the cost of living, 1938–1948

12
THE COST OF LIVING, I938-194 The information contained in the latest National Income White Paper' and in some calculátions published by Professor R. G. D. Allen makes it possible to refine and extend previous estimates on the change in the cost of living since pre-war for different social strata. The second part of this paper examines the bias in the official cost-of-living index in the light of these estimates, and the third part contains a recalculation of the middle-class cost-of-living index for the post-war years. THE COST OF LIVING OF DIFFERENT SocIAl.. STRATA Professor Allen has published a series of index numbers showing the cost of buying the pre-war working-class purchases (as shown by the Ministry of Labour enquiry in 1937-8) in each year of this period. Full details of calculations are not given, but there is good reason to trust the reliability of estimates by Professor Allen.3 The procedure that will be used to arrive at the rise in the overall middle- class cost of living is the same as that adopted for previous calculations,' the principle being that the national price indices implied or shown in the National Income White Papers must be the weighted average of working- class and middle-class price indices, if we use the colloquial definition of 'middle class' as covering all those other than working-class. The first step is to make the 'working-class' and national price indices as nearly comparable as possible, which is done in the Appendix. Once we have roughly comparable price indices the 'middle-class' cost- of-living index for the war years can be obtained as a residual. The latest Cmd. 7649. * London and Cambridge Econo,nic Service. Vol. XXVII, BULLETIN I. p. 15. 'Professor Allen was a member of the Technical Committee advising the Minister of Labour on the construction of the new Interim Index of Retail Prices. It has already been shown that the construction of the new Index involved a very careful calculation of the increase in the working-class cost of living between 1937-S and 1947. (BULLETIN, Vol. 10, p. 140 and p. 156, bottom. For the convenience of those who have the booldet' Changes in the Cost of Living and the Distribution of income since 1938 ',the references there are pp. 19 and 35). If it is worth anything as a check, Professor Allen's estimates correspond closely to some rough ones of my own (BULLETIN, Vol. 9, p. 248). and reach almost exactly the same total rise up to 1947 as I inferred the Ministry of Labour had estimated (BULLETIN, Vol. 10, ¡oc. cil.). They also give monthly the same increase for each groupas the Ministry appear to have estimated. For j one 1947. in fact, it is certain that the price relatives refer to the cost of working-class purchases. A note on page 32 of the L.C.E.S. Bulleiin states, as is implied by Professor Allen's text, that the price relatives for 1938 to 1947 were derived from calculations 'based on National Income White Papers'. They were not, however, based entirely on the White Pape:s. If they had been, the group price relatives would be nalional rather than working-class. A detailed working-class weighting system has been used (Allen ¡oc. cii. p. 15), and the price relatives do not quite correspond with those in the White Papers. The relatives for food implied by Cmd. 7099 and Cmd. 7371 for 1942, 1943 and 1944 were 130. 131 and 136, for example, compared to relatives of 125 shown by Pro- fesser Allen for each year; and the relatives implied by C,nd. 7099 for the average of 1946 were 174 for footwear, 187 for men's clothing, and 188 for women's clothing, compared to 171 shown by Professor Allen for the clothing group, Note also the discrepancies in drink and tobacco relatives shown in the appendix to this paper. So although White Paper relatives have been used, they must have been supplemented by other information, and since the weighting system is working-class, I conclude that Professor Allen's index givès an .ccurate estimate of the working-c1*cs cost of living. 'Buu.Krw, Vol. 10, P. 255. Booklet, p. 41.

Upload: dudley-seers

Post on 02-Oct-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: THE COST OF LIVING, 1938–1948

THE COST OF LIVING, I938-194

The information contained in the latest National Income White Paper'and in some calculátions published by Professor R. G. D. Allen makes itpossible to refine and extend previous estimates on the change in the cost ofliving since pre-war for different social strata. The second part of this paperexamines the bias in the official cost-of-living index in the light of theseestimates, and the third part contains a recalculation of the middle-classcost-of-living index for the post-war years.

