Published by World Advertising Research Center Farm Road, Henley-on-Thames, Oxon RG9 1EJ, UK Tel: +44 (0)1491 411000  Web: www.warc.com

Vol. 21, No. 4, 2002

The influence of advertising on the demand for chocolate confectionery

Bob Eagle
Independent Consultant

and

Tim Ambler
London Business School

This paper addresses the impact of advertising on the size of five European chocolate confectionery markets: Belgium, France, Germany, the Netherlands and the UK. We are not concerned with brands or market share. No doubt brand owners advertise to the extent that their particular brands are advantaged. The question here is whether it is likely that a reduction in advertising would decrease the demand for chocolate products as a whole. The literature implies that total market effects are unlikely but it is not one–sided. Legislators remain inclined to ban advertising for various categories where it seems that they, at least, expect total consumption to be reduced as a result.

Concerns over the nation's diet are an example, especially with respect to children's health. The Department of Health (2000) reported that four out of five children aged between 4 and 18 regularly ate snack foods but very little fruit. This prompted calls from the Food Commission[1] for stricter government regulation on the promotion of so–called 'unhealthy' foods. TV advertising is seen by critics as responsible (e.g. Lewis & Hill 1998; Holmquist 2000).

This paper is organised as follows. After reviewing the literature on the effects of advertising on market size and the continuing calls for banning advertising in some sectors, we review briefly the development of the chocolate market in Europe. We then analyse data from the five largest European chocolate confectionery markets: Belgium, France, Germany, the Netherlands and the UK. After discussing the findings, we propose future research and draw conclusions.

Previous research

First, we consider the generic research evidence, much of which is associated with foods. If advertising never changes market size, at least for mature markets, there would be no case to answer. As the answer appears to be context–specific, we need to look a little closer. We found no previous work for confectionery markets but some inferences may be drawn from the work on the alcohol and tobacco markets.

Advertising and market size

Borden (1942), in a pioneering review, concluded that advertising had no effect on mature market size. Sturgess (1982) tested causality between total advertising expenditure and aggregate consumption using Box–Jenkins time series analysis. For the period 1969 to 1980 he found no relationship between them and, rather robustly, dismissed their linkage as a myth, albeit at the aggregate level of consumption.

Henry (1984, 1996) and Duffy (1999) studied food markets and found no meaningful relationship between changes in advertising spend and market size. Henry's conclusion was that advertising could be an effective tool of competition between brands, but in mature food markets there was no discernible effect on the market size.

We need to distinguish between brand advertising in static markets, where there can be no market effect by definition, and using advertising to stimulate the market as a whole. In the latter case, the category becomes, in effect, the brand. Forker and Ward (1993) suggest that the category effects of generic advertising could be different from the cumulative effects of brand advertising because it is a coordinated effort designed specifically for one purpose – to increase total category consumption. Table 1 summarises representative findings.

Table 1: category effects of generic advertising

Of course, advertising may fail to increase the category because the advertising was poor and not because it cannot be done. The importance of this literature is that at least the advertising was trying to increase the category which is not necessarily true for brand advertising. In any case, no conclusion can be drawn.

Ambler et al. (1998) analysed 156 winners of the UK IPA Advertising Effectiveness Awards between 1981 and 1994. Krona margarine increased the margarine sector at the expense of butter, leaving the broader yellow fats category unaffected. Viennetta ice cream dessert similarly increased dairy product sales but not total desserts. On the other hand, advertising Sanatogen tablets appeared to increase the multivitamins category without cannibalising other vitamin sectors.

In total, fewer than 20% of all the cases indicated any kind of market size effect. Importantly, these prize winners are unusually successful examples of advertising and might be expected to have unusually powerful effects. One explanation for the differences in results is that it depends on how the 'market' is defined. Ambler et al. (1998) argue that in order to find an answer to whether advertising does affect the overall market, it is critical to define the market first:

Markets are of different types. There are cases where an effect clearly exists, and there are others where careful examination has failed to find effects and advertising, as many manufacturers usually suppose, affects only brand share (p. 267).

