David Fortunato and Matt W.Loftis explained by expected durations.We also add that,if empirical correlates of cabinet stability.Thus,to gen- cycling under electoral endogeneity was the norm,then erate our measure of a government's life expectancy past PBC research would not have failed to recover ev- which we will impute to the cabinet,we accept the col- idence of cycling behavior in these countries as often as lective wisdom of the discipline and derive our mea- it has. sure from the extant literature.More specifically,we A second possibility is that cabinets privileged by estimate the survival model described in Chiba,Mar- some exogenous shock creating a windfall of popular- tin,and Stevenson (2015),which accounts for poten- ity may call early elections without having to ramp up tial selection bias induced by strategic formations by spending in the manner we predict-a stochastic oc- jointly modeling formation and duration.Note that currence that may confound assessment of our the- as King et al.(1990)explain,this duration model cap- oretical argument.Of course,this second possibility tures "events"in its stochastic component(distributed would not produce our predicted relationship,rather, Weibull in this case).while including"attributes"as the this type of opportunistic behavior would bias against measured covariates in its systematic component.For recovering support for our hypothesis.And,because all cabinets in our data,we use the model estimates to our focus is the effect of cabinet durability on pub- predict the number of days the cabinet will last before lic spending,rather than the relationship among eco- dissolution. nomic inputs (e.g.,spending,inflation),economic out- Importantly,we do not pool the risks of dissolution puts(e.g.,growth,unemployment),and electoral timing and replacement.15 We make this choice for two rea- and performance,opportunistic elections of this type sons.First,the risk type that is theoretically salient is are stochastic nuisances rather than threats to design dissolution and,as such,we do not want replacement credibility.We also note that if the true process guiding terminations contributing to our durability estimates spending patterns is some combination of systematic Second,cabinets have no incentive to prepare for re- cycling under electoral endogeneity and stochastic op placement rather than dissolution.In the event of a portunism,then,again,true durations would provide a replacement,the incumbent loses its governing status better fit to our spending data vis-a-vis expected dura- and is consigned to opposition.Given what we have tions.We return to this issue and other potential con- learned about the collective memory of voters,when founders in a short section on robustness following our the next election arrives,it is not the displaced incum- empirical analysis bent that is the focus of competency evaluations,but its successor and any choices made by the displaced RESEARCH DESIGN incumbent are likely to be irrelevant (or nearly so)to the election outcome (Healy and Lenz 2014).16 Taken To test our hypotheses,we first develop a model of together,these factors imply that a cabinet's best strat- duration expectations for the cabinet.We assume that egy would be to ignore replacement risk when generat- cabinets take into account the observable characteris- ing its expectations for durability.Fortunately,the data 是 tics of their government and accordingly generate their allow us to bring evidence to bear on the question of expectations for its duration,with greater or lesser de- expectation formation under competing risks and we grees of certainty.That is,because governments cannot discuss this in the robustness section. reasonably be expected to foresee the precise timing of Our public spending analysis is conducted at the S5.501g "critical events"such as economic downturns,political country-year level.Therefore,to capture each cabinet's scandals,or abrupt fluctuations in public opinion,the expected remaining time in office when the budget is best information they have to generate their expecta- set,we subtract from the cabinet's total predicted life tions are the life cycles of governments past,the observ- the number of days the cabinet has already served at able characteristics of those governments,and the at- the time it presents its annual budget to the legisla- tributes of their own cabinet-a form of Muth's (1961) ture.This estimate of the cabinet's remaining life ex- classic rational expectations.14 pectancy is the covariate of interest in our models of The most obvious starting point for estimating these public spending(explained in detail below).We expect expectations is the rich political science literature on this to have a negative relationship to public spending the durability of governments discussed above.After in general-that a cabinet will spend more as its ex- the work of,for example,King et al.(1990),Laver and pected duration dwindles-and a negative relationship Shepsle (1998),Diermeier and Stevenson(2000),and to deficit spending-that a cabinet will spend less re- Chiba.Martin,and Stevenson(2015),there can be little sponsibly as its expected duration dwindles.Of course. 四 doubt that political scientists have amassed an impres- our measure of the cabinet's duration expectations is sive understanding of the theoretical foundations and estimated with error,and,as such,we take care to model this error structure. 14 For the purposes of research design,there is much to recommend estimating durations from the observable characteristics of the cabi- 1s In estimation,replacements are right censored. net at the time of formation.Most importantly,however,is that these 16 A third consideration would be that,in addition to being much L durability estimates are not endogenous to the changing political less common than dissolution terminations,replacement hazards are economic climate (e.g.,growth,scandal,militarized dispute,etc.)and effectively flat,indicating that they are the product of a much more are therefore free of potential "feedback effects,"where spending is stochastic process(Diermeier and Stevenson 1999).As such,the de- conditional upon on durability expectations,and durability expecta- gree to which cabinets are able to forecast relatively accurate expec tions are then updated conditional on the effects of spending.and tations of replacement terminations vis-a-vis dissolution would be s00n. impaired. 