Local Unemployment Index, 1927-39.

Table ID:

LUI     (1251396)

Contents:

Local Unemployment Index, 1927-39.

Approx. number of rows:

41,831

Table type:

Raw Data

Documentation Author:

Humphrey Southall

Chronology:

The data cover the period 1927 to 1939.
Dates and times are identified by:
   Day
   Month
   Year

Sources:

1. The data were transcribed from official tabulations of the Local Unemployment Index, published by the Ministry of Labour from 1927 onwards. The original figures were published monthly, but this transcription is quarterly (Jan/Apr/July/Oct) for January 1927 to July 1939 inclusive.
2. The computerisation of this table formed part of research into the geography of economic distress funded by the Nuffield Foundation, and was carried out by Maureen Watters at QMW in late 1995.
3. The Ministry of Labour's Local Unemployment Index was produced monthly between January 1927 and July 1939, and issued to subscribers only. It was advertised as being 'issued primarily for the guidance of manufacturers and distributors wishing to adjust their sales activities to the changing prosperity of the various local areas, and to make due allowance for such changes in judging the effectiveness of their advertising and salesmanship.'

---

Notes:

1. These data are much the most geographically-detailed source for unemployment in Britain between the wars, but are limited as they provide only a count of the total number insured, and then percentages unemployed for different groups: it is possible to estimate total number unemployed from that percentage, but not numbers by age or gender.
2. The reports state that with the exception of places within 10 miles of central London, where an estimate based on the locally resident population was used, the 'numbers insured' was a count of the 'number of insured persons whose unemployment books were exchanged at Employment Exchanges serving the respective areas'.
3. Between 1927 and 1936, the LUI was known to be imprecise because the count of the unemployed included (a) workers insured under the Agricultural Scheme, (b) workers aged between 14 and 16 and (c) workers aged 65 and over, while the count of insured workers used as a divisor in calculating the unemployment rate did not. In January 1937 a 'New Series' of the LUI was begun in which the number recorded as unemployed excluded the above categories. All reports included figures for change in the total percentage unemployed over the preceding twelve months, and those given in the 1937 reports are based on an adjusted figure for the previous year; it is therefore possible for 1936 to compare the rates originally published with corrected rates for the same places and dates. This comparison was used to compute the adjusted rates through the following steps:
   1. Calculate, for each area in each month of 1936 the percentage change between the original rate as reported during the year and the adjusted rate as reported in the 1937 listing for the same month.
   2. Compute for each area the average change over all four months for which we have data.
   3. For counties for which this average change is missing, i.e. those for which the data are derived following the procedures described above, compute it as the average change for all the towns/exchange areas defined as being within it, weighted by their numbers of insured.
   4. If the adjustment factor is positive, replace it by zero.
   5. Apply the area's adjustment factor to the rates originally reported in all years up to 1936.
   6. If the area does not have an adjustment factor, apply the adjustment factor for the containing county.
   7. For 1937 and later years, simply set the adjusted rate equal to the original reported rate.
   8. At the end of this procedure, all rows which had an original reported rate had an adjusted rate, except for some 'special' areas which reported on only before 1936.
4. The reports provide data on several geographical levels, recorded here via the row_type column:
   • N = National: The table includes data for Great Britain, England, Scotland and Wales for all dates (but not for "England and Wales").
   • C = County: As detailed below, the "counties" listed in the report, and recorded in the container_name column, are not always the official counties of the time. We have therefore given rows of original data which were not actually counties the type S, and computed new rows which are more closely comparable to census data, with row_type = C and derived_flag = Y.
   • E = Exchange: It is unclear exactly what these units are, as many separate places were listed within London while other large towns, which clearly contained multiple exchanges, have just a single row of data. NB the "Urban Labour Market" unit type within the AUO was created to provide an ID for all these rows of data.
   • S = Special: See next note. These rows always have g_unit and county_unit set to null, while all other types of row have g_unit set.