Their second conclusion was that the category effect depends on its maturity and size. Increase is less likely if the market is large and mature. Kotler (1991) defines the initial two phases of the product life cycle in terms of triallists and these would be expected to be influenced by advertising. Thereafter repeat purchases tend to dominate and the market turns to brand switching.

Ambler and Walters (1994, pp. 15–16) define mature market as

a market where the addition of new products and new consumers is relatively insignificant to the whole. Product innovation is a feature of any market and as consumers age through their own life cycles, they too come and go. Nevertheless the characteristics of the market will not be dominated by first time buyers. Advertising will be more concerned with reminder than carrying new information. Most of the competition that intends to join the market will have joined. It is apparent that, using these characteristics, a category may grow or decline whilst remaining in the mature stage.

Before returning to chocolate we will consider two other markets which would appear to be similar in terms of maturity and level of aggregation: alcohol and tobacco.

Advertising and alcohol

Published studies show no evidence for a causal relationship between advertising spend and an aggregate demand for alcoholic products, although a weak correlation may exist (Fisher 1993). Prest (1949) showed that advertising did not significantly increase primary demand for homogeneous packaged goods such as distilled spirits. Bourgeois and Barnes (1979) analysed the relationship between alcoholic beverage advertising and consumption in Canada between 1951 and 1974. This study examined the relationship between the consumption of beer, wine, distilled spirits and total alcohol, with a variety of marketing and non–marketing variables, and found little or no correlation between alcoholic beverage advertising and per capita alcohol consumption. The authors concluded that total alcohol consumption is driven more by environmental factors than by those that firms could control in the short run.

A number of more recent studies reached similar conclusions. Saffer (1992), of the US National Bureau of Economic Research, found little consistent evidence that alcohol advertising affects alcohol sales and drinking, as have other studies (FTC 1985; Duffy 1989; Smith 1990; Lee & Tremblay 1992). Calfee and Scherage (1994) looked at France, the UK, the Netherlands and Sweden and found similar results. Conversely, research in France showed that increases in total alcohol advertising correlated with reduced sales. This is presumably explained by a switch from cheap generic wines to more expensive, branded wines and spirits. The policy option that advertising should be increased in order to reduce consumption further does not appear to have been seriously considered by the French government.

This literature supports the significance of defining the 'market' noted above: advertising did seem to be linked with the sub–sectors (e.g. wines), but not with the alcohol sector as a whole.

Advertising and tobacco

Schmalensee (1972) and Hamilton (1972) both concluded that the impact of cigarette advertising on aggregate consumption was statistically insignificant. The widespread proposals to ban tobacco advertising and two voluntary arrangements, in 1985 and 1986, stimulated a significant amount of research assessing the impact of advertising on tobacco consumption. Waterson (1984, 1986, 1990) concluded that advertising has little or no effect upon aggregate UK consumption of cigarettes.

On the other hand, McGuinness and Cowling (1975), using quarterly UK data for 1957 to 1968, found a significant relationship between advertising and consumption. Their linear and log–linear regression models included cigarette consumption and prices, the level of disposable income, the last quarter's level of cigarette expenditure, advertising and health education. Bishop and Yoo (1985) obtained similar results for the USA. These results were disputed by the industry, and Godfrey (1986) found technical flaws in both sides of the debate which related mainly to whether annual (no relationship) or shorter time periods were used. Since the argument is strategic, we would side with the use of annual time periods.

The Smee Report (Department of Health 1992) used data from 1958 to 1987. The total number of cigarettes was regressed against implicit average real cigarette price, personal disposable income per head at 1970 prices, lagged advertising with a quarterly carry–over of 0.7 for ten periods, and a lagged dependent variable term. The results showed no significant correlation between advertising and consumption.

