Introduction
In this article, we highlight the importance of accounting for time in the study of pledge fulfillment, effectively adding a significant element to the ongoing academic discussions of the factors that influence the fulfillment of party promises. It is generally accepted that government decision makers, constrained by limited attention and resources, must prioritize certain issues ahead of others (e.g.
Baumgartner and Jones, 1993;
Jones and Baumgartner, 2005;
Kingdon, 1984). This process occurs in cycles as policy-making is not continuous but rather structured around decision points where the agendas are aggregated and priorities are established (
Bevan et al., 2011). One recurring key aggregation moment is when party platforms are made public during election campaigns. Election platforms are a key element in the establishment of policy priorities because they are the transmission belt between party promises and subsequent government action.
Bevan et al. (2011) state as follows:
Transmission [between policy promises and statuses] is not perfect since legislative priorities and outputs are susceptible to changes in public opinion or media coverage, unanticipated events in the external world, backbench rebellions, changes in the political parties, and the practical constraints of administering policies or programmes (
Bevan et al., 2011: 395).
One approach in the promise-to-policy linkage literature looks at the congruence between the emphases that parties place on different policy issues in their election platforms and the number of programs enacted on those issues by the party that wins the election (see, for example,
Klingemann et al., 1994). Meanwhile, the approach subscribed to in this article looks directly at the promise-to-policy linkage, one specific promise at a time (
Naurin, 2011;
Pétry and Collette, 2009;
Royed, 1996;
Thomson et al., 2017). Previous studies have considered the influence of time on the fulfillment of specific party promises, either as a resource that works in favor of longer-lasting governments (
Thomson et al., 2017) or as a long-term trend through which parties appear to be making (and fulfilling) ever larger numbers of election pledges (
Håkansson and Naurin, 2016). But as far as we know, no published study has ever documented the timing of the fulfillment of specific pledges for the duration of a given government and correlated it with other determinants of pledge fulfillment.
This study is innovative as it takes into account the moment at which individual pledges become fulfilled during a mandate. The traditional methods of logistic and linear regression used in previous pledge fulfillment studies are not well suited to incorporate the timing of pledges. That is why we use a discrete time duration approach to pledge fulfillment. This approach aims to discover whether and explain why some pledges are more likely to be fulfilled than others during a government mandate. The research question is not only whether or not a pledge has been fulfilled but also when the fulfillment occurred. The discrete time approach that we use highlights yet unobserved dynamics. More precisely, we see that if the government does not enact pledges quickly, the probability of these pledges ever being fulfilled drops drastically. The drop in the probability of pledge fulfillment after the first half of the mandate seems understandable in view of the various institutional and political factors that may erode a government’s policy-making capacity.
The discrete time approach also allows us to conduct a more detailed analysis of the importance of a key explanatory variable in the pledge fulfillment literature: budget balance. Our research also extends the study of pledge fulfillment to a new case, the province of Quebec, for the period of 1994–2014, a period which encompasses six successive governments. We also conduct analyses of pledge fulfillment by Canadian parties over seven successive governments from 1993 to 2015. This study analyzes a total of 1431 manually coded election pledges. The rest of the article is organized as follows. First, we briefly cover the scientific literature on pledge fulfillment and then discuss the importance of the timing of election pledges, before moving onto the methods and data utilized in the study. We then discuss the results and their implications before concluding with our general remarks and takeaways.
Theory and Hypotheses
None of the above studies account for time dependence. It is possible that the timing of pledge fulfillment is a uniform or random process that does not affect the likelihood of fulfilling specific pledges as these studies implicitly assume. However, there are theoretical reasons to believe that the time passed since the government has been elected affects pledge fulfillment although it is not clear at the outset whether these effects should delay pledge fulfillment until later or accelerate it at the beginning of a government’s mandate. One theory, inspired by the political budget cycle approach, suggests that governmental policy outputs should accelerate as the next election draws closer. Political budget cycle scholars assume that voters are myopic: they base their retrospective policy evaluations of candidates more on late changes than on earlier changes in the electoral cycle (see
Campbell and Garand, 2000;
Lewis-Beck and Stegmaier, 2000 for reviews). With near-sighted voters, vote-maximizing politicians have a clear incentive to manipulate the political budget cycle to delay the implementation of popular macroeconomic and transfer policies until the last year of an electoral cycle (
Nordhaus, 1975;
Tufte, 1978). The same reasoning can be easily transposed to campaign pledge fulfillment. Assuming that near-sighted voters tend to better remember the fate of the most recent promises, rational vote-maximizing parties have an incentive to manipulate the policy cycle to delay the fulfillment of their popular promises until the year that immediately precedes the next election.
