13
ADVERTISING AND THE PROPENSITY TO CONSUME* M. M. Metwally and H. U Tamaschke INTRODUCTION Advertising was not mentioned by Keynes as one of the objective factors determining the propensity to consume.1 This exclusion, which might have been justified at the time The General Theory was written, cannot be taken for granted in recent years where advertising is playing an increasingly important role in the economic life of almost every market economy. Australia, for example, spent over one billion Austra- lian dollars on advertising in 1979. As a proportion of national income, advertising expenditure in this country averaged approximately 1.5 per cent during the period 1960-78. Claims are often made that most, if not all, these expenditures are an economic waste in the sense that they do not benefit the consumer. Some go even further and suggest that advertising results only in a reallocation of the market positions of existing firms with no effect on total consumption or the propensity to consume. In other words they claim that the informative role of advertising is not strong enough to motivate potential customers to expand their income (through increased effort) in order to finance increased consumption expenditure. Despite these claims and despite the importance of evaluating the role of advertising in modern economies, literature in this field has been almost neglected? *The authors are grateful to Garnsey Clemenger Pty. Ltd., Brisbane, Australia for supplying the necessary data on advertising and to the BULLETIN's referees for their comments. 'Keynes listed six objective factors: changes in the wage (and price) level; changes in accounting practice with respect to depreciation; windfall gains or losses; changes in fiscal policy; changes in expectations and changes in the rate of interest. See Keynes (1936), The General Theory of Employment, Interest and Money, pp. 91-5. The bibliography to this paper provides additional references to the consumption function literature. 2 As far as the authors are aware there have only been three studies in the area in the last decade. Taylor and Weiserbs (1972), examined the relationship between levels of total ad- vertising and total consumption using only the 'Houthakker-Taylor Model' of consumer be- haviour. In a recent article Ashley, Granger and Schmalensee (1980) use Box-Jenkins methods to analyse relationships between levels of aggregate consumption and levels of aggregate ad- vertising only. No other economic variables, including the a priori important income are con- sidered. Earlier, one of the co-authors, R. Schmalensee (1972), had also experimented with levels of aggregate consumption and levels of aggregate advertising in 'incomplete' models. In addition, none of the studies attempted disaggregation to examine the effects of advertising on different commodity groups on the consumption function. The lack of disaggregation prevented 273

ADVERTISING AND THE PROPENSITY TO CONSUME

Embed Size (px)

Citation preview

Page 1: ADVERTISING AND THE PROPENSITY TO CONSUME

ADVERTISING AND THE PROPENSITYTO CONSUME*

M. M. Metwally and H. U Tamaschke

INTRODUCTION

Advertising was not mentioned by Keynes as one of the objectivefactors determining the propensity to consume.1 This exclusion, whichmight have been justified at the time The General Theory was written,cannot be taken for granted in recent years where advertising is playingan increasingly important role in the economic life of almost everymarket economy. Australia, for example, spent over one billion Austra-lian dollars on advertising in 1979. As a proportion of national income,advertising expenditure in this country averaged approximately 1.5 percent during the period 1960-78.

Claims are often made that most, if not all, these expenditures arean economic waste in the sense that they do not benefit the consumer.Some go even further and suggest that advertising results only in areallocation of the market positions of existing firms with no effecton total consumption or the propensity to consume. In other wordsthey claim that the informative role of advertising is not strong enoughto motivate potential customers to expand their income (throughincreased effort) in order to finance increased consumption expenditure.Despite these claims and despite the importance of evaluating the roleof advertising in modern economies, literature in this field has beenalmost neglected?

*The authors are grateful to Garnsey Clemenger Pty. Ltd., Brisbane, Australia for supplyingthe necessary data on advertising and to the BULLETIN's referees for their comments.

'Keynes listed six objective factors: changes in the wage (and price) level; changes inaccounting practice with respect to depreciation; windfall gains or losses; changes in fiscalpolicy; changes in expectations and changes in the rate of interest. See Keynes (1936), TheGeneral Theory of Employment, Interest and Money, pp. 91-5. The bibliography to thispaper provides additional references to the consumption function literature.