Finally, Andrews and Franke (1991) employed meta–analysis to summarise the econometric findings on the relationships between cigarette consumption and advertising, price and income of 48 studies from 1933 to 1990. Twenty–five of these studies were from the USA, 13 from the UK and ten from other countries including Australia, Finland, Papua New Guinea and Germany. The results indicate that there is a significant relationship between advertising and cigarette consumption that is independent of study design, although the magnitude of this relationship varies. The positive impact of advertising on cigarette consumption has declined over time – an observation consistent with the market maturing.

Conclusion

Although the literature supports both points of view to some extent, the balance of expectation is that, for a market as specific and mature as chocolate confectionery, advertising will not affect market size. If, however, they are correlated, then we should explore the situation at the level of confectionery as a whole where we would have even less expectation of seeing linkage. We now consider the empirical evidence for chocolate confectionery in Western Europe.

European chocolate confectionery

Chocolate was first imported to Europe from the Americas by Spanish gold hunters, and the first chocolate houses emerged in England in the 1600s (Field Museum 2002). In the nineteenth century, brands such as Cadbury and Nestlé opened up chocolate to the mass market.

In the past 15 years, the cumulative volume growth in Western Europe has been 23%, and 60% in value terms (Table 2). More recently, however, total and per capita growth has stalled. The difference between volume consumption and value growth is explained by inflation and changes in currency exchange rates. Predictions are inevitably speculative but the same source forecasts 4.7% growth for 2001 to 2005, or about 1% volume growth per annum.

Table 2: European chocolate confectionery: retail sales, 1987–2001

The countries chosen for this study account for 68% of the total European chocolate confectionery market (Table 3). We now analyse the growth in these five chocolate confectionery markets.

Table 3: chocolate confectionery: market size and category development in selected countries, 2000

Empirical study

The data

We selected Germany, France, Belgium, the Netherlands and the UK as distinct markets for the 11 years from 1990 to 2000. Variables used in similar studies are shown in Table 4.

Table 4: variables used in previous research

Our dependent variable was taken to be the annual percentage growth in the chocolate market and, after consideration of those above, five independent variables were chosen (Table 5).

Table 5: variables used in this research

The data on GDP and inflation rate were taken from Euromonitor, the Global Marketing Information System. The consumer–related data across all the countries were extracted from the database of AC Nielsen. Data on advertising expenditure and advertising inflation were collected with the help of media departments of local advertising agencies.

Analysis

An initial analysis showed an anomalous growth in the German chocolate market at the time of that country's reunification. To avoid undue and misleading influences on the analysis, the data for Germany at that time were excluded.

Plotting growth in advertising expenditure against market growth suggests little association, as shown in Figure 1. On the other hand, the elliptical shape of the relationship between price and market growth shown by Figure 2 provides an expectation of significance.

First, multiple regressions were made for each individual country separately. In all models, R–squared was around 0.027 and country parameters were insignificant (Table 6).

Table 6: significance of individual countries

Examination of advertising, per capita income and price coefficients across countries revealed no significant differences. As a result, we eliminated the country parameter and pooled all countries. Retail price changes and time were significantly associated with chocolate confectionery consumption (p < 0.043 and p < 0.003) but changes in advertising expenditure were not (Table 7).

Table 7: pooled results

The size of the constant, relative to the coefficients of the variables, confirms that this is a stable market where even the significant variables have a relatively small effect.

To examine the possibility of delayed effects, i.e. advertising expenditure growth affecting sales in the subsequent year, the model was re–run with the previous year's weight of advertising as an additional variable. The fit of the model barely changed and there was still no significant effect of advertising weight on market growth (Table 8).

Table 8: pooled results with delayed advertising effect

Table 9 shows that excluding the advertising variable did not damage the overall model but slightly improved the fit for the remaining variables.

