13
Household Hamilton, Expenditure Patterns: New Z eal1and"t Very little is known about the pattern of comumer expenditure in New Zealand, and still less about income (or total expenditure) elas- ticities of demand for different items. The object of this paper is to give some information about consumer expenditure in the city of Hamilton on the basis of a family budget survey carried out in October 1968.l Data were collected from 50 families covering income, in- dividual outlay on various items, and demographic and financial char- acteristics? Section I describes some of the economic characteristics of house- holds covered by the survey. Section I1 ishows the monthly distribu- tion of household incomes and expenditure on different items. In Section I11 a multivariate relation is estimated for this particular sample. In Section I V estimates are given of the total expenditure elasticities of demand for nineteen items; the marginal propensities to consume and save are also calculated. Finally, Section V considers the validity of the hypotheses and inferences to be drawn from the analysis. I. Some Economic Characteristics of Households Covered A total of 46 out of 50 households replied to the questionnaire-a 92 per cent rate of response. The number of reliable replies received was 40.s Despite the smallness of the sample (a fact that needs to be kept continuously in mind when interpreting the hdings of the study), the range covered was wide enough to permit statistical analy- sis. * This research was financed by the University of Waikato. The interviewing of households was carried out by students of the School of Social Sciences at the University. f I wish to express my thanks to Professw J. T. Ward and to Mr C. Pugh of the Department of Economics at the Universia of Waikato. I wish also to thank Professor A. R. Bergstrom of Auckland University for his valuable suggestions and comments. Any remaining errors are, of cours;e, my own responsibility. A copy of the questionnaires circulated to households is available on demand. zThe families were chosen at random. Four families refused to reply and six were excluded on grounds of the inadequacy of their response. It is interesting to note that poorer households were more co-operative than richer. 73

Household Expenditure Patterns: Hamilton, New Zealand,

Embed Size (px)

Citation preview

Household Hamilton,

Expenditure Pat terns: New Z eal1and"t

Very little is known about the pattern of comumer expenditure in New Zealand, and still less about income (or total expenditure) elas- ticities of demand for different items. The object of this paper is to give some information about consumer expenditure in the city of Hamilton on the basis of a family budget survey carried out in October 1968.l Data were collected from 50 families covering income, in- dividual outlay on various items, and demographic and financial char- acteristics?

Section I describes some of the economic characteristics of house- holds covered by the survey. Section I1 ishows the monthly distribu- tion of household incomes and expenditure on different items. In Section I11 a multivariate relation is estimated for this particular sample. In Section IV estimates are given of the total expenditure elasticities of demand for nineteen items; the marginal propensities to consume and save are also calculated. Finally, Section V considers the validity of the hypotheses and inferences to be drawn from the analysis.

I . Some Economic Characteristics of Households Covered A total of 46 out of 50 households replied to the questionnaire-a

92 per cent rate of response. The number of reliable replies received was 40.s Despite the smallness of the sample (a fact that needs to be kept continuously in mind when interpreting the hd ings of the study), the range covered was wide enough to permit statistical analy- sis.

* This research was financed by the University of Waikato. The interviewing of households was carried out by students of the School of Social Sciences at the University.

f I wish to express my thanks to Professw J. T. Ward and to Mr C. Pugh of the Department of Economics at the Universia of Waikato. I wish also to thank Professor A. R. Bergstrom of Auckland University for his valuable suggestions and comments. Any remaining errors are, of cours;e, my own responsibility.

A copy of the questionnaires circulated to households is available on demand. zThe families were chosen at random. Four families refused to reply and six

were excluded on grounds of the inadequacy of their response. I t is interesting to note that poorer households were more co-operative than

richer.

73

74 THE ECONOMIC RECORD MARCH

Table I shows distribution of families by size, families of three to five persons accounting for three-quarters of total interviews. The arithmetic mean of family size in this sample is not very different from that found in an earlier social study in Hamilton4

Of the households covered by the sample, 65 per cent owned their own accommodation while 20 per cent had one or more full-time boarders. The average number of rooms occupied was 3.75 per household.

