# Useful Resources for Proportional SIC Mapping

Before using resources for proportional mapping, please ensure that you are certain that a proportional mapping is suitable for your purposes. Further discussion can be found on my LFS SIC Mapping page.

### 1. Office for National Statistics proportional mapping between SIC 1992/SIC 2003 and SIC 2007

**ONS Excel spreadsheets with proportions, mapping SIC 2007 to SIC 2003 (1992) and vice versa**. (Note that SIC 2003 is essentially identical to SIC 1992 at the 3-digit aggregation level and above. See the ONS Table of Changes Between SIC 1992 and SIC 2003, for example.)

- 2003-to-2007 spreadsheet with proportions mapping each SIC 2003 category to one or more SIC 2007 categories (with proportions describing the likelihood of each such correspondence) - at each of the 2-, 3-, 4- and 5-digit levels (Division, Group, Class and Sub-Class levels, respectively).
- 2007-to-2003 spreadsheet with proportions mapping each SIC 2007 category to one or more SIC 2003 categories (with proportions describing the likelihood of each such correspondence), at each of the 2-, 3-, 4- and 5-digit levels.

These spreadsheets are based on mappings created using the Inter-Departmental Business Register (IDBR), the continually-updated universe of UK VAT- or PAYE-registered businesses maintained by the ONS. They are based on IDBR data from December 2009 and will not be updated.

The ONS provide information about SIC 2007.

The ONS also provide information about previous SICs (prior to SIC 2007) on their website of Archived SICs.

### 2. Jennifer Smith's proportional mapping between SIC 1980 and SIC 1992

**I have produced a proportional mapping between SIC 1980 4-digit Activities and SIC 1992 2-digit Divisions based on information on job stayers.**

- This Excel spreadsheet gives my suggested proportional mapping and explains how it was produced.
- To implement this proportional mapping, Mapper98 software could be used.
- This proportional mapping uses information based on workers remaining in the same job (stayers) between Autumn 1993 and WInter 1993/94.
- Autumn 1993 is the last dataset coded according to SIC 1980, and a full set of classification levels are available. Winter 1993/94 is the first dataset coded according to SIC 1992, and again a full set of classification levels are available. Note that in the more-readily-available calendar-quarterly dataset 1993 Q4, no SIC information is available due to the change in classification during the calendar quarter. Seasonal datasets are available on request to the Economic and Social Data Service (ESDS).
- Given the quarterly data, a job stayer has to be coded as someone whose job started more than three months ago, using variables CONMON and CONYR. Note that CONMON is not available when the job started some time ago, so it is necessary to impute a value for CONMON to maximise data availability. For example, month CONMON could be imputed as July in years prior to the current one, and as January in the current year.
- This stayer-based proportional mapping is created during Part 2 of my STATA do-file to generate SIC80-SIC92 mappings.

- A proportional mapping between SIC 1980 Activities and SIC 1992 Sections could be generated using only Autumn 1993 data, in which SIC 1980 Activity and SIC 1992 Section are simultaneously observed.
- A proportional mapping between SIC 1980 Activities and SIC 1992 Sections could also be generated using data on job stayers between Autumn 1993 and Winter 1993/94, as described above for SIC 1992 Divisions.

### 3. Jennifer Smith's proportional mapping between SIC 1968 and SIC 1980

**I have produced a proportional mapping between SIC 1968 3-digit Minimum List Headings and SIC 1980 4-digit Activities**.

- This Excel spreadsheet gives my suggested proportional mapping and explains how it was produced. The mapping is based on SIC68-SIC80 correspondences published by the ONS (see below), and proportions are calculated based on population weights of each SIC 1980 Activity (as measured in 1981). (Note that inability to simultaneously observe SIC 1968 and SIC 1980 categories means that it is not possible to use proportions based on weights of each SIC 1980 Activity by each SIC 1968 Minimum List Heading.)

A spreadsheet documenting SIC 1968 - SIC 1980 correlations - between SIC 1968 3-digit "Minimum List Headings" and SIC 1980 "Activities" - is available from the LSE RLAB.

These correlations correspond to ONS documents available in the ONS SIC Archive:

- ONS pdf document including SIC 1980 - SIC 1968 correspondences between SIC 1980 4-digit "Activities" SIC 1968 3-digit "Minimum List Headings" and NACE 4-digit "Sub-groups".
- ONS pdf document including SIC 1980 - SIC 1968 correspondences between SIC 1980 2-digit "Classes", 3-digit "Groups" and 4-digit "Activity Headings" and SIC 1968 3-digit "Minimum List Headings".

### 4. Mapper98 mapping software

The excellent proportional mapping software Mapper98 is available, along with very helpful information on implementation, from Maarten Goos's webpage at Catholic University Leuven.

- Mapper98 can generate mapping matrices or proportionate mapping correspondences from two simultaneously-observed classifications.
- If there are no periods when the two classifications are observed simultaneously, a proportional mapping can still be created using Mapper98 if panel data are available. It is possible to use data on 'stayers' (workers remaining in the same job between two observed periods)