Table 9: effect of dropping advertising from variables

Results

Time is a significant negative variable, as would be expected from the downward trend in the overall volumes over the past ten years (Table 2). Whether Euromonitor is right to project the last five years' uplift of 1% per annum into the next five is beyond the data in this paper. We can speculate, however, that two demographic forces are competing. Children make up a declining share of the population in these countries, but that is offset by increasing per capita income which was significant at the 90% level.

The other significant variable was price with – as would be expected – a negative effect on market growth. Again that must be in competition with incomes. In other words, if incomes are increasing faster than price inflation the net price effects are unlikely to be material.

No significant effects were found for advertising, lagged or otherwise, per capita income or country. While we expected the results for advertising, the absence of apparent country effects is more surprising. Although the countries share a number of characteristics such as maturity, adjacency, membership of the EU and per capita disposable income, we had not expected such consistency in the results.

At the same time, the model explains only about 24% of the variance. This result is significant at the p < 0.01 level and is typical for analyses of this type (e.g. Vacker & Wilcox 1992; Calfee & Scherage 1994; Duffy 1999; Wilcox 2001). We were, of course, seeking to explain changes in market size and therefore the lagged effects of the sales of one year on the next do not arise. They are, of course, reflected in the relatively large constant.

Discussion

This paper was prompted by suggestions that European public health could be improved by reducing chocolate confectionery consumption which in turn could be achieved by reducing brand advertising in that sector. This echoes previous calls to ban alcohol and tobacco advertising along with any advertising to children.

As noted in the limitations section below, data were not available to analyse the children's market separately. Much of the purchasing and consumption is within the family context. Advertising no doubt includes younger consumers, but it would, in general, be inefficient to target children with separate ads. Table 10 shows that there has been a slight decline in the proportion of under–15s.

Table 10: under–15s proportion of the population (%)

During this time period, the population of the five countries increased by 4% to 228.5 million. About half of the increase was due to longevity: the population of over–65s increased by 4 million. The actual number of under–15s, however, was static at 40 million. It seems unlikely therefore that changes in chocolate consumption can be attributed to any change in the population of under–15s.

No doubt excessive consumption of chocolate is harmful along with excessive consumption of almost anything. Apparently carrots can be dangerous. At the same time, moderate consumption of most of these products, including alcohol, carrots and chocolate, is beneficial. Identifying the benefits and dangers of any such commodity, and the borderline between the two, is beyond this paper but we need to recognise the context in order to position the role of advertising.

While individual campaigns can not only move brands but their markets too, as witness the winners of the IPA Awards, others have only a weak impact at the brand level and some have none at all. Ehrenberg (1974) (and subsequently) has argued that advertising is usually a weak force reinforcing the brand experience. In other words, it maintains share in a competitive market and brand shares are remarkably stable.

Yet the literature shows that advertising does sometimes affect market size, most obviously when the market is immature. When mature markets are thought to be flagging, generic campaigns may reinvigorate sales but here again the picture is mixed, even though the objective, unlike brand advertising, is specifically to increase market size.

It seems likely that those calling for curtailing advertising are seeking a convenient scapegoat rather than attempting to understand either how advertising works or when it increases category demand, or when it increases the dangers of the product relative to the benefits. This is not to say that their solution is wrong but only that alternative solutions to a perceived problem, such as public health, ought to receive more careful analysis. In this particular case, namely chocolate confectionery, curtailing advertising does not appear to be a valid solution.

Limitations and further research

A number of other variables could be examined to explain changes in consumption to the extent that data are available. For example, Yasin (1995) considered the impact of private label. Similarly under retailer control are the number and depth of promotions that may vary from year to year. Total advertising expenditure is only a proxy for its effectiveness. Examination of advertising at brand level would be needed to establish effectiveness, but it is possible that there was some variation in total impact from year to year.

Mela et al. (1998) argue that omitted variable bias is controlled by a macroeconomic variable (GDP in our case) and a lagged dependent variable. Since we took the growth in sales as the dependent variable, we suggest that the lagged variable is implicit.