Five household heads were receiving pensions of one sort or another. Twelve members had more than one job. Table I1 gives the occupational classification in major groups : using the chi-square test it was found that the distribution in this table agrees with the national distribution at the 0.01 level of ~ignificance.~

There were 56 wage-earners in the sample-35.82 per cent of total numbers. It was found that larger families usually contained more earners.

TABLE I Dktm’butwn of Families b y dize

Size of family 1 No. of families I Per cent

1 2 3 4 5 6 7

0 7 10 8 12 2 1

0.0 17.5 25.0 20.0 30.0 5.0 2.5

Total I 40 I 100.0

Note: Arithmetic mean of family size = 3.875 * 0.406. (Student’s t distribu- tion was used to estimate the population mean with 95 per cent confidence co- efficients.)

TABLE I1 Occupational CZawification in Major aroups

Major group 1 % of total

Professional, technical & administrative workers Clerical workers Sales workers Farmers, fishermen, hunters, miners & related workers Workers in transport & communications Craftsmen, production process workers & labourers n.e.i. Services, sport & recreation workers Not classified

14.7 ii.9 8.0

13.8 6 .1 36.7 6.2 1.6

Total I 100.0

4 See ‘Survev of Hamilton’. School of Social Sciences. Victoria Universitv of < -

Wellington. 5 See New Zealand Ojicial Yearbook B67, p. 947.

75

Income (or total expenditure) and family size tended to be sig- nificantly correlated. The simple correlation coefficient between h- come and family size was 0.3347, and thah between expenditure and family size 0-4307 ; both are significant at the 0.5 level of significance.

Table III shows the age structure of household members. Using

1970 HOUSEHOLD EXPENDITURE PATTERNS

Age group (years) total members

No. of members -

0-9 10% 20-29 30-39 40-49 50-59 60 & over

39 27 19 20 17 13 20

25.16 17.42 12.26 12.90 _- _.

, 1o.w

I 12.90 8.39

Total 1 155 - 1 100.00

the chi-square test it was found that this distribution agrees with the national distribution revealed in the 1961 population census at the 0.01 level of signi.6cance.6

Members of households were also classified according to their educational attainment. It turned out that only 8 per cent of those aged 15 years or over had received higher education (i.e. were mi- versity graduates) ; 58 per cent had received secondary education (in- cluding technical education) and 34 per cent merely primary educa- tion.

Families were asked about the contribution of teenagers to family housekeeping. The average contribution per week was found to be about $6.75, with a wide variation from one family to another.

Table IV shows the distribution of households according to monthly disposable income.? It can be seein that the top 20 per cent of households received about 35 per cent of total income.

;TI. Household Expenditure A family budget survey generally provides information on ex-

penditure and income for a number of families during some fhed period of time. The data covered in the present study related to one month. However, the questionnaire was designed to get information not for that particular month but for ‘a month’.

One basic assumption made in the analysis of cross-section data is that Meren t people are homogeneous except for the dserences in the variables being considered and a raindom error. This is not an

Family (monthly) disposable incomes rangie from $150 to $800, the household 6 See ibid., p. 84.

average being $301.62.

76 THE ECONOMIC RECORD MARCH

TABLE IV Distribution of Households by Size of Monthly Disposable Income

7 12 7 2 4

Monthly income $

17.5 9.82 30.0 21.56 17 .5 1 5 . 4 6 4.0 5.40

10.0 12.23

150-199 200-249 250-299 300-349 350-399 400-449 450-499 500 and over

Total

"14 of total income

Households

No. I $6

2 5.0 6.83 8.04

4

altogether valid assumption. I n family budget surveys dealing with in- dividual family units there are many variables that must be taken into account in the statistical estimation of families and such variables may change only gradually over time. Economic applications of estimated demand functions which are also averages in their final form may consequently attach little significance to these factors. They can, however, be quite important in obscuring the underlying relation between expenditure and income because such demographic and other personal characteristics do vary widely among individuals in a single cross-section sample.

