Hazardous Waste Policy Management - Institutional Dimensions. Chapter 2: Risk Assessment of Technological Systems - Dimensions of Uncertainty

NOT FOR QUOTATION WITHOUT PERMISSION OF THE AUTHOR

HAZARDOUS WASI'E POLICY MANAGEMENT

-

INSI'ITUTIONAL DIMENSIONS

CHAPr]ER 2:

RISK

ASSESSMENT

OF TECHNOLOGICAL SYSI'EMS

-

Dimensions of Uncertainty

B.

Wynne

May 1984

WP-84-42

Wo'rking Papers a r e interim reports on work of the International i n s t i t u t e for Applied Syslems Analysis and have received only limited review. Views or opinions expressed h e r e i n do not necessarily r e p r e s e n t those of t h e Institute or of i t s National Member Organizations.

INTERNATIONAL iNST1TUTE FOR APPLIED SYSTEMS ANALYSlS 2361 Laxenburg, Austria

PREFACE

This paper has been produced as part of IlASA's hazardous waste manage- m e n t work, which is t h e main component of the Institutional Settings and Environmental Policies project. The overall aim of this work, reflected in this paper, is t o systematize our understanding of interactions between institu- tional and technical factors in policy making and implementation. The influence of institutional processes upon technical knowledge built into policy has been increasingly recognized. However, it has yet t o be adequately clarified in comparative r e s e a r c h on different regulatory systems. Institutional struc- t u r e s canot be easily transplanted from one culture to another. Nevertheless, through t h e normal flux of policy, institutional development slowly occurs any- way, in more or less ad hoc fashion. Comparative insight may help to direct reflection and adaptation in more deliberate and constructive ways.

This paper forms one draft c h a p t e r of a n intended book on hazardous waste management. The r e a d e r will therefore notice references t o other draft chapters in this study which a r e also being circulated separately, and which a r e available from IIASA. A full list is given overleaf. At this stage t h e papers are drafts, and are not intended for publication in present form. They a r e being circulated for review a n d revision.

I would like t o thank those policy makers and others who have exchanged papers and information with us, and those who generously gave of their time and experience in t h e many interl:iews -.vhleh form a substential input to this work. A full list of acknowledgements will eventually be published.

Brian Wynne Research Leader

Institutional Settings and Environmental Policies

HAZARDOUS WASlX

POLICY

WWAGEMENT

-

INSIlTUTlONAL DIMENSIONS

INTRODUCTION CHAPTER ONE CHAPTER TWO

CHAPTER THREE

CHAPTER FOUR CHAPTER FIVE

CHAPTER SIX CHAPTER SEVEN

B.

Wynne

Hazardous Waste

-

What Kind of Problem?

B. Wynne

Risk Assessment of Technological Systems

-

dimensions of uncertainty

B. Wynne

Risk Assessment and Regulation for Hazardous Wastes B. Wynne

The

Listing

and Classifying of Hazardous Wastes M. Dowling and J. Linnerooth

Government Responsibility for Risk: The

Bavarian

and Hes- sian Waste Disposal Systems

J. Linnerooth and G. Davis

Enforcement of Hazardous Waste Legislation in the

UK E.

Ley and

B.

Wynn e

Summary. Implications. and Further Problems B. Wynne

Fbther

Cme

Studies

Hazardous Waste Management in Hungary

- E.

Kiss Hazardous Waste Management in t h e Netherlands

Central processes in policy and implementation

-

J. Dirven

Dutch policies from a local perspective

-

J. van ELndhoven, R. fiortensius, C.

Nauta, C. Worrel

CONTENTS

INTRODUCTION DEFINING RISK

DEFINING THE RISK GENERATING PROCESS

Dimensions of Uncertainty

-

Ignorance and Perception LEG Facility Risks

From Technical Imprecision t o Social Contradiction UNCERTAINTY BY STRATEGIC DESIGN

(a) Data Uncertainties

(b) Institutional Diffraction of Key Data and Key Terms CONCLUSIONS

REFERENCES

c-2:

RISK ASSEXiMCNT

OF TECHNOLOGICAL

SYSI'EMS

-

Dimensions of Uncertainty

Brian Wynne

INTRODUCTION

The main aim of this chapter is to identify t h e fundamentally different kinds of uncertainty in risk assessments, especially t h e difference between con- ventional technical uncertainty, and incompatible socially influenced definitions of the risk-generating system. This distinction is crucial, yet t h e second ldnd of uncertainty is often very subtle. Recognizing t h e extent of this second ldnd of uncertainty in the technical and institutional context of risk assessment, regulation a n d implementation has far-reaching implications.

The following chapter t h r e e examines the conflict between t h e need, on t h e one hand, for stacdardized technical formulae and methods in risk assessment a n d regulation, and t h e contradictory logic of tailoring risk assessment and regulatory controls to risks arising in real situations in all their diversity and instability. This conflict in technical frameworks straddles a deeper institu- tional conflict as t o where to allocate responsibility and power to i n t e r p r e t

regulatory aims. I t parallels t h e disparity which exists between central policy formulation of regulatory rules, standards, etc.. and t h e often very different informal realities of their diverse local implementation, when other influences and logics come into play.

This c h a p t e r develops these questions through several examples, and shows how they a r e connected, for example in the ways in which genuine situa- tional risk-variation overlaps with and often looks identical t o varying percep- tions of t h e s a m e risk-situation or process.

Perceptual differences are often treated a s r a t h e r exotic m a t t e r s of public

"irrationalities" only, having little t o do with technical realities and discrimina- tions. The present analysis concludes to t h e contrary, t h a t perceptual differences of what a technology is, what are its significant components and connections (in detail and in t h e large) influence experts and t h e i r rigorous technical risk assessments also. Yet this influence is usually unrecognized, and conflicting analyses attributed instead merely t o un-closed technical imprecisions and uncertainties.

These assumptions or perceptual commitments underlying technical analysis for regulation are p a r t of a tissue of informal judgments in science which ultimately cannot be justified by tight, unambiguous rules of inference, method or logic. This 'informalist' model is sometimes regarded as criticism of science: in fact i t is a tribute t o its flexibility and resilience. Yet t h e opposite model dominates public attitudes, and policy m a k n g institutions [I.]. As zmphasized in this Chapter, the pervasiveness of this intrinsic, informal dirnen- sion of science complicates t h e requirements of formal accountability and standardization for authoritative regulation. This i s especially t r u e where t h e real world c h a r a c t e r of t h e issue is so ill-defined and extremely variable, and where public skepticism and more elaborate justification - especially on siting

and transport -is increasingly being demanded of regulation [Z].

The impossibility of objective definition of risk problems and of assessment or regulation decision-rules is stressed in this Chapter. However, t h e aim is not to suggest t h a t formal risk assessment should not be pursued, but to lay bare the extreme fragility of t h e authority of such decision processes t o public s k e p ticism, if this begins t o a s s e r t itself. This inherent vulnerability is multiplied by the large unknowns and indeterminate, ill-defined n a t u r e of t h e policy field in the hazardous waste case, properties which undermine a t t e m p t s t o discrim- inate and even rank with scientific precision t h e risks associated with different regulatory options. In these circumstances, administrative cultures and insti- tutional arrangements which fragment the overall process of risk management and regulation a r e more likely t o find their policy implementation picked apart and undermined or paralyzed due to the interacting uncertainties, complexi- ties and conflicts involved, than systems which manage to coalesce and absorb the different phases into more unitary institutional forms.

