lancaster university school of management risk redistribution in technology decision making m sedmak...

14
Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Upload: stewart-richard

Post on 25-Dec-2015

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Risk redistribution in technology decision making

M SedmakJ S BusbyLancaster University School of Management

Page 2: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Background

Risk distribution as organising principle for society (Beck 1992)Fair risk distribution a value in itself (Renn 1992)Distributional questions often central to risk regulation (Hood et al 2001)

YetDistinctions between risks people care about get lost (Sunstein 2005)In ‘standard model’ benefit > cost the decisive criterion (Hermansson 2005) Perceived equity in risk distribution not well modelled (Slovic 1998)Institutions unprepared for redistribution; risks take path of least resistance(Halfacre et al 2000)

Page 3: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Purpose

To study how (if at all) risk assessment processes…associated with basic technology decisions…take account of changes in distribution of risks…across categories of general, societal significance

…social groups, future/current generations, rich/poor regions…high probability-low impact/low probability-high impact…simple-understood/complex-opaque failure modes…hazard types eg safety/security, immediate/delayed…harm domains eg safety/environment/economy…etc

Context• railway engineering eg ERTMS• offshore engineering eg FPSOs

Page 4: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Method

Qualitative analysis of:• people talking about processes of risk assessment• organisational norms governing risk assessment• products of specific exercises of risk assessmentAt the level of:• risk assessment producers (Firms: Atkins Rail, AMEC Oil and Gas)• risk assessment consumers (Regulators: HSE, HSL, MCA and others)Involving:• grounded selection of text referring to (re)distribution• interpretation of how redistribution dealt with by organisations

Page 5: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Example

(Interview)“… it’s not okay in [our] eyes for railways to argue that removing a train with a minor defect from service will be risk increasing because more people will travel on the roads as a result… this is seen as an excuse...”

(Interpretation)• redistribution as device to justify reduction in standards• rejecting this appears as insensitivity to redistribution

Page 6: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Example

(Policy)“…For existing installations the occupational and travel risk account for a contribution to IRPA [Individual Risk Per Annum] of about 3 x 10-4, the contribution of hydrocarbon events is between 1 and 2 x 10-4, giving an IRPA averaged over all personnel on an installation of the order of 4 x 10-4. New installations should show an improvement …”

(Interpretation)• individuals’ risk bearing as basic unit of analysis• significance given to average of this

Page 7: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Example

(Product)“……generates absolute forecasts for the measures of harm arising from the design and operation of [X] to Passengers, Workers and Neighbours. These are measured in minor, major injuries and fatalities and normalised with respect to the population at risk. The consequent individual risks are subsequently compared with the safety targets for the on-board and trackside ERTMS constituents in order to establish tolerability for the Core Hazards and the constituents failures…”

(Interpretation)• assessment sensitive to distribution among basic social groups• only tests each group’s exposure to absolute standard• doesn’t test whether significant redistribution from status quo

Page 8: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Early findings: social groups

Some categories are preserved• eg social groups like workers/passengers/bystanders• eg vulnerable groups particularly at risk like those disabled…but redistribution isn’t analysed

Individual exposure often finer-grained than social groups• eg assess all for track-side hours instead of ‘track-side workers’• test risk borne by most exposed individuals in their ‘bucket of risk’…but certain groups excluded eg train enthusiasts

Use of ‘hypothetical person’ and ‘reference environment’• often to make progress where data is absent• allows specific cases as derivations…but again averages out distributions eg over ‘good’/’bad’ firms

Page 9: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Early findings: social groups

Certain distributions attributed to risk bearers themselves• eg behaviour around lifting operations• eg child/adult trespassing…so assessments sometimes prescriptive about behaviour

Page 10: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Early findings: social groups

Usually all groups benefit but zero-sum-game redistributions do occur• eg worker exposure vs public accident hazard…and heuristics used to resolve

Static risk standard can become redistributive• eg reduce safety standard in fuel crisis to maintain generating capacity…decision elevated to governmental level

Transparency can hamper redistributive reasoning• eg land use decision in which redistribution of risk onto roads considered…under FoI inspector unlikely to act ultra vires…and intractability of redistribution analysis made it uninformative

Possible low-level individual disease hazard to employee better than possible high-level catastrophic failure hazard to public

Page 11: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Early findings: other categories

Some categories are preserved• eg types of consequence like injury/fatality…but redistribution isn’t analysed

Categories often used to facilitate assessment• eg chemical/physical/biological/psychological/social…rather than to reflect societal significance

Unknown risks occasionally redistributed into known risks• eg worker exposure to reduce very low level emission…where impetus is political rather than analytical

Page 12: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Tentative conclusions

Processes insensitive to redistribution…• Not due to bounded rationality or political cowardice• But other effects: eg avoiding legitimising reduction in standards

Difficulty of dealing with redistribution inherent…• Institutions/organisations built around the categories we use• So redistribution questions become inter-organisational problems

Risk analysis involves selection from fundamental opposition of…• Aggregation & elision (reduction to one variable for rational choice)• Differentiation & distinction (revelation of properties for political choice)

“You can optimise the [shift] system for people to be optimally alert at work but at the cost of lack of alertness at home… people come off shift at end of 14 nights then and they’re dead for 3 days… we’re the regulator for people’s safety at work – we appreciate it might lead to higher risk at home but [this is irrelevant to us]…”

Page 13: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Preliminary implications

Redistributive risk assessment might be a better process…avoiding absolute estimations…sensitising firms & regulators to potential controversiesMight consist of asking:1) is risk reduced across all categories? 2) is the percentage across categories changed? 3) is there a new categorisation that becomes material?4) does anyone care?

“The received wisdom is that looking for absolute numbers from risk assessment is pointless, but provided you’ve been consistent they do help with relative risk”

Page 14: Lancaster University School of Management Risk redistribution in technology decision making M Sedmak J S Busby Lancaster University School of Management

Lancaster University School of Management

Preliminary implications

Use risk assessment as tool of differentiation/detailing…as well as aggregation/averagingMight consist of asking:1) what are all the ways we can characterise the risks we identify? 2) how might these generate controversy or interest? 3) could this contradict the outcome of the aggregating process?4) what process can deal with such contradiction?