2 As far as the authors are aware there have only been three studies in the area in the lastdecade. Taylor and Weiserbs (1972), examined the relationship between levels of total ad-vertising and total consumption using only the 'Houthakker-Taylor Model' of consumer be-haviour. In a recent article Ashley, Granger and Schmalensee (1980) use Box-Jenkins methodsto analyse relationships between levels of aggregate consumption and levels of aggregate ad-vertising only. No other economic variables, including the a priori important income are con-sidered. Earlier, one of the co-authors, R. Schmalensee (1972), had also experimented withlevels of aggregate consumption and levels of aggregate advertising in 'incomplete' models. Inaddition, none of the studies attempted disaggregation to examine the effects of advertising ondifferent commodity groups on the consumption function. The lack of disaggregation prevented

273

Page 2: ADVERTISING AND THE PROPENSITY TO CONSUME

274 BULLETIN

This paper attempts to examine the relationship between advertisingintensity (as defined in Section I) and the propensity to consume. Thehypothesis tested here is that informative advertising would causeshifts in the propensity to consume. This would happen if: (a) theadvertising messages persuade potential customers to cut down ontheir rate of savings in order to finance advertising-induced consump-tion expenditure and/or (b) the messages persuaded potential customersto run into debt to finance their expenditure. In both cases, a higherlevel of consumption will be associated with a given level of incomedue to the roles of advertising. Advertising messages which are directedat attracting existing customers to a particular brand (i.e.competitiveadvertising) and those which aim at protecting a particular brand'smarket share (i.e. defensive advertising) could be expected not to havemuch significant effect on the community's propensity to consume.

The relationship between advertising and the propensity to consumeis, however, a dual one. An increase in the propensity to consume islikely to cause an increase in advertising expenditures for two reasons:(a) the increase in consumption can attract new entrants into themarket. These newcomers may use advertising as a means of com-petition to establish themselves; (b) existing firms may try to maintainor even expand their market shares in the face of expansion in con-sumption. These two effects would be more important in situationscharacterized by non-price competition.

This paper has four sections. In Section I, a single-equation modelis developed to test the relationship between the propensity to con-sume and total national advertising intensity. Section II considers asimultaneous-equation model where the interrelationship between thetwo variables is investigated. In Section III advertising expenditures aredisaggregated according to different product groups and an econometricanalysis (using a single-equation model and a simultaneous-equationmodel) is carried out to evaluate the role of different advertisingexpenditures (classified according to the degree of product differentia-tion) on the propensity to consume. Finally, Section IV summarizesthe main findings and outlines the basic conclusions of the paper.

I

In this section we test a model of consumption where advertisingintensity enters an independent variable, The model tested takesthe form:

the authors from reaching any positive conclusions regarding the informative effects of advertis-ing as opposed to its escalation effects. The differences between this study and the work ofTaylorWeiserbs, Schmalensee, and Ashley, Granger and Schmalensee will become apparentlater in this paper.

Page 3: ADVERTISING AND THE PROPENSITY TO CONSUME

ADVERTISING AND THE PROPENSITY TO CONSUME 275

(C/Y) = a0 + ai( C/ Y) + a2(A/C)r + a3Yt + a4(W/S)t + asrt

+ ct6G + Ut

where:

(C/Y) The propensity to consume in period t where C is realprivate final consumption and Y real household disposableincome.

(C/Y)_1 Lagged propensity to consume. This variable is introducedto allow for lagged effects as well as being a proxy forwealth?

(A IC) = Advertising intensity; where A is total national avertising.L(W/S) = Changes in income distribution; where W is income in

wages, salaries and supplements and S is gross operatingsurplus.

r The real rate of interest.G Changes in fiscal policy measured by government deficit!

surplus.U = An error term.

All variables were deflated using the implicit consumer price index.The above variables represent very closely those objective factors

mentioned by Keynes, in addition to advertising. The choice for ad-vertising intensity (i.e. A/C) and not advertising levels is based on thegrounds that businessmen think more of advertising/sales ratios ratherthan of the absolute amount of money spent on advertising and thatit is more appropriate to think of the effectiveness of advertising interms of consumption-unit-cost.