5. The original data for the following areas were given the row type 'S' (for Special), either because they were not counties or because their figures needed to be replaced by re-computed data for reasons given in the next note.
   • Cambridgeshire (including Isle of Ely)
   • Cheshire (1927 only)
   • Hampshire (1927-38)
   • Lancashire (1927 only)
   • Lincolnshire (all three parts)
   • London (Greater) (data for the County of London are always also provided)
   • Northamptonshire (including Isle of Ely)
   • Suffolk (combining East and West)
   • Sussex (combining East and West)
   • York (combining all three Ridings plus York city)
   • "Rest of Wales"
   • Within Pembrokeshire: "Rest of County"
   • Aberdeenshire (including Aberdeen city)
   • Caithness and Sutherland
   • Forfar, or Angus (including Dundee)
   • Lanarkshire (including Glasgow)
   • Midlothian (including Edinburgh)
   • "Rest of Highlands"
   • "Rest of Lowlands"
6. The container_name column holds the "county" listed after each place in the report, but these frequently do not correspond to the actual Administrative Counties that existed at the time, and which were reported on by the 1931 Census. The county_name column has been added by us to identify the Administrative County containing the place, and we have added additional rows of data giving totals for these. These additional rows have row_type = C, and derived_flag = Y:
   • Cambridgeshire/Isle of Ely: The LUI includes the Isle of Ely within Cambridgeshire. We re-assigned Ely and Wisbech to Isle of Ely, used these to create a county total for the Isle of Ely, and subtracted those data from Cambridgeshire.
   • Cheshire/Lancashire: In 1927, but not in any later year, the LUI lists Bredbury under Lancashire. Therefore for just that year we created new county totals for both county, subtracting the Bredbury data from Lancashire and adding them to Cheshire.
   • Hampshire/Isle of Wight: Up to the end of 1938, the LUI includes the Isle of Wight within Hampshire. We re-assigned Cowes, Newport, Ryde and Sandown to Isle of Wight, used these to create a county total for the Isle of Wight, and subtracted those data from Hampshire.
   • Lincolnshire: The LUI treats Lincolnshire as a single county. We assigned the towns of Lincolnshire to the relevant Parts, of Holland, Kesteven and Lindsey, and computed new totals for those actual Administrative Counties.
   • Northamptonshire/Soke of Peterborough: The LUI includes Peterborough within Northamptonshire. We re-assigned it to the Soke, used its data as a total for that County, and subtracted its data from Northamptonshire.
   • Staffordshire/Worcestershire: For 1927-33, the LUI includes Cradley Heath under Worcestershire, then Staffordshire for 1934-9. Cradley Heath was not a local government district or a Civil Parish, and it was right on a boundary, so we have NOT made any adjustment.
   • Suffolk: The LUI treats Suffolk as a single county. We assigned Felixstowe, Ipswich, Lowestoft and Stowmarket to East Suffolk, Bury St. Edmunds and Newmarket to West Suffolk, and computed new totals for those actual Administrative Counties.
   • Sussex: The LUI treats Sussex as a single county. We assigned Bexhill, Brighton, East Grinstead, Eastbourne, Hailsham, Hastings, Haywards Heath, Lewes, Newhaven and Rye to East Sussex, Bognor, Chichester, Crawley, Horsham, Littlehampton, Shoreham and Worthing to West Sussex, and computed new totals for those actual Administrative Counties.
   • Yorkshire: Almost all places in Yorkshire have 'county' set to the appropriate Riding, but 'York City' has 'county' set to 'YORKSHIRE'. The 1927 reports do not divide Yorkshire into Ridings, but the 'county' field has been set to the appropriate Riding using the January 1928 report as a guide.
   • Caithness: The LUI provides a single total for Caithness and Sutherland, but from 1931 onwards also includes data for the town of Wick. We used this to create a series for Caithness.
   • Scottish Cities: The LUI included the four main Scottish cities as parts of counties, whereas Scottish census reports treated them as essentially separate. We therefore extracted Aberdeen from Aberdeenshire, Dundee from Forfar/Angus, Edinburgh and Leith from Midlothian, and Glasgow from Lanarkshire. In each case, we created new county-level data for the city, and subtracted the city data from the county.
7. The reports for 1927 do not include national totals, but figures have been added by (a) computing the total 'insured' by summing the county populations for January 1927 (excluding 'London (Greater)' and adding 'York City' for England), and (b) computing the total unemployed rate by using the figures for the equivalent month in 1928, adjusted using the figure for change since previous year. January 1927 total insured were used for all months as the totals for April were identical, and the July and October data looked erratic.