Table V shows the average family income and expenditure per month in Hamilton. It also shows the pattern of family monthly ex- penditure and the percentage of income (and total expenditure) allotted to each of the major groups. It is readily apparent that the largest item of expenditure is food (including non-alocoholic

TABLE V Household Monthly Expenditure

Items

Food and non-alcoholic drinks Alcoholic drinks and tobacco Apparel and personal accessories Household durables Housing (incl. rent, fuel and light,

household operations) Transportation and communication Medical and personal care Education, culture and entertainment Other goods and services savings

s

61.57f5.2 20.06k2.3 35.31 f3.8 26.84k3.2

58.91 k6.3 47.73&7.6 4.70f0.51 15.49k4.2 6.09&1.8 27.42 f8.7

% of 1 total expenditure

22.25 7.25 12.76 9.70

21.29 17.25 1.70 6.60 2.20 -

I 276.70f36.9 I 100.00 Total

% of total

income

20.41 6.65 11.71 8.90

19.53 15.82 1.56 5-14 2.02 8.26

100~00

Note: Student's t distribution was used to estimate the population mean with 95 per cent confidence coefficients.

1970 HOUSEHOLD EXPENDITURE PATTERNS 77

drinks). The second item is housing (including rent, imputed rent, fuel and light, and household operations). The third biggest item is transportation and communication. Families in Hamilton spend more on alcoholic drinks and tobacco than on education, culture and entertainment. (Education here refers only to the private sector.) Net personal savings average just over 8 per cent of disposable in- come.

Table VI shows the pattern of consumer expenditure by five broad categories in a number of countries. I t would seem that the New

Food

23.7 23.7 29.4 30.3 31 -1 46.6

TABLE VI

Beverages andtobacco

5.7 8 .3 9 .2 8 - 1

14-1 10.7

Structure of Consumer Expenditure in Sedected Cvu?&t&es, 1950-59 (percentages)

Clothing

10.1 10-2 13.9 16-7 11.7 11-6

- Samples

from Housing

15.7 21.2 13.9 10.2 12.8 5.2

U.S.A. Canada Sweden Norway U.K. IMY

Others

44.8 36.6 33-6 34.7 30 -1 26.0

Source: S . Kuznets, ‘Quantitative Aspects of the Economic Growth of Nations: VII’, Economic Developmewt and Cultural Change, Vol. X, January 1962.

Zealand pattern of expenditure is quite similar to that of other coun- tries.

An important factor which affects household expenditure is family size. We have classified the sample into Merent groups ac- cording to family size and examined the expenditure-income re- lation in each group.* Broadly speaking, the figures indicate a regular growth of expenditure on non-durables a,s income increases and as family size increases. They show a regular growth of expenditure on durables as income increases, and a slight decrease of expenditure on durables as family size goes up. They also show a regular growth of expenditure on services and family savings as income increases, and a regular decline as family size expands.

The statistics also reveal that the peircentage spent on non-dur- ables falls &s income increases but rises as family size increases, and that the percentage spent on durables remains constant or rises slightly as income increases and shows a slight expansion as family size expands. Finally, the percentage spent on services and the per- centage of savings both show a regular growth as income increases and a regular decline as family size becomes larger.

We have also calculated the average expenditure on sub-groups among all consumer units within a given income class when the se-

a Interested readers may obtain further statistics relevant to this section by contacting the author.

MARCH 78 T H E ECONOMIC RECORD

lected units are restricted to a given size (two, three, four, five, or six or more persons). The figures show a regular growth of expenditure on all the different items as total expenditure increases; a regular growth of expenditure on food, alcoholic drinks and tobacco and m e d i d and personal care as total expenditure increases; a regular decline of expenditure on housing, education, culture and entertain- ment as family size increases; a slight decline of expenditure on ap- parel and personal accessories and transportation and communication as size increases; and a slight growth of expenditure on household durables as size increases.

As to the pattern of household expenditure on sub-groups, it can be concluded from the Gndings of the study that : the percentage spent on food and non-alcoholic drinks declines as income increases and rises as family size goes up; the percentage spent on apparel and household durables rises slightly as income increases but remains more or less the same regardless of family size; the percentage spent on housing decreases with expanding income and family size; the per- centage spent on transportation and communication rises as income increases, while decreasing very slightly the greater the family size; the percentage spent on education, culture and entertainment rises as income increases and falls as family size increases; the percentage spent on medical and personal care decreases slightly as income in- creases but goes up as family size increases ; and the percentage spent on alcoholic drinks and tobacco does not show a regular pattern in relation to income but rises with family size.