THE COST OF LIVING OF DIFFERENT SocIAl.. STRATA

Professor Allen has published a series of index numbers showing the costof buying the pre-war working-class purchases (as shown by the Ministryof Labour enquiry in 1937-8) in each year of this period. Full details ofcalculations are not given, but there is good reason to trust the reliability ofestimates by Professor Allen.3

The procedure that will be used to arrive at the rise in the overall middle-class cost of living is the same as that adopted for previous calculations,' theprinciple being that the national price indices implied or shown in theNational Income White Papers must be the weighted average of working-class and middle-class price indices, if we use the colloquial definition of'middle class' as covering all those other than working-class. The firststep is to make the 'working-class' and national price indices as nearlycomparable as possible, which is done in the Appendix.

Once we have roughly comparable price indices the 'middle-class' cost-of-living index for the war years can be obtained as a residual. The latest

Cmd. 7649.* London and Cambridge Econo,nic Service. Vol. XXVII, BULLETIN I. p. 15.'Professor Allen was a member of the Technical Committee advising the Minister of

Labour on the construction of the new Interim Index of Retail Prices. It has already beenshown that the construction of the new Index involved a very careful calculation of theincrease in the working-class cost of living between 1937-S and 1947. (BULLETIN, Vol. 10,p. 140 and p. 156, bottom. For the convenience of those who have the booldet' Changesin the Cost of Living and the Distribution of income since 1938 ',the references there arepp. 19 and 35). If it is worth anything as a check, Professor Allen's estimates correspondclosely to some rough ones of my own (BULLETIN, Vol. 9, p. 248). and reach almost exactlythe same total rise up to 1947 as I inferred the Ministry of Labour had estimated (BULLETIN,Vol. 10, ¡oc. cil.). They also give monthly the same increase for each groupas the Ministryappear to have estimated. For j one 1947. in fact, it is certain that the price relatives referto the cost of working-class purchases. A note on page 32 of the L.C.E.S. Bulleiin states,as is implied by Professor Allen's text, that the price relatives for 1938 to 1947 were derivedfrom calculations 'based on National Income White Papers'. They were not, however,based entirely on the White Pape:s. If they had been, the group price relatives would benalional rather than working-class. A detailed working-class weighting system has beenused (Allen ¡oc. cii. p. 15), and the price relatives do not quite correspond with those in theWhite Papers. The relatives for food implied by Cmd. 7099 and Cmd. 7371 for 1942, 1943and 1944 were 130. 131 and 136, for example, compared to relatives of 125 shown by Pro-fesser Allen for each year; and the relatives implied by C,nd. 7099 for the average of 1946were 174 for footwear, 187 for men's clothing, and 188 for women's clothing, compared to171 shown by Professor Allen for the clothing group, Note also the discrepancies in drinkand tobacco relatives shown in the appendix to this paper. So although White Paperrelatives have been used, they must have been supplemented by other information, andsince the weighting system is working-class, I conclude that Professor Allen's index givèsan .ccurate estimate of the working-c1*cs cost of living.

'Buu.Krw, Vol. 10, P. 255. Booklet, p. 41.

Page 2: THE COST OF LIVING, 1938–1948

iz8 THE BULLETIN

National income White Paper shows no change in the estimate of wages in1938. Hence the calculations of total working-class consumption previouslymade' can stand, and the small adjustments made to the estimates of 1938consumption can be treated as adjustments to middle-class consumption.2The division of national consumption in 1938 was therefore approximately£2,361 millions working-class and £'95o millions middle-class. Using theseas weights we obtain the middle-class cost-of-living indices shown in Table I.The caution made in previous papers about the diversity of the 'middleclass', as defined for this purpose, should be borne in mind.

TABLE I. Working-class and Middle-class Cost of Living (1938=100)

*From 1946 foreign travel prices are taken into account which raise the national priceindex by rather less than 1 point, and the middle-class cost of living by more than I point.

The adjusted index of Professor Allen's should not be taken as accurate to half a unit,but if the correction for beer strength is made in half-units the possibility of cumulativeerror involved in rounding is reduced.

