The advertising sensitivity of different product classes to business cycles: an empirical investigation

(1)

The advertising sensitivity of different product classes

to business cycles: an empirical investigation

(2)

The effect of business cycles in advertising sensitivity for different product classes Page 2

The effect of business cycles in advertising sensitivity different product classes:

an empirical investigation

by

Erwin Strijker

University of Groningen

Faculty Economics and Business Administration Msc Marketing Research November 2012 Bossemaat 4 7933 RJ Pesse (0528)241595 (06)18035436 E.Strijker@student.rug.nl Student number: 1736361

(3)

The author conducted an empirical investigation into the cyclical comovement elasticity of advertising expenditures for different product classes and product types in the Netherlands on a weekly aggregation level. This aggregation level is never used in previous (similar) research which made use of annual data. 76 product types were used and bundled into six product different classes; food & drinks, care product, household products, durables, services and retail. It was attempted to provide a proper answer to the following problem statement:

What kind of effect do business cycles have on the advertising level on different product classes such as food & drinks, care products, household, durables, services and retail?

In order to investigate this, the Cristiano and Fitzgerald full sample filter was used in order to remove the linear trend of the advertising level and the level of GDP. Then, a regression analysis was performed to investigate what the effect of business cycles is on the six different product classes. Three product classes showed an elastic countercyclical effect. These product classes are food & drinks, care products and durables. One product class showed a inelastic counter cyclical comovement elasticity (retail). Household products and services are the product classes which have a pro-cyclical effect. Thus, it can be seen that there are large differences between different product classes instead of taking the whole industry as aggregation level as is done by previous research. Furthermore, some reliability test are performed in order to look if the resulted values were reliable. These tests consist of another base year of the GDP, an examination of extreme values and a lagged effect, because it is assumed that the sales of a previous period is important for the advertising budget of next year. From these tests, the results are considered as reliable because the values were robust. Moreover, the comovement elasticities from the lag effect were much extreme that they do not were accepted as reliable.

(7)

The effect of business cycles in advertising sensitivity for different product classes Page 7 3. INTRODUCTION

Advertising is the most visible and also one of the most important marketing instruments for a firm. (Deleersnyder et al., 2009.; Sethuraman, 2011). Advertising has several advantages for a firm. Firstly, advertising contributes toward building strong brands that enable a firm to earn a price premium relative to competing brands and reduces a firms vulnerability to competition (Keller, 1998). Moreover, strong brands serve as market entry barriers for potential competitors. Thus, advertising could lead to a more recognized brand to consumers and it increases competition for competitors. Also, advertising has a significant long term effect on a firm's sales by influencing the attitudes of consumers and changing their consumption behaviors (Aaker, 1996; Keller, 1998). The arguments mentioned above, could all lead to positive effects why a firm should decide to invest in advertising.

Each firm maintains a different level of advertising. Some companies spend a lot of resources on advertising while some firms do not spend anything on advertising at all. According to Danahan (1992), there are several methods that could be used to determine the level of advertising spending for a firm. Examples of methods that are used by firms are: A certain percentage of last year’s sales, matching competitors percent of sales, setting share of voice equal to targeted market and advertising intensiveness curves. Each firms maintains their own method on how much resources they will invest on advertising.

As said before, every firm maintains a different method to determine the amount of advertising spending and the level of advertising differs between firms. Although advertising has positive signs towards a firm or brand, it is one of the first cuts for a firm when there is an economic downturn. This results in a widespread reduction in advertising activity by many companies during a recession which translates into a considerable drop in a country’s aggregate advertising spending. Conversely, when the economy goes up, advertising expenditure accelerates (Frankenberger and Graham 2003). Thus, according to Frankenberger and Graham (2003) the advertising level correlates along with the economy. This is confirmed by Tellis and Tellis (2009) who state that in their review on advertising in a recession, that advertising is strongly related to economy.

(8)

The effect of business cycles in advertising sensitivity for different product classes Page 8 consume less and their ability is decreasing due to shrinking purchasing power and a lack of resources, is it still useful to advertise in economic recessions? Why would a firm invest in advertising while their turnover drops because of a decrease in consumption? These are important questions, because firms are in uncertainty if the investments in advertising will be earned back. Apparently, concluded by previous research, (Frankenberger and Graham 2003; Tellis and Tellis (2009) most firms will not take this risk and will cut down their advertising spending.