Although t h e immediate problems facing policy makers have been about t h e establishment of an effective industrial t r e a t m e n t and disposal (T & D) infrastructure, this focus has been complicated by t h e increasing need

-

aris- ing out of growing public concern

-

to address the risk management issue more explicitly and systematically. Thus a circular obstacle has tended to con- found a t t e m p t s to develop the appropriate infrastructure.

Due to basic ignorance, t h e particular configuration of wastes and t h u s risks is badly defined, and cannot be better defined anti1 a better knowiedge of waste arisings, properties a n d specific environmental dispositions is gained. An industrial T & D infrastructure needs to be developed to control these waste arisings now, but may require adaptation a n d thus possibly costly abandonment or changes of large investments when risk-estimates a r e revised. However,

whatever regulations a r e established will directly affect public acceptability of those plants, and t h e size of their markets, t h u s their viability in two dimen- sions. Hence t h e r e is a very great reluctance on the part of private industry t o t a k e initiatives or be involved in t h e T & D field. Thus t h e interaction between

"industrial innovation" and Risk h s e s s m e n t

(RA)

definitions is strong, and not necessarily free of contradictions.

Risk assessment requires reliable estimates of t h e toxicity or hazard causd by exposure t o a given waste, and estimates of t h e chance of exposure. This is a combination of intrinsic material properties and situational variations

-

how it is packaged, mixed, treated, confined, etc. Unfortunately, physical. chemical and behavioural heterogeneity, and unpredictable behavioural freedom in t h e system, mean t h a t "downstream" unpredictables may swallow up putative risk- differentials.

Yet however underripe t h e field may be for it, a RA management framework s e e m s inevitable. It is therefore necessary to explore what t h e possibilities and implications a r e for using formal risk assessment approaches t o hazardous waste management. There is currently a lively debate amongst policy makers a s t o how elaborate RA c a n and should be for hazardous waste management.

Even in t h e United Kingdom, t h e traditional stronghold of non-quantified, informal methods of decision making on issues involving risks, a r e c e n t Royal Society Study Group and t h e Royal Commission on Environmental Pollution both expressed strong support for more quantification of'risk assessments [3].

Not only in t h e US therefore but also in Europe t h e r e is already, and will con- t i n u e t o be, growing pressure to adopt formal RA methods in hazardous waste management. This c h a p t e r will therefore review t h e possibilities and limita- tions of present risk assessment methods applied to hazardous waste manage- ment.

An initial question is whether formal RA should apply a t a central policy level where decisions may be more discrete 'events' (such as whether to develop local or regionalized facilities), or a t more routine but perhaps equally significant levels of regulatory implementation (such a s siting or licensing con- ditions of given facilities; design of particular processes, including containment devices; or trigger standards filtering hfferent materials into different levels of regulation and different levels of t r e a t m e n t and disposal). Once, one could say t h a t the former may have involved justification as well a s internal technical analysis, whilst the l a t t e r were purely technical, with no symbolic justificatory dimension. Nowadays, even these tend to require justification as external scru- tiny and scepticism advance. This changes t h e role of Rk

The US Office of Technology Assessment Report reflects a typical view, of

"technical optimism", t h a t formal hazardous waste RA should produce hazard classiflcations ranking degrees of hazard and indicating appropriate T & D routes [4]. Being based on t h e s a m e scientific knowledge, t h u s would presum- ably generate consistency amongst definitions a n d classifications. Yet a s chapters 3 and 4 show as well as this chapter. "scientific" definitions of hazard a r e ambiguous: they a r e not merely physically uncertain, but actively incor- porate different social assumptions reflecting different, and even incompatible administrative purposes commitments and needs in different systems. Chapter 4 describes in detail some of t h e origins of such different mixes of "science"

(including uncertainty) and other factors in different hazard classification schemes. As we shall see in this chcpter, risk assessments, even for relatively uniform, well defined technologies let alone hazardous chemical wastes, have suffered large intrinsic uncertainty and inconsistencies due t o implicit differences in the assumptions structuring technical analysis.

DEFINING

RISK

The conventional definition of risk is t h e product of t h e degree of h a r m a given e v e n t would c a u s e , a n d t h e probability of t h a t event's occurring,

This would express a risk as, s a y e s t i m a t e d n u m b e r of a t t r i b u t a b l e d e a t h s o r o t h e r damage p e r u n i t t i m e of operation of a given activity. But, say, a chemi- cal plant m i g h t accidentally e m i t l e t h a l clouds of toxic gases every y e a r in a r e m o t e region, a n d c a u s e z e r o h a r m . O r a given chemical waste m a y be e x t r e m e l y toxic, a n d t h u s in principle of high hazard, b u t environmentally highly immobile a n d r e m o t e , therefore of low risk. Hazard m a y t h e r e f o r e describe t h e i n t r i n s i c "worst-case" damage a process or m a t e r i a l could c a u s e whilst t h e above definition of r i s k incorporates variable situational qualifications which r e d u c e t h e probability of t h i s worst case damage [5].

In t h e c a s e of industrial plant, s o m e s u c h qualifications a r e t h a t : properly designed, c o n s t r u c t e d a n d o p e r a t e d e q u i p m e n t h a s a low c h a n c e of failure;

m a n y p a r t s of processes have reserve p a r t s i n c a s e of failure; have fail-safe o r r e d u n d a n c y built i n t o t h e system; a n d have monitoring s y s t e m s which automatically r e a c t t o early signals so a s t o prevent major failures. Other fac- t o r s affecting P m i g h t be t h a t operating staff a r e of g r e a t e r or l e s s e r profes- sional expertise, t h a t regulation a n d inspection is l a x or tight, t h a t t h e r e is g r e a t e r o r l e s s e r p r e s s u r e economically t o c u t c o r n e r s , t h a t t h e r e is m o r e o r l e s s design, c o n s t r u c t i o n a n d operating experience, r e m o t e siting, e t c . e t c .

In t h e case of hazardous chemicals, some equivalent qualifying f a c t o r s m i g h t be t h e physical form of a chemical, (e.g., if i t is a n inhalation danger, is i t i n Ane powder form?); c h e m i c a l s t a t e (e.g., i s i t in a soluble compound valency s t a t e ) a n d form of containment; volume; local disposition (is i t accessi- ble t o e n v i r o n m e n t a l pathways back t o h u m a n populations); s t a t e of mixing

with o t h e r materials; t h e kind of h u m a n handling it receives, e t c . , e t c . A typical s c h e m a t i c form of Risk function would be a s in figure 2.

" z e r o - i n f in i t y "

r i s k s

1 1

I

1

c2 Consequences, C (e.g

. ,

d e a t h s / t i m e l

FIGURE 1. A Typical risk-function (schematic).