This Table shows us how price movements were favouring the workingclass relative to the rest of the comrnunitywhat I shall call the pro-gressive' effect of price changes. It should be emphasised that we are dis-cussing here the effect of price movements only, not of all economic develop-ments. VThat is really of interest to the economist is the change in realincomes, which will be investigated in a further paper. Relative pricemovement is only one of the forces affecting relative real incomes.3

We can measure the progressive effect of price movements from year toyear by the following ratio

Middle-class cost of living in year T Working-class cost of living in year TMiddle-class cost of living in year T-1 Working-class cost of living in year T--1This ratio indicates the extent to which the working-class was sheltered

from price increaseswhether because of a socially discriminating pricepolicy or not. A high ratio indicates a high 'progressive' effect, and viceversa.' The value of this ratio was as follows :-

$ Using middle-class index excluding foreign travel in 1946, for comparability with 1945.

'BULLETIN, Vol. 10, p. 187. Booklet, p. 11. Wage3 arc, of course, by far the mostimportant determinant of working-class consumption.

All expenditure on foreign travel is considered middle-classIncome movements with a 'redistributive' effect may even imply to somc extent price

movements with the opposite effect, e.g., the fall in the relative share of salaries in thenational income may involve a cheapening of services (which are more highly consumed bythe middle classes) relative to goods.

4 These calculations are on a 1938 base throughout. This means that they do notnecessarily measure the cost of a given amount of satisfaction. Indications are that post-war Paasche indices would show a relatively higher increased cost of living for the workingclass, and hence less total progressive effect,

1939 1940 1941 1942 1943 1944 1945 1946 1947 1948

National price index 102 119 133 143 148 152 156 160e 171 184Middle-class cost of

living ...... 103 120 136 146 153 157 164 171* 183 194Working-class cost of

livingt ...... 102 119 1301 140 1441 147f 149 15L1 164 175k

1938-9 1939-40 1940-1 1941-2 1942-3 1943-4 1944-5 19456 1946-7 1947-8

1.01 1.00 1.03 1.00 1.01 1.00 1.03 1.01 1.00 0.98

Page 3: THE COST OF LIVING, 1938–1948

THE COST CF LIVING, ¡938-1948 129

The main progressive effect clearly took place in two steps, ¡940Iwhen price controls and subsidies were greatly extended, and ¡944-5 whendrink prices actually fell, while prices of luxury goods and services rosesharply (see Table VII). 1947-8, with its big increases in food and drinkprices, compared to other goods, saw the first major step in the other direction.It is interesting to note that price movements had on balance a more pro-gressive effect under the Coalition Government than under the presentadministration,' prices tending in fact to change to the advantage of theworking-class under the Coalition Government and to the advantage of therest of the population. after 1945.2 Anyone tempted to draw rather oddconclusions about the association between political shade of Government andprogressive effect of price movements should bear in mind the warning thatprice changes are only part of the economic history of this period. Otherrelevant points are that such price changes were not always deliberate andthat the lives of each Government were subject to very different economicforces.

These calculations can be extended to sections of the middle classes bythe same methods and subject to the same cautions as have been outlined inprevious essays.3 Any corrections which are necessitated by new consumptionestimates in the latest White Paper are made to ' upper ' middle-classestimates.4 First dividing the middle class into' upper' and' lower' accord-ing to whether incomes in 1938 were above or below £500, we have theresults in Table II.

TABLE IICost-of-Living Indices for' Upper' and ' Lower' Middle Class (1938=100)

From 1946. foreign travel is included in these figures.

Much the same development is evident as was shown by comparison be-tween working and middle classes. ¡940I saw a substantial increase in uppermiddle-class living costs relative to lower middle, and a fortiori relative toworking-class, followed by another fanning out of indices in ¡942-3 andagain in ¡944-5, when the increased prices mainly affected the highest incomegroups, whose living costs have apparently increased at a fairly steady ratesince 1941.

Data from Tables I and II are given graphically in the diagram, whichshows that price indices have notably accelerated their upward movement

The total progressive ratio for 1940-5 was 1.05, and for 1945-8 it was 0.98, givinggeometric mean annual ratios of 1.01 in the first period and 0.99 in the second.

This tends to confirm previous surmises (BULLETIN, Vol. 10, P. 157. Booklet, p. 43).BULLETIN, Vol. 10. pp. 265 et seq. Booklet, pp. 51 et seq.

'In addition foreign travel spending is wholly allocated to the upper middle class forwant of any information on its division.