We may conclude this section by developing the following ex- planations for these results :

(i) Large families economize in the consumption of necessities, such as food, by buying in ‘family sizes ’ and perhaps by pur- chasing inferior quality goods.

(ii) Family size seems to have no significant effect on the ac- quisition of some durables. The purchasing of such items as furniture and certain other durables seems to follow a pat- tern related to the age and marital status of the head of the household. Large families will be faced with other demands and will normally hold reasonable stocks of those com- modities.

(iii) Large families are more likely to substitute domestic services for the purchase of the services on the market. The demand for labour-saving devices in the home is a function of the family composition as well as of its income.

(iv) Consumption of alcoholic dr inks and tobacco depends more on tastes and other personal (non-measurable) factors than on income. But large families with numbers of adults are likely to spend more on alcoholic drinks and tobacco than small families or than large families with many children.

HOUSJCEOLd EXPENDITURE PATTERNS 79 1970 (v) Clothing in New Zealand is not generally considered a

luxury item despite its high price. Part of the explanation lies in the fact that New Zealanderrs do not need all that much clothing. Style and taste in dress affect consumer behaviour to a large extent in this respect.

I I I . Partial Elasticities for Expenditure Groups with Bespect to Total Expenditure and Family Size

The principal criterion by which the similarity of expenditure patterns in different countries is usually judged is the elasticity of particular items of expenditure With respect to total expenditure. In what follows this elasticity will often be ireferred to briefly although not entirely accurately as the income elasticity.

It is essential to choose an adequate .mathematical form for the relation between particular expenditure and total expenditure, for calculated income elasticities depend on the type of function fitted. Three types of formulae are available to choose from: linear, semi- logarithmic and double-logarithmic. Linear functions do not provide an adequate fit ; the semi-logarithmic was recommended by Prais and Houthakker for necessities? I n the present study, however, the double-logarithmic function was preferrerd.10 Expenditure on each item and total expenditure were expressed as logarithms (to the base e ) . By using the method of least squares, the following equation was estimated for each item :

where Y4 = expenditure on the ith item j X = total expenditure ; Z = family size ; ar, pi and yc are constants to be estimated ; and Ud is a dis- turbance term. The logarithmic relationship has the property of con- stant elasticity. The coefficient p is the totarl expenditure elasticity and the coefficient y is the partial elasticity with respect to family size. Households were cross-classified by total expenditure and family size. This overcame the diffhdty arising from the fact that total expendi- ture and family size are significantly correlated.

Table VII summarizes the results for ten major groups and sub- groups of expenditure. In addition to the coefficients of elasticities, the table gives the partial correlation coefficients between expenditure and family size. It also gives the coefficient of multiple determination.

It can readily be seen that the coefficient of multiple determina- tion is quite high in all cases, which means that expenditure on each item may be explained to a considerable extent in terms of changes in total family expenditure and family size.

s. J: Prais and H. S . Houthakker, The AiwolysiJ- of Family Budgets (Carn- bridge University Press, London, 1955).

lo See H. S. Houthakker, ‘The Econometrics of Family Budgets’, / o w m l of the Royal Statistical Society (Series A), Vol. CXV, Part I, 1952, for the ad- vantages of using the double-logarithmic function. Some of these advantages are good fit, ease of computation and automatic correction for heteroscedasticity.

log Y4 = ai + pc log x + y4 log z + 0 4

TABL

E VII

Re

latio

nshi

p be

twee

n Fa

mily

Exp

endi

ture

on D

iffer

ent I

tem

s, T

otal

Eltl

mily

Exp

endi

ture

0.97

48

0.9816

0~

09

80

Item

0.89

73

0~

18

S6

-0

.952

0

Non

-dur

able

8 D

urab

le8

Serv

ices

Food a

nd n

on-a

lcoh

olic

0'

9788

0.98

57

0.98

56

0.97

42

0.95

38

0.94

37

0 ' 98

38

drinks

App

wel

and

per

sona

l ac

cess

orie

s H

ouse

hold

dur

able

s H

ousl

ng :

rent

, fue

l and

lig

ht. h

ouse

hold

op

erat

lons

T

rans

port

atio

n an

d co

mm

unic

atio

n M

edic

al &

per

sona

l car

e E

duca

tion,

cultu

re an

d en

tert

ainm

ent

0.87

92

0.04

49

-0.0

646

-0,6

355

-0~1800

0,40

54

-0.9

068

I I

I I

I

0.09

73

0'07

58

0.82

70

0.02

33

0.02

70

0.00

001

--

0.04

6 0.