Chocolate consumption is seasonal and more associated with cold weather; this raises the possibility that average temperature, or even rainfall, may have an impact. Limitations arise from the data which are subject to the accuracy of the source and the selection of years as the unit of analysis. As Godfrey (1986), among others, has shown different time periods can affect the associations.

Finally, lack of data prevented us from isolating the advertising seen by children, and their consumption of chocolate, from the market as a whole. Since such consumption is likely to be predominantly in a family context, we see no reason to expect that our findings would be different. On the other hand, balanced research into solutions to reduce excessive consumption, without damaging the benefits from moderate consumption, would require more detailed research.

Conclusion

Chocolate confectionery in Western Europe is a mature market where the rate of growth is in slow decline. This study looked for correlation primarily between advertising and the year–to–year changes in market size of five countries: Belgium, France, Germany, the Netherlands and the UK. Since we found no significant association, advertising cannot be driving changes in market size. We also found no lagged effects.

As was expected, there was a negative correlation between market size and price. The results were consistent across the five countries. The perhaps surprisingly low (90%) level of correlation significance for per capita income might be strengthened with more data. The underlying trend, which works against the price and wealth effects, was directionally of gentle decline.

These results are consistent with prior research for other categories.

Acknowledgements

The authors are grateful to Julia Tishchenko and to executives at Masterfoods for their valuable contributions to this paper.

References

Ambler, T. & Walters, C. (1994) Advertising effect on mature market size. PANAGRA Working Paper, London Business School, February.

Ambler, T., Broadbent, S. & Feldwick, P. (1998) Does advertising affect market size? Some evidence from the United Kingdom. International Journal of Advertising, 17(3), pp. 267–300.

Andrews, R.L. & Franke, G.R. (1991) The determinants of cigarette consumption: a meta–analysis. Journal of Marketing Research, 21(2), pp. 65–74.

Bishop, J.A. & Yoo, J.H. (1985) Health scare, excise taxes and advertising ban in the cigarette demand and supply. Southern Economic Journal, 52(10), pp. 402–411.

Bjornal, T., Salvanes, K.G. & Andreassen, J.H. (1992) The demand for salmon in France – the effects of marketing and structural change. Applied Economics, 24(9), pp. 1027–1034.

Borden, N.H. (1942) The Economic Effects of Advertising. Chicago, IL: Richard D. Irwin.

Bourgeois, J.C. & Barnes, J.G. (1979) Does advertising increase alcohol consumption? Journal of Advertising Research, 19, pp. 19–29.

Calfee, J.E. & Scherage, C. (1994) The influence of advertising on alcohol consumption: a literature review and an economic analysis of four European nations. International Journal of Advertising, 13(4), pp. 287–310.

Department of Health (1992) Effect of tobacco advertising on tobacco consumption: a discussion document reviewing the evidence (the Smee Report). Economics and Operations Research Division, London.

Department of Health (2000) National Diet and Nutrition Survey results. Press release, 1 June.

Duffy, M. (1989) Measuring the contribution of advertising to growth in demand; an econometric–accounting framework. International Journal of Advertising, 8(2), pp. 95–110.

Duffy, M. (1999) The influence of advertising on the pattern of food consumption in the UK. International Journal of Advertising, 18(2), pp. 131–168.

Ehrenberg, A.S.C. (1974) Repetitive advertising and the consumer. Journal of Advertising Research, 14 (April), pp. 25–34.

Euromonitor (2001) Chocolate confectionery market report. Market Research GB, February.

Federal Trade Commission (FTC) (1985) Recommendations of the staff of the Federal Trade Commission re: Omnibus petition for regulation of unfair and deceptive alcoholic beverage advertising and marketing practices. Docket #206–246, March, Washington, DC.

Field Museum (2002) Chocolate. www.fieldmuseum.org.