The s a m e Risk, R on this formulation, can be given by different combina- tions of

P

a n d

C,

for example,

P I C l = PZCZ,

b u t t h e s e m a y r e p r e s e n t radically different .events and experiences. Thus a compelling criticism of t h e R

= PC

formulation h a s been t h a t t h e universal dimensions t h u s produced. t a k e n o a c c o u n t whatever of o t h e r , perhaps m a j o r differences in t h e kinds of d a m a g e u n d e r consideration - i t does not a t all compare like with like. "Risk" a s con- ventionally defined i s t h u s an artificially narrowed concept which m a y o r m a y n o t c a p t u r e t h e essential f e a t u r e s of a n issue o r decision problem which i t s different participants define.

A r e l a t e d difficulty of compound r i s k approaches is t h a t they m a y conceal value c o m m i t m e n t s in t h e i r definition. Thus risk expressed as p r o d u c t ,

&C.

may incorporate different kinds of h a r m

-

mortality, morbidity, o t h e r losses

-

without specifying t h e s e , a n d be m e a s u r e d implicitly against different yardst- icks. Thus a comparison of risks per u n i t t i m e m a y be very different from a comparison of t h e s a m e risks per unit of o u t p u t (or capital or labor input) if one process is m o r e productive t h a n a n o t h e r in t e r m s of t i m e , capital o r labor.

A work-force m a y wish to know risks per u n i t of work time; a m a n a g e r per capi- tal input or output; a n d a local resident, per u n i t of residence time. These o f t e n implicit yardsticks can suddenly c h a n g e t h e a p p a r e n t scale and impor- t a n c e of risks very considerably [ 6 ] .

Another problem with t h e conventional approach h a s been t h a t t h e proba- bility of a given h a r m actually occurring usually depends upon a compound of probabilities: (i) of a s e t of necessary, or facilitating sub-events occurring; (ii) in a complex plant o r with hazardous m a t e r i a l s t h e s a m e damage could be c r e a t e d by many different possible accident sequences. involving different chains of events in different components. Thus even i n t h e s a m e plant or sys- t e m , t h e s a m e "risk" may be posed by different sequences a n d combinations.

In o t h e r words, t h e u s e of t h e formula R

=

P x C m a y often conceal m o r e t h a n it illuminates, for one thing because i t m a y confuse different r o u t e s t o t h e s a m e risk end-point; a n d secondly because i t may conceal different assumed e n d points which should be carefully distinguished. Thus for example a highly hazardous facility m a y be recorded as a low risk facility because i t is remotely sited, when i t is t h e .remote siting which is low risk, n o t t h e facility.

In a n y reai anaiysis, t h e P of a given e n d poifit. C i s actually t h e integral of products of probabilities P, of e a c h s e t of sub-events which could e n d up in C.

If o n e includes e x t e r n a l doses a n d health damages a s e n d points ( r a t h e r than, say, releases from a plant) t h e network of e v e n t s and chains proliferates. Given t h a t risk analysis is supposed t o be a policy decision aid, t h e r e is a t r a d e off t o

be made between (i) decomposition of risk calculations, which can show sensi- tive points ( e i t h e r of ignorance o r failure probability) in t h e overall risk sys- t e m , but which m a y n o t give policy actors end points which a r e meaningful to their decision language; o r (ii) composite risk t e r m s which have the opposite pros and cons.

In t h e light of t h e s e problems, other experts therefore advocate t h a t t h e t e r m risk be used to define only the probability of o c c u r r e n c e of a s p e c i f i e d e n d - p o i n t or harmful e v e n t [7], so as t o clarify the distinctions between (possi-' bly multiple) i n t r i n s i c hazards of any material or activity, a n d qualifying fac- tors such as siting, c o n t a i n m e n t , operating rules, t r e a t m e n t , etc., which reduce the probability of a given hazard's being realized in practice.

We can therefore distinguish between what might be called t h e "fundamen- talist" approach (in t h a t it a t t e m p t s t o distinguish between "fundamental"

hazard c h a r a c t e r i s t i c s which i t is supposed a r e invariant), a n d a situational approach which recognizes variations according to different physical situations, chemical forms, environmental conditions and h u m a n actions and decisions.

This distinction is analyzed in Chapter Three.

Situational discrimination s e e m s to represent a n overall improvement in clarity of risk definition, b u t i t should be noted t h a t t h e distinctions a r e n o t absolute. For example if a chemical waste is t r e a t e d (deliberately o r not) t h u s reducing i t s hazardousness by making i t say less soluble therefore less environ- mentally mobile ( a s well a s less gut-ingestible), is t h i s an i n t r i n s i c change or a situational one (especially if it is a reve-sible change)? Containment or back- u p devices for n u c l e a r r e a c t o r s or o t h e r hazardous installation . m a y be regarded as 'intrinsic' p a r t s of plant design in some countries, but optional extras, t h u s 'situational' elsewhere

[a].

There is no clear-cut 'natural' s t a t e of a material or technology by which to define its intrinsic hazards and which

could a c t as a definitive base for explicating all situational risk-qualifiers. We r e t u r n to this point later. Nevertheless, with this concept in mind the distinc- tion may still be a valid methodological principle, and an improvement over compound Risk terms.

Formal, quantified risk assessment was first developed in military and nuclear systems especially in engineering reliability for design and construc- tion standards. Although t h e human components in s u c h systems have recently received increased attention, t h e approach was dominated by proba- bilistic estimation of failure r a t e s in mechanical components. More recently however, this engineering s t r a n d has been complemented by developing biologi- cal approaches t o analyzing t h e hazards of released materials e i t h e r from large accidental discharges following s u c h mechanical failures or from routine emis- sions [9].

As is discussed below, even risk analysis of relatively standard technologies such as liquid energy gas terminal facilities or nuclear power station, suffer colossal differences according to different implicit process and problem- definitions and underlying assumptions.

DEFINING

THE RISK

GENERATING PROCESS

Formal probabilistic risk assessment of complex. potentially hazardous systems examines component reliability and t h e knock-on effects of failure through a causal chain in t h e system t o some harmful consequence. Thus

"fault tree" analysis begins with a h y p ~ t h e s i z e d failure a t some chosen point, then identifies the possible branching sequences, attaching an estimated proba- bility t o each sequence aiming at a composite estimated probability of occurrence for a range of harmful consequences. A sister-technique, event t r e e analysis s t a r t s the other way about, analyzing t h e various possible chains

of e v e n t s in t h e s y s t e m t h a t could lead t o a given e n d point, t h e n estimating t h e probabilities of e a c h linking failure, leading t o overall probabilities for e a c h identified h a r m f u l e n d point.

Various controversies a r o u n d such risk analyses, notably t h a t over t h e US Rasmussen Nuclear Reactor safety analysis

[lo],

have shown how deep a r e t h e u n c e r t a i n t i e s a n d opportunities f o r choice a t almost every s t e p in s u c h ana- lyses:

(i) t h e r e a r e m a n y different potential release-events from a given plant, e a c h of which has t o be analyzed for i t s e s t i m a t e d work-force a n d e x t e r n a l consequences. In o r d e r t o r e d u c e t h e consequence- estimation t o feasible scale t h e s e a r e usually grouped into a smaller n u m b e r of families. For example 14 failure or release categories were u s e d in t h e Rasmussen Study, a s t h e i n p u t s t o analyses of external consequences.

