The analysis of the family budget data was carried out on the basis of per head consumption, in groups of families which were defined by the average income per head in the family, i.e., the total family income from all sources divided by the number of persons, irrespective of age and sex, supported by that income. It was then necessary to make some estimate of the proportions of the population falling into these several income groups. The earnings of the head of the family are no sure guide; the earnings of other members, together with supplementary income—unemployment benefit, pensions of various kinds, public assistance, investments— have to be brought into the account, and the total related to the number of persons to be maintained. An average per head income of say, 60s, per week may be reached in many ways—by a skilled worker at £6 per week with only a wife to support; or by a worker and his wife both in employment with earnings of 75s. and 45s. per week respectively; or by a man earning £1,100 a year with a wife and four children and a domestic servant; or by a variety of other combinations of earners and non-earners.
The method adopted in estimating the approximate proportions in the different income groups may be very briefly described as follows :—
(i) Tables were constructed separating married men, single men and women, and male and female juveniles into groups according to their estimated weekly incomes on the following basis :—
- 35s. and under per week.
- 35s. to 45s. per week.
- 45s. to 55s. ,,
- 55s. to 65s. „
- 65s. to 75s. „
- 75s. to 85s. „
- Over 85s. „
For this purpose the occupation tables in the 1931 Census Report were used, together with such published information as could be traced regarding wages and earnings in different occupations and different areas. Those returned as out of work in the 1931 Census were classified separately. A number of independent estimates of this nature were made by different authorities, and as the results showed a satisfactory degree of agreement it was felt that any of them might be accepted as a reasonable approximation to the facts.
(ii) Tables were constructed showing the estimated proportionate distribution of families according to the numbers in the family, and the numbers of earners or recipients of income from other sources. In addition to one table covering all private families, separate distribution tables were made for those families which included a married couple, and those in which no married couple occurred. “One-person” families, and families of which the head was “retired” were excluded from these subsidiary tables.
These tables were derived from figures given in the Report on Housing and from the General Tables in the 1931 Census. The former volume includes tables showing the numbers of private families of different sizes; and an analysis of private families according to constitution of family (married couples, adult males, adult females and children) for families of each size from 1 to 15 persons in two boroughs (Camberwell and Sheffield). The General Tables give, in addition to population by ages and marital condition, information regarding the number and (partially) the sexes, ages and marital condition of persons living in hotels, boarding houses, schools and institutions of various kinds; and a comparison of the Census and resident populations.
(iii) By courtesy of the Registrar General a random sample of 23,000 returns of private families was taken from the original records of the 1931 Census and frequency distribution tables, by size of family and numbers of earners, were constructed for seven different groups—those in which the head of the family fell into the following classes :—
- agricultural workers.
- unskilled labourers.
- other manual workers.
- unemployed.
- no earner.
- remainder.
- all families.
These figures included as earners only the occupied, retired and out of work, whereas the tables referred to under (ii) above included as ” earners ” all persons dependent upon social and investment income. Allowing for this difference, the table showing distribution of all families was found to be in sufficient agreement with the similar table constructed from the Census publications to warrant acceptance of the latter. Moreover, the Census sample revealed differences between the various occupations in such matters as average size of family and proportion of non-earners to earners, small enough to justify the application of the frequency distribution tables to all families, irrespective of the income or occupation of the head of the family.
(iv) The estimated number of private families including a married couple (roughly 8,000,000) agreed fairly closely with the number of married men in the country (8,500,000), the disparity being accounted for by families including more than one married couple, and by married occupants of hotels, boarding houses and institutions. The married men in the various income groups were consequently allotted families of earners and dependants in accordance with the frequency distribution tables for “married couple” families. Subsidiary earners (varying from an old age pensioner at 10s. and a juvenile at perhaps less, to an adult male earning many times that sum) were distributed among the families upon a simple mathematical basis in accordance with the estimated numbers in the various income groups, after deducting married men. Aggregating the incomes and dividing by the numbers in family give the numbers both of families and of persons in the following ” per head ” income groups :—
- Up to 10s. per head per week.
- 10s. to 15s. „ ,, ,,
- 15s. to 20s. ,, ,, ,,
- Over 20s.
