6.3 Social Welfare Administrative Data in the EU
8.1.4 Methodological Approaches to Estimate the Cost of Specialist Services
For the specialist service sector, the review found that three methodological approaches had been used to es- timate costs:
A unit cost of service multiplied by the number of service users (unit cost model);
The total operating cost of a service for intimate partner violence (total operating cost model); The proportion of an agency’s total operating costs
attributable to intimate partner violence services (proportional operating cost model).
The particular methods and service for which they were applied for each of these models are presented in turn.
Unit Cost Model
Four studies had employed a unit cost model to estimate the economic impact of crisis helplines and support centre services (Canada: Zhang et al 2012) and refuges (Denmark: Helweg-Larson et al 2010; England & Wales: Walby 2004; Finland: Piispa et al 2001). The methods employed by each of these studies are summarised in Table 8.1.4.1 below.
Table 8.1.4.1: Methods employed for unit cost model (specialist services)
Study Unit Cost Unit Cost Data
Source Multiplier Multiplier Data Source
Canada
(Zhang et al 2012)
Average hourly cost of crisis line phone call CAD 20.
Job advertisements for
crisis line workers. Number of women that contacted crisis lines. Average length of phone call (24 minutes) and aver- age of 5 calls per victim to a crisis line = 2 hours telephone time).
General Social Survey. Qualitative research with crisis line workers.
Support Centres: Assumed operating cost per hour: CAD 30.
Authors’ constructed
unit cost. Number of victims of partner violence that used support centre services. Assumed that each woman uses the centre a number of times. Aver- aging 15 hours of use.
General Social Survey. Assumption.
Victim of Crime services: Average cost of victim services per client. CAD 453.
Victim Services Survey Number of victims of partner violence that used victim of crime services.
General Social Survey.
Denmark (Helweg-Larson et al 2010)
Shelter rate per night (includes cost of spe- cialist counselling).
Administrative data from shelters/central social services.
Number of shelter night
stays per year. Administrative data from shelters/central social services. England & Wales
(Walby 2004)
Cost of rent per week
per refuge place. Weekly rent data sourced from survey of five refuges.
Number of refuge places. Administration data. Women’s Aid Federa- tion of England. Finland
(Piispa et al 2001)
Cost of a shelter place
per day. Administration data from shelters. Estimated number of shelter places and client days per year.
Administration data. Federation of
Mother & Child Homes and Shelters.
Three studies (Denmark: Helweg-Larson et al 2010; Eng- land & Wales: Walby 2004; Finland: Piispa and Heiskanen 2001) estimated the unit cost of refuges by establish- ing a daily or weekly rate of a refuge place which was then multiplied by the number of respective shelter stays per year. The average unit costs and refuge places were estimated from information from specialist service providers. This model for estimating refuge costs dem- onstrates reasonable accuracy and is relatively simple, enhancing its replicability; its feasibility is dependent upon administrative data from specialist service refuge provision.
Walby (2004) reports that refuges provide many more services than accommodation and these include le- gal and emotional advice and counselling, outreach
support, telephone helplines and children’s support, but data was not available to estimate these separately. Similarly, Zhang et al (2012) undertaking economic anal- ysis in Canada found little administrative data to support an estimate of telephone helplines and specialist victim support centres. The duration of helpline and centre contacts for underpinning impact in this study (ibid) was estimated from qualitative research with crisis line workers and an assumed number of visits per victim to a support centre. In the absence of information, Zhang et al (2012) constructed unit costs with reference to the advertised salaries of crisis line workers and an assumed support centre operating cost per hour, per victim con- tact. The accuracy of estimates based on constructed costs and assumed impacts hold a significant degree of uncertainty and would not be a preferred model.
Total Operating Cost Model
Three studies in the review (Switzerland: Stern et al 2013; Finland: Piispa et al 2001; Sweden: Envall et al 2006) op- erationalised Total Operating Cost Model methodolo- gies in their economic analysis and this approach was
employed to estimate the impact of refuge and victim support combined, non-residential victim support, and other forms of support groups such as those involving perpetrator programmes. The methods employed in formulating Total Operating Cost Model methodologies are summarised in Table 8.1.4.2 below.