This raises the question whether it is better or not for firms to decrease their advertising spending during economic recessions? Most researchers argue that firms ought to increase their advertising expenditures while there is an economic recessions (American Business Press [ABP] 1993; Biel 1998; Dhalla 1980; O’Toole 1991; Frankenberger and Graham, 2003 and Srinivasan and Lilien (2005). Graham and Frankenberger (2011) state that ‘’ advertising and promotion during such economically stressful times could serve to make consumers feel more comfortable and secure at a time when consumer confidence is low.’’ Some other argue that decreasing advertising spending in recessions does not affect firm performance (Kijewski 1982). To conclude, most research state that the firm performance (for example, turnover and profits) of firms which advertise in economic downturn is better compared to firms who do not advertise during recessions.

(9)

The effect of business cycles in advertising sensitivity for different product classes Page 9 advertising level. These studies are the only three recent empirical studies which measured the effect of business cycles on (aggregate) advertising spending. Thus, there has not been done much research on this field yet. There are some studies who conducted a study on the advertising level of different market categories such as consumer, industrial and service markets (Farris and Buzzell (1979); Balasubramanian and Kumar (1990). However, these studies did not include the effect of business cycles on the advertising level.

(10)

The effect of business cycles in advertising sensitivity for different product classes Page 10 The next problem statement will be assigned during this study:

What kind of effect do business cycles have on the advertising level on different product classes such as food & drinks, care products, household products, durables, services and retail?

(11)

The effect of business cycles in advertising sensitivity for different product classes Page 11 4. THEORETICAL FRAMEWORK

There is lots of research whether or not it is better for a firm to increase their advertising level (during recessions) or not and what kind of impact it has on different kind of firm measurements such as sales or market share. First empirical research is done by Vaile (1927) who investigated the effect of advertising on sales. The author used data of 250 firms on a time span of four years. The conclusion of the study was that increasing advertising will lead to increasing sales for each of the investigated four years. Moreover, the author found that a decrease in advertising also leads to a decrease in sales. A last conclusion was that firms that did not advertise at all, had stable sales during that certain time period of four years. Therefore, a better firm performance is achieved when firms decide to invest in advertising.

Last decades, there has been more research on the advertising level and the state of the economy. It became more and more a question whether or not to invest in advertising during economic downturn. One of the first studies to this topic has been conducted by King and Biel (1990) who linked the advertising level with market share. They also include the state of the economy to advertising as they used several time periods such as a recessionary, normal of expansionary period. The authors concluded that during a recession, advertising will increase market share. Moreover, when a firm maintained a large increase of their advertising level, the market share of that firm was increasing even more. Furthermore, advertising during a recession is more productive compared to advertising in an expansion. Thus, according to King and Biel (1990), firms ought to increase their advertising level with a large amount when the economy is in state of recession in order to increase their market share.

Frankenberger and Graham (2003) also conducted a study on the effect of advertising in recessions. The conclusion was in the same as the research by King and Biel (1990); an increase in advertising during recessions resulted in significant increases in earnings. This result is greater when the advertising level increased during recession than economic expansion. This holds that the return of advertising during a recession is higher compared to advertising during economic expansion.

(12)

The effect of business cycles in advertising sensitivity for different product classes Page 12 that the first-difference operator emphasize short term fluctuations. Moreover, some studies investigated the long term relationship between the advertising level and the economy (Jung and Seldon, 1995). These studies provided important and interesting insights on the effect of economy to the advertising level. One further limitation of these studies is the used time span of used data which were too short in order to provide proper conclusions on several business cycles.

According to Christiano and Fitzgerald (1998), business cycles last between one and a half and eight years. There are a few recent empirical studies which linked the advertising level with the business cycles and satisfy to this requirement to contain at least one or several business cycles. These researches have been conducted by Deleersnyder et al. (2009) and Lamey et al. (2007 and 2008). The above mentioned studies capture a time span of at least 20 years. Data with a time span of 20 years are therefore large enough to capture one or multiple business cycles which could result in more precise and better conclusions.

Deleersnyder et al. (2009) focused on the aggregate advertising level and business cycles. They also investigate the role of national cultures to the advertising level during recessions. Lamey et al. (2007 and 2008) