 (January 2012) 62ndash90

3 See the SEDEPE network datasets available at httpsedepenet4 Octavio Amorim Neto and Kaare Stroslashm ldquoBreaking the Parliamentary Chain of Delegation Presi-

dents and Non-Partisan Cabinet Members in European Democraciesrdquo British Journal of Political Science36 (October 2006) 619ndash43 Pedro Tavares de Almeida Antonio Costa Pinto and Nancy Bermeo edsWho Governs Southern Europe Regime Change and Ministerial Recruitment 1850ndash2000 (London FrankCass 2004)

5 See Miguel Angel Centeno and Patricio Silva eds The Politics of Expertise in Latin America (New YorkSt Martinrsquos Press 1998)

6 Wilma Bakema and Ineke Secker ldquoMinisterial Expertise and the Dutch Caserdquo European Journal ofPolitical Research 16 (March 1988) 153ndash70

7 See Kaare Strom Ian Budge and Torbjorn Bergman eds Cabinets and Coalition Bargaining TheDemocratic Life Cycle in Western Europe (Oxford Oxford University Press 2008)

8 Carlos de la Torre ldquoEl Tecnopopulismo de Rafael Correa iquestEs Compatible el Carisma con laTecnocraciardquo Latin American Research Review 48 (Spring 2013) 25ndash43 Maria Escobar-Lemmon andMichelle Taylor-Robinson ldquoGetting to the Top Career Paths of Women in Latin American CabinetsrdquoPolitical Research Quarterly 62 (December 2009) 685ndash99

9 John Huber and Cecilia Martinez-Gallardo ldquoReplacing Cabinet Ministers Patterns of MinisterialStability in Parliamentary Democraciesrdquo American Political Science Review 102 (May 2008) 169ndash80Indridi Indridason and Christopher Kam ldquoCabinet Reshuffles and Ministerial Driftrdquo British Journal ofPolitical Science 38 (October 2008) 621ndash56

10 Arend Lijphart Patterns of Democracy (New Haven Yale University Press 2012) Octavio AmorimNeto ldquoThe Presidential Calculus Executive Policy Making and Cabinet Formation in the AmericasrdquoComparative Political Studies 39 (May 2006) 415ndash40 Jorge Lanzaro ldquoPresidencialismo con Partidos ysin Partidosrdquo in Jorge Lanzaro ed Presidencialismo y Parlamentarismo Ameacuterica Latina y EuropaMeridional (Madrid Centro de Estudios Poliacuteticos y Constitucionales 2013) 15ndash62

11 Magna Inaacutecio ldquoEscogiendo Ministros y Formando Poliacuteticos Los Partidos en Gabinetes MultipartidistasrdquoAmeacuterica Latina Hoy 64 (2013) 41ndash66

12 Eric Raile Carlos Pereira and Timothy Power ldquoThe Executive Toolbox Building Legislative Supportin a Multiparty Presidential Regimerdquo Political Research Quarterly 64 (June 2011) 323ndash34 Selena GrimaldiI Presidenti Nelle Forme Di Governo Tra Costituzione Partiti e Carisma (Rome Carocci 2012)

13 Notice that because departures from the cabinet reflect an overall reallocation strategy they do notnecessarily indicate the failure of an individual occupying the specific portfolio

14 Lofti A Zadeh ldquoFuzzy Setsrdquo Information and Control 8 (June 1965) 338ndash5315 Thomas Poguntke and Paul Webb eds The Presidentialization of Politics in Democratic Societies

A Framework for Analysis (Oxford Oxford University Press 2005)16 The survey item reads ldquoIn general do you approve or disapprove the work the national government is

doingrdquo Because the number of categories in the response scale changed in 2004 we applied a correction forlater data equivalent to the one used by Carlos Gervasoni in ldquoA Rentier Theory of Subnational Regimes FiscalFederalism Democracy and Authoritarianism in the Argentine Provincesrdquo World Politics 62 (April 2010)


332

302ndash40 Mora y Araujo conducted surveys since October 1984 to anchor the interpolation we employed adata point from a survey conducted by Edgardo Catterberg in April 1984

17 Marcelo Camerlo and Aniacutebal Peacuterez-Lintildeaacuten ldquoMinister Turnover Critical Events and the ElectoralCalendar in Presidential Democraciesrdquo ICS Working Papers 2 (University of Lisbon 2013)

18 We estimated additional models including controls for ministers of Finance Interior Foreign AffairsDefense and the Chief of Staff In addition and in order to deal with unobserved sources of heterogeneity weestimated a frailty model with latent risk varying by portfolio The results available on request remainedconsistent with the findings presented below

19 INDEC figures for the consumer price index have been seriously questioned since 2007 We do notattempt to adjust the series for this (downward) bias in the analysis

20 James Heckman ldquoDummy Endogenous Variables in a Simultaneous Equation Systemrdquo Econometrica46 (July 1978) 931ndash59

21 To save space we do not report the first-stage model but results are available in the replication filefor this study

22 Ryan Carlin Cecilia Martiacutenez-Gallardo and Jonathan Hartlyn ldquoExecutive Approval Dynamics UnderAlternative Democratic Regime Typesrdquo in Douglas Chalmers and Scott Mainwaring eds ProblemsConfronting Contemporary Democracies Essays in Honor of Alfred Stepan (Notre Dame University ofNotre Dame Press 2012) 203ndash26

23 To assure exogeneity for this instrument the indicator does not code a honeymoon for second termsin office for Eduardo Duhalde (appointed after de la Ruacutea) and for Cristina Fernaacutendez de Kirchner (whoseinauguration was seen as a continuation of Nestor Kirchnerrsquos administration)

24 The R2 for the approval model using only the exogenous variables in Model III is 55 while the additionof the excluded instruments increases the R2 to 69 We have offered substantive reasons to argue that theexcluded instruments are exogenous Moreover in a hazard model akin to equation III but including thefour instruments these variables showed insignificant effects


333

Capa_Comparative politics

Indice_Comparative politics

Notes_on_ContriComparative politics

ICS_MCamerlo_Politics_ARI

involves the preservation of technocrats responsible for successful government policiesbut not necessarily of outsiders who become less necessary in a context in which allministers align with the head of the government

From Cabinet Appointments to Minister Survival

Portfolio reallocation has been increasingly depicted in the literature as a tool todeal with difficult situations such as the presence of ldquoundesirablesrdquo and ldquoopportunistsrdquowithin the cabinet the decline of government popularity low policy performance andeconomic or socio-political conflicts1 The emerging research agenda departs fromthe well-developed literature on cabinet formation In the classic approach portfolioallocation was conceptualized as a tool to obtain legislative support for the establish-ment performance and survival of governments This perspective focused on thepartisan composition of the cabinet underscoring the role of institutional conditions(eg legislative majorities and the nature of coalitions) as the main explanatory vari-able Initially developed to explain the politics of European parliamentary democraciesthe approach later travelled to the realm of Latin American presidentialism2

Alternatively portfolio reallocation studies focus on the appointment and removalof individual ministers within cabinets and privilege explanations that underscore therole of critical events and individual attributes above and beyond partisanship Theemphasis on individual traits has led to a renewed interest in gender technical skillsand extra-party affiliation of ministers as well as the development of more exhaustiveand sophisticated datasets on individual backgrounds3

A central theme in this emerging line of research has been the role of non-partisansin government The issue of ldquoindependentrdquo ministers has been tackled by the literatureson parliamentary and presidential regimes although from different angles Research onEuropean democracies has focused on the increasing presence of ldquooutsidersrdquo in thecabinet over the last decade demonstrating a deviation from traditional career pathsThe need to recruit based on technical skills to face the growing complexity of multi-level governance and the organizational decline of political parties has been posedas the main explanation for the new phenomenon in addition to factors such as theldquopresidentializationrdquo of politics and the institutional separation between executive andlegislative powers4 In presidential democracies where rulers are expected to nominatemore ministers without party backgrounds scholarly attention has focused on expertswithout political experience but with considerable political power Initially triggeredby the performance of successful ldquotechnopolsrdquo during the nineties this research haslinked technocratic politics to the increased impact of economists neoliberalism andstructural reforms5

Despite the relevant contributions of these research agendas three implicit assump-tions present in the literature hamper an adequate consideration of minister profiles andtheir role in contemporary governments First portfolio reallocation is conventionallydepicted as a reactive tool used by weak heads of government to confront negative


316








317







318








319











320







321








322

Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333












317







318








319











320







321








322

Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333











318








319











320







321








322

Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333












319











320







321








322

Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333















320







321








322

Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333











321








322

Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333












322

Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333





Data and Method







323






Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333










Results



324



(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333







(009)976(009)

973(011)

972deg(017)


806(574)

810(578)


378(325)

373(256)


983(017)


1037deg(020)


(443)1239(428)

1175(411)

1167(412)

Coalition 1188(703)

988(593)

1025(598)

0974(692)


(699)1776(771)

2014(878)

2095(1122)


1931deg(731)

1856(702)

1914(862)

Calendar 1000(000)

1000deg(0000)

1000deg(000)

1000(000)


2525(866)

2451(847)

2455(849)


(005)1011(005)

1010deg(005)

1010deg(006)

Growth 971(023)

972(023)

975(023)

977(026)


867(257)

865(256)

861(258)


(016)993(015)

996(015)

996(015)

Woman 821(320)

905(355)

886(348)

878(350)

Economist 678(229)

- - -


- - -


- - -





325








326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333












326





-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333









-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

-08

-04

00

40

8

Mar

gina

l Eff

ect o

f A

ppro

val

0 2 4 6 8 1 0 2 4 6 8 1Technocrat





327





-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333









-4-2

02

4

Mar

gina

l Eff

ect o

f T

echn

ocra

t

0 20 40 60 80Approval

-4-2

02

4

Mar

gina

l Eff

ect o

f O

utsi

der

0 20 40 60 80Approval




328









329








330

Conclusion







331


NOTES



















332











333













329








330

Conclusion







331


NOTES



















332











333












330

Conclusion







331


NOTES



















332











333





Conclusion







331


NOTES



















332











333






NOTES



















332











333















333





comparative politics - ulisboa€¦ · comparative politics april 2015 316. situations. we show in...

Documents