(ii) e a c h r e l e a s e event e n d point usually h a s m a n y , possibly i n t e r a c t i n g chains of possible failures t h a t c a n lead t o t h e s a m e end-point. There i s no g u a r a n t e e t h a t all possible significant release end-points or o t h e r pathways increasing t h e probability of even a known e n d point, have been identified.

(iii) even before composite probabilities a r e estimated, t h e description of possible chains of e v e n t s is in itself so complicated t h a t t h e r e i s room even a f t e r a r e a l e v e n t s u c h a s t h e Three Mile Island accident, for dispute as t o whether or not t h e r e a l world event sequence was actu- ally described in t h e preceding analysis [ll].

(iv) t h e basis of probability e s t i m a t e s of c o m p o n e n t failures and cascading sequences of events is highly variable. In s o m e cases good empirical experience exists for reasonable statistical extrapolation; in o t h e r s

t h e applicability of historical data is questionable (e.g.. samples of boiler failures

-

can data on conventional s t e a m boilers of smaller size. thicker or t h i n n e r metal etc. apply to nuclear pressure vessels?);

in others theoretical estimation has t o suffice but may be incapable of experimental validation; in yet others, s h e e r ignorance prevails and either expert subjective judgments have t o be o r c h e s t r a t e d using Bayesian statistical methods [12], or crude guesses a r e made.

(v) normally, such uncertainties could be expressed as confidence limits or error bars around each component probability. But just a s the pro- babilities (if independent) multiply through t h e analysis, so do t h e uncertainties surrounding each figure. Furthermore, in many real cases, t h e specific events considered do not have independent proba- bilities. This so-called "common-mode" [13] escalation of normal failure probabilities is especially prominent where h u m a n actors a r e more influential in t h e system. Legitimately different implicit judge- ments about these have large effects on analytical outcomes. This factor h a s not been widely recognized until recently.

Dimensions of Uncertainty

-

Ignorance and Perception

There a r e several basically different sources of conflict or uncertainty in such risk analyses. These a r e often confused. There is of course t h e possibility t h a t one analyst or a n o t h e r has simply been less careful or competent than another. More important however, a r e rhose apparent!^ frequent cases when equally competent analysts reach vastly different conclusions as t o t h e risks from ostensibly t h e s a m e process or system. Sometimes this is attributable t o very different but equally plausible (or implausible) guesses as t o component behavior or process connections about which little or no knowledge exists. This

would c r e a t e different risk e s t i m a t e s even if precisely t h e s a m e process or sys- t e m were being defined. J3u.t consider t h e case in which t h e e s t i m a t e d probabil- ity of one link in an event-chain leading t o a n a c c i d e n t is infinitesimally small in one analysis, a n d significant in a n o t h e r . In t h e l a t t e r case i t may become t h e critical path t o an accident, in t h e former i t m a y b e placed amongst t h o s e conceptually possible pathways t h a t a r e regarded a s virtually negligible. Notice now t h a t strictly speaking, each different analysis m a y described a d i n e r e n t detailed risk-process as t h e one leading t o t h e m o s t significant risks.

This point may be generalized, because in m a n y r e a l cases ignorance about causal events s u c h as: hydrodynamics in complex pipe-work u n d e r e x t r e m e conditions; h u m a n behavior affecting risk-pathways; m a t e r i a l s failure u n d e r very specific. often e x t r e m e conditions; a n d t h u s of pathways t o s u b s e q u e n t releases, is overwhelming. Assumptions by analysts therefore about t h e boun- dary of t h e technology or process u n d e r analysis, a n d also about i t s i n t e r n a l s t r u c t u r e , may be legitimately very variable. There i s a c o n t i n u u m from sys- t e m s where t h e different choices between analysts m a y be very narrow, detailed a n d technical (though still highly significant) t o s y s t e m s where t h e differences may include implicit behavioral judgments, large-scale differences over system-boundaries, i n t e r n a l cause-effect s t r u c t u r e s , etc. Whether broad o r detailed, t h e s e differences of system- o r problem-definition may be d e t e r m i n e d by social positions of analysts r a t h e r t h a n freely chosen.

LEG

Facility Risks

I will give some examples ranging from narrow t o broad differences: The IIASA s t u d y of t h e risk analyses produced during t h e four different national sit- ing decisions for LEG t e r m i n a l facilities can be used a s a n example of narrow differences [14]. In all t h e r e were fifteen different risk analyses, but although

t h e technology a n d process analyzed was very similar in all four cases,. t h e y r e a c h e d very varied conclusions. In p a r t , t h i s was d u e t o different analytical definitions of what is m e a n t by risk

-

what kind of potential cost? Given t h a t s u c h definitions vary according t o social positions a n d values, i t is not surpris- ing t h a t t h e initial, often unreflective process of narrowing down t h e analytical 'problem' t o one kind of cost o u t of t h e wide choice in principle available (e.g..

population-risk, critical group; per day, p e r t o n n e of gas, p e r job provided, etc.), produced s o m e t i m e s incompatible s t a r t i n g points, let alone finishing points.

However t h e r e were deeper problem a n d u n c e r t a i n t i e s t h a n this.

As Mandl a n d Lathrop n o t e [15]:

...

several decisions m u s t be m a d e in t h e course of performing a risk a s s e s s m e n t , s u c h a s how t o c h a r a c t e r i z e risk, what presentation for- m a t s t o use, what gaps to fill with assumptions, what assumptions t o adopt, which of several conflicting models t o use, how to indicate t h e d e g r e e of confidence of t h e r e s u l t s , a n d which events simply t o omit from t h e analysis. These decisions c a n push t h e r e s u l t s in any direc- tion

...

Thus for example, s o m e studies included shipping collisions or grounding a n d spills, o t h e r s focused only on storage t a n k r u p t u r e , o t h e r s included t r a n s f e r spills, none analyzed poLentia1 sabotage. Even on specific events, esti- m a t e s varied without a n y explanation. One study a s s u m e d t h a t for a typical l a y o u t of six t a n k s s u r r o u n d e d by dikes, a valid e s t i m a t e for a credible spill size was 15% of t h e c o n t e n t s of one tank, whilst o t h e r s took a t least t h e full c o n t e n t s of o n e t a n k a s a conservative e s t i m a t e . The e s t i m a t e d probability of a spill a t one s i t e varied by a factor of lo3 t o in t h r e e s e p a r a t e analyses.

When t h e analysis is extended to cover e f f e c t s of a release, t h e conflicting a s s u m p t i o n s multiply. Different models of dispersion a n d ignition were used,

The major difference was that three facilitjes were for liquefied natural gas, methane, which requires very low temperatures (-161.5"c), the other was for liquefied petroleum gas, mainly propane and butane, which are less vo!ati!e and can be stored a t nearly ambient temperature and pressare. In fact this technical difference was dominated by other differences introduced by the analysts in the studies.

different causes of damage were assumed - some took secondary blast effects to be t h e sole cause of deaths while others took thermal radiation.

It may be initially tempting to say t h a t analysts chose t h e i r detailed problem-definition t o suit t h e conclusion they wanted; but n o t all s u c h assump- tions have identifiable effects on t h e conclusion, and one m u s t also accept t h a t some of t h e shaping of problem-definition is unconscious and d e t e r m i n e d by social positions, specific intellectual traditions to which analysts belong, etc.

