Comparing Blunders in Government

Much attention has been paid to ‘blunders’ and ‘policy disasters’. Some argue, on the one hand, that the UK’s political and administrative system is disproportionately prone to generating disasters, but offer no systematic evidence on the record of failures of policies and major public projects in other political systems. On the other hand, research on cognitive biases and other failures of collective decision-making has developed highly generic frameworks that are used to assess cases of perceived policy failure. Both of these perspectives rely on post-hoc assessments of failure and intentions, often from those actors involved in the process. This paper develops a comparative perspective on ‘blunders’ in government. It does so by (a) developing theory-driven expectations as to the factors that are said to encourage ‘failure’, and (b) by devising a systematic framework for the assessment of policy processes and outcomes. The paper applies this novel approach to a set of similar ‘failures’ in particular domains (i.e. in public buildings, transport infrastructure, IT projects, benefits/tax systems, and aerospace/defence projects) to assess whether different politicaladministrative systems are prone to different kinds of ‘blundering’ or whether there are universal patterns in the occurrence of costly and avoidable policy mistakes across policy domains. Paper for the UK Political Studies Association annual conference, Sheffield, 2015 Why do governments blunder? Political science is inherently interested in the question as to why things go wrong in government. Ranging from the ‘adversary politics’ hypothesis concerning British politics (Finer 1975), to studies of implementation (Pressman and Wildavsky 1973) to the more comparative attempts at explaining ‘policy fiascoes’ (Gray and t' Hart 1998), scholars have been pre-occupied with identifying policy episodes perceived as cases of poor performance characterized by flawed planning and management. These episodes are typically considered in the light of their particular political features or context. An earlier example might be the call for a German-type administrative system by Woodrow Wilson (1887) which was seen as a remedy to advance the administration of the emerging US political system. This text was ‘discovered’ in the 1920s when, again, the capacity of the US political system was under challenge. Regardless of era, the conclusion of these various accounts of blunder-prone government is, in turn, advocacy of political regime change. For example, changes in electoral systems are put forward to encourage change in political style, administrative reform, such as calls for more ‘Weberianism’, is seen as enhancing quality, or calls for wider changes in the political economy of particular countries are seen as essential in advancing economic performance. Among the most recent, and prominent incarnations of this interest in the question of ‘why we are so badly governed’ is King and Crewe’s (2013) The Blunders of our Governments. Their account suggests that well-intentioned individuals are inevitably led down the path of policy blundering as the British political system discounts deliberation and over-centralises executive power. In doing so, they stand in a long line of accounts that suggest that policymaking in Britain is let down by its own constitutional features. As noted, this tradition can be traced back to Finer (1975; also see Hailsham 1976) who suggested that the tendency to strong one-party government led to continuous U-turns and policy reversals (which, empirically, however remained rather elusive), as governments tend to pursue partisan agendas to satisfy activists and party elites. More recent formulations of this argument are Dunleavy with his unique brand of commentary criticising the Westminster system as offering the ‘fastest law of the west’ and an incomplete set of checks and balances (Dunleavy 1994) and Moran’s (2001; 2003) analysis of political and institutional causes of policy catastrophes in the age of The British Regulatory State. What is common to this tradition of thinking is the view that the British political system is prone to large-scale, avoidable and expensive policy mistakes which stem from over-confidence, ambitions of high politics, and the clubbish nature of decision-making across governing institutions. 1 The problems with the literature on government ‘failure’ and ‘blunders’ are multiple. Take the literature on the British state and its multiple diagnosed failings as an example. First, the literature usually reflects on the political position of the author within the British political system. An emphasis on blunders and other dysfunctionalities of the British political system are typically put forward by those sceptical of the majoritarian characteristics of the Westminster system. Second, the frequent recourse to anecdotal reminiscence of key actors, rationalizing their own role in decision-making after the event is arguably flawed as a method. Regardless of method, a pure reliance on the wisdom of hindsight may betray the uncertainties that shaped the context of decision-making. Third, assertions are made that Britain's political system is particularly blunder-prone without offering real (or