شناسایی و کنترل خطر :مسئله عدم قطعیت در طرح امور مالی شخصی در خدمات بهداشت ملی بریتانیا

Risk estimation and uncertainty recognition are key elements in decision-making. This paper analyses decision-making processes, highlighting the way quantitative risk estimation is given a privileged position in decision processes thus ‘silencing’ qualitative uncertainties. This we argue is because of the dominance of ‘accounting logic’ in decision processes. The paper explores these relationships through analysis of the decision-making processes in the Private Finance Initiative (PFI) in the UK's National Health Service (NHS). To undertake a PFI option, in health or in other areas, involves deciding whether to contract with a private sector operator to provide a range of services (including often an infrastructure element) in exchange for long-term regular payments. Four general propositions are developed and amplified through analysis of the PFI decision-making processes. First that PFI, like other forms of decision-making, is dominated by quantitative risk estimation. Thus PFI provides a unique and significant example of these processes with its emphasis on transferred risks to the private sector. This quantitative dominance, we argue, is because of the pervading power of ‘accounting logic’ and this is illustrated through our research in 17 PFI health projects. Second, that the result of this dominance is that more qualitative concerns, such as uncertainty recognition, are downplayed even though they may be recognised at the pre-decision stage, very largely because at that point there are maximum anxieties surrounding the process of deciding. The 17 PFI cases provide a number of key examples of this thinking through their introduction of what are referred to as ‘shared risks’, which we see as qualitative uncertainties. Third, that at the post-decision project evaluation stage only some of the quantification concerns in the pre-decision stage are given visibility whilst all uncertainties are ignored. Again we link this analytically to the dominance of accounting logic. Eight of the 17 PFI cases are analysed and provide a powerful illustration of this general characteristic. Their sophisticated measurement of facility management issues ignores all ‘shared risks’ and even some of the quantitative transferred risks. Finally, the paper argues that unless there can be a fundamental reshaping of the mind set and/or the removal of the dominance of accounting logic there will be little hope of allowing equal emphasis of risks and uncertainties in decision-making to occur. In the PFI case there is evidence of a small but growing concern for recognising the importance of uncertainty and non-quantitative analysis. The outworking of this and its combination with quantitative analysis has the potential to provide important leads as to how to bring these important considerations into decision-making more generally. The paper is equivocal as to whether this development will occur very largely because of the resistance of accounting logic to change.

Decision-making often involves facing an unknown future. Decision makers often prefer to find ways to analyse and rationalise the future implications of pursuing a particular action before deciding to act. This process is easily achieved in a certain world. However as this does not exist we must make decisions in an environment that is fraught with uncertainty and so we have no certainty about outcomes and their implications. In order to mitigate this, consistent attempts have been made to develop sophisticated models to reduce or capture and express these uncertainties in more concrete ways. A key assumption of this paper is that this process has, in turn, been captured by what has been referred to as ‘accounting logic’ (Broadbent, 2002) or ‘accountingization’ (Power and Laughlin, 1992). This assumes that it is possible to find ways to capture everything of importance through measurement technologies. In turn, this leads to a reduction of the importance of everything that cannot be subject to these technologies. In other words a key role for accounting logic in decision-making is to shape key uncertainties into definable and measurable risks through an analysis of the past as a basis for projecting the future (McGoun, 1995). This analysis, based on accounting logic, is what Burchell et al. (1980) calls an ‘ammunition machine’ for decision-making. It provides a level of security and confidence in the actions taken since ‘accountingization’ gives a particular and seemingly more certain depiction of the outcomes and implications of any decision. Accounting logic is particularly influential in decision-making processes even though it is not always practiced by qualified accountants.1 Accounting logic may, however, be misleading as it is informed by an implicit claim that it is possible to define measurable probabilities for a future state. A particular