This has been widely found to occur in science generally [16].

As Mandl and Lathrop conclude [17],

...

what is striking about t h e estimates is t h e magnitude of t h e differences. Societal risk, individual risk, and t h e risk of one or m o r e fatalities vary over four orders of magnitude across sites, and t h e risk of ten or more fatalities varies over eight orders of magnitude across sites. It is hard to imagine a n o t h e r a r e a of political c o n c e r n where performance measures receiving a s much attention as t h e s e did could vary over such a wide range. Yet even more striking a r e t h e differences between t h e t h r e e reports prepared for Point Conception.

There is about a factor of t e n difference in both societal and individual risk

....

There is a difference of four orders of magnitude in t h e risk of ten or more fatalities. A policy maker faced with s u c h variations could conclude t h a t all t h r e e reports are based on very limited knowledge of the risks of LEG.

furthermore,

....

Each report poses a s a representation of t h e c u r r e n t s t a t e of h o w l e d g e regarding LEG risks, but because t h a t knowledge is incom- plete, some of the reports represent it using probabilistic t e r m s or e r r o r bounds. Yet each report is based on a different s t a t e of howledge: different assumptions a r e made, models used, probabili- ties estimated, etc. No one report in fact represents a comprehensive representation of the c u r r e n t s t a t e of knowledge. When SAI gives a probability of 9.9~10". a n d FERC gives a probability of 8.1x10-~, for t h e same event, t h e policy maker is likely to be somewhat a t a loss as to t h e appropriate figure u p o ! ~ which to base his 01. her decisions.

...

each represents only a subset of t h e total s t a t e of know-ledge. Yet n e i t h e r report acknowledges t h a t t h e other estimate exists!

...

The implication, not fully spelled out, is t h a t formal risk assessments may be unreflectively pretending t o contain ignorance and "uncertainty" within apparently probabilistic bounds, which then appear t o be analytically

manageable, therefore definable a s 'risk,' [18]: whereas t h e real scale of uncer- tainty is m o r e properly c h a r a c t e r i z e d by (i) ignorance ( t h e r e a r e factors and combinations t h a t a r e j u s t n o t even identified, l e t alone 'estimable'), and (ii) such a wide scope of legitimate analytical choice in defining t h e r e l e v a n t sys- t e m s t r u c t u r e t h a t t h e resulting knowledge i s n o t characterized only by 'pas- sive' uncertainty, t h a t d u e t o t h e effects of i m p r e c i s e l y k n o w n q u a n t i t i e s , b u t also by 'actively' shaped u n c e r t a i n t y ( a n d c e r t a i n t y -five studies did not even mention uncertainties!) This l a n d of uncertainty, although usually perceived as t h e 'imprecision' kind, is actually i m p l i c i t conflict. It may or m a y n o t be redu- cible by n e g o t i a t i o n between t h e analysts, but it will n o t be resolvable by m o r e precise observation or analysis which i s t h e usual fallacy. The conflict is d u e t o the effects of i m p l i c i t a n a l y t i c a l choices e v e n in defining w h a t the 'technology' is. When technology is viewed a s i t should be, as a social-organizational e n t i t y (embodying 'hardware' b u t also behavioral relationships), t h i s point can be m o r e clearly s e e n [19].

Prom Technical Imprecision to Social Contradiction

Cox has given a useful discussion of unrecognized u n c e r t a i n t i e s underlying r i s k assessments, c a u s e d by variation in t h e actual processes being evaluated, when fixed processes a r e being a s s u m e d in t h e risk analysis [20].

In order t o simplify his example Cox t a k e s t h e evaluation of only work- force risks, t h u s excluding for now t h e f u r t h e r domain of problems of external emissicns and associated risks. Although h e discusses risks of e l e c t r i c i t y pro- duction technologies, t h e point applies t o all technological process. Following is an outline of his argument:

Modern analysis c a n define technology a s a network of stages c o n n e c t e d by input-output flows. A given s t a g e is defined by i t s input-output s t r u c t u r e . For

example: s t a g e s of mining, smelting, refining, manufacturing and finishing in a typical metallurgical industry; waste arising, in-plant mixing o r t r e a t m e n t , 'packaging,' t r a n s p o r t , storage, possible t r a n s f e r a n d f u r t h e r mixing, final t r e a t m e n t a n d disposal, in t h e case of hazardous waste. More detailed models c a n be m a d e of single plants.

The occupational risk p e r overall u n i t of o u t p u t associated with a s e t of s t a g e s , J in t h e process, is

where t h e s e t of s t a t e s , J, i s defined a s a technology (say, incineration) which i s a s s u m e d t o be well-defined. with a c o n s t a n t input-output s t r u c t u r e ;

Qj

i s t h e n u m b e r of u n i t s of o u t p u t from s t a g e j p e r y e a r ; Lj is t h e n u m b e r of m a n - h o u r s of labor u s e d in t h e production of one u n i t of o u t p u t from stage j ; r j is t h e n u m b e r of d e a t h s p e r employee-hour in s t a g e j ; a n d a, is t h e fraction of t h e a n n u a l o u t p u t from s t a g e j (e.g., x t o n n e s of e n r i c h e d u r a n i u m from fuel repro- cessing) n e e d e d t o support whatever overall production unit is used as r i s k yardstick (e.g., p e r 1 GWe of electricity produced or consumed).

Conventional risk a s s e s s m e n t u n c e r t a i n t i e s arise, a n d multiply in t h e m u l - t i l i n e a r combination of values, e a c h of which is a product of o t h e r u n c e r t a i n e s t i m a t e s a n d data. a n d so on. However, t h e r e a r e m o r e basic u n c e r t a i n t i e s i n defining t h e 'system,' 'process,' or 'technology' in t h e first place. For example every process h a s t o be an open s y s t e m , with i n p u t s from a n d o u t p u t s t o a n out- side environment. Therefore a process o r technolcgy h a s t o be d e f i ~ e d by plac- in.g l i m i t s on it, thereby also defining i t s environment: b u t t h e n a r i s e s t h e t h o r n y question of appropri.ate s y s t e m o r problem boundary, or - p u t a n o t h e r way - a t t r i b u t i o n of risk responsibility:

Should nuclear r e a c t o r risk assessment include t h e risks of reprocessing, transport, waste d s p o s a l , even possibly horizontal nuclear weapons prolifera- tion since t h e s e a r e arguably associated with it as inevitable entailments?

Should t h e risk assessment of fluorine contaminated hazardous waste from aluminum smelting incorporate a n element of t h e risks of coal mining or nuclear risks because of t h e intensive use of electricity in aluminum produc- tion? Does one include in coal risks, t h e risks involved in t h e manufacture of, say, t h e trucks found a t mine heads, even though t h e same trucks would have been made had t h e r e never been such a coal mine? Once begun, t h e possibili- ties of such connections a r e limitless and paralyzing.