any) comparative evidence, a point originally made by Gray (1996). In other words, the implied argument is that different political systems are in some kind of ‘blunder’-race with Britain leading the way, whereas other countries subject to more ‘gridlock’ (or more consensual styles of politics) are lagging behind and enjoy greater policy success (or at least fewer catastrophic failures). However, it is not clear whether Britain is lonely at the top when it comes to government blundering and dysfunction. For example, readers of Hacker and Pierson’s (2010) Winner-Take-All-Politics might similarly suggest that the US system is particularly flawed (and arguably with more highly significant societal consequences), while Jones and Baumgartner (2005) point to the erratic pattern of political attention across institutions. Fourth, what is actually a blunder is also difficult to define as relevant time periods are hard to establish /define, and official and intended outcomes may conceal other motives and side-effects. Peter Hall's (1980) seminal Great Planning Disasters opened with the example of the blunder-ridden construction and public financing of the Sydney Opera House, before acknowledging that the project ultimately had been a great triumph when measured against more subjective and long-term iconic grounds. Finally, the reasons for ‘blunders’, ‘disasters’, ‘fiascos’, ‘catastrophes’ and ‘failure’ are rarely investigated in any major systemic way; at best, lists of the different ‘failure mechanisms’ are presented and then associated with a variety of cases. 1 Somewhat differently, in the 1970s, it was argued that it was the lack of capacity of the state in being able to govern the economy that explained the British relative decline and the absence of major projects. This was either due to ‘overload’ or due to the lack of linkages between government and finance/industry (King 1975; Hayward 1976). This is not to say that we dispute the diagnosis of 'failure', 'fiasco' or 'disaster' in any one particular case. We also do not want to appear as apologists for one particular political system over another; however. Instead, the following should therefore be seen as a call for a genuinely comparative analysis of the cases, and therefore identification of particular mechanisms. Whether political systems appear to generate more or less, or particular types of failure is, ultimately, an empirical question. THEORETICAL PERSPECTIVES ON GOVERNMENT BLUNDERING In the following discussion, we develop four complementary perspectives on governmental and policy ‘blundering’. As starting point, we assume that decision-making always takes place under conditions of ambiguity and uncertainty. In other words, we assume that probabilities of particular events occurring cannot be easily calculated, and neither can their impact. Decision-making processes are characterised by pre-existing norms and worldviews, fluid participation, and strategic and self-interested actors. Furthermore, we assume that most decision-making in government deals with ‘wicked issues’ (Rittel and Webber 1973), they are inherently about value trade-offs and multiple contradictory dimensions and objectives. In order to explore different dimensions of blundering we distinguish between sources of blundering on two dimensions. We distinguish between those sources for blundering according to whether a lack of resources or a high degree of resource depletion can be associated with blundering. Furthermore, we can distinguish sources for blundering in terms of whether supposed blunders are caused by the (tragic) choices of individuals, or by actors whose actions are channeled through the inherent characteristics of politico-administrative systems (such as the ‘Westminster system’). Table 1 points to these two dimensions and the four resultant sources for policy blundering. We explore each of these four perspectives in more detail below. Table 1: Four Types of Government Blunder No Resource Issues High Resource Issues Blunder due to particular choices Hyper-Excited Politics Instrument Choice Blunder due to underlying structures/systemic features Constitutional System Administrative Capacity The four perspectives on blundering point to various reasons as to why policy decisions can go ‘wrong’. In the case of hyper-excited politics the source of failure lies in intentional choices by individuals or groups of individuals. This choice is associated with ‘moral panics’, Pavlovian politics and other external pressures that make politicians and official rush policy announcements only to repent at leisure when these commitments turn out to be counterproductive, more costly than expected, fail to achieve intended outcomes, or generate no interest (see Cohen 1972; 2002; Lodge and Hood 2005; Breyer 2003; Sunstein and Zeckhauser 2011). Examples include policy over-reactions to perceived risks such as mobile phones or high profile announcements to provide public funding for flood victims which are never taken up. Similarly, the context of iconic or symbolic politics is often associated with hyper-excited blundering (Moran 2001; Jennings 2013). Politicians (and senior administrators) in the realm of high politics are motivated by the possibility of creating signature projects or legacies (whether this is in terms of buildings, enduring reforms to welfare or tax systems, events or other high-profile projects) and, therefore, are prone to optimism bias in discounting potential disadvantages and costs. Specifically, ‘fast-thinking’ (Kahneman 2013) by key decision-makers acting under political pressure may make them vulnerable to cognitive biases, such as disproportionately ‘locking-on’ to particular bits of information (see Jones 1994 for a wider discussion of bounded rationality in policy-making). 