example of this is the increasing use of techniques of risk assessment which belittles or ‘silences’ (cf. Buhr, 2001 and Young, 2003) those uncertainties that cannot be interpreted in this way. This paper analyses this complex dynamic teasing out the dangers of accounting logic in these processes. The role of accounting logic is explored in and through the context of the Private Finance Initiative (PFI) in the UK with a particular emphasis on PFI in the National Health Service (NHS). PFI involves the private sector in contractual relationships requiring the supply of public services over periods of up to 25 years and sometimes longer. This approach is justified by the claim that PFI is a ‘less risky’, long-term alternative than public sector procurement. Accounting logic informs the techniques that are key in the definition and assessment of these risks. The focus of this paper, therefore, is on how accounting logic can shape, enable and constrain the nature of decision criteria. The particular emphasis of the paper is on risk. The extent to which risk is correctly defined and the way other uncertainties have been ignored in PFI decision-making has been addressed by Froud (2003). In this paper we seek to elaborate and develop some aspects of Froud's thesis both generally as well as specifically by explicating the crucial role of accounting logic in these processes. The wider intention is to draw from this analysis a more general critique of the role of accounting logic in decision-making. To do this requires further consideration on how to define risks and uncertainties. As Froud (2003) indicates there is little consensus on what scholars mean by these terms, although there seems to be a general recognition that the two are different. Froud (2003, p. 569) draws from Humpty Dumpty (‘when I use a word … it means just what I chose it to mean—neither more nor less’) and Keynes to highlight the problems of defining risk and also to make a clear distinction between risk and uncertainty. As Froud indicates, the distinction is centrally concerned with ‘calculable probabilities’, a view reinforced by McGoun's (1995) historical analysis of risk measurement. Where there is no possibility of placing a numerical probability on something occurring or not, the unclear future state is referred to as an ‘uncertainty’ rather than a ‘risk’. Risk, therefore, involves the possibility of placing some ‘calculable probability’ on a future event occurring. Our argument is that this calculation is informed by accounting logic. Other elements are uncertainties and are downgraded in importance in the context of the technologies applied because they cannot be measured. Risk and uncertainty are inherent problems in all decision-making but are not always recognised or considered in decision processes. Froud's (2003) study highlights three different recognition models or ‘concepts’, as she calls them. The first, the ‘technicist’, is where risks are recognised and attempts are made to ‘manage’ them through estimation, whilst uncertainty is ignored. The second, the ‘postmodern’, refers to the ‘risk society’ drawing from Giddens (1998) and Beck (1992). This argues that our increasing and complex decision processes remain inadequate to predict or control risks and, more importantly, that risks/uncertainties are generated by the decisions we take (what Beck refers to as ‘system risk’). Third, the ‘radical’ model is one, which encourages the recognition and active engagement with uncertainties in decision-making not just risks. Our view, like Froud's, is that the ‘technicist model’ dominates PFI decision-making whereas the ‘radical’ alternative should be predominant. Our thesis in this paper is that the reason for this ‘technicist’ domination is because of the significance of accounting logic, which will prevent a move to the preferred ‘radical’ model. Froud's (2003) analysis of risk and uncertainty draws a number of key conclusions about PFI. Froud (2003, p. 568) argues that there is a tendency in current PFI decision processes to conflate risks and uncertainties often leading to the non-recognition of the latter. This leads her to the view that PFI is ‘… inappropriate for the provision of public services over a medium term horizon’. Froud's (2003, p. 582) argument is that ‘… the PFI contract reduces the ability of the public sector to deal with uncertainty, by locking the state into contracts typically over 20 or 30 years (or longer—our addition) and reducing the flexibility to respond to a dynamic environment’. Froud argues that in Beck's (1992) terms there is a considerable ‘system risk’ in pursuing public sector service developments through PFI since it generates a (presumably unintended) ‘system risk’ for the State. This results from the reduced possibility to adapt to future societal changes due to ongoing commitments to honour these long-term contracts. Similar conclusions are arrived at by Lonsdale (2005). He challenges PFI on the basis that the decision to pursue PFI fails to recognise the ‘system risk’ that the public sector is ‘locked in’ with the private sector partner, which restricts the public sector in many different ways. Lonsdale's focus is on risk transfer, which he sees as the foundational reason for justifying PFI. With the public sector ‘locked in’ to a contract tied often to ‘asset specific’ buildings, his argument is that there is little scope for overall risk transfer to occur in the final analysis. This paper develops an alternative critique of PFI decision-making which will feed into the longer term evaluation of PFI which both Froud and Lonsdale point us towards. Our view is that whilst Froud's and Lonsdale's conclusions are logical possibilities, this does not necessarily mean that either the Government is restricted in the way Froud suggests or that risk transfer cannot occur because of ‘lock in’ based on the Lonsdale's view. These potentialities will only become apparent or otherwise empirically over the duration of the contract. Given that most contracts have been running for only a few years, a more limited set of evaluatory conclusions are all that can be drawn at this point in time. In our view what is needed if we are to drive changes in PFI is a more detailed empirical analysis of what we know to date. This paper develops this analysis and explores how PFI decisions are taken, what uncertainties are recognised and converted into measurable risks, what uncertainties are ignored and the role of accounting logic in these processes. It has a particular critical accounting focus, which will provide one part of a more long-term post-decision project evaluation.2 In order to develop this argument, the paper will focus upon the extent to which risk and uncertainty is reflected in both the guidance given to NHS Trusts undertaking PFI as well as in the pre- and post-decision processes. We will look at the pre-decision processes of PFI and draw from the Full Business Cases of 17 NHS ‘first wave’ PFI decisions. We will also look at the post-decision project evaluation systems of eight of these PFI projects. Whilst the focus of the paper is to provide empirical detail about the workings of PFI, our intention is not only to understand this specific area but to demonstrate how accounting logic is deeply embedded in these risk and uncertainty recognition and measurement problems. We also wish to explore what this says about the role of ‘accountingization’ more generally. The paper has four key propositions that inform both the description and analysis. First, that there is an overwhelming tendency to what Froud refers to as ‘technicist model’ in PFI decision-making that relies on only variables that can be expressed through ‘accountingized’ measures. Secondly, that this leads to a reduced consideration of the other important uncertainties, which cannot be handled by measurement technologies, in the pre-decision processes. These might be recognised peripherally due to the ‘what if’ anxieties that decision-making engenders but they are not dealt with systematically. Third, at the post-decision project evaluation (PPE) stage there will be even greater concentration on the measurable elements identified at the pre-decision stage leading to yet more distancing of uncertainties. Finally, unless we can reshape accounting logic to break out of its measurement bias, there will be little hope of what Froud calls a ‘radical model’ of decision-making ever becoming reality. These propositions, drawn from the discussion above and from wider consideration of accounting logic and ‘accountingization’, are not intended to be tightly defined scientific hypotheses. Rather they provide a theoretical language to frame the empirical example of PFI that follows in a way that allows empirical ‘surprises’ as befits the ‘middle range thinking’ (cf. Broadbent and Laughlin, 1997, Laughlin, 1995 and Laughlin, 2004) that informs this paper. The paper is divided into four further substantive sections. The next (second) section explores some of the key Government and National Audit Office (NAO) Guidance that has been issued to provide direction for how PFI decision-making should be undertaken with a concentration on risk recognition and its importance. The third section looks at the way PFI decisions have been taken in the NHS, drawing from the Full Business Cases (FBCs) of the 17 NHS Trusts. The fourth section explores some of the PPE systems that are in place in eight NHS Trusts that are fully operational. Exploring PPE systems provides pointers to the way these are undertaken in practice and more specifically what risks and uncertainties are deemed to be important to control once the PFI project is operational. The fifth and final section reflects on some of the more general lessons on understanding risk, uncertainty and accounting logic which flow from our analysis of PFI.