---

Checking:

1. When inputting the 1937 figures for changes over the previous twelve months, a number of places had no signs given; these have been input as negative values, as positive numbers were rare.
2. Consistency checks were run to seek out large changes in the unemployment rate, for example:
   
   select c.year, c.month, c.x_place, c.x_county, c.total,
   diff = (c.total - p.total) * 100. / p.total
   from lui c, lui p
   where (((c.total-p.total)/p.total) > 0.2 or
   ((c.total-p.total)/p.total) < -0.2)
   and ((c.year = p.year and c.month = p.month + 3) or
   (c.year = p.year + 1 and c.month = p.month - 9))
   and c.x_place = p.x_place and c.x_county = p.x_county

---

Indices:

Index	Type	Column(s) indexed
lui_idx	Unique	area_name, container_name, row_type, year, month

---

Columns within table:

Column	Type	Contents
row_type	Text string (max.len.=8).	Type of geographical unit. See note above.
derived_flag	Text string (max.len.=6).	Set to 'Y' where the row is a new county total we have created, as explained in the notes above, and null otherwise.
year	Integer number.	Year from which report taken.
month	Integer number.	Month from which report taken.
day	Integer number.	Precise day on which data were gathered.
data_date	Database date value.	Year, month and day combined into a database date value.
area_name	Text string (max.len.=40).	Name of geographical unit to which the data refer.
g_unit	Integer number.	ID number of the nation, county or town/urban labour market to which the data relate, as defined in the AUO. If the area is a county, identical to county_unit.
container_name	Text string (max.len.=36).	Name of the "county" as given after the place-name in the original listing. Where the data are for a county, this will be identical to area_name.
county_name	Text string (max.len.=36).	Name of the Administrative County, Scottish County or Scottish City actually containing the place, as researched by us. See note above.
county_unit	Integer number.	ID number of the Administrative or Scottish County containing the reporting area, as defined in the AUO.
nation	Text string (max.len.=6).	Code indicating nation: 'E' = England, 'S' = Scotland, 'W' = Wales.
insured	Integer number.	Number of insured individuals.
urate_men	Floating point number.	Percentage of men unemployed.
urate_women	Floating point number.	Percentage of women unemployed.
urate_juvenile	Floating point number.	Percentage of juveniles unemployed. Null from Jan. 1937 onwards.
urate_total	Floating point number.	Overall percentage unemployed.
chan_last_yr	Floating point number.	Change since 12 months previously. Entered only for 1937 reports and used in calculating adjusted figures.
urate_total_adj	Floating point number.	Adjusted total percentage unemployed (see note above).
implied_tot_count	Floating point number.	Estimated total number of persons unemployed, computed from the insured and the overall percentage. Data for counties not listed in the reports have been computed using this implied county, and numbers insured.
n_flag	Text string (max.len.=10).	Flag indicating a footnote. Those appearing are: '+', '*', '$', 'X', and 'XX'. These codes attempt to replicate those appearing in the original reports to indicate notes and footnotes. During processing, the relevant notes were added to the comment field hence these codes are of limited interest.
notes	Text string (max.len.=260).	Comment. Where this is an expansion of information given in the 'note' field, it appears inside square brackets.
rec_num	Integer number.	Sequence number added on loading, to keep rows in order.