JSNA Living Well

Statistical Neighbours

Where statistical neighbours are used as a comparator, they will be, depending on the topic area, either Chartered Institute of Public Finance and Accountancy (CIPFA) Nearest Neighbours or Children’s Services Statistical Neighbours Benchmarking Tool (CSSNBT) Nearest Neighbours

Both of these measures use various indicators to group local authorities with those most statistical similar to them, in order that benchmarking can be applied against those authorities most similar, rather than just those geographically nearest.

The CIPFA nearest neighbours for Hartlepool are:

1	Redcar and Cleveland
2	South Tyneside
3	Middlesbrough
4	North East Lincolnshire
5	Sunderland
6	Gateshead
7	Stoke-on-Trent
8	North Tyneside
9	Stockton-on-Tees
10	Blackpool
11	Wolverhampton
12	Kingston upon Hull
13	Halton
14	Walsall
15	St. Helens

 

The CSSNBT nearest neighbours for Hartlepool are:

1	Redcar and Cleveland
2	Halton
3	North East Lincolnshire
4	South Tyneside
5	Blackpool
6	Barnsley
7	Tameside
8	Sunderland
9	St. Helens
10	Gateshead
11	Doncaster
12	Middlesbrough
13	Rotherham
14	Wigan
15	Darlington

 

Below is the criteria that the is used to produce each grouping.

 

CIPFA

The Chartered Institute of Public Finance and Accountancy (CIPFA) Nearest Neighbours model seeks to measure similarity between Local Authorities. This is done by following the traditional ‘distance’ approach whereby a selection of variables (see below) is standardised (with a mean value of zero and a standard deviation of one) and the Euclidian distance between all possible pairs of local authorities is calculated1. These distances are then summed across every single subject and ‘rebased’ (by assigning a distance of 1 to the farthest neighbour meaning all overall distances will lie between zero and one) to calculate the final distance.

It should be noted that the output returned by these calculations is a simplistic way of presenting complex underlying data. Broadly speaking, the results are what might be expected, though the outcome ultimately relies on the indicators and mathematical procedures used.

 

For further information please see http://www.cipfastats.net/resources/nearestneighbours/.

 

2018 model

CIPFA have updated their Nearest Neighbours Model in 2018, resulting in amended peer groups for local authorities. These new benchmarking groups have been implemented in Fingertips from April 2018.

 

The CIPFA groups used in Fingertips are those provided by CIPFA as their “default” groupings – these include in benchmarking groups the 15 nearest neighbours selected only from local authorities of the same type.

 

For upper tier LA comparisons, LAs are only compared within one of these groups:

•           Counties

•           London boroughs

•           All other unitary authorities (including metropolitan districts)

 

For lower tier LA comparisons, LAs are only compared within one of these groups:

•           Non-metropolitan districts

•           London boroughs

•           All other unitary authorities (including metropolitan districts)

 

The CIPFA model allows selection of the indicators used to define the benchmarking groups, but again the groups used in Fingertips are those based on the “default” selection of indicators.  These are:

•           Population

•           Proportion of population aged 0 to 17

•           Proportion of population aged 75 to 84

•           Proportion of population of working age

•           Output area density

•           Output area based sparsity

•           Taxbase per head of population

•           Proportion of population unemployed

•           Retail premises (m2) per 1,000 population

•           Housing benefit caseload (Proportion of population in receipt of housing benefit)

•           Proportion of population born outside the UK and Ireland

•           Proportion of households with less than four rooms

•           Proportion of households in social rented accommodation

•           Proportion of persons in lower NS-SEC (social) groups

•           Standardised mortality ratio for all persons

•           Authorities with coast protection expenditure

•           Non-domestic rateable value per head of population

•           Proportion of properties in council tax bands A to D

•           Proportion of properties in council tax bands E to H

•           Area cost adjustment (other services block)

 

CSSNBT

Statistical neighbour models provide one method for benchmarking progress. For each local authority (LA), these models designate a number of other LAs deemed to have similar characteristics. These designated LAs are known as statistical neighbours. Any LA may compare its performance (as measured by various indicators) against its statistical neighbours to provide an initial guide as to whether their performance is above or below the level that might be expected.

 

The CSSNBT was originally produced by the National Foundation for Educational Research (NFER) in 2007. There is an accompanying practitioner user guide and final report which explain the tool and its development in more detail.

 

The background variables used in the tool are listed in table 1 below. These are the variables used to define each LAs statistical neighbours.

 

We have updated the background variables derived from Census data using more recent information from the 2011 Census.

 

No other aspect of the original CSSNBT methodology used to derive sets of statistical neighbours has been updated.

 

Table 1: Background Variables used in the Models Variable

Mean Weekly pay - gross

% of pupils known to be eligible for FSM

% of vehicles that are three years old or less

% dependent children in household with occupancy rating of +2 or more

% dependent children in overcrowded household

% dependent children in households with 2 or more cars

% dependent children in one adult household

% dependent children in household where HRP is in any managerial or professional occupation

% dependent children in household where HRP is in any routine occupation

% people with mixed ethnicity

% people with Indian ethnicity

% people with Pakistani ethnicity

% people with Bangladeshi ethnicity

% people with Other Asian ethnicity

% people with Black Caribbean ethnicity

% people with Black African ethnicity

% people with Other Black ethnicity

% of working age people with higher qualifications

% people in good health

% households owned outright or owned with mortgage

% households with 3 or more dependent children

% of the population living in villages, hamlets or isolated settlements