1939 1940 1941 1942 1943 1944 1945 1946 1947 1948

All middle class ... 103 120 136 146 153 157 164 171* 183 194Tpper middle class... 103 121 139 148 157 160 170 179e 193 203Lower middle class ... 102 119 133 144 148 152 154 158 168 181

Page 4: THE COST OF LIVING, 1938–1948

t'

¡c1 3't 140 L.I

/

b

b J' /

7/$ '/

5 b

s tI I5%

t'-. //

't_t

s

b

I

t) 43 4L t4S Ljk.

130 TITlE BULLETIN

DIAGRAM

Cost-of-living Indices, 1938-48. (1938 ioo)

= Upper Middle ClassLower Middle ClassNational Price Index

- Working Class= Official Index

t.-

Page 5: THE COST OF LIVING, 1938–1948

THE COST OF LIVING, 1938-1948 131

since 1946, the last 3 years having been responsible in fact for a large part ofthe increase since pre-war. The official cost-of-living index is also shown forcomparison.'

Indices are shown in Table III for the fairly homogeneous lower middle-class salaried stratum (whose expenditure is taken from the Massey budgets,and which is considered subject to national average price movements forindividual commodities), and for all salaried staff, allowing for those belowL250 per annum in 1938, considered subject to working-class living costs,and those above £Soo per annum in 1938, considered subject to uppermiddle-class living costs. It is only towards the end of the period that thesetwo indices diverge at all noticeably.

TABLE IIICost-of-Living Indices for Salaried Lower Middle-class and all Salaried Employees

(1938 = 100)

II. PRICE MOVEMENTS AND INDEX DISTORTION

\Ve can now revise some previous estimates of the bias in the official cost-of-living index, when it is considered as a measure of the increase in the cost ofliving since 1938.2 In order to simplify subsequent analysis, the adjustedindex of Professor Allen will be used, although it would make little differenceto the conclusions if we used the unadjusted index. Table IV shows a com-parison between official and actual working-class cost-of-living indicestogether with an index of bias in the official index .3

TABLE IVOfficial and Aetual Working-class Cost-of-Living Indices (1938=100)

1939 1940 1941 1942 1943 1944 1945 1946 1947eOfficial working-class cost-of-

living index ......101 1171 127 128 127 1284 130 130 130Actual working-class cost-of-

living index ... ... 102 119 1304 140 1444 1474 149 1514 1624tIndex of bias 99 99 974 914 88 87 87 86 80

Mid-June. tSame correction made for beer strength as for annual average of 1947.Officiai index XloOActual index

The increase in bias between 1941 and 1942 was so marked as to indicateclearly that this was the period when subsidies were directed specificady to

'The values shown on the graph against the mark for each year are the annual averages.except that the 1947 value of the officiai working-class cost-of-living index is the mid-Juneone. $ BOLLETIN, Vol. 9. p. 248.

I This is also an index of the error in the calculations of those who have used the officialworking-class cost-of-living index as jilt measured the increase in the cost of living, e.g.Pigou, E. J., 1948, p. 204, and E. Victor Morgan, 'The Conquest of Unemployment'. p. 4.Calculations of real wage changes between 1938 and 1945, on the basis of the old index shouldfor example, be multiplied by 87/100 if they are to have any meaning at ail.

1939 1940 1941 1942 1943 1944 1945 1946 1947 1948Salaried lowc,r middle-

class ... ... 102 119 133 145 149 153 155 159 170 182All salaried staff ... 102 119 133 144 149 153 156 161 172 184

Page 6: THE COST OF LIVING, 1938–1948

132 THE BULLETIN

the over-weighted items in the old cost-of-living index, and fiscal policy wasframed with one eye on what were in fact errors in weights (when the indexwas used as an official measure of changes since pre-war, as was encour-aged by the use of September Ist, 1939 as a base). Subsequent developmentsbecome clearer if we consider the ratio :-

Actual Index in Year T Olficial Index in Year TActual Index in Year T-1 Olticial Index in year T--1

This ratio shows the extent to which cost-of-living changes exceededindex changes, and its value was as follows :-

There was little fresh distortion in fact for some years after 1943thoughthe low ratios indicate not that the index was truly indicating the rise inliving costs since pre-war, but merely that there was no additional distortion.1946 and 1947 saw fresh achievements in index stabilisation, however, thelatter year being really remarkable in that a rise of over io points in theactual cost of living found no reflection at all in the official index, which wasmaintained quite steady, increasing the already substantial public suspicionof official statistics. Over this period the index ceased to hear even a slightrelation to the actual cost of living and became a measure of administrativeagility. Any upward movement merely represented a lapse of efficiency,and was rapidly reversed.'