However, the political budget cycle theory has important limitations when it comes to explain the timing of pledge fulfillment. First, not all campaign promises are popular with the electorate. By the theory’s own logic, unpopular pledges should be fulfilled early, not later, in the cycle. Second, several studies cast doubt on the plausibility of the assumption that voters are myopic (
Hibbs, 1989;
Wlezien, 2015). If voters react to what happens over the entire cycle, the incentive for politicians to delay policy change until later in the cycle is removed, and so is the incentive to delay the fulfillment of campaign promises. More directly relevant to pledge fulfillment is a finding by
Pétry and Duval (2017) that the accuracy of Quebecers’ perception of the fulfillment of specific pledges is not correlated to the time at which they were fulfilled during the cycle.
1 Third, in contrast to claims by the political budget cycle literature, the policy agenda-setting literature indicates that governments intending to enact policy changes typically do so shortly after the election and not later in the cycle.
Reasons why one could expect a drop in the rate of pledge fulfillment with time are related to likely changes in the political and societal environment (change in public opinion, change of the economic circumstances, and change of political leader abroad) during the course of a mandate. We elaborate on these reasons, drawing from the legislative policy-making literature, in the following paragraphs.
There is evidence from the policy agenda literature that governments perform the most extensive policy changes soon after their transition to power and not toward the end of their mandate.
Bevan and Greene (2016) show that most dramatic policy changes occur immediately following a party’s transition into power, especially when economic conditions are bad. After this initial period of change, things tend to stabilize (
Mortensen et al., 2011). Researchers generally agree that a party’s policy agendas are typically consistent throughout a government’s duration but that their partisan goals can sometimes be mitigated by political and economic conditions (
Green-Pedersen and Mortensen, 2010). They generally emphasize the role of institutional structures that limit parties’ ability to fulfill their preferred policies (
Bevan et al., 2011;
Jennings et al.,2011a,
2011b). A good example of this is
Bevan and Greene’s (2016) recent article that highlights the effects of different predictors (party organizations, economic conditions, and size of the parliamentary delegation) on the timing of issue attention. Even more telling, given our interest for electoral pledges fulfillment, is
Huber’s (1996) finding that new governments focus on their electoral priorities early in the new legislative cycle. In short, parties are not always capable of creating the policies that they desire past an initial period of dramatic policy change. Essentially, we expect a government’s capacity to enact policies, and to fulfill its election pledges in particular, to be higher at first and then decrease for external reasons (
Bevan and Greene, 2016) and/or for institutional reasons (
Bevan et al., 2011;
Jennings et al., 2011a;
2011b).
Furthermore, the legislative decision-making literature also highlights that parties are not immune to demands from multiple groups both internal and external to the party (
Ceron, 2015;
Harmel and Janda, 1994;
Greene and Haber, 2016), and that it is highly unlikely that these demands remain unchanged throughout the full mandate. The pledges contained in electoral platforms are said to have important intra-party functions (
Rose, 1984;
Thomassen, 1994;
Thomson et al., 2017), but these pledges can’t be retrospectively changed to address changes in the demands of the groups pressuring the government party. These demands will cause a change in the present policy agenda and the pledges that were not enacted when change occurs will become irrelevant when weighed against new priorities. This is amplified by the fact that parties have limited resources and that they will not attach those resources to issues they hold unimportant or that would require significant ideological compromise (e.g.
Döring, 2003). Disagreement within a party is also a key factor that affects the timing of the legislative process. Disagreements slow down the enactment of legislative proposals already in the pipeline (
Haber, 2015), which can mean that some pledges will die on the order paper. Intra-party demands also become more important at later points in the government’s mandate, as shown by Ceron, as the impact of the preferences of intra-party groups constrains party leaders closest to the election (
Ceron, 2015). Looking at inter-party legislative dynamics,
Martin and Vanberg’s (2011) demonstrate that less ideologically contentious legislation is often passed before more contentious politics.
Greene (2016) adds that this has implications for coalitions’ ability to maintain their majority in parliament.
All those factors, and their possible combinations, highlight the diminishing return of pledge fulfillment throughout the mandate. Circumstances in which governments will prefer to fulfill promises later rather than sooner in their mandate do not easily come to mind, which leads us to hypothesize as follows:
H1. The likelihood of election pledges being fulfilled is higher at the beginning of a mandate than at later stages.