944David Fortunato and Matt W. Loftis explained by expected durations. We also add that, if cycling under electoral endogeneity was the norm, then past PBC research would not have failed to recover ev￾idence of cycling behavior in these countries as often as it has. A second possibility is that cabinets privileged by some exogenous shock creating a windfall of popular￾ity may call early elections without having to ramp up spending in the manner we predict—a stochastic oc￾currence that may confound assessment of our the￾oretical argument. Of course, this second possibility would not produce our predicted relationship, rather, this type of opportunistic behavior would bias against recovering support for our hypothesis. And, because our focus is the effect of cabinet durability on pub￾lic spending, rather than the relationship among eco￾nomic inputs (e.g., spending, inflation), economic out￾puts (e.g., growth, unemployment), and electoral timing and performance, opportunistic elections of this type are stochastic nuisances rather than threats to design credibility.We also note that if the true process guiding spending patterns is some combination of systematic cycling under electoral endogeneity and stochastic op￾portunism, then, again, true durations would provide a better fit to our spending data vis-à-vis expected dura￾tions. We return to this issue and other potential con￾founders in a short section on robustness following our empirical analysis. RESEARCH DESIGN To test our hypotheses, we first develop a model of duration expectations for the cabinet. We assume that cabinets take into account the observable characteris￾tics of their government and accordingly generate their expectations for its duration, with greater or lesser de￾grees of certainty. That is, because governments cannot reasonably be expected to foresee the precise timing of “critical events” such as economic downturns, political scandals, or abrupt fluctuations in public opinion, the best information they have to generate their expecta￾tions are the life cycles of governments past, the observ￾able characteristics of those governments, and the at￾tributes of their own cabinet—a form of Muth’s (1961) classic rational expectations.14 The most obvious starting point for estimating these expectations is the rich political science literature on the durability of governments discussed above. After the work of, for example, King et al. (1990), Laver and Shepsle (1998), Diermeier and Stevenson (2000), and Chiba,Martin, and Stevenson (2015), there can be little doubt that political scientists have amassed an impres￾sive understanding of the theoretical foundations and 14 For the purposes of research design, there is much to recommend estimating durations from the observable characteristics of the cabi￾net at the time of formation.Most importantly, however, is that these durability estimates are not endogenous to the changing political economic climate (e.g., growth, scandal, militarized dispute, etc.) and are therefore free of potential “feedback effects,” where spending is conditional upon on durability expectations, and durability expecta￾tions are then updated conditional on the effects of spending, and so on. empirical correlates of cabinet stability. Thus, to gen￾erate our measure of a government’s life expectancy, which we will impute to the cabinet, we accept the col￾lective wisdom of the discipline and derive our mea￾sure from the extant literature. More specifically, we estimate the survival model described in Chiba, Mar￾tin, and Stevenson (2015), which accounts for poten￾tial selection bias induced by strategic formations by jointly modeling formation and duration. Note that, as King et al. (1990) explain, this duration model cap￾tures “events” in its stochastic component (distributed Weibull in this case), while including “attributes” as the measured covariates in its systematic component. For all cabinets in our data, we use the model estimates to predict the number of days the cabinet will last before dissolution. Importantly, we do not pool the risks of dissolution and replacement.15 We make this choice for two rea￾sons. First, the risk type that is theoretically salient is dissolution and, as such, we do not want replacement terminations contributing to our durability estimates. Second, cabinets have no incentive to prepare for re￾placement rather than dissolution. In the event of a replacement, the incumbent loses its governing status and is consigned to opposition. Given what we have learned about the collective memory of voters, when the next election arrives, it is not the displaced incum￾bent that is the focus of competency evaluations, but its successor and any choices made by the displaced incumbent are likely to be irrelevant (or nearly so) to the election outcome (Healy and Lenz 2014).16 Taken together, these factors imply that a cabinet’s best strat￾egy would be to ignore replacement risk when generat￾ing its expectations for durability. Fortunately, the data allow us to bring evidence to bear on the question of expectation formation under competing risks and we discuss this in the robustness section. Our public spending analysis is conducted at the country-year level. Therefore, to capture each cabinet’s expected remaining time in office when the budget is set, we subtract from the cabinet’s total predicted life the number of days the cabinet has already served at the time it presents its annual budget to the legisla￾ture. This estimate of the cabinet’s remaining life ex￾pectancy is the covariate of interest in our models of public spending (explained in detail below).We expect this to have a negative relationship to public spending in general—that a cabinet will spend more as its ex￾pected duration dwindles—and a negative relationship to deficit spending—that a cabinet will spend less re￾sponsibly as its expected duration dwindles. Of course, our measure of the cabinet’s duration expectations is estimated with error, and, as such, we take care to model this error structure. 15 In estimation, replacements are right censored. 16 A third consideration would be that, in addition to being much less common than dissolution terminations, replacement hazards are effectively flat, indicating that they are the product of a much more stochastic process (Diermeier and Stevenson 1999). As such, the de￾gree to which cabinets are able to forecast relatively accurate expec￾tations of replacement terminations vis-à-vis dissolution would be impaired. 944 Downloaded from https://www.cambridge.org/core. Shanghai JiaoTong University, on 26 Oct 2018 at 03:53:05, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0003055418000436