(v) Similar tables were constructed for families without a married couple, the appropriate frequency distribution table being used, and the earners again being distributed proportionately throughout all families. A rough estimate was made of the probable distribution among the income groups of persons living alone, and separate estimates were also made for married couple families in which the head was “retired”—on the ground that such families have fewer dependants than married couples still in occupations.
(vi) The addition of the numbers in the different per head income groups (up to 20s.) accounted for practically one half of the population in private families. The provisional figures arrived at were as follows :—
| No of Families | No of Persons | Proportion | |
|---|---|---|---|
| Up to 10s. per head per week | 701,000 | 2,935,000 | 7.7% |
| 10s. to 15s. | 1,649,000 | 6,826,000 | 18.0% |
| 15s. to 20s. ,, ,, ,, | 2,026,000 | 8,356,000 | 21.9% |
| Over 20s. | 5,854,000 | 19,923,000 | 52.4% |
| Total | 10,230,000 | 38,040,000 | 100% |
Beyond 20s. per head per week the first analysis did not go, but income tax statistics indicate that roughly 10 per cent, of the incomes in the country are in excess of £250 per annum, and if the ratio of dependants to earners in that group is taken at about 1-1 (as compared with rather less than 1-0 for the country as a whole) then 10 per cent of the population may be taken as having a per head income of 45s. per week and upwards. The 42 per cent, of the population between 20s, and 45s. per head per week were then divided into two equal groups, at 20s. to 30s. and at 30s. to 45s. Further analysis of the available material suggests that the two 5s. ranges between 20s. and 30s. may, however, embrace as much as 27 per cent, of the total population (17 per cent, between 20s. and 25s., and 10 percent, between 25s. and 30s.) while the next three sub-divisions, from 30s. to 45s., may comprise only 15 per cent.
Estimates were also made on the same bases of the numbers of children of and below school age falling into the respective groups. From these estimates it appears that children comprise 49 per cent, of the persons in group I, 35 per cent, of those in group II, 25 per cent, of those in group III, 14 per cent, of those in group IV, and about 12| per cent, of those in groups V and VI.
(vii) The analysis described above related only to private families in England and Wales. The figures are very rough, and, moreover, take no account of reduced incomes owing to sickness and short-time employment, nor, on the other hand, of casual earnings, overtime and soldiers’ disability pensions and allowances. In view of these factors, and in view also of the probability that the proportions in the lower groups in Scotland are larger than those in England and Wales, it was considered reasonable to round up the proportions in the two lowest groups to 10 and 20 per cent, respectively and to apply them to the whole of Great Britain. Finally, the results were assumed to apply to the inhabitants (staff, residents and inmates) of hotels, boarding houses and institutions as well as private families. The figures finally adopted for the purposes of this report are as follows :—
| Group | Income per head per week | Estimated average expenditure on food per week | Estimated population numbers | |
|---|---|---|---|---|
| I | Up to 10s. | 4s. | 4,500,000 | 10% |
| II | 10s. to 15s. | 6s. | 9,000,000 | 20% |
| III | 15s. to 20s. | 8s. | 9,000,000 | 20% |
| IV | 20s. to 30s. (a) | 10s. | 9,000,000 | 20% |
| V | 30s. to 45s. (a) | 12s. | 9,000,000 | 20% |
| VI | Over 45s. | 14s. | 4,500,000 | 10% |
| Average | 30s. | 9s. |
(a) Further analysis suggests that the upper limit of group IV and the lower limit of group V should be somewhat below 30s., but the average expenditure upon food and the consumption of individual foods in these two groups would not be materially affected by this alteration.
Included in the table is a column showing average food expenditure per head per week. This has been computed from the analysis of family budgets, and represents an average outlay on food amounting to rather over 45 per cent, of income for the lower three groups, the proportion falling sharply above the third group. Expenditure upon food includes the value, at retail prices, of meals taken at restaurants, etc., but excludes the cost of service of such meals.
The foregoing description of the various steps taken in arriving at a rough estimate of the proportions of the population falling within certain per head income groups gives a very summary and imperfect indication of the mass of calculations involved and the many considerations which had to be taken into account. Comparison with particulars of persons below the poverty line given in recent social surveys in London, Merseyside and Southampton and with data regarding the dispersion of family incomes in these surveys, suggests that, at any rate as regards the two lowest groups, the results are probably not seriously in error. But more information is needed in respect of earnings and the constitution of families before the population can be divided into per head income groups with a satisfactory degree of precision.