Risk, uncertainty and accounting and PFI: an overview This paper has attempted to demonstrate the distinctions and inter-relationships between risk, uncertainty and accounting in PFI processes. Central to PFI is an understanding of risk and uncertainty. Whether PFI is viewed as a macro-fiscal arrangement or a micro-procurement process, issues related to risks and uncertainties are always dominant. Our focus has been primarily at the micro-organisational level, even though there are connections to the macro-fiscal concerns given that the risk assessment tends to dominate the decision to proceed with PFI. The issue of risk and uncertainty is also important in the context of the link to the PPE that must be undertaken once the decision to proceed with PFI has been made. Throughout the paper we have concentrated on three key aspects of risk and uncertainty at two different levels. The three key aspects of risk are: first, the important and close linkage between management systems and risk; second, the key distinction between risk and uncertainties; and third, the concept of ‘system risk’ (Beck, 1992). We have explored these at the macro-conceptual level in terms of Government Guidance that has been issued in relation to PFI pre- and post-decision processes and in the micro-empirical outworking in the actual PFI decisions undertaken in NHS Trusts. The different characteristics and levels of analysis described above meld together to provide three key observations/conclusions. First, that risks and uncertainties (although not clearly distinguished) are recognised and considered at the pre-decision stage in PFI, both in the Guidance and actual practice. Some systems risks are also considered at the pre-decision stage in all these contexts. Only some of these elements are quantified,16 and it is these that play a dominant role in deciding whether to proceed with this procurement route. Second, the evidence of the eight PFI projects in the NHS that we have analysed suggests that the management of a limited number of the risks (but not the uncertainties) is the central characteristic of all PPEs. Uncertainties are largely ignored. Third that accounting technologies and accounting logic play a pivotal role in defining what is important in guiding the pre- and post-decision criteria and concerns. (ii) PFI pre-decision processes: some conclusions Two key conclusions can be drawn in relation to pre-decision processes. First, that it is invariably a quantitative analysis, involving costed transferred risks, that has dominated the decision-making processes as to whether PFI is demonstrating value for money (v-f-m). Nevertheless, some changes that will involve more consideration of qualitative issues are currently being actively developed, particularly in the context of the national Guidance. The risks that are deemed to be transferred to the private sector in the comparison of the PFI to a PSC have been the key variable in decision-making. It is these risks that have allowed the PFI to be viewed as cheaper than the PSC and that are used as the justification to proceed with the PFI procurement. Despite the extensive criticisms levelled against the calculation and content of the PSC, this form of justification has not been abandoned but rather has been developed in the new ‘Green Book’ (HM Treasury, 2002). Here the change in the discount rate, reducing it from 6% to 3.5% has been introduced. At the same time, there is more emphasis on transferred construction risk taking into account what is called ‘optimism bias’ that has been seen to have lead to a conservatism in estimating design and building costs. The emphasis on the importance of a full consideration of construction costs is reinforced even further with the recently published NAO Report on construction performance (NAO, 2003). This indicates how the vast majority of PFI projects are all delivered on time and to cost relative to those under public sector procurement. It makes a compelling case for construction risk transfer being valued even higher than at present. Together, these changes are likely to make the case that the PFI alternative is more cost effective. The approach still retains the ‘technicist’ comparison with the PSC. But the changes adopted also seek to meet the criticisms levelled at this decision. Whether these changes are sufficient remains a moot point. Andrew Wynne of the ACCA, expressed the matter in the following critically direct manner: ‘The proposed revisions to the Green Book will make the basis of the decisions to adopt a PFI project even more complex, more subjective and less information will be available to substantiate the decision. In practice it will mean that any PFI project can be accepted and shown to provide value for money.’ (Unpublished ACCA Commentary by A. Wynne, 2002) Whilst these technical changes are occurring, the NAO's recent Report on the PFI contract of West Middlesex University Hospital (NAO, 2002) seems to be leading the way on a wider approach to the value for money (v-f-m) question. The Executive Summary of this Report provides a flavour of a shifting emphasis away from using only the PSC comparison in judging v-f-m. So for instance the Executive Summary states: ‘In this project the financial comparison was not clear-cut. The attention given by the Trust to the figures shown by the financial comparison may have masked evidence of important wider benefits that the PFI approach was expected to secure.’ (Paragraph 13) ‘Both the Trust and its advisers KPMG considered the PFI option would deliver value for money taking all benefits and disbenefits into account. But they had concerns about the accuracy of the initial financial comparison and whether its results might prevent the project being approved by the Department [of Health].’ (Paragraph 11) ‘The Department [of Health] told us that it would not necessarily withhold approval for a PFI project that appeared slightly more expensive than conventional procurement if there were convincing value for money reasons for proceeding with the deal.’ (Paragraph 11) These softening of views and emphasis on the ‘technicist’ criteria that the NAO have seemingly adopted previously17 is now being mirrored in recent Government proposals (HM Treasury, 2003) where, as they indicate: ‘The Government believes that a rigorous economic assessment is important to ensure that the right procurement option is chosen on the basis of value for money. The Government agrees with the NAO, however, about the dangers of putting disproportionate emphasis on a single figure comparison. It therefore believes that the PSC continues to have an important role but as the second stage in a three stage process, and needs some reform itself.’ (HM Treasury, 2003, Paragraph 7.10, p. 80) The first stage is similar to current Guidance related to option appraisal in the OBC but stresses the need for ‘… a comprehensive appraisal of both the financial and non-financial factors of each procurement option’ (HM Treasury, 2003, Paragraph 7.15, p. 81). However, it is the formal introduction of a third stage, following the comparison with the PSC where the non-quantitative concerns are starting to become apparent: ‘The final stage focuses, prior to going out to tender, on the potential market capacity to deliver the project, and then on the quality of the competitive interest in it’ (HM Treasury, 2003, Paragraph 7.13, p. 81). The importance of ‘delivery’ to time and to budget is now clear but so is a return to a belief in market competition.18 These proposals are clear: the PSC will remain of central importance but will be accompanied by other (more qualitative) criteria in the final judgement of value for money. Whether this rhetoric becomes reality depends on many factors, not least, we would argue, on whether there are adequate accounting technologies to realise this transition (see below for more on this). Our second conclusion in relation to pre-decision processes concerns actual practices. We note that, whilst the PSC comparison is dominant, many other risks and uncertainties are considered at the contract formation stage, but none of these are costed or determine decisions. Whilst the macro-Guidance seems to be changing, the 17 cases we investigated were finalised long before these developments and their procurement relied on previous advice related to the importance of the comparison with the PSC. Even so qualitative concerns related to certain uncertainties were recognised but were more apparent in contract negotiations rather than in the processes of decision-making. As we indicated when examining Dartford and Gravesham, the most significant of these uncertainties – (which continue to be called risks) – are those which are ‘shared’. Many of these are systemic risks that inherently arise from undertaking a PFI option. Notable amongst these are ones that arise if medical practice should change and the hospital becomes obsolete. For Froud (2003) and Lonsdale (2005) arguing at the macro-economic level, this danger and the inherent inflexibilities that PFI procurement produces is the reason why the PFI option should not be pursued. As we saw in the case of Dartford and Gravesham at the micro-organisational Trust level, this danger was recognised and, in effect, accepted on the grounds that they required delivery of new facilities now rather than risk waiting for a public procurement they judged unlikely to succeed. In the case of West Middlesex, this systems risk was explicitly considered and led to an attempt to overcome any potential problems. According to the NAO, West Middlesex addressed this as follows: ‘Long term planning is difficult in the health service because healthcare is changing over time and the local demography may also change. This may affect the optimum type and location of facilities that are required. This issue is not limited to PFI hospitals, but the long-term service contract of a PFI deal makes termination likely to be more expensive. In the West Middlesex deal there is some flexibility to accommodate these uncertainties. Up to six additional wards can be provided or alternatively bed numbers could be decreased. The Trust believes the contract provides sufficient flexibility to address future uncertainties in long-term healthcare.’ (Paragraph 5 of the Executive Summary of NAO, 2002) Whether this flexibility in design is sufficient to address the uncertainties will only become apparent in the future. (iii) PFI post-decision project evaluation: some conclusions Two observations can be made about the PPE system by way of introduction: First, our study of the PPEs in operation in 8 Trusts (selected from the overall set of the 17 cases we considered) are all strongly managerial, regulatory, and are ongoing. The mode of PPE that is in operation in all Trusts deals with the ongoing monitoring of the set of risks associated with FM. These ‘proactive’ PPE systems act to monitor FM systems to ensure payments that are made for availability and maintenance of the building and for the provision of other ‘softer’ services (e.g. catering, laundry, portering, etc.) are only made when the provider delivers to specified