~ n h a b e r ' s use [23.] of essentially t h e s a m e approach to analyzing energy system risks for example found high total risks for wind and solar power. But closer examination showed t h a t these high risks resulted from an arbitrary assumption t h a t dirty coal would be used as back up for these ( i n t e r m i t t e n t ) technologies used as base-load supply systems. Thus Inhaber's definition of 'solar technology' included dirty coal technology too! A normal definition of solar and wind technologies has t h e m organized with storage systems, o r with clean back-up. This is a different definition of t h e technology a s a social-organizational unit.

A s Cox emphasizes, the real economy and real technologies a r e far less simple t h a n t h a t implied in fixed internal s t r u c t u r e s of technologies, processes or even industries, and thus Axed technological coefficients. Both the boun- daries and interns1 s t r u c t u r e of a technology can (a) vary in t h e real world. a ~ d (b) be defined variably by the risk analyst (and others) as "the" technology or

"the" risk problem in question. I t is important to note t h a t these a r e uncer- tainties in risk analysis over and above, a n d q u a l i t a t i v e l y different from, those associated with imprecise m e a s u r e m e n t in analysis. I t is suggested t h a t t h e y

a r e also key unrecognized variables in t h e strong dislocations of risk percep- tions between different 'experts' and different public groups which is now a major concern in policy making.

Another s e t of examples enlarging t h e s a m e point come from scientific disputes o i e r t h e e n v i r o n m e n t a l risks of t h e proposed MacKenzie Valley pipe- line from Arctic Canada t o t h e USA [22]. Implicit, and eventually revealed, in t h e analyst's conflicting scientific conclusions were different social-behavioral judgments which c r e a t e d different problem-definitions. Thus some scientists assumed t h a t one pipeline could realistically be evaluated for i t s effects i n iso- lation from f u r t h e r pipelines, roads, telegraph lines, airfields, residential service-towns a n d o t h e r developments ( t h e "corridor") which o t h e r scientists a s s u m e d would inevitably follow. a n d which should therefore, t h e y believed be a 'natural' p a r t of t h e system t o be evaluated.

In a n o t h e r p a r t of t h e s a m e dispute, t h e damage t o t u n d r a f r o m construc- tion work was assumed by s o m e a n a l y s t s t o be limited to t h a t within o f i c i a l lim- itations of construction t o winter m o n t h s , when t h e t u n d r a was hard-frozen.

Other analysts assumed this was unrealistic because t h e y believed t h e p r e s s u r e of deadlines a n d h u g e i n v e s t m e n t s would inevitably c a u s e t h e s e limitations t o be broken in practice, with summer-season construction leading to far g r e a t e r damage.

N u c l e a r Technology

In t h e s e cases, as before, l n e r e is n o objeclive, sinzular problem-deiinition or technological system which c a n be m o r e a n d m o r e precisely "revealed." A f u r t h e r example showing a slightly different b u t essentially similar dimension arose in t h e Wiadscale lnquiry in 1977 i n t o a. proposed oxide n u c l e a r fuel repro- cessing plant [23]. The inquiry c h a i r m a n , t h e n u c l e a r industry and g o v e r n m e n t

agencies defined t h e risk a s s e s s m e n t decision a s t h a t c o n c e r n i n g a single reprocessing plant, a n d nothing more. Objectors on t h e o t h e r h a n d , a s s u m e d t h a t t h e plant, which g e n e r a t e d plutonium and u r a n i u m for f u r t h e r rounds of n u c l e a r power s y s t e m s , would c r e a t e institutional m o m e n t u m for m o r e n u c l e a r developments including widespread f a s t breeder r e a c t o r s a n d plutonium com- m e r c e . The risk a s s e s s m e n t question, a n d t h e associated technological system, was deflned a s m u c h l a r g e r a n d m o r e diffuse, and was dismissed as "emotive"

nonsense by the c h a i r m a n .

Here was a conflicting choice of technology- or problem-definition which was not a 'facts' v e r s u s 'emotions' division. Nor was i t clearly perceived a n d debated in t h e Inquiry a s a conflict of founding problem-definitions. Yet t h e conflicting, equally l e g i t i m a t e definitions were a s y m m e t r i c a l pair based upon t h e different behavioral j u d g m e n t s a n d objective social experiences of t h e con- tending groups. To m e m b e r s of t h e establishment, i t was rational t o draw a boundary round t h e p r e s e n t plant, because they could objectively expect t o influence a n d identify with t h e subsequent decisions w h e t h e r or not t o m a k e f u r t h e r commitments. These decisions, a n d t h e technologies involved, could be logically fenced off a n d n e g l e c t e d For outsiders t o t h e decision making estab- l i s h m e n t however, a n incompatible, b u t equally objective logic prevailed. From t h e i r objective social position, with t h e i r social experience, i t was rational t o a s s u m e t h a t t h e y would have n o real p a r t in any of those s u b s e q u e n t decisions, a s t h e y had been excluded in t h e past. It was therefore rational to condense all

~ o s s i b l e foreseeable f u t u r e developments onto t h e p r e s e n t single plant deci- sion. The technology o r risk-syst.em was t h u s defined t o take t h e s e extensive f u r t h e r probabilities i n t o a c c o u n t .

The i m p o r t a n t point i s t h a t each position, 'expert' or otherwise, was based upon behavioral j u d g m e n t s a n d social experiences which were ~ l e c e s s a r y t o

f r a m e a problem a t all. But e a c h was equally defensible, or illogical, according t o one's social position. No deeper, m o r e objective definition of ' t h e technol- ogy' existed. Nevertheless, t h e language of t h e Inquiry was totally t h a t of an 'objective' technology with 'objective' effects, which could be ' d i s c o v e r e d ' t h r o u g h t h e conflict by more rigorous analysis.

One s t r u c t u r a l variable which h a s become increasingly p r o m i n e n t even in highly a u t o m a t e d technologies is t h e role of "the h u m a n factor" in bringing about accidents. To t h e e x t e n t t h a t t h i s h a s been systematically examined a t all, i t t e n d s to have taken a mechanical, individual operator emphasis, a t t e m p t - ing t o draw upon empirical experience of "failure rates" for probabilistic extra- polation. Organizational distractions, a n d dislocations brought about by collec- tively induced "mind-sets" have been less fully i n t e g r a t e d i n t o risk analysis [241.

P e s t i c i d e s

Again in t h i s general c a s e , t h e institutional origins of t h e risk analysis influence t h e definition of t h e ' s t r u c t u r e ' of t h e technology, which is neverthe- less p r e s e n t e d a s if fixed, n a t u r a l , a n d 'objective.' "Expertise" in defining "the"

s y s t e m may be open to surprisingly wide dispute. For example, t h e official government scientific Advisory Committee on t h e Safety of Pesticides (PAC) in t h e UK, evaluated t h e risks associated with 2,4,5-T in t h e l a t e 1970s, when pub- lic suggestions about i t s pervasive h a r m were accumulating [25]. Having analyzed t h e scientific evidence, t h e Carnrnittee d~ c:eed t h a t 2.4,5-T could ccn- t i n u e in i t s widespread use. After a t t e m p t s t o reopen t h e issue by t h e National Union of Agricultural a n d Allied Workers, t h e main labor union involved in spraying 2,4,5-T for farm and o t h e r employers (including many local authorities a n d government agencies such a s British Rail), t h e PAC r e a s s e r t e d t h e safety of

2.4,5-T, dismissing t h e large NUAAW dossier of admittedly circumstantial clini- cal and o t h e r evidence of actual h a r m , a s unscientific.