Table 8.1.4.2: Methods employed in the total operating cost model (specialist services)
Study Total Cost Data Source
Switzerland (Stern et al 2012)
Total operating costs for shelter, counselling
and therapy. Organisation of Women’s Shelters. Finland
(Piispa et al 2001)
Total annual cost of (non-residential) spe-
cialist services. Federation of Mother and Child Homes. Total additional budgets awarded for dis-
cussion groups for victims and perpetrators. Finland Slot Machine Association. Sweden
(Envall et al 2006)
Total expenditures for refuge and victim
support. Refuges and Rape Crisis centres. National women’s shelters organisations.
Two of the studies (Switzerland: Stern et al 2013; Swe- den: Envall et al 2006) presented combined estimates for specialist refuge and victim support services. How- ever, even for this, data was not readily available; these studies (ibid) surveyed specialist services for information about budget expenditures and the extent of services provided. This method would probably warrant con- siderable additional work; for example, in the Switzer- land study (Stern et al 2013) 16/19 agencies provided information about their service activity and budgets. However, if the budget for a particular service is isolated as reported by Piispa and Heiskanen (2001) for non-res- idential victim support, and other support groups, then this ‘total budget per service’ information can be easily transposed into an impact assessment.
Proportional Operating Cost Model
Three studies (Canada: Zhang et al 2012; Australia: (Ac- cess Economics 2004; Finland: Piispa et al 2001) present- ed estimates in the form of a proportional operating costs model. This methodology was used to estimate the economic impacts of refuge services (Canada: Zhang et al 2012), counselling services (Australia: Access Economics 2004), victim support services, and rape cri- sis services for intimate partner violence (Finland: Piispa et al 2001). The methods employed by the studies are summarised in Table 8.1.4.3 below.
Table 8.1.4.3: Methods employed in the proportional operating cost model (specialist services)
Study Unit Cost Unit Cost Data
Source Multiplier Multiplier Data Source
Canada
(Zhang et al 2012)
71 % of total operating
costs of shelters. Statistics Canada’s Transition Home Sur- vey (assumption that all were female places).
Data from 593 shelters that 71 % (3 298) of the 4 645 female residents reported abuse as the main reason for seeking refuge.
Statistics Canada’s Transition Home Survey.
Australia (Access
Economics 2004) Counselling: Estimate of Annual NGO fund- ing scaled from NSW data.
NSW, Department of Community Ser- vices Family/IPV Grant Information.
Assumption that 75 % of service use was attribut- able to IPV.
Author’s estimate based on informa- tion available and consultations. Finland
(Piispa et al 2001)
Proportion of victims of crime budget for women subjected to IPV.
Total budget: Finnish Service for Crime Victims.
Proportion of contacts to victims of crime service (90 %) because of IPV.
Statistical data Finnish Service for Crime Victims.
Proportion of rape crisis budget attribut- able to IPV.
Rape crisis centre
administration data. Proportion of contacts to rape crisis centre attribut- able to IPV (50 %).
Rape crisis centre ad- ministration data and research.
Switzerland (Stern et al 2013)
Total estimated
budget. Consultancy study. Estimated budget for ‘domestic violence’ victim support multiplied by 0.75 for estimate of intimate partner violence.
Survey of cantonal agencies.
Total expenditures. Administrative data
(Statistik Schweiz). Cantonal expenditures for emergency aid and longer term assistance. Estimated budget for ‘domestic vio- lence’ multiplied by 0.75 for estimate of intimate partner violence.
Administrative data (Cantonal agencies, Statistik Schweiz).
In this model data is provided from specialist services to indicate the proportion of the services’ total budget that was expended on intimate partner violence ser- vices. This approach is useful, for example, in the case of the victims of crime service in Finland (Piispa and Heis- kanen 2001). This service (ibid) provided some services for non-intimate partner violence, however administra- tive data identified that 90 % of the service expended was attributable to intimate partner violence. Adminis- trative data was used to underpin the proportion that should be attributable to intimate partner violence in four of the five impacts assessed in this way indicating a reasonably replicable and robust methodology. How- ever, in the Australia study (Access Economics 2004)
administrative data was not available and the estimate was calculated on an informed assumption.
8�1�5 Overview of Data Sources of Methods