Methods
First, we manually coded each individual election pledge. We identified election pledges as defined by
Thomson et al. (2017). All members of the CPPG also employ this method. Election pledges are statements that contain unequivocal support for proposed government policy actions or outcomes that are testable. The “unequivocal support” part of the definition stipulates that a statement must imply an explicit promise to do something. Some language is too soft to be considered “unequivocal support.” Statements in which parties promised to “consider” or “look into” specific policy actions do not qualify as pledges under
Thomson et al.’s (2017) definition. The second part of the definition stipulates that pledges contain “proposed government policy actions or outcomes” that are testable. This clause demands that pledges describe the proposed policy or outcome in an explicit way so that a criterion is provided on the basis of which the fulfillment of the pledge can be judged. Regarding actions, these criteria consist of the passing of particular legislation or executive orders. Pledges may also refer to outcomes: for example, statements such as “we will strive toward reducing inflation,” or “our program will reduce unemployment by 100,000” are considered pledges. In other words, the writers of election programs provide the criteria used to judge the fulfillment of pledges, not the researcher (see also
Thomson, 2001).
The coding scheme is the same as the CPPG’s. At first, pledges are classified in three categories: (1) pledges kept, (2) pledges kept in part, and (3) broken pledges. To be classified as “kept,” a pledge has to be followed by a subsequent government action (a law, a regulation, a treaty, or an agreement) that has been passed. A pledge is rated as “kept in part” when the corresponding action is a compromise (the action is completed, but it does not go as far as what was initially promised). A pledge is classified as “broken” when it is not followed by a governmental action. Those categories are then regrouped as 1 = “At least kept in part” or 0 = “not carried out” for analysis purposes and for comparability with other members of the CPPG. All of this was done manually for every government in the study.
To account for the timing of pledge fulfillment, we opted for the Event-History Analysis approach (
Allison, 1984;
Yamaguchi, 1991). Event-History Analysis seeks to explain the likelihood that certain subjects (pledges) will experience an event of interest (fulfillment), determine when this interest will occur, and determine what are the covariates influencing that likelihood (
Gottman, 2013). Following
Beck et al. (1998), we chose to model the fulfillment of pledges under a time-series cross-section analysis with a binary dependent variable (Binary Time-Series Cross-Section (BTSCS)). In essence, modeling the fulfillment of pledges in this way allows us to show not only when a pledge is fulfilled or not but also to identify when the series “switches” from not fulfilled to fulfilled, how long does it take for the fulfillment to happen, and what are the covariates associated with this change. This means that we are looking at the “status” (fulfillment) of each pledge at a given time. In this case, quarters for the duration of the mandate. Our first dataset consists of 603 pledges made over six governments over the period of 1994–2012, resulting in a total of 8100 temporal observations for Québec. Our second dataset consists of 828 pledges made over seven governments between 1993 and 2015, resulting in a total of 10,052 temporal observations for Canada.
This means that our model includes a time spell variable, which is a count of the number of quarters since the mandate started or since the fulfillment (event) occurred, whichever is most recent. This time spell is included in the model under the form of a cubic polynomial approximation following
Carter and Signorino (2010).
2 Including the cubic polynomial approximation is a trivial implementation; it simply consists of including the time spell, time spell squared, and time spell cubed in the regression models (
ts, ts2, and
ts3). Most of the work that comes with the choice of using a BTSCS approach instead of the regular pledge fulfillment modeling is the manual addition of a fulfillment date to every pledge coded as
at least kept in part in the datasets; the rest can be automated very straightforwardly. The fulfillment date corresponds to the day (which is later recoded as quarter) when the condition to be coded as either kept or kept in part were achieved. For example, promises that required a law to be passed were coded as kept and dated the day of the assent. Promises that required an investment were coded as kept the day the budget was passed and so on.
We also control for factors found in the pledge fulfillment literature: namely, the type of government—majority or minority, whether the government party is an incumbent government or not and, at the pledge level, whether each specific pledge was in agreement with the opposition’s campaign pledges and the type of pledge.