This r a t h e r patronizing scientific r e b u t t a l only polarized t h e gathering conflict even m o r e , a n d eventually, in t h e face of f u r t h e r Union action, t h e PAC advanced t h e explicit qualification t h a t i t s assertion of t h e safety of 2,4,5-T was conditional upon i t s proper m a n u f a c t u r e , distribution a n d use. These condi- tions were precisely where t h e farm workers' a n d others' direct experience and evidence was focused. In t h i s behavioral reality of t h e technology of 2,4,5-T pro- duction a n d use, t h e y were t h e e x p e r t s and n o t t h e PAC scientists. This 'behavioral' reality was not m e r e l y social, it was also physical - i t shaped t h e a c t u a l physical processes t h a t led t o r e a l damage. The laboratory controlled t e s t s in t h e scientific l i t e r a t u r e examined by t h e PAC produced Risk Assess- m e n t s which excluded a priori t h e realities of distribution a n d u s e of 2,4,5-T, a n d which potentially radically a l t e r e d i t s risks. F u r t h e r m o r e , t h e s e were behavioral or institutional conditions

-

" t h e h u m a n factor" - j u s t a s objective a s m a n y physical p a r a m e t e r s included. For example d r u m s of 2,4.5-T often arrive with defaced o r removed labels supposed t o describe proper conditions of use. Even if t h e s e a r e lmown, t h e organizational realities of farm life often do n o t allow a f a r m worker t o refuse t o spray just because t h e climate i s not c o r r e c t , o r because specified protective equipment is defective or non-existent.

Also, t h e c u l t u r a l reality of such work life does not encourage a m a n t o s a y h e is concerned about t h e possible risks of s u c h materials.

The point of t h i s example Is t o d ~ r n o n s t r e t e again t h e analytical options available in specifying t h e process or technology for risk assessment. The n a r - row, unrealistic technical definition in t h i s case produced arguably false risk assessments, arid corresponding social a n d technical dislocations which have ramified beyond t h e specific issue in question into general issues of regulatory

credibility, even good faith.

Below, we explore t h i s perspective f u r t h e r s i n c e i t produces additional i m p o r t a n t questions about t h e kinds of u n c e r t a i n t y underlying formal risk a s s e s s m e n t a n d policy management.

To s u m m a r i z e t h e p r e s e n t section, different socially influenced s t r u c t u r a l definitions of a 'technology' or 'technological s y s t e m ' [including i t s i m p l e m e n - tation] form t a k e n for g r a n t e d but different problem-definitions underlying for- mal, rigorous risk analysis. There is n o single objective definition of a technol- ogy which supercedes all others. Risk analysts have t o make c o m m i t m e n t s t o o n e definition o r another, b e f o r e analysis begins. These c o m m i t m e n t s m a y differ. This point has been addressed a t length, because i t clarifies a key confu- sion frequently found between two q u i t e different types of u n c e r t a i n t y in risk analysis:

Type 1 i s t h e most commonly recognized, for example in t h e estimation of t h e probability of a given event, s a y a key component failure, which combined with o t h e r events may lead t o h u m a n h a r m or o t h e r unacceptable end-points.

There m a y be (is always) u n c e r t a i n t y about s u c h factors because t h e y a r e genuinely indeterminate, o r because although d e t e r m i n a t e , we have inadequate knowledge for a c c u r a t e estimation; o r d u e t o complex m i x t u r e s of both. For all i t s inscrutability o r a t best, partial scrutability, o n e c a n a t least s a y t h a t t h i s type 1 u n c e r t a i n t y is 'there,' in t h e system.

All t h i s usually generates u n c e r t a i n t y enough. However type 2 uncertainty intermingles with type 1, but is fundamentaily differel~t. It is t h a t induced by different perceptions, different definitions even of t h e risk problem. But t h e s e may be s o subtle or socially ingrained a s t o be unrecognized a s s u c h by t h e analyst, even when they a r e e x t r e m e [ a s in t h e 2,4.5-T case]. These a r e actively, though often sub-consciously c r e a t e d u n c e r t a i n t i e s built i n t o t h e very

s t r u c t u r e of the analytical domain, and influenced by institutional factors, s u c h a s t h e analyst's social a n d professional background, institutional position in a decision making network, etc. This type of difference can give rise t o different estimations of the same factor, t h e r e f o r e to recognized bands of type 1 uncer- tainty. But it does not s t o p t h e r e . Institutional u n c e r t a i n t y e n v e l o p s t h e sys- t e m r a t h e r t h a n merely resides within it, because "the system" or "the starting problem" is itself subject t o conflicting definitions. This u n c e r t a i n t y is not in t h e definitions, b u t all around t h e m . This may have radical implications for poticy a n d risk assessment.

UNCERTAINTY BY SIRATEGIC DESIGN

In t h e previous part of this c h a p t e r we have distinguished i n technical risk assessment between 'orthodox' imprecision, type 1 (which may include real sys- t e m indeterminacy), and s t r u c t u r a l , or institutional u n c e r t a i n t y , type 2, brought about by (frequently subtle) differences and t a c i t conflicts of problem- definition.

In this section we will examine different kinds of uncertainty underlying t h e a t t e m p t to define t e r m s and d a t a which a r e normally regarded a s absolutely c e n t r a l to a regulatory s c h e m e . We will progress from 'orthodox' uncertainties in measuring hazardous waste forms, t o active socially g e n e r a t e d uncertainties in measuring hazardous waste forms, t o active socially g e n e r a t e d uncertainties in even defining hazardous wastes. The point will be t o show t h a t a s a policy issue, hazarcious waste c a n n o t be managed by an approach based on conven- tional notions of 'decision making under uncertainty' alone. In risk issues this approach tends to locate t h e origins of u n c e r t a i n t y only in biophysical reality [26]. More technical analysis, and decision-insurance against a n y i n t r a c t a b l e biophysical imprecision, is t h u s t h e conventional way t o solutions.' Our con-

clusion is t h a t t h e c e n t r a l properties of hazardous waste m a n a g e m e n t make i n s t i t u t i o n a l l y g e n e r a t e d uncertainties t h e dominant h n d . These u n c e r t a i n t i e s include: t a c i t social conflicts or dislocations over t h e problem being addressed;

over the key t e r m s used for defining t h e problem and satisfactory m a n a g e m e n t mechanisms; and t h e key data. The first has already been tackled, s o we now address t h e last two aspects.

(a) Data Uncertainties

Even if t h e definitions of "hazardous" and "waste" were universally agreed, a n d t h e r e were also no i n t e r m e d i a t e interests diffracting "real" quantities and kinds of waste arisings i n t o regulatory data, t h e r e would still be more problems t h a n often recognized, simply in making t h e "correct" technical observation.