Also as the data used are necessarily annual (covering the period1960-78), there is an additional reason for using advertising intensity(and not advertising levels). Advertising intensity is less likely to haveits full effect on the propensity to consume for periods substantiallyshorter than one year; bearing in mind the normal reaction lags inherentin economic behaviour. The estimated equation using ordinary leastsquares estimation is:4

(C/Y) = 0.646 + 0.085(C/Y)1 + 0.098(A/C) - 0.006)'(2.282) (2.351) (2.649) (-2.935)

+ 0.054(W/S) + 0.002r + 0.00 lG.(0.855) (2.635) (0.568)

3See Koyck (1954), and Spiro (1962).for multicollinearity were performed in this and other equations and gave no undue

cause for concern. See Farrer and Glauber (1967). For what it is worth (in the context ofN = 19), tests for serial correlation with Durbin's 'h' statistic were also performed in equationswith lagged dependent variables. All of the statistics were not significant at well below the10 per cent level. For details of the test see Durbin (1970).

Page 4: ADVERTISING AND THE PROPENSITY TO CONSUME

276 BULLETIN

R2 = 0.911; F 27.131; d.f. = 11.

The values in parenthesis are t statistics.The regression results of the single-equation model would seem to

suggest that:(j) Advertising intensity is an important determinant of the pro-

pensity to consume. The coefficient of (A /C is positive and significantat the 1 per cent level of significance. This suggests that an increase intotal national advertising intensity leads to an increase in the propensityto consume. This conclusion would have two implications: (a) nationaladvertising as a whole can not be regarded as being an economic waste.Its significant effect on the country's propensity to consume suggeststhat at least some part of national advertising has a strong enoughinformative role to motivate potential customers to increase theirconsumption either by expanding their incomes (through increasedefforts) or by reducing their propensity to save or by running debts.5(b) National advertising should, in future studies of the theory ofconsumption, be considered as an important factor effecting theconsumption function.

Variables representing wealth, income and interest and laggedeffects seem to be important factors affecting the propensity to con-sume in Australia.

Changes in income distribution and fiscal policy do not seemto have had much effect on the propensity to consume in Australiaduring the period of study.

II

The following simultaneous relationships, known as structural equations,have been developed. The first equation examines the relationshipbetween the propensity to consume as the dependent variables, ad-vertising intensity, real national income, the real rate of interest andwealth (measured by the propensity to consume in period tl) asindependent variables.6 Thus this equation is the same as that testedin the previous section. The present equation does not, however,include fiscal policy and changes in income distribution. These twovariables are excluded on the ground that they did not play any signifi-cant role in determining the propensity to consume in Australia duringthe period of study, as demonstrated in the single-equation model.

The second equation in the system investigates the relationshipbetween advertising intensity as a dependent variable and the propensity

fact that advertising changes the propensity to consume does not necessarily indicatethat it has not led to socia1 waste'. This depends on the 'weight one puts on savings'.

6 As previously the (CI variable also allows for lags generally. On this see fn. 3.

Page 5: ADVERTISING AND THE PROPENSITY TO CONSUME

ADVERTISING AND THE PROPENSITY TO CONSUME 277

to consume as an independent variable to test the interaction betweenthe two variables. Advertising intensity in this equation is also assumedto be a function of its level in the previous period and of the degree ofconcentration in the manufacturing sector.7

We also tested the hypothesis that advertising and concentrationare interrelated simultaneously. Thus the third equation expressesconcentration as a function of adveritising intensity. However, con-centration depends on the height of entry barriers, which are deter-mined in part by technical factors such as the extent of productioneconomies of scale relative to the size of the market, the amount ofcapital required to operate a plant of minimum efficient scale and otherproduction cost disadvantages of new entrants. The height of entrybarriers may be affected by advertising. Product differentiation viaadvertising affects entry barriers in three ways: (1) Fligh prevailinglevels of advertising create additional costs for new entrants whichexist at all levels of output. This cost is necessary to attract the'attached' customers of existing brands. (2) Established firms withbigger market shares are likely to enjoy economies of scale in advertis-ing which result from the increasing effectiveness of advertising messagesas well as from decreasing costs of each advertising message purchased.(3) To enjoy fully available economies of scale in advertising, the newentrants could need more capital than is necessary to provide physicalplants and equipment.8 It follows, therefore, that plant size, capitalinvestment and advertising should enter as independent variables inthe concentration equation.

Structural Equations(C/Y) = a0 + a1(C/Y)1 + a2Y + a3(A/C)t + ct4r + u1,(A/C) = ßo + ßi(A/C)t_i + ß2(C/Y) + 1335t + U2,Si ='Yo ±71(A/C)+72K+y3Q+bt3.Endogeneous Variables(C/Y) = The propensity to consume in period t defined as above.(A/C) = Advertising intensity in period t defined as above.S = Degree of seller concentration (estimated as the ratio

of value-added of manufacturing establishments employing500 or more to total value-added).9

Predetermined Variables(C! Y)1 = The propensity to consume in period t-1.