2 Blundering due to particular features of the constitutional system is often put forward as accounting for the record of failure in British central government (see Dunleavy 1994; King and Crewe 2013), as noted earlier. Similarly, the wider literature on ‘credible commitment’ and ‘veto points’ has suggested that some political systems are more likely to witness policy reversals than others, and therefore to be more prone to U-turns and other policy inconsistencies (e.g. Weaver and Rockman 1993; Pierson 2001; Tsebelis 2002). The propensity towards ‘strong government’ is seen as problematic as it reduces deliberation opportunities and therefore also the information available to decision-makers, as well as failing to check over-enthusiastic and over-confident executives. Elsewhere, Janis (1972) famously notes insularity of collective decision-making as a source of blundering in foreign 2 Note that this view of 'fast' and 'slow' thinking is disputed. In areas of genuine uncertainty, fast thinking (or heuristics) is the superior strategy. It might be argued that the large scale projects represent a large series of loosely coupled and decentralised decisions, made under conditions of ambiguity (i.e. uncertainty). However, initial choices in terms of embarking on a particular mega-project might be seen as somewhat different to genuine choices in uncertainty that might apply to financial meltdowns, natural catastrophes or military invasions, for example. policy although this insularity was arguably less a result of the constitutional system. While, therefore, majoritarian systems are seen, especially from a British perspective, as problematic, a similar case can be made for consensus systems. Here, the need to bargain and seek consensus is seen as delaying decisions, requiring side-payments and pork-barreling, and slowing down the business of government. Instrument choice as a source of blundering is associated with explanations that suggest the ‘wrong tools for the job’ have been chosen for the policy problem or objective in question. Here blunders are caused, or could be avoided, because an inappropriate or sub-optimal method, such as a regulatory standard or economic incentive (e.g. taxes, subsidies) is chosen to achieve the desired effect. Policy instruments differ in terms of their degree of ‘coerciveness’ and ‘depletability’ (Hood 1983; Hood and Margetts 2007). For example, using advertising campaigns to change behaviours is somewhat ‘cheaper’ than extensive policing, backed by a highly punitive enforcement approach. Similarly, the use of highly punitive and visible punishment might not just require the creation of a particular enforcement machinery, but it is also likely to trigger resistance and avoidance, requiring, in turn, further repressive organisational tools. In other words, the less the degree of social acceptance of the chosen policy and its instruments, the more likely it is that there will be extensive gaming and counter-learning (Hood 1976). Furthermore, the use of long-term fixed contracts, such as in the case of public-private partnerships, may shift the financial burden from current to future generations (something which could interact with cognitive biases, such as the hyperbolic discounting of future costs). Indeed, instrument choice also links to wider organisational features, namely the dispersion of responsibility for particular activities. Hood (1976) has noted problems associated with ‘multi-organisational sub-optimisation’ -where the activities of multiple organisations across multiple domains are interdependent without necessarily being aware of, or sharing, each other's objectives. Similarly, Pressman and Wildavsky’s (1973) classic implementation study noted how the ‘wrong’ instrument, namely one with too many ‘clearance points’ was inevitably leading to a disappointing policy outcome. Finally, blunders are often associated with a lack of administrative capacity. Administrative capacity is related to bureaucratic organisation and competence. This factor therefore differs from instruments in that the latter relate to the resource implications of particular modes of policy intervention. Administrative capacity, in contrast, relates to the skills and capabilities that are expected to exist within bureaucracies. In other words, a lack of administrative capacity is defined by the unavailability of relevant and suitable skills within public and private organizations, issues with communication and co-ordination within and across organisations in the design and delivery of policies, and the ability to monitor ongoing policies and projects. Information-gathering, sharing, and -processing are inherently demanding and resource-intensive, especially in the context of governments’ scarce attention (see Jones and Baumgartner 2005; 2014). A lack of co-ordination capacity might be said to lie at the heart of failures of collaboration between different government agencies and private and para-public bodies. Co-ordination capacities are also at the heart of problems of ‘turf’ in government. For example, silos in cross-agency intelligence sharing were identified by the 9/11 Commission as a key factor in the failures that led to the attack on the US on September 11 th . Elsewhere, in the British context, Dunleavy (1994) identifies the loss of institutional memory due to an almost never-ending cycle of administrative restructuring and reform as a major limit on administrative capacity. A lack of administrative capacity might also be linked to a lack of staff in policing certain activities (such as undertaking food inspections).For example, a lack of civil service expertise in procurement is often attributed as a factor in cost overruns in major IT projects and large defence contracts (e.g. Margetts 2012). More generally, one can distinguish administrative capacity in a variety of types: (i) regulatory, (ii) delivery, (iii) co-ordination, and (iv) analytical (Lodge and Wegrich 2014). Accordingly, blundering may emerge due to (i) dysfunctional oversight activities, (ii) insufficient resources to engage at the policy frontline, (iii) an inability to bring together and maintain a diverse set of actors, or (iv) failures of imagination due to a lack of technical capacity to process information and forecast trends. How does this framework for analysis differ from existing studies of blundering governments? The preponderance in the literature places the state and the underlying constitutional framework at the heart of its analysis. For Dunleavy (1994), for example, the ‘fastest law in the West’ facilitated the passage of major public sector reforms, therefore, worsening the potential effects of determined decision-making within a more or less vetopoint free context. For Moran (2001; 2003), the ‘incomplete penetration of the regulatory state’ logic within the informal logics of the Westminster system has generated its own dysfunctionalities and ‘catastrophes’. For King and Crewe (2013), the constitutional system facilitates rather than hinders over-excited politicians and civil servants. We do not suggest that such reinforcing processes do not matter, but we suggest that in order to establish whether the ‘constitution’ is to blame, we need to focus on these aspects in more detail, and cross-nationally, in order to understand the nature of hyper-excited politics in nonWestminster systems. We also need to better understand the role of instrument choice, administrative capacity and hyper-excited politics, in order to assess the contributing or enabling role of the constitutional setting of policy blunders. A METHOD FOR ANALYSIS How might one approach analysis of the conditions associated with policy disasters? Given that even in apparently blunder-prone settings policy disasters are relatively rare events compared to the usual politics of the mundane, it is necessary to find a way of sampling them. One option would be to examine the properties of all policies of a given government for a defined period, evaluate which might be considered disastrous against a defined set of criteria and then record the associated set of contributing factors that led to the blunder. This would enable us to consider how blunders differ from the wider population of policies that are less (visibly) problematic. Such an approach would be highly resource intensive, however, since policy disasters make up just a small proportion of the overall population, and because collecting comparable data for all policies would be impractical. In order to properly understand the causal factors leading to each disaster one would need to conduct some sort of process-tracing on each case. It would also require us to compare disasters within populations of policies across multiple systems in order to determine whether constitutional design features are a source of blundering. Unlike comparative studies of outcome factors, such as cost overruns (e.g. Flyvbjerg et al. 2002), we cannot simply use indicators to determine the underlying cause of policy failure. It quickly becomes apparent, then, that a classic comparative study is out of the question. What are the alternative options? The problems of selecting on dependent variable are well-established (Geddes 1990; King et al. 1994), as this gives rise to no-variance in the outcome variable. For our purposes, however, examining a sample of prominent policy disasters from across a range of contexts provides a test of the necessary and sufficient conditions (Dion 1998) for the occurrence of policy disasters. In other words, selecting a set of known blunders as our case enables us to determine whether there is an observable pattern of contributing factors that reveal similarities or differences across a range of contexts and underlying conditions. Our cases are, therefore, drawn from a number of policy domains and countries. First, while there is no single agreed definition of blunders, disasters, fiascos or catastrophes, what is shared across most studies is the sense that these are large-scale, foreseeable