Majority and Minority Governments
It is unclear if the absence of a formal coalition in a minority context makes it harder to create policy. While some assume that minority governments are ineffective, others such as
Strøm (1990) provide evidence that this is often not the case; minority governments in Europe are surprisingly effective. In Canada,
Godbout and Høyland (2011) show that
informal voting coalitions are a common occurrence under minority governments. In fact, evidence indicates that Canadian minority governments fulfill as many if not more pledges as majority governments while they last. They appear to fulfill fewer pledges than majority governments simply because majority governments last longer (
Pétry and Birch, 2016;
Pétry and Duval, 2015). Nonetheless, majority governments are coded 1 and minority governments are coded 0.
First and Recurring Mandates
Reelected parties are often better positioned than newly elected parties to fulfill their election pledges. This is due in part to the fact that they have “the opportunity to obtain public-service advice and expertise in the day-to-day operations of government” as well as “within the process of drafting their election platforms being partially responsible for these outcomes” (
Flynn, 2011: 249). They may also take advantage of the projects in the legislative “pipe-line” that died on the order paper when the election was called. These elements benefit reelected parties once they return to government relative to newly elected governments. We argue that first mandate governments are at a considerable disadvantage and will therefore fulfill fewer pledges than recurring governments. The variable for first mandate governments is coded 1 for first governments and 0 for returning governments.
Inter-Party Agreement
Agreement between government and opposition indicates that the pledge is in response to widely shared societal demands, rather than reflecting narrow partisan interests. It also seems likely that the government will find very little opposition when the time comes to enact these pledges. Previous research on pledge fulfillment found that pledges made by the party in government were more likely to be acted upon when they were in agreement with pledges by opposition parties (
Pétry and Duval, 2015;
Thomson et al., 2017). Agreement is a categorical variable with the following two levels: 0 = “No agreement or disagreement” and 1 = “in agreement with at least one party.” For Canadian pledges, the comparisons were made among the electoral pledges of the LPC, the
Progressive Conservative Party (PCPC), and the
New Democratic Party (NDP) from 1993 to 2004 and the CPC, PLC, and NDP from 2006 to 2015. For Quebec pledges, the comparisons were made among the
Parti Québecois (PQ), the
Parti Libéral (PLQ), and
Action Démocratique du Québec (ADQ) from 1994 to 2008 and among the PQ, PLQ, and
Coalition Avenir Québec (CAQ) for 2012.
Type of Proposed Change
One property of pledges likely to affect their fulfillment is the type of change they promise. Pledges that promise change may be more difficult to fulfill than pledges that promise to maintain the status quo, given the incrementalism of large governments. There is evidence from past studies that pledges to initiate change are less likely to be fulfilled than pledges to maintain the status quo (see, for example,
Thomson, 2011). The variable for the type of proposed change has four categories aside from maintaining the status quo (the reference category): (1) to expand governmental spending, (2) to contract governmental spending, (3) to introduce other non-budget-related changes, and (4) to achieve any type of policy outcome not related to governmental spending or taxation (e.g. to create jobs).
3
Budget Balance
Budget balance is our key economic indicator reflecting governments’ financial capacity to fulfill their election pledges. This is operationalized as the previous year’s budget (expressed in $B). This variable is often absent from existing pledge fulfillment studies; however, we think it is an important factor as the surplus or deficit generated in previous years directly relates to their pledge fulfillment capacity. One anecdotal illustration is the first Liberal government of Jean Chrétien elected in 1993. Chrétien’s “Red Book” election program contained several promises to expand social programs (childcare, social security); however, the fact that the Liberal government was facing a budget deficit forced Jean Chrétien to abandon several of these promises during his first mandate (
Prince, 2006). The data were collected through the
Institut de Statistique du Québec’
s website for Quebec and through
Statistics Canada’s website for Canada.
Results
We coded a total of 828 government pledges for Canada with levels of fulfillment ranging from 53% for the first Chrétien government elected in 1993 to 84% for the third Harper government elected in 2011. Of those pledges, 565 were fulfilled at least in part, with a relatively high overall fulfillment rate of 68%. The low pledge fulfillment score of the first Chrétien government stands as an exception here. The low score can be attributed to the fact that the 1993 elections were essentially an adjustment period for the Liberal Party, which made several very ambitious promises that it could later not fulfill (
Flynn, 2011).