Advertising expenditure of the manufacturing sector (producers, wholesalers and retailers)amounts to over 70 per cent of total national advertising in Australia.

8See Borden (1947).manufacturing sector, and hence the measure of concentration, is disaggregated in the

disaggregated analysis of Section III. As the data used in this paper are time series, the 'crosssection' (Herindahl-Hirschman) F/ index could not be used.

Page 6: ADVERTISING AND THE PROPENSITY TO CONSUME

278 BULLETIN

= Real household disposable income in period t.rt = Real rate of interest in period t.(A /C)1 = Advertising intensity in period t-1.K = Average (real) fixed capital expenditure per manufacturing

establishment in period t.Qt = Average (real) value-added per manufacturing establish-

ment in period t.u1,u2,u3 = Error terms.

The above system of structural equations is mathematically completein the sense that it contains as many equations as endogenous variables.Applying the order and rank conditions of identification, we noticethat each equation is over-identified. It is logical, therefore, to use themethod of two-stage least squares to estimate the parameters of theequations. The regression results are:

(C/Y) = 0.635 + 0.058(C/Y)1 + 0.I56(A/C) 0.0004Y(11.994) (2.410) (7.677) (-4.188)+ 0.005r,

(2.55 2)

R2 = 0.970; F 127.769; d.f. = 13.

(A /C) = 0.023 + 0.2l3(A/C)_1 + 0.055(C/Y) + O.023S.,(-4.467) (1.964) (3.462) (6.057)

R2 = 0.926; F = 50.252; d.f. = 14.

= 0.120 - 0.0004Kg + 0.0004Q + 7.292(A/ç,(7.629)(-0.288) (7.692) (5.6201)

R2 =0.931; F57.984; d.f. 15.

Three main conclusions can be derived from the statistical results:The pro'ensity to consume is not only affected by advertisingintensity but is also an important variable determining this in-tensity.10 A rise in the propensity to consume (partially due to anincrease in advertising intensity) results in an increase in advertisingper each unit of consumption.Market concentration is an important determinant of advertisingintensity and, conversely, advertising is an important determinantof market concentration.Barriers to entry (in terms of economic size of plant but not interms of capital investments) seem to play an important role indetermining the degree of concentration in manufacturing industriesin Australia.

'° results suggest that simultaneous equation bias could have biased the (A ¡C) co-efficient downwards in the estimated single equation using ordinary least squares estimationin Section I. In relation to this, see also fn. 4.

Page 7: ADVERTISING AND THE PROPENSITY TO CONSUME

ADVERTISING AND THE PROPENSITY TO CONSUME 279

Thus the simultaneous-equation model results seem to support thoseof the single-equation model: advertising is a significant determinantof the propensity to consume. This reinforces our suggestion that infuture studies of the consumption function due consideration shouldbe given to the role of advertising as an important determinant of thepropensity to consume. Also, the simultaneous equation results wouldseem to confirm our previous conclusion that at least some part ofadvertising has a strong enough informative effect to motivate potentialcustomers to increase their rates of consumption.

III

Both the single-equation and simultaneous equation model analyses ofthe previous two sections support the view that aggregate advertising isan important determinant of the propensity to consume and converselythe propensity to consume is a determinant of aggregate advertising.

This section examines whether the above conclusion holds fordisaggregated advertising. In other words, this section attempts toanswer the question: Do all types of advertising lead to an increase inthe propensity to consume? An answer to this question could helpdetermine the role of informative advertising and the magnitude ofadvertising escalation effects.

In our analysis we used advertising expenditures on sixty consumergoods segmented into four broad classifications, defined in terms ofthose product characteristics generally agreed by economists to berelevant to optimal advertising expenditure. In ascending order ofdifferentiability by means of advertising, the four groups of consumergoods are distinguished as follows:11

Group 1: Relatively standard goods, which, in general, are low inprice. Included in this group are fluid milk, butter, cheese, ice cream,flour, biscuits, bread, sugar, canned fruits and vegetables, men's andboys' apparel, women's clothing, children's dresses, women's under-wear, hats and caps, millinery, knitted goods, men's shoes, women'sshoes, luggage and other leather goods, paints and varnish, polishesand sanitation goods.