We can now, with the aid of measures of price increases for all classesand the ratios of progressive effect and index distortion, get a rough pictureof the social implication of price movements over this period. From 1938to 1940 all indices, including even the official cost-of-living index, moved upat about the same rate. In 1940-1 the effects of an economic policy designedto achieve a stabilisation of the cost of necessities became apparent, the actualcost-of-living indices fanning out. After 1941 the official cost-of-living indexwas still prevented from rising as rapidly as national prices, and was in factactually forced downward. But this was not achieved so much by ' progres-sive' price movements as by index manipulation. Thenceforward the work-ing-class index was much nearer to the national price index than to the officialindex. In 1943-5, however, index movements once more reflected changes inworing-class living costs, and in 1944-5 the second big progressive set ofprice changes occurred, which affected the top social stratum particularly.After 1945 index distortion grew on the one hand, while on the other handprices were moving so as partly, but only partly, to reverse the previousprogressive effect.

It should not be assumed that index distortion was always conscious, anymore than that the progressive price effect was entirely the result of deliberatepolicy. The aim of subsidy policy was primarily to restrain the wage-price

'The geometric mean annual distortion ratio for successive Chancellors, indicating theextent to which additional subsidies were directed towards over-weighted items, were:Sir Kingsley Wood, 1.04; Sir John Anderson, 1.00; Dr. Dalton, 1.04.

1938-9 1939-40 1940-1 1941-2 1942-3 1943-4 1944-5 1945-6 1946-7

1.01 1.00 1.01 1.06 1.04 1.01 1.00 1.02 1.07

Page 7: THE COST OF LIVING, 1938–1948

THE COST OF LIVING, 1938-1948 133

spiral, the progressive effect being largely a by-product. Subsequently, how-ever, it seems to have occurred to those in charge of economic policy that thisrestraint could be achieved by keeping the official cost-of-living index steadynearly as effectively as by spreading subsidies over all wage-goods. Theregressive effect of price movements after the war resulted partly from theworld sellers' market for food, and partly from a budgetary policy designedinter alia to cover concessions in direct taxes and increased social securitybenefits by increasing indirect taxes (particularly on tobacco), which wouldrestrict consumption. One must emphasise once more that this is not thewhole economic story of this period, and that though it is useful for analyticalpurposes to study price movements in isolation from other economic changes,a misleading impression can be gained if the other factors involved are leftout of account.

III. THE POST-WAR MIDDLE-CLASS COST-OF-LIVING INDEX

\Ve can also now revise the calculations which were the basis of theconstruction of the middle-class cost-of-living index. If we treat eachapproximate Laspeyre type price relative for commodity groups as theweighted average of working-class and middle-class relatives, and use theworking-class relatives for each group already calculated,' we obtain asresiduals the middle-class price relatives shown in Table V.

TAaLE VlVorkil1g-rlass and iIiddle-class price relatives for 1947 (1938=100)

It will be seen that the total indices reached this way differ slightly fromthose obtained from Professor Allen's index, which were i6i for the working-class and 183 for the middle-class. It is not possible, however, to splitProfessor Allen's index into the necessary detail, and the above table will beused as the basis of the calculations that follow.

'BULLETIN, Vol. 10, p. 261. Booklet, p. 47. Where the national price relative is taken,an adjustment is made for new price relatives implied in Cmd. 7649.

Cornmodty group Nationalprice relative

Working-classprice relative

Middle-classprice relative

Food 151 138 186Drink... 250 253 246Tobacco ... 335 337 331Rent and rates 111 111 111Fuel and light 150 151 144Household durables 220 206 234Other household goods 156 156 156Clothing 189 174 202Reading matter ......... 118 118 118Motoring and travel 146 135 153Communications 134 134 134Entertainments 176 178 172Other services 149 149 149Other goods 221 185 231Income in kind of H.M.F 154 154 154Trav3l abroad ......... 260 - 260

171 162 182

Page 8: THE COST OF LIVING, 1938–1948

134 THE BULLETIN

Orouping into categories suitable for linking to the Retail Prices Indexwas done by the a;uc ithods as used for the previous calculations, and thisgave prke relatives for the Index groups. From these could be obtainedestimated midçlk-class weights for the base date of the Index.