0208

0.

0216

0.

0323

0.

0333

0.

062

0.02

91

0'03

01

0,04

9 0.

0211

0.

0218

1.04

48

0.00

59

0.06

2 0.

0265

0,

0274

1.

0556

-0

.009

0.

066

0,02

82

0.02

92

0.78

42

-0.1

548

0.05

7 0.

0243

0.

0251

1.44

29

-0.0

609

0.15

80

0.06

74

0.06

97

0.88

23

0.14

19

0.10

75

0.04

58

0.04

74

2.37

83

-1.2

848

0.16

41

0.06

57

0.06

80

Not

e: S

ee te

xt f

or e

xpla

natio

n of

form

ula

adop

ted

and

mea

ning

of s

ymbo

ls.

1970 HOUSEHOLII EXPENDITURE PATTERNS 81

The income elasticity for food is significantly less than one, as should be the case with necessities if Engel’s Law about the decline in percentage spent as income rises is valid. ‘The elasticity for apparel and personal accessories is greater than unity, but not markedly SO. In the technical sense, apparel is therefore a lluxury although only to a very moderate degree. The same applies to household durables. For housing (which includes fuel and light, and household operations) the elasticity is less than one, and for medical care just under one : housing and medical care are thus necessities in the technical sense. Income elasticities for transportation and communication, and also for education, culture and entertainment, are well above one.

Table VII indicates that the partial elasticities for food and medical care with respect to family size are significantly positive. The coefficient of family size shows the influence of economies of scale, an influence which operates regardless of the method of measuring the sizell of the family and of the type of expenditure considered.

The partial elasticities for transportation and communication, housing and education, culture and entertainment are significantly negative. The partial elasticities for apparel and household durables are not significant: the elasticity for apparel is positive and for household durables negative.12

Family size affects consumer behaviour in two ways: (a) As the family increases in size there will be more need for all commodities, and so expenditure on different items can be expected to increase. This is a positive effect. (b) As the family increases in size, members of households become relatively poorer if total family income is held constant. This is an income effect which is negative. The family may be forced to cut its expenditure on items which are considered luxuries in order to increase its expenditure on the others which are considered necessities. If the income effect is stronger than the family effect, yr will be negative ; otherwise it will be positive.

IV. Individual Expenditure and Estimation of Income Elasticities of Demand fin- Certain. Itemsla

Average expenditure per head as well as the pattern of in- dividual expenditure are of great importance in studying the struc- ture of consumer expenditure in any country. The data collected in our survey permit us to estimate average monthly expenditure per head on twenty sub-groups of items (Table VIII) and the pattern of individual expenditure for different individual incomes.

llFamily size is measured by the number of persons without any weighting according to age or sex.

12The elasticities with respect to family size are related to each other by an identity, just as the elasticities with respect to total expenditure. Since the YC are partial elasticities, the total effect of a change in family size on all expenditure must be zero and hence zy4YC = 0. Consequently it is not possible for the 7 4 to be all positive or a11 negative. See Houthakker, op. cit.

l3 Further statistics relevant to this section may be obtained by contacting the author.

82 THE ECONOMIC RECORD MARCH

It can be seen from Table VIII that rent and transport occupy the biggest share of the individual's monthly total expenditure, f ol- lowed by apparel, household durables, alcoholic drinks and tobacco, meat, and entertainment.

The figures show that the sum spent on each item rises as income increases although at different rates. The percentage of individual expenditure on noa-durables falls significantly as income increases, that on durables stays more or less the same or increases very slightly, while that on services rises substantially. The percentage spent on non-durables decreases from 42 per cent to 26 per cent as total ex- penditure expands from $39 to $170, that on durables rises from 20.8 per cent to 22.37 per cent, while that on services goes up from 36-89 per cent to 51.63 per cent.