Group 2. Non-standard commodities which are relatively highly priced.The group includes wines, distilled liquor, books, furniture, rubbertyres, hand tools, household and service machinery and instruments.

'1These groupings follow Greer (1971), and Metwally (1977). To preserve degrees offreedom, we combined Groups 1 and 2 of these studies to form our Group 1: our Groups 2, 3and 4 are equivalent to their Groups 3, 4 and 5. It is acknowledged that it might sometimesbe difficult to decide whether some item should be a member of one or other group. In thisrespect the results of some subsidiary regressions run by the authors (in which groupings weremodified slightly) suggest that the results stated in the paper are not unduly sensitive to slightvariations in the groupings.

Page 8: ADVERTISING AND THE PROPENSITY TO CONSUME

280 BULLETIN

Group 3. Luxury or semi-luxury commodities with hidden qualities.The group includes furs, carpets, domestic refrigerators, domesticwashing machines, domestic vacuum cleaners, domestic electric ranges,electric fans, heaters, sewing machines, radio and television sets, motor-cycles, motor vehicles, motor parts and accessories, clocks and watchesand jewellery.

Group 4: Commodities with a high potential for differentiation bymeans of advertising. This group includes items which are non-necessities,but not luxuries either; they are also relatively frequently purchased.In the group are included beer, confectionery products, cereals, cigars,cigarettes, soap, chewing and smoking tobacco, smoking accessories,detergents, toothpaste, aspirin types, perfumes and comestics, andcostume jewellery.

We begin our disaggregated analysis by considering a single-equationmodel which takes the form:

(C/Y) = o + i(C/Y)ti + a2Y + a3r + ct4(A1/C) + ct5(A2/C)+a6(A3/C) + o7(A4/C) + Ut

where (C/Y), (C/Y)1, Yt and rt are the same variables as before.(A 1/C); (A 2/C)r; (A 3/C) and (A 4/C) refer to advertising intensity forgroups 1, 2, 3 and 4 respectively, where A1 refers to national advertisingexpenditures on products included in the ith group and C refers tohousehold consumer expenditures.

The statistical results of the tested model give:

(C/Y) = 0.513 + 0.204(C/Y)_1 0.0003Y + 0.590rt 0.017(A1/C)1(5.289) (2.501) (-2.400) (1.1 909)(1 .455)+ 0.049(A2/C)1 + 0.1 25(A3/C)1 + 0.009(A4/C),

(1.897) (4.269) (1.073)R2 = 0.986; F= 105.138; d.f. = 10.

The above statistical results suggest that not every type of advertisinghas a significant effect on the propensity to consume. The 't' values forgroups 1 and 4 are not statistically significant whereas those for groups 2and 3 are significant at the 10 per cent level and the 1 per cent level ofsignificance respectively. Total advertising expenditures on those twogroups amount to approximately one-third of total national advertisingin Australia in 1977. This suggests that an increase in the advertisingintensity related to non-standard commodities which are relativehighly priced and considered luxuries or semi-luxuries do have a signifi-cant effect in motivating people to increase their propensity to con-sume. On the other hand an increase in the advertising intensity relatedto relatively standard low-priced goods and on commodities with a highpotential for differentiation by means of advertising has no significanteffect on the propensity to consume.

Page 9: ADVERTISING AND THE PROPENSITY TO CONSUME

ADVERTISING AND THE PROPENSITY TO CONSUME 281

The disaggregate analysis, therefore, suggests that a good part oftotal advertising expenditures in Australia (approximately one-third)may be regarded as having an important informative role while asubstantial part of these expenditures results only in a reallocationof th market positions of existing firms without any significant effecton the propensity to consume.