The total index differs by four points, as does the middle-class index inTable V, from that previously calculated.R This is due partly to the fact thatit is now possible to allow for foreign travel, but mainly to the revision on theapproximate national Laspeyre price relative for focd, which meant a greaterrevision of the middle-class food price relative obtained as a residual (from168 to 186). This caused a corresponding increase in the weight given tofood in the revised weighting system shewn on the second column. Thissystem can be used to link the middle-class index to the new Retail PricesIndex, to obtain an approximate measure of short-period current movementsin the middle-class cost of living and doing this we obtain a revised seriesfor the middle-class cost of living, as shown in Table VI, replacing theslightly lower series previously published in the BULLETIN.2

The rise from mid-June I9i to the average of i.ç8 obtained in thisway is 7 per cent, compared to a rise of 6 per cent indicated by Table I.This difference seems to indicate that even within commodity groups working

'BULLBUN. Vol. 10, p. 264. Booklet. p. 50.'Errors will, of course, occur in as far as the working-class price relative for each group

differs from the middle-class, whether because of wrong internal weighting or because ofprice movements affecting purchases by different strata to a different extent.

TABLa VIMiddle.da.s Cost-of-Living (1938 100)

1947 1948 1949January - 190 197February - 193 198March - 193 19SApril - 195 198May - 195June 182 197July 183 195August 183 196September 184 196October 185 197November 188 197December 189 197

(;i.u./' tIj4Id1fCl4. price rdi/it'e.jj,I4lc-cj,.cfoi /so'e 1947 (1938' 100) weiqhtfor Index

I Food 186 183Il Rent and rates 110 61

Ill Clothing and footwear 202 1381V Fuel and light 138 21V Household durables 234 78

VI Miscellaneous goods ... 205 106VII Miscellaneous services 159 280'11I Drink and tobacco ... ... 265 133

Total ... 182 1000

Page 9: THE COST OF LIVING, 1938–1948

TITE COST OF LIVING, 1938-1948 135

class prkcs 'nay have been rising faster than rniddk-class, hut it is withinthc hounds of statistical error, and tot) much should not be read into it.'['he middle-class index will not be corrected for this difference here, silice theindex at June 1947.may vell have been a point or so too low, and we nowobtain for the average of 1948 a relative of 195, compared to ¡94 obtained inTable I from the adjusted index of Professor Allen and the National PriceIndex.1

We can also show the approximate middle-class cost of living for March1949, compared to ¡938, by commodity groups :-

.l1idd1e.d.,s cost of liviagIndex Group March 1949 (1938=100)

I Food 201II Rent 110

III Clothing 2371V Fuel 154V Durables 253

VI Other gonds 224VII Sereices 167

VIII Drink and Tobacco 294

Total 19S

If we could allow for the decline in quality compared to pre-war, we wouldprobably find that for the whole of the middle-class the cost of living wasin March, before the Budget, almost exactly twice pre-war.

DUDLEY SEERS.

APPENDIXADJUSTMENT OF INDICES FOR COMPARABILITY

A major adjustment has to be made to Professor Allen's series to make itcomparable with the 'White Paper series. Beer prices are calculated for theInterim Index of Retail Prices in such a way as only to allow partly forreduced strength,2 whereas for the National Income White Paper they are ona 'standard barrel' basis, which allows fully for changes in strength. It canbe plesumed that Professor Allen's series is calculated on the same bisis asthe Interim Index, for he links his series to that shown in the Interim Index,and he would not be likely to show without comment an Index not fullycomparable throughout. In addition, his series for Drink and Tobacco is forthe whole period less than the series given by applying the base-year weightsof the Interim Index3 to the price relatives implied by the fixed-price tablesof the National Income White Papers.'

'This series is, therefore, not quite comparable to that shown in Table LVide Supplement No. 2 to the Industrial Relations Handbook. p. 26.