standards. They also provide rational decision tools to allow suitable deductions to be made if the availability and service standards are not achieved. Second, all the FM (PPE) systems that are in place deal only with transferred risks. However they do not address all these risks. The FM (PPE) systems are concerned only with service and maintenance risks that are costed and transferred to the private sector. Based on our analysis this accounts for approximately 43% of the total transferred (costed) risks. FM systems do not deal with construction risks or ‘other’, often financial and legal risks. None of the other (shared) uncertainties that are considered as carefully at the pre-decision contract stage are given any attention. This partial consideration of risks that seemingly ignore the (shared) uncertainties is a key matter that needs to be addressed in developing a more robust design for PPEs.19 The view implicitly adopted is that the transferred risk that relates to construction risk (50% of the transferred risks based on the Dartford and Gravesham case) is taken into account through ensuring completion to time and cost. However this is too simplistic. There remains considerable potential for problems to arise. For example, undertaking variations at the building stage may be cheaper than adopting them later. There may be disputes about where liability lies for these ‘snags’ identified on handover of the building, both between the private and public sector partners and between the construction companies and the service companies with which they are associated.20 Whilst building to cost and time is important and avoids the embarrassment of schemes running into extensive overspends and delays, the fact that a building was handed over on time and to cost may not demonstrate that construction risk was avoided. Of even more significance is the omission of any consideration of the handling of the shared uncertainties. No evidence has been found that extensive thought has been given to these and how they will actually be handled. Notable amongst these is the handling of the periodic (usually 5 years) contract renegotiation for FM provision and the effect of evolving medical care on the effectiveness of this very different form of contractual relationship. The FM contract renegotiation will be complex, possibly difficult and very expensive for the public sector. Of even more significance is the nature of the negotiations that will need to be undertaken to accommodate the effect of changing approaches to medical care where this has ramifications for the design of the building of PFI hospitals. There is a need to ensure that the social needs identified in the outline business case, such as public accessibility, are achieved. This suggests that at both the pre- and post-decision stages a qualitative analysis of non-financial benefits and ‘costs’ must be undertaken. The lack of attention to qualitative elements in PPEs mirrors the emphasis on the quantitative analysis at the pre-decision stage. The NAO, as demonstrated by their West Middlesex Report, are overtly changing their views as to what constitutes a ‘good deal’ (NAO, 1999 and NAO, 2002) at the pre-decision stage. Equally, the Government appears to be following this lead (HM Treasury, 2003). This will hopefully have an effect on both the pre-decision processes as well as the design of PPEs to ensure they reflect this developing emphasis. Then we could perhaps move to a more discursive stakeholder based approach to PPEs, as Broadbent et al. (2003) suggest, which can supplement the FM (PPE) systems that are now becoming well established in the PFI projects that are underway. This positive outcome is again dependent upon accounting logic breaking out of its quantitative emphasis to encompass this qualitative concern. (iv) Risk, uncertainty and accounting logic: some reflective thoughts Perhaps we could end with some reflections on what this case adds to established understanding of decision processes with a particular emphasis on risk, uncertainty and accounting logic. First, as is generally the case in decision processes, quantitative risk analysis is a dominant form of justification in the decision-making process of PFI. Many approaches to decision-making seek to develop accurate costs for potential projects and this involves attempts to bring some order to an uncertain future. Quantitative risk analysis with various probabilities attached to possible outcomes is advocated as a key aspect of many decision processes. PFI is no exception to this but the processes attempt to add a concern with uncertainties, as we will argue later. However, at this juncture we wish to stress that PFI decision-making is similar to other forms of decision-making and relies on established risk estimation technologies to assist its decision process. This bears out the first proposition highlighted in Section 1 to this paper, that there is an overwhelming tendency towards what Froud (2003) refers to as a ‘technicist’ model in decision-making that relies on only variables that can be expressed through ‘accountingized’ measures. This leads to our second observation. Second, risk estimation technologies for decision-making, including those for PFI, are driven by accounting logic. In this context accounting is particularly being used in its role as an ‘ammunition machine’ (Burchell et al., 1980). Moreover, and of key importance, is the point that the use of accounting technologies has a constraining influence. In every case that we investigated, major professional firms of accountants or their associated consulting companies were hired to provide assistance with the estimation of the costs and risks for the PSC—particularly the estimation of transferred risks.21 They use models that are adaptations of other more general forms of cost and risk estimation they have developed for use in different decision situations. Put simply the nature of these estimation processes is accounting driven and accounting determined. The outcome of the processes is limited by what constitutes and is acceptable as knowledge as defined by accounting logic. The measures that accounting logic provides, therefore, can be seen as a constraint to moving beyond these quantitative ‘safe harbours’ related to estimation processes. This, of course, depends on the adaptability of the accounting profession (as the originators of accounting logic) as well as the flexibility of those inheriting its thinking to move beyond the boundaries of the seeming certainty quantitative approaches supplies. The dominance of the accounting-led approaches leads to a reduction in the importance of taking into account other uncertainties, which cannot be made subject to measurement technologies in pre-decision processes. It also links directly to and reinforces the first part of the second proposition presented in Section 1. Third, there are indications from the PFI case that uncertainty recognition and non-quantitative analysis are recognised at the pre-decision stage but not at the post-decision project evaluation stage. This development, if fully integrated into the decision processes, could lead to moves away from the highly ‘technicist’ emphasis which still dominates decision-making more generally. As indicated above, shared risks – or, as we have described them in this paper, uncertainties – are very important in PFI. However, they are more important in contract negotiation than in decision-making. The changing nature of the national Guidance may lead to increasing concentration on non-quantitative criteria as well as uncertainty recognition in decision-making and contract formulation. The PFI case is at an early stage of this development. If the current trajectory is maintained and a clear combination of quantitative and qualitative aspects to guide decision-making is adopted and accepted for PFI this could, in turn, provide a rich case for developing other forms of decision-making. This remains an empirical question but the indications are not good with Treasury officials talking about ‘quantifying the qualitative’ and continuing their reliance on the PFI/Public Sector Comparator analysis. It still does not address the problem of the invisibility of qualitative aspects in PPE of PFI – an aspect that links directly into the second part of the second proposition we raised in Section 1 as well the third proposition. These propositions raised the possibility that at the pre-decision stage, due to the unease about making decisions in an uncertain environment, qualitative ‘what if’ concerns will be considered but these will be ‘silenced’ (cf. Buhr, 2001 and Young, 2003) by the certainty the quantitative measures provide when final decisions are made. Further, at the PPE evaluation stage where ‘what if’ concerns are less prevalent, all qualitative concerns disappear. Fourth and finally, breaking out of the hold that ‘accountingization’ has over decision-making requires accounting logic changing from its fundamental emphasis and belief in the power and importance of measurement. Here we are far from optimistic. This links directly into the fourth proposition in Section 1. Whether accounting logic's success in the reliance given to quantitative thinking is due to society's fixation with numbers (cf. Boyle, 2001 and Power, 2004) or is actually the generator of this mystique is debatable. The more important question is whether or not the accounting profession (as the orginators of accounting logic) or those other non-accountants (who are custodian of this logic) are willing to make the step change that is required to address what is an important challenge for decision-making. Here we come full circle back to PFI decision-making where at least there are clear moves to involve both quantitative as well as qualitative analysis in the pre-decision processes. The outcome of this promise remains to be seen, but the indications are not very encouraging. If the accounting profession does not take the lead needed, it is questionable whether the National Audit Office can, through their audit and value for money function, produce the change required in accounting logic. Until their PFI Report on the 18 PFI hospitals finally appears this remains an open question. If the uncertainties and system risks remain largely ignored because of accounting logic's inflexibility then this does indeed raise long-term evaluatory concerns and makes Froud's (2003) and Lonsdale's (2005) predicted concerns a real possibility of becoming reality. However, whether their macro-projections occur is an empirical question. This again reinforces the need for a careful monitoring of all aspects (risks, uncertainties and system risks) of PFI projects over the duration of the contracts to be in a real position to judge merit or worth of this form of procurement. Similar points apply to the evaluation of all decisions. Ideally this should be complemented with ensuring that risks, uncertainties and system risk play a major and equal part in both pre- and post-decision processes but this is dependent on accounting logic evolving. Our analysis provides far from encouraging pointers to this possibility.