errors/mistakes of either omission or unintended consequences (see Hall 1980; Gray 1996; King and Crewe 2013). We therefore identified cases which might be categorised as having been ‘disastrous’ in terms of cost and time in the following sorts of public policies or projects: (i) public buildings and stadiums, (ii) transport infrastructure, (iii) IT projects, (iv) benefits/tax systems and financial regulation, and (v) aerospace/defence projects. Admittedly, none of the decision-making in these cases is likely to have directly caused death or disease. These areas are also somewhat different to other examples of ‘failure’, such as ‘poor’ regulation, military adventures, or misdirected, complex welfare programmes. Additionally, the severity of blundering varies considerably across the cases, which range from extreme outliers in terms of completion and cost overruns to modest failures of management or decision-making that incurred smaller costs but rather more in the way of political embarrassment. Our cases are accordingly given a score between 0.0 and 1.0 for the degree of blundering that was involved. This is not purely a measure of losses since it is impossible to calculate a universal unit of policy disaster (i.e. is a 50% cost overrun better or worse than a one year delay in project completion?). Rather it indicates the overall severity of the failings of planners and managers (such as the extent to which cost inflation was foreseeable or the frequency with which warnings were missed). Our cases were drawn from six countries: Britain, France, the US, Germany, Canada, Australia; with some of the policies falling under the jurisdiction of national government and others being controlled by subnational, regional or local government. Through this selection of countries, we are able to observe the factors associated with policy disasters across federalpresidential and semi-presidential systems, and in unitaryand federal-parliamentary systems. This approach allows us to test whether the presence or absence of particular conditions (i.e. instruments, system-features, hyper-excited politics or administrative capabilities) is common across all cases. This will further enable us to demonstrate the degree of contingency associated with government blundering if common patterns are not observed. It also will enable us to determine whether particularly severe policy blunders are associated with particular causes whereas more modest blunders are associated with a different set of conditions. Categorising Hyper-Excitement, Instruments, Constitutional Systems and Administrative Capacity Having identified a set of cases we classify each of the conditions associated with the occurrence of policy disaster. Was hyper-excited politics involved? What was the instrument chosen to deliver the policy? What features of the constitutional system exacerbated or restricted the blunder? Where the capabilities of bureaucrats at fault? This requires us simply to determine whether or not a particular factor is present in the case of a given policy disaster. For each case, then, we complete the relevant information for each of the quadrants in Table 2 (see below). Having addressed this dichotomous question, we then assign a weight to the variable (taking a value between 0 and 1.0) which provides a subjective evaluation of how significant the factor was in leading to the blunder. This draws upon Ragin’s (2000) method of qualitative comparative analysis (QCA). On the basis of inspection of contemporaneous accounts and subsequent inquiries, this serves to indicate how much each factor contributed to the policy disaster. As we will show, not only does this enable systematic comparison across cases it also enables the identification of multiple permutations of combinations of covariates that are associated with a particular output value. Table 2: Examples of Four Types of Government Blunder No Resource Issues High Resource Issues Blunder due to particular choices Hyper-Excited Politics: Short-term political response to a tabloid newspaper campaigns Political backing for a project designed to win over voters Ideological preference for a programme of privatisation leads to disregard of evidence and/or opposition Project seen as a legacy for the politician making the decision to adopt or abandon it Instrument Choice: Use of a private finance initiative imposes a long-term financial burden on government Public/private partnership creates moral hazard where government bears the financial risk Reliance on private financing for a high risk venture with uncertain revenues Blunder due to underlying structures/systemic features Constitutional System: Policy/project not subject to sufficient checks and balances Fast-tracking of legislation possible due to the political system Complex political distribution of responsibility between national and local government creates ambiguity in implementation and blame avoidance behaviour Administrative Capacity: Bureaucrats inattentive to the development of problems with a project. Lack of capacity for subjecting initial budget forecasts to scrutiny. Government reliant on private firms and industry sectors for delivering/administering policy (e.g. security for major events) QUALITATIVE COMPARATIVE ANALYSIS OF TWENTY-THREE POLICY