Fisher, J.C. (1993) Advertising, Alcohol Consumption and Abuse. Westport, CT: Greenwood Press.

Forker, O.D. & Ward, R.W. (1993) Commodity Advertising. The Economics and Measurement of Generic Programs. New York: Lexington Books.

Godfrey, C. (1986) Government policy, advertising and tobacco consumption in the UK. A critical review of the literature. British Journal of Addiction, 81, pp.339–346.

Green, R.D., Carman, H.F. & MacManus, K. (1991) Some empirical methods of estimating advertising effects in demand systems: an application to dried fruits. Western Journal of Agricultural Economics, 16(1), pp. 63–71.

Hamilton, J.L. (1972) The demand for cigarettes: advertising, the health scare and the cigarette advertising ban. Review of Economics and Statistics, 56, pp. 401–411.

Henry, H. (1984) Does advertising affect total market size? Admap, 20(11), pp.524–532.

Henry, H. (1996) Does advertising affect total market size? Admap, 32 (January), pp. 16–19.

Holmquist, K. (2000) Children who munch their way to ill health. Irish Times, 11 March, p. 10.

Jensen, H.H. & Schroeter, J.R. (1992) Television advertising and beef demand: an econometric analysis of split–cable household panel scanner data. Canadian Journal of Agricultural Economics, 40, pp. 271–294.

Kotler, P. (1991) Marketing Management. Analysis, Planning, Measurement and Control. 7th edition. Englewood Cliffs, NJ: Prentice Hall.

Lee, B. & Tremblay, V.J. (1992) Advertising and the US market demand for beer. Applied Economics, 24, pp. 69–76.

Lewis, M.K. & Hill, A.J. (1998) Food advertising on British children's television: a content analysis and experimental study with nine–year–olds. International Journal of Obesity, 22 (3 March), pp. 206–214.

McGuinness, T. & Cowling, K. (1975) Advertising and the aggregate demand for cigarettes. European Economic Review, 6, pp. 311–328.

Mela, C.F., Gupta, S. & Jedidi, K. (1998) Assessing long–term promotional influences on market structure. International Journal of Research in Marketing, 15
(2 May), pp. 89–107.

Prest, A.R. (1949) Some experiments in demand analysis. Review of Economics and Statistics, 31, pp. 33–49.

Saffer, H. (1992) Alcohol advertising bans and alcohol abuse: econometric evidence Paper presented at NIAAA Working Group on the Effects of the Mass Media on the Use and Abuse of Alcohol, Washington, DC, 24–25 September.

Schmalensee, R. (1972) The Econometrics of Advertising. Amsterdam: North Holland.

Smith, D.I. (1990) Consumption and advertising of alcoholic beverages in Australia, 1969–1986. Drug and Alcohol Review, 9, pp. 33–43.

Sturgess, B. (1982) Dispelling the myth: the effects of total advertising expenditure on aggregate consumption. International Journal of Advertising, 1(3), pp. 201–212.

Vacker, B. & Wilcox, G. (1992) Cigarette advertising and consumption in the United States: 1961 to 1990. International Journal of Advertising, 11(3), pp.269–278.

Venkateswaran, M. & Kinnucan, H.W. (1990) Evaluating fluid milk advertising in Ontario: the importance of functional form. Canadian Journal of Agricultural Economics, 38(3), pp. 471–488.

Waterson, M.J. (1984) Advertising on Cigarette Consumption (5th edn). London: The Advertising Association.

Waterson, M.J. (1986) An Analysis of the Evidence Concerning the Impact of Advertising on Cigarette Consumption. London: The Advertising Association.

Waterson, M.J. (1990) Advertising and tobacco consumption: an analysis of the two major aspects of the debate. International Journal of Advertising, 9, pp. 59–72.

Wilcox, G. (2001) Beer brand advertising and market share in the United States: 1977 to 1998. International Journal of Advertising, 20(2), pp. 149–168.

http://www.warc.com © Advertising Association