The concept of elasticity is essential for forecasting consumption ; yet virtually nothing is known about the income (or total expendi- ture) elasticity of demand for different items in New Zealand. To help remedy this, total expenditure (income) elasticities of demand fo r nineteen groups of commodities have been estimated from our survey using five different f0 rm~lae . l~ The use of different formulae gave different results but not in the technical sense.15

The following conclusions can be drawn : (i) According to all formulae, all food items except biscuits and

cakes have income elasticities of demand of less than unity and can therefore be considered necessities in the technical sense.

(ii) According to all the formulae, alcoholic drinks and tobacco, soaps and detergents, fuel and light, medical care, and rent have income elasticities of less than unity and are also neces- sities in the technical sense.

(iii) According to all the formulae with the exception of (3) , ap- parel and household durables have income elasticities of more -although not much more-than one. Thus in a technical sense they are luxuries but only to a moderate degree.

(iv) Education, culture and entertainment, transportation, com- munication, and household operations have income elasticities of more than one and so fall technically into the luxury class.

The formulae are arranged in Table I X according to adequacy. It can be seen that formulae (1) and (2)-simple regression and double-

14 These formulae (employing the notation already used) are : (1) yC = a4 + p a X + U'. (2) y 4 = a c X / 3 , + U c . (3) YC = a4 -& + UC.

X (4) yC = e a* - W S +uC. (5) y~ = a4 + BC log, X + UC.

15 The application of different formulae did not change the elasticity co- efficient in such a way as to make the item considered a necessity instead of a luxury or the other way round.

83 1970 HOUSEHOLn EXPENDITURE PATTERNS

TABLE VIII Average Monthly Expenditwe Per Head on Different I t e m , 1968

Items

Non-durables Cereals Sugar & sweets Biscuits & nuts Vegetables Fruits Meat, fish, eggs & poultry Dairy products & fats Non-alcoholic drinks Alcoholic drinks &tobacco Soaps & detergents Fuel & light Others Total food Total non-durables Durables Apparel & personal accessories Household durables Total durables Services Medical & personal care Education, culture & entertainment Transportation Communication Rent Household ooerations Others Total services Total

1.28 2 0.07 0.80 f 0.06 0.74 f 0.14 2-78 f 0.13 2.022 0.25 5.06 .t 0.62 2.12 f 0.09 0.43 f 0.03 5.77 -t 0.64 1.10 -t 0.05 2-32 f 0.25 1-08 16.3% 25.55

9.38 -t- 1.28 7.56 f 1.00 16.94

1.28 f 0.15 4.45 f 1-41 11-32 f 2-05 1.54 f 0-28 11.79 f 1.43 2.71 & 0.80 0.53 33.62 76.11

%

1.68 1.05 0.97 3.66 2.66 6.66 2.79 0.63 7-59 1.45 3.05 1.42 21 55 33.61

12.32 9-93 22.25

1.68 5.85 14.89 2.03 15-51 3.58 0.70 44.24 100~00

Note: Student’s t distribution was used to estimate the population mean with 95 per cent confidence coefficients.

logarithmic-are more adequate than the other three. Formula (1) is more adequate than all others in seven case13 out of nineteen and comes second best in six instances. Formula (2) is more adequate in seven cases and second best a further seven timea. Formula (3)-reciprocd transformation-is more adequate in only one case (soap and deter- gents) and second best in another (dairy products). Formula (4)- logarithmic, reciprocal transf ormation-is more adequate in one in- stance (dairy products), second best in three cases and third in seven others. Finally, formula (5) -semi-logarithmic transformation -is more adequate in three cases, second best twice and third seven times. So we may rank the five formulae according to adequacy as:

Table X presents estimates of the marginal propensity to consume at various levels of income, employing the same five formulae. Again there are substantial differences in resultrs according to the formula used. Ranking the formulae according to the same criterion of ade- quacy, we have : (2), (l), (4), (5) , and (3)l.

(21, (11, (51, ( 4 > , a d (3>.16

16For an explanation of ranking, see note tab Table IX.

84 THE ECONOMIC RECORD MARCH

TABLE IX Comparison of Estimates o f Expenditure Elusticities Obtained

Item

Cereals Sugar Biscuits Vegetables Fruits Meat Dairy products Non-alcoholic drinks Soaps and detergents Alcoholio drinks and

Fuel and light Apparel Household durables Medical & personal care Education and culture Transportation Communication Rent Household operations

tobacco

from Different Pormulae*

Estimates of expenditure elasticities

0-3622 0-5601 1-3146 0.3472 0.8352 0.8682 0.3011 0.4433 0.3109

0.7467 0.6093 1.0360 1.0211 0.8247 2-3502 1.2649 1.3127 0.7391 2.3029

* See footnote 14 for the det:

0.361 76 0.60599 1.41790 0-37696 0.92585 0.89872 0.35260 0.48280 0.39538

0.69429 0.68913 1 *01610 1.02890 0.76780 2.42732 1.29335 1.18670 0.86226 1.97883

(3)

0.4914 0.5489 1.2652 0.3444 0.8336 0.8413 0,3295 0.4469 0.3538

0.6511 0.6133 0.9488 0.9344 0.7975 1.7889 1.1726 1'1394 0.7672 1.8848

s of the formulae.

0.3097 0.5386 1 .2980 0.3415 0.8493 0.8169 0.3334 0.4371 0.3754

0.6011 0.6553 0.9256 0.9351 0.7837 2.2595 1.1826 1 * 0584 0.8156 1.7661

( 5 )

0.3864 0-6766 1.3299 0.3748 0.9336 0-9500 0.3532 0.5014 0.3752

0.7816 0.6766 1 .0982 1.0831 0.9048 2.4260 1-3561 1.3610 0-8370 2-3064

Ranking of formula

adequacyt by

?Formulae were ranked according to the value of ~ ( Y c - YC est)2. The most adequate equation is the one with the lowest value.

TULE X Estimation of Marginal Propensity t o Consume hy Different

FO9-!mUla

MontNv income I Marginal propensity to consume

50 75

100 125 150 I 0.81 1 0.76 I 0.25 175 0.81 0.75 0.18 200 1 0 - 8 1 1 0.74 1 0.14

V . Conclusions

(4)

1.43 0.96 0.72 0.57 0.48 0.41 0-36

1.25 0.84 0.59 0.42 0.32 0.36 0.20

This study is a first step along an endless road-and a very narrow road at that.

A basic limitation was imposed by the smallness of the sample covered by the study. Had the sample been larger, it might have been possible to widen the coverage of the study in several respects :

1970 HOUSEHOLD’ EXPENDITURE PATTERNS 85

(i) Stratification is a common feature o € household surveys. The totality of sampling units includela in the frame can be divided into groups or strata, each stratum being sampled separately and independently, so that a specified number of sampling units or a specific fraction of all such units can be obtained from each stratum. Stratification also ensures efficient estimation of parameters €or the population as a whole. Unfortunately the smallness! of the sample covered by this study made stratification impossible.

(ii) The smallness of the sample made it difficult to obtain rela- tionships for detailed items of expeioditure and represents a major obstacle to forecasting the future consumption of these items.

(iii) The enlargement of the sample would probably not have given different coefficients but would certainly have reduced the sampling errors.

The other limitation of this study is imposed by the peculiar characteristics of Hamilton. With its high ra,te of mobility and rapid expansion, Hamilton is not an altogether typical New Zealand city. It is thus questionable whether the findings3 have relevance for the whole country.

In spite of these limitations, and keeping in mind as well that family differences are substantial and that large numbers of variables must be allowed for to measure the relevant income elasticities pro- perly, the results of this study nevertheless deserve to be treated seriously. As well as demonstrating some significant relationships between family composition and consumer behaviour, the study could perhaps provide a weighting pattern for a cost-of-living index in New Zealand based on the pattern of household expenditure (in so far as the conclusions can be generalized over the whole society). Moreover, the coefficients of elasticities could possibly aid in the formulation of development programmes in the Hamilton area and particularly in the forecasting of demand fo r certain groups iin the near future.

U n k r s i t y of Waikato M. M. METWALLY