O. cit., p. 6.'All the relatives used in this paper are based on the latest available datai.e. Cnmd.

7649 for 1948. 1947 and 1946, C,nd. 7371 for 1945 and 1944, and Cmd. 7099 for earlier years.

Page 10: THE COST OF LIVING, 1938–1948

136 THE BULLETIN

Working-class Drink and Tobacco Price Indices (1938 = 100)

1939 1940 1941 1942 1943 1944 1945 1946 1947Professor Allen's Series ... 107 141 159 197 225 236 234 241 274Series derived from National

Income White Papers ... 108 142 165 207 236 249 242 255 288

In order to put Professor Allen's index on a comparable basis, therefore,with that taken from the National Income White Papers, the derived serieswill be substituted for his Drink and Tobacco series. Since the base-yearweight for Drink and Tobacco is approximately one-eighth of the totalweights, the total Index must be adjusted by one-eighth of the differencebetween the two series above, i.e. by i or z points, giving the working-classcost-of-living index shown in Table II. An index number for the wholeindex in 1948 must also be found. Professor Allen gives one of x 6i for June1947.2 If we multiply this by 107.7, the average of the new Interim indexfor 1948, we obtain a value of 173, uncorrected for beer strength. TreatingDrink and Tobacco similarly, it would seem that the average price relativefor 1948 would be 309,3 compared to 331 obtained from the White Paper asindicated above. This suggests a correction of 2 points for beer strength,bringing the cost-of-living index for 1948 to I75.

Adjustments also have to be made to the price relatives implied in theNational Income White Papers. These relatives are obtained by dividingcurrent cost of current purchases (Ptt) by the 1938 Cost of the same pur-chases (pi93sqt), and are hence current-weighted or 'Paasche '-type indices,whereas what we want for comparison with the 1938-weighted working-classcost-of-living indices are 1938-Weighted or 'Laspeyre '-type indices. Themain step is to weight the' Paasche ' price series for sub-groups shown in theNational Income White Papers by expenditure in 1938. The resultantindices for commodity groups will only be approximately what we want,since 'Laspeyre' series should be used for the sub-groups, but the errorshould be small.'

The food group must be given special treatment. The change in thepattern of consumer expenditure and in the pattern of prices mean that forthe whole food group a Paasche Index might differ appreciably from theLaspeyre. The latest National Income White Paper gives sufficient informa-tion to allow the following Paasche-type price series to be calculated for eachfood sub-groups :-

1BULLEnN, Vol. 10, p. 148. Booklet, p. 27.'Cf. my own estimate for June 1947 of 161+ (1937-8=100). vide BULLETIN, Vol. 10, pp

140 et seq. Booklet. pp. 19 et seq.283 for June 1947, multiplied by 109.1, the Index group average for 1948 (June 1947

=100).'Because within sub-groups the difference in price relatives for individual commodities

will be small, the changes in relative purchases will be small, and the correlation betweensuch changes and the individual price relatives will be small.

'The White Paper conspicuously refrains from showing price relatives for individualfood groups itselí, and hence these series should be used with caution. However, theynidicate the right order of magnitude, as indicated by a comparison with relatives calculated

li«erent basis and shown in Butiarni, Vol. 10, p. 260. Booklet, p. 46.

Page 11: THE COST OF LIVING, 1938–1948

we weight these series by consumer expenditure in ¡938 we have anapproximation to a Laspeyre index for the total (since for food sub-groupsLaspeyre and Paasche type indices should not differ greatly). These arecompared below with the Paasche type indices calculated directly from theWhite Paper totals for food

Index for Ioal food cxpcizditure (1938=100)

Lack of information prevents our making approximate estimates ofLaspevre indices for earlier years, for which we have only Paasche. We can,however, indicate the magnitude of the difference between the two types ofindices. If we examine the food price relatives for 1946 fl the table above, itis noticeable that there are really two classes of sub-groups, those whichhave increased in price by 50-70 per cent and the remainder which haveincreased less than 40 per cent, and if we exclude Beverages less than 30per cent. The first class consists of Bread and Cereals (which include cakeand biscuits), Sugar Products, Fruit, Potatoes and Vegetables, and mis-cellaneous manufactured foods, and there is little reason to doubt that thesame sub-groups would show substantially larger price rises for the waryears too. Now the difference between Paasche and Laspeyre indices dependsprimarily on the proportionate weights attached to the groups whose priceshave risen most rapidly, i.e. largely on the percentage of total food expenditurethat is spent on this class of foodstuffs which have become relativelydear. Now this percentage can be calculated from the National IncomeWhite Papers.I It has been as follows :-

No division of food expenditure shown.

If we'look first at the figures for 1946, 1947 and ¡948 taken from Cina'. 7649we can see that a sharp increase in the proportion of expenditure on the class

'1 These figures refer, of course, only to household food expenditure. They would not showa greatly different story for total food expenditure, to which the conclusions refer. Thechanges from one White Paper to another are largely changes in the scope of the food sub-groups.

From 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948Cmd. 7099 43 s 48 48 49 48 47 47 49 - -Cmd. 7371... 42 45 47 48 55 -Cmd. 7649... 41 48 54 54

THE COST OF LIVING, 1938-1948Food Price Series (1938=100)

'37

Sub-Group 1946 1947 1948Bread and cereals 165 176 180Meat and bacon 123 132 133Fish 129 133 138Oilandfats 119 113 111Sugar products 162 176 213Dairy products 128 131 150Fruit 158 186 188Potatoes and vegetables 160 1S6 197Beverages 139 148 155Other manufactured food 164 167 158

1946 1947 1948Paasche Index 141.5 153.9 162.8Approximate Laspeyre Index 140.7 151.1 159.7

Page 12: THE COST OF LIVING, 1938–1948

138 THE BULLETIN

of relatively dearer foods occurred after 1946, paralleling the rise in thedifference between Paasche and Laspeyre indices from nearly i point in 1946to z.8 points in 1947 and 3.' points in 1948, as shown above. It can also beseen that before 1946 the proportion of expenditure on foods whose pricehad increased most was at all times below the 1946 proportion.' This indi-cates that the difference between the Paasche and Laspeyre indices in thewar years would also be below the estimated 1946 difference of o.8 points.This result is on other grounds not unacceptable. It was not until after thewar ended that these relatively dearer foods became at all readily available.However, the rise in the proportion since 1938 indicates that the Paascheindex would be slightly higber throughout, and the adjustment made hasbeen to reduce the Paasche index by 0.5 points throughout to obtain approxi..mate Laspeyre index numbers.

TABLE VIIApproximations to Laspeyre National Price Îndices (1938= 100)

Nationni Income White PaperGroup 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948

I Food ... 103 119 128 130 130 135 139 141 151 1602 Drink ... 104 136 160 202 225 240 230 245 250 2893 Tobacco .........112 146 161 202 235 250 251 257 335 3904 Rent and Rates ... 101 102 103 103 103 103 103 107 111 1125 Fuel and Light ......100 111 117 123 141 136 142 146 150 1636 Durable Household Goods ... 102 129 173 218 215 216 212 207 220 2397 Other Household Goods ... 102 114 122 127 128 140 142 138 156 1658 Clothing ... ......103 133 165 179 174 180 185 185 189 2109 Reading Matter ......100 106 111 115 117 116 117 115 18 119

10 Private Motoring ......101 126 137 135 138 138 148 161 170 18011 Travel .........100 108 114 114 114 114 113 116 128 13312 Communications ......100 122 138 126 129 129 130 163 134 13513 Entertainments ......100 115 116 136 157 166 171 176 176 17114 Other Services ... 101 111 117 121 124 129 136 142 149 15315 Other Goods .........102 115 144 169 173 178 191 202 221 23216 Income in kind of H.M Forces 104 119 129 130 133 136 140 142 154 16319 Foreign Travel 252 260 260

* Data for 1939-43 from C,nd. 7099, for 1944-5 from C,nd. 7371, and for 1946-8 fromCmd. 7649. Hence ail the series may not be fully comparable throughout.

t Includes foreign travel.

The approximation to Laspeyre-type indices so obtained are shown inTable VII. The series for groups lo, 13, 14, 15, i6 and 19 are pure Paascherelatives which, however, are treated, for lack of alternative, as Laspeyrerelatives in these calculations.

In 1942 the proportion was 48.6, compared to 49.3 in 1946.

Total ... 102 119 133 143 148 152 156 160t 171t 184t