4
For Québec, we coded a total of 603 pledges in the six electoral platforms in our data. Of those pledges, 362 (60%) were at least partly fulfilled and 241 (40%) were not. The PQ made 99 pledges during the 1994 election campaign and fulfilled 71 at least in part (fulfillment rate of 72%). The PQ government elected in 1998 fulfilled 94 of the 127 pledges made during the election campaign (fulfillment rate of 74%). The fulfillment score for the PLQ government elected in 2003 is 53% with 56 pledges at least partly fulfilled out of 106; the PLQ government fulfilled 55 of the 98 pledges made during the election campaign of 2007 (with a fulfillment rate of 56%). The score for the PLQ government elected in 2008 was 47% (29 out of 62 pledges at least partly fulfilled). The PQ government elected in 2012 fulfilled at least in part 57 pledges out of its 113 campaign pledges (50%; see
Figure 1).
The federal context is quite comparable to the provincial one given that we are in the presence of the same electoral system and that there were three relatively short-lived minority governments during the period (2004, 2006, and 2008). Despite the many similarities, Québec’s scores are significantly lower, the difference being 8% points (or 12% points if we exclude the 1993 federal government). We can also compare these results with other countries; the proportion of fulfillment in the United States from 1976 to 2000 was 64.4%, 85.4% in the United Kingdom from 1974 to 1997, and 50.3% in Ireland from 1977 to 2007 (Thomson et al., 2014).
Figure 2 illustrates the variation in the rate of fulfillment of pledges as time unfolds for the duration of a mandate. It can be seen that the fulfillment of pledges during a mandate is far from uniform. As expected, there is a concentration in the fulfillment of pledges early on in the mandate, followed by a significant drop. Of our 565 fulfilled pledges in Canada, 286 (51%) were fulfilled within the first year of their respective governmental mandate and 165 (29%) were fulfilled the following year. Only 86 (15%) were fulfilled in the third year, and the remaining 5% of the pledges were completed in the last year. For Quebec, out of 362 fulfilled pledges, 151 (42%) of the pledges were fulfilled in the first 12 months of their respective governmental mandates, 124 (34%) were fulfilled in the second government year, 62 (17%) in the third year, and 25 (7%) after that.
This suggests that the traditional way of accounting for time in the pledge fulfillment literature so far (e.g.
Thomson et al., 2017) as the number of months that a government lasted only unveils a small part of the story. While it is true that governments that last longer typically fulfill more pledges, the difference once the 20-month threshold in mandate duration has been reached is marginal. The asymmetrical nature of the fulfillment of pledges through time is well captured by the skewness of our two distributions; both are heavily skewed to the right with a skewness of 1.02 (with a kurtosis of 3.56) for Canada and 0.90 (with a kurtosis of 3.22) for Quebec.
The bivariate results that have been presented so far strongly suggest that our expectation regarding the changing likelihood of pledge fulfillment through time is well founded. We now turn to a multivariate model, in which the variables relating to budget balance, first mandate, majority governments, and inter-party agreement are added. As mentioned above, we do so using BTSCS regressions models with a cubic polynomial approximation of time where our dependent variable is the fulfillment of pledges (1 = fulfilled, 0 = unfulfilled). The results are displayed in
Table 1, with Model 1 for Canada and Model 2 for Québec.
This brings us to the central point of the article, time. As
Carter and Signorino (2010) point out, trying to interpret the time-spell coefficients in
Table 1 directly is a rather futile enterprise. To facilitate interpretation of our results, we draw a hazard plot (
Figure 3). This is essentially a graphical representation of the probability of pledges being fulfilled as time passes. As we can see in
Figure 3, time matters. Holding everything else constant, the likelihood of election pledges being fulfilled will be higher at first and then decreases substantially as hypothesized. More specifically, after roughly 2 years in power if a pledge has not been fulfilled its probability of later being fulfilled drops at a fast rate. Canada’s slope is constantly dropping whereas Québec’s remains stable for about 2 years before dropping at a faster rate. In Canada, the predicted probability of a pledge being fulfilled dropped by 0.02 (from 0.96 at Q0 to 0.94 at Q3) the first year, 0.07 the second year (from 0.93 at Q4 to 0.86 at Q7), 0.11 the third year (from 0.86 at Q8 to 0.75 at Q11), and subsequently dropped to 0.52 by our last time point (Q14). In short, a very slow decrease for the first half of the mandate followed by and an abrupt one for the second half. In Quebec, the predicted probability of a pledge being fulfilled remains pretty stable, hovering around 0.58 for the first 2 years (from 0.59 at Q0 to 0.57 at Q3 and 0.59 at Q7) and then drops by 0.05 the third year (from 0.59 at Q8 to 0.54 at Q11), and by 0.49 subsequently to find itself at 0.10 at our last time point (Q17). Essentially, the drop in probability is quite dramatic as the next election approaches, highlighting what we believe could be a pledge fulfillment cycle. The probability of a pledge being fulfilled is relatively stable at the beginning of a mandate, but at later stages of the mandate, each quarter that goes by is accompanied by a drop in that probability. This supports our hypothesis.
The dotted lines in
Figure 3 present the same probability if we run the models without our minority governments. This is done to show that our results are not mere artifacts of having a few shorter minority governments in our data. The shape of the distributions is highly similar, the only difference being that the probability for majority governments is slightly higher at times. The probability of pledges being fulfilled looking only at majority governments is slightly higher at first in Canada but only by 0.04 on average for the first 2 years (0.93 compared to 0.97 for the same Q0–Q7 period), after which the probability drops much more quickly for majority governments. At Q12, both curves are at the same probability (0.71) but the majority government only estimates drop to 0.25 and meanwhile the curve regrouping all governments only drops to 0.52. Meaning majority governments tend to shift their focus away from election pledges quicker than minority governments. In Quebec, there are barely any differences.
Overall, we find strong support for H1 in both cases. The likelihood of election pledges being fulfilled is much higher at the beginning of a mandate than at later stages.
Figure 4 presents the results for the interaction between the effects of time (measured by our cubic polynomial terms), the budget balance. In Canada, higher budget balance correlates with much more stable and higher odds of fulfillment. When the previous year budget surplus is at its highest, the likelihood only drops from 0.96 to 0.54; meanwhile when it’s at its lowest, it drops to 0.42. In Quebec, we find a similar effect, only stronger. When the budget balance is at its highest, the probability of a pledge being fulfilled increases for the first 2 years (0.60 at Q0 to 0.70 at Q8) before it drops to 0.37 over the rest of the period. With the biggest deficit, the probability drops fairly quickly from 0.60 to 0.51 (Q4) and then hovers around that value until Q11 (0.50) after which it drops to nearly 0 by the end of the period (0.05). Both the Canadian and the Quebec cases support our expectation that the likelihood of pledge fulfillment drops faster over time when the budget balance is low than when it is high.
Conclusion and Discussion
Time is an integral part of the phenomenon we are trying to explain. As highlighted in the policy agenda-setting literature and legislative behavior literature, policy-making is not a continuous process (e.g.
Bevan et al., 2011). In this article, we show that the probability of pledges being fulfilled is high at first and then drops considerably during the mandate. More specifically, our results indicate that the drop in probability occurs around the middle of a typical Canadian/Quebec mandate that runs its full 4-year course. Whether our broad expectation of a decrease in probability over time and the more specific findings we uncover in Canada and Quebec are supported in general is still unknown. Further studies are needed before we can know if there is in fact such a thing as the pledge fulfillment cycle that was described here.
Whether our findings about the importance of time are generalizable or not, our way of accounting for time allowed us to unveil interesting and not so obvious findings. First, the likelihood that governments fulfill their campaign pledges decreases over time as previous findings in the policy agenda-setting literature suggest and contrary to what the political budget cycle literature would lead us to believe. Second, the finding that there is little difference between the hazard plots for minority and majority governments in Canada and in Quebec (
Figure 3) sheds light on the somewhat puzzling finding in previous studies of pledge fulfillment that minority governments do not fulfill fewer pledges than majority governments (
Artés and Bustos, 2008;
Naurin, 2014;
Thomson et al., 2017). Third, our research design allows us to uncover new and interesting interactions between pledge fulfillment and budget balance.
This first foray in the topic of the relationship between time and pledge fulfillment raises more questions than it answers. One interesting avenue of research left unexplored so far is whether certain types of pledges are more likely than other types to be fulfilled at a particular time in the cycle. Are pledges that appear more important to the government (or to the public) likely to be fulfilled early in the cycle? Another avenue of research would be to examine whether the timing of pledge fulfillment behaves differently in single party governments such as Canada and Quebec and in coalition or divided governments. A definitive answer awaits further tests of the timing of pledge fulfillment to see if it remains constant or even increases over time in coalition and divided governments. However, the possibility appears unlikely. Our general argument is that the unpredictability of future political developments motivates elected politicians to fulfill campaign pledges early in their mandate. As this article shows, the argument is well supported by the evidence in single party governments in which there is a relatively small degree of political unpredictability. It logically follows that the argument should be supported at least equally if not better in coalition or divided governments in which the level of political unpredictability is usually larger (
Lupia and Strøm, 1995;
Warwick, 1994).