There was a t r e a t m e n t a n d disposal company in t h e UK which had been criti- cized for not controlling t h e composition of the wastes delivered t o it. The company contracted experts a t Harwell to help devise an a c c u r a t e analytical sampling device for just o n e of i t s m a n y consignments, an oil-water emulsion delivered in 4,000 gallon t a n k e r s [ 2 7 ] . A standard vertically sectioned thief- tube was recommended. Trials found even this simple two-phase physical sam- pling impossible t o perform except by very rough estimation. There were n o t two phases but four

-

water; sludge; oil; sludge,

-

with very indistinct boun- daries. Samples along t h e t a n k e r (which was only baffled, not compartmental- ized) showed variations by *50%, though t h e y should have been identical. This was the !east diEcult kind of s a m ~ l i n g a n d analysis

-

a very simple, merely physical analysis with no chemical complications

-

y e t it proved impossible to perform anywhere n e a r accurately. This was also for only one load, of only one

Obviously there are different goals and preferences entering into policy decisions and these are dealt with by the standard approaches. However the dominant framework as- sumes that a unitary factual basis can be found (even if this is probabilistic), from which those preferences can take off.

type of consignment among many different sorts.

I t

is not surprising in t h e light of s u c h realjties why even in t h e detailed d a t a survey of waste arisings in Hungary (see c a s e study), it was admitted t h a t experts frequently had t o r e s o r t t o estimation t o obtain figures a t all.

D e t e r m i n i n g the Amount of Hazardous Waste in Massachusetts

A t t e m p t 1 : The GCA Stlldy. In 1976, t h e Division of Water Pollution Control of t h e Commonwealth of Massachusetts commissioned a study from an environ- m e n t a l consulting firm, t h e GCA Corporation t o "survey t h e quantities, t h e geo- graphic distribution, and t h e c u r r e n t practices of hazardous waste disposal in t h e commonwealth" [29]. As a first step, to d e t e r m i n e t h e quantity of waste g e n e r a t e d in t h e s t a t e , GCA reviewed t h e Division's file containing t h e p e r m i t applications a n d monthly r e p o r t s from waste t r a n s p o r t e r s licensed t o o p e r a t e in Massachusetts. These reports were required under a Massachusetts law prior t o t h e e n a c t m e n t of t h e federal RCRA regulation. The reports were supposed t o include monthly s u m m a r i e s of where t r a n s p o r t e r s picked up a waste, where it was s e n t , waste type, and m e t h o d s of t r e a t m e n t a n d disposal. However, GCA found t h e i r information incomplete a n d difficult t o t r a c k or compile [30]. They decided therefore t o conduct a telephone survey of a selected n u m b e r of firms.

Some 446 plants responded t o their telephone r e q u e s t s for information con- cerning type of waste, a m o u n t g e n e r a t e d per m o n t h , etc. This information usu- ally r e p r e s e n t e d t h e "best guess" of t h e plant m a n a g e r or t h e plant's environ- m e n t a l engineer.

In order t o yield statewide totals t h e waste figures reported by t h e firms were simply extrapolated on t h e basis of n u m b e r of e m p l o y e e s in t h e firms sur- veyed, compared t o total n u m b e r of employees in t h e industries state-wide.

The firms surveyed r e p r e s e n t e d 36Z of t h e S t a t e ' s manufacturing employees.

This procedure assumed a linear relationship between waste g e n e r a t e d and n u m b e r of employees in a particular firm which GCA admittedly had no evi- dence was correct. But they felt t h a t the estimates of waste s o generated were good lower limits, "probably a c c u r a t e t o within a factor of two" [31]. With t h i s methodology GCA e s t i m a t e d t h a t 37.57 million gallons of waste were being pro- duced per year in t h e s t a t e .

A t t e m p t 2: The N e w E n g l a n d Regional C o m m i s s i o n S t u d y . In 1979, t h e New England Regional Commission employed Arthur D. Little (ADL) consultants t o develop estimates of hazardous waste generation for t h e six s t a t e New England Region [32]. ADL performed no new analyses, r a t h e r it used t h e d a t a of previous s t a t e studies including t h e GCA Report in Massachusetts. Taking GCA's raw d a t a a n d performing the s a m e extrapolation based on waste generated per employee ratios, ADL estimated t h e total wast generated for Massachusetts in 1979 was 49.2 million gallons. a n increase over GCA's total of approximately 30%. presum- ably due to changes in employee statistics [33].

The difficulty in using waste p e r employee ratios for extrapolation is shown by t h e wide range of ratios ADL found in New England.

S a t e

Connecticut Maine

Massachusetts New Hampshire Rhode Island Vermont

Average Std. Dev.

The r e p o r t admitted t h a t

Waste g e n e r a t e d p e r e m p l o y e e p e r y e a r

"variations between t h e s t a t e s a r e n o t readily explained on t h e basis of industry differences" [34].

In addition to this e s t i m a t e , ADL provided a "high sludge" e s t i m a t e on t h e assumption t h a t introduction of planned waste water t r e a t m e n t programs

would lead t o a n i n c r e a s e in h a z a r d o u s waste generation. Figures from Connec- t i c u t , which already h a d s u c h a program were used t o e s t i m a t e "high sludge"

a m o u n t s for t h e o t h e r s t a t e s . This a m o u n t for Massachusetts was r e p o r t e d a s 84.9 million gallons per y e a r [35].

E s t i m a t e s B e c o m e Fhct.

I t

was this crudely e s t i m a t e d r a n g e of g e n e r a t e d hazardous waste, 49.2-84.9 million gallons p e r year, t h a t b e c a m e t h e official s t a t e s t a t i s t i c for hazardous waste generation. I t was published however in u n i t s of t o n s , where 240 gallons were a s s u m e d t o equal one English ton (based on t h e density of water). These values were 200,000-350,000 t o n s of waste p e r year. J u s t looking a t one assumption alone, sludge o r solid waste could be several t i m e s heavier t h a n water, leading t o t o n n a g e figures several times g r e a t e r t h a n t h e estimates. With little reference t o t h e i r u n c e r t a i n t i e s , t h e figures were used t o a r g u e for t h e e n a c t m e n t of a s t a t e hazardous waste control program modeled a f t e r RCRA.

A t t e m p t 3: D e p a r t m e n t of ~ v i r o n m e n t a i M a n a g e m e n t : Obviously n o t satisfied with t h e s e a t t e m p t s a t e s t i m a t i n g hazardous waste g e n e r a t i o n in t h e s t a t e , t h e Massachusetts Department of Environmental Management (DEM) decided in 1981 t o do i t s own survey [36]. They hired yet a n o t h e r consulting firm, Urban Systems r e s e a r c h a n d Engineering, to computerize a n d compile t h e information c o n t a i n e d in t h e s t a t e ' s t r a n s p o r t e r r e p o r t s ( t h e s a m e r e p o r t r e j e c t e d a s too incomplete by GCA). This study calculated 170,000 t o n s of hazardous waste being produced in t h e s t a t e . In addition. DEM reviewed t h e EPA notification list of potential g e n e r a t o r s (compiled u n d e r RCRA) a n d a n industrial directory in o r d e r to identify "potential" g e n e r a t o r s n o t reporting t h e i r wastes. Interviews c o n d u c t e d on s i t e a n d by phone, a n d reviews of out-of- s t a t e manifest t o t a l s for waste from Massachusetts delivered t o o t h e r s t a t e s ,

"revealed a n additional 17,000 t o n s of hazardous nTaste not reflected in or