We now test if this conclusion still holds in the face of the inter-action between advertising intensity and the propensity to consume.For this purpose, we used the following simultaneous-equation model:

Structural Equations(C/Y) = c + 1(C/Y)ti + o2Y + o3(A1/C)t + a4(A2/C)

+ c5(A3/C) + o6(A4/C) + u1,

(A1/C) = ¡3 + j31(A1/C)_1 + ß2(C/Y) + ß3Slr + u2,

(A2/C) = 'o + y1(A2/C)1 + y2(C/Y) + y3S2 + u2,(A3/C) l + 1(A3/C)_1 + ¿52(C/Y) + ö3S3 + u4,(A4IC) = Xo + X0(A4/C) + X2(C/Y) + X3S4 + u5,Sit = a0 + ai(Ai/C)t +a2Ki + a3Qit + 146,

S2 = b0 + b1(A2 /C) + b2K2 + b3Q31 + u7,S3r = g0 + g1(A3/C) + g2K3 + g3Q3 ± u8,S4 h0 + h1(A4/C) + h2K4 + h3Q4 + u9,

Endogenous Variables(CI Y) = the propensity to consume in period t,(A1/C) = advertising intensity of commodities in the ith group (j

1,.. .,4)inperiodt,St = concentration ratio of commodities in the ith group (1 =

I,.. .,4) in period t.Predetermined Variables(C! Y)-1 = the propensity to consume in period t 1,

= real household disposable income in period t,(A/C)1 = advertising intensity of commodities in the ith group

(j = 1,. . ., 4) (in period (t-1)),= average (real) fixed capital expenditure per manufacturing

establishment in industries producing commodities of theith group (j = 1, . . ., 4) in period t,

= average (real) value-added per manufacturing establishmentin industries producing commodities of the ith group(i 1,..., 4)in period t.

The above system of structural equations is mathematically completein the sense that it contains as many equations as endogenous variables.Applying the order and rank conditions of identification, we noticethat each equation is over identified. It is logical, therefore, to use themethod of two-stage least squares to estimate the parameters of theequation. The regression results are:

Page 10: ADVERTISING AND THE PROPENSITY TO CONSUME

R2 = 0.930 F 53.671; d.f. = 15.

The above statistical results suggest that:-(a) The disaggregate simultaneous equation model supports the results

of the disaggregate single equation model. Only advertising intensi-ties in groups (2) and (3) seem to be significant determinants ofthe propensity to consume and vice versa. These two groups seemto fall into a high rate of product-change category. The continuousintroduction of new models (or styles) with additional (in somecases superior) features requires advertising to inform potential

282

(C/Y) =

BULLETIN

0.602 + 0.15 1(C/ Y)ri - 0.0004Y - 0.01 5(A1/C)(6.260) (2.860) (--3.100) (-1.319)

+ 0.070(A2/C) + 0.l05(A3/C) + 0.0O5(A4/C),(1.858) (3.629) (0.468)

R2 = 0.989; F = 139.400; d.f. = 11.

(A1/C) -5.240 + 0.007(A1/C)1 + 0.O27(C/Y) + 3l.663S(-15.202) (2.209), (0.085) (20.566)

R2 = 0.978; F= 173.525; d.f. = 14.

(A2/C) = -5.746 + 0.133(A2/C)_1 + 0.762(C/Y) + l3.248S.,(-6.573) (2.775) (2.639) (5.688)

R2 = 0.918; F = 44.548; d.f. = 14.

(A3/C) -4.222 + 0.016(A3/C)1 + 0.348(C/Y) + 7.73lS.,(-10.741) (2.865) (3.548) (6.093)R2 = 0.986; F = 289.932; d.f. = 14.

(A4/C) = -2.264 + 0.055(A4/C)1 - 0.055(C/Y) + 8.227Sf,(-12.649) (3.435) (-1.435) (31.711)

s4( =

(3.749)(14.021) (2.984) (3.072)R2 = 0.956; F 93.174; d.f. = 15.

0.207 + 0.075(A4/C) + 0.003Kg + O.00 1Q.,(2.150) (9.943) (1.654) (3.324)

R2 = 0.991; F = 453.809; d.f. = 14.

sir = 0.134 + 0.030(A1/C) + 0.002Kg + 0.00020Q,(1.859) (4.762) (0.333) (0.382)

R2 = 0.805; F 16.533; d.f. = 15.

s2r 0.411 + 0.03 2(A2/C) + 0.00 1K + O.000SOQr,(55.385) (4.833) (1.894) (5.073)

R2 = 0.963; F 102.699; d.f. = 15.

s3t 0.351 + 0.111(A3/C) + 0.002Kg + 0.0003Q,

Page 11: ADVERTISING AND THE PROPENSITY TO CONSUME

ADVERTISING AND THE PROPENSITY TO CONSUME 283

customers of these changes. These advertising expenditures couldin turn affect the propensity to consume. Also a good number ofthe products included in groups (2) and (3) have relatively high'income elasticities of demand'.12 Thus as income rises a largerproportion of consumers' spending would be devoted to theseproducts. Business firms would be expected to compete for thesepotentially new markets through advertising. Hence, for groups(2) and (3), these two factors, namely the high rate of productchange and the relatively high income elasticity of demand mayoffer a plausible explanation for the disaggregated simultaneousresults obtained in this paper.Advertising intensities in all groups seem to be important deter-minants of concentration and vice versa.Economies of scale and capital intensities seem to be importantdeterminants of concentration in some cases but not in others.Advertising budgets seem to be determined partially by last year'sadvertising.

Thus the disaggregate simultaneous results support the previousconclusion that a good part of total advertising expenditure has animportant effect in stimulating the propensity to consume; conversely,changes in the propensity to consume have a significant impact on theadvertising intensities of a number of groups.

IV

This paper despite its limitations, has some important findings:The statistical results suggest that the intensity of aggregate advertis-ing is an important determinant of the propensity to consume andvice versa. This conclusion has two implications: (a) in future studiesadvertising should be included explicitly in models of consumer be-haviour; (b) the claim that all advertising is an economic waste anddoes not play an informative role is not completely sustained.The disaggregate analysis suggests that a substantial part of ad-vertising has no effect on the propensity to consume. This partrelates to relatively standard low-priced goods and to commoditieswith a high potential for differentiation by means of advertising.Thus it could be argued that though not all advertising is an economicwaste; at best, a substantial proportion, results only in a reallocationof market position of the firms.The simultaneous-equation aggregate and disaggregate analysessuggest that advertising intensity is an important determinant ofmarket concentration and vice versa.

University of Queensland

12See Metwally (1973).

Page 12: ADVERTISING AND THE PROPENSITY TO CONSUME

284 BULLETIN

REFERENCES

Ando, A. and Modigliani, F. (1963). 'The "Life Cycle" Hypothesis of Saving:Aggregate Implications and Tests', American Economic Review, Vol. 53, No. 1,Part 1, January, pp. 55-84.

Ashley, R., Granger C. W. J., and Schmalensee, R. (1980). 'Advertising and Aggre-gate Consumption: An Analysis of Causality', Econometrica, Vol. 48, No. 5,July, pp. 1149-68.

Bain, J., (1952). Price Theory, New York, Henry Holt.Bain, J. (1956). Barriers to New Competition, Cambridge, Harvard University

Press.Borden, N. H., (1947). The Economic Effects of Advertising, Chicago, Richard

Irwin.Comanor, W. S. and Wilson, T. A. (1967). 'Advertising Market Structure and

Performance',Review of Economics and Statistics, Vol. 44, No.4, pp. 423-40.Cowling, K. (1973). Market Structure and Economic Behaviour, London, Gray-

Mills.Cowling, K. et al. (1973). Advertising and Economic Behaviour, New York, Holmes

and Meier.Doyle, P. (1968). 'Advertising Expenditure and Consumer Demand', Oxford

Economic Papers, Vol. 20, No. 2, November, pp. 394-414.Duesenberry, J. S. (1949). Income, Savings and the Theory of Consumer Behaviour,

Cambridge, Harvard University Press.Durbin, J. (1970). 'An Alternative to the Bounds Test for Testing for Serial Correla-

tion in Least-Squares Regression', Econometrica, Vol. 38, No. 3, May,pp.410-29.

Edwards, C. D. (1968). 'Advertising and Competition: An Evaluation of Exhorta.tive Programs',Business Horizons, No. 1, February, pp. 59-76.

Farrar, D. E. and Glauber, R. B. (1967). 'Multicollinearity in Regression Analysis:The Problem Revisited', Review of Economics and Statistics, Vol. 49, No. 1,February, pp. 92-107.

Feilner, W. A. (1949). Competition Among the Few, New York, Knopf.Friedman, M. (1957). A Theory of the Consumption Function, Princeton, Princeton

University Press.Greer, D. (1971). 'Advertising and Market Concentration', Southern Economic

Journal, Vol. 38, No. 1, July, pp. 19-32.Gupta, J. K. and Krishnan, K. S. (1967). 'Mathematical Models in Marketing',

Operations Research, Vol. 15, No. 6, November-December, pp. 1040-50.Johnston, J. (1972). Econometric Methods, 2nd ed.. New York, McGraw-Hill.Kaldor, N. (1951). 'The Economic Aspects of Advertising', Review of Economic

Studies,Vol. 18,No. "pp. 1-27.Kaldor, N. and Silverman, R. (1948). A Statistical Analysis of Advertising Ex-

penditure and the Revenue of the Press, Cambridge, Cambridge UniversityPress.

Keynes, J. M. (1936). The General Theory of Employment, Interest and Money,New York, Harcourt Brace.

Koyck, J. M. (1954). Distributed Lags and Investment Analysis, Amsterdam, North-Holland.

Mann, H. M., Henning, J. and Meehan, J. W. (1967). 'Advertising and Concentration:

Page 13: ADVERTISING AND THE PROPENSITY TO CONSUME

ADVERTISING AND THE PROPENSITY TO CONSUME 285

An Empirical Investigation', Journal of Industrial Economics, Vol. 16, No. 1,November, pp. 34-45.

Metwally, M. M. (1973). 'Australian Advertising Expenditure and Its Relation toDemand',Economic Record, Vol. 43, No. 128, pp. 290-9.

Metwally, M. M. (1973). 'Economic Strategies of Firms Facing Asymptotic De-mand: A Case Study',Applied Economics, Vol. 5, No.4, December, pp. 271-80.

Metwally, M. M. (1975). 'Advertising and Competitive Behaviour of SelectedAustralian Firms', Review ofEconomics and Statistics, Vol. 57, No. 4, November,pp. 417-28.

Metwally, M. M. (1976). 'Profitability of Advertising in Australia: A Case Study',Journal of Industrial Economics, Vol. 24, No. 3, March, pp. 221-31.

Metwaily, M. M. (1977). 'Market Concentration and Advertising: The AustralianExperience',Management Science, Vol. 23, No. 6, February, pp. 557-66.

Nerlove, M. and Arrow, K. J. (1962). 'Optimal Advertising Policy Under DynamicConditions', Economica, Vol. 29, No. 114, May, pp. 129-42.

Palda, K. S. (1964). The Measurement of Cumulative Advertising Effects, Engle-wood Cliffs, Prentice-Hall.

Rao, C. R. (1965). Linear Statistical Inference and Its Applications, New York,Wiley.

Reekie, W. D. (1975). 'Advertising and Market Structure: Another Approach',Economic Journal, Vol. 85, No. 337, March, pp. 156-64.

Reekie, W. D. (1974). 'Advertising and Market Share Mobility', Scottish Journalof Political Economy,Vol. 21, No. 2, June, pp. 143-58.

Rees, K. D. (1975). 'Advertising, Concentration and Competition: A Commentand Further Results',Economic Journal, Vol. 85, No. 337, March, pp. 165-72.

Schmalensee, R. (1972). The Economics of Advertising, Vol. 80, Contributions toEconomic Analysis, Amsterdam, North-Holland.

Spiro, A. (1962). 'Wealth and the Consumption Function', Journal of PoliticalEconomy, Vol. 70, No. 4, August, pp. 339-54.

Sutton, C. J. (1974). 'Advertising, Concentration and Competition', EconomicJournal, Vol. 84, No. 333, March, pp. 56-9.

Sutton, C. J. (1975). 'Advertising, Concentration and Competition: A Reply andSome Further Comments', Economic Journal, Vol. 85, No. 337, March,pp. 173-6.

Taylor, L. D. and Weiserbs, D. (1972). 'Advertising and the Aggregate ConsumptionFunction',American Economic Review, Vol. 62, No.4, September, pp. 642-55.

Telser, L. (1964). 'Advertising and Competition', Journal of Political Economy,Vol. 72, No. 6, December, pp. 537-62.

Vernon, J. M. and McElroy, N. B. (1973). 'A Note on Estimation in Market Struc-ture Performance Studies', American Economic Review, Vol. 63, No. 4,September, pp. 763-7.

Vernon, J. M. and Nourse, R. E. M. (1973). 'Profit Rates and Market Structure ofAdvertising Intensive Firms', Journal of Industrial Economics, Vol. 22, No. 1,September, pp. 1-20.

Weiss, D. L. (1968). 'Determinants of Market Structure', Journal of MarketingResearch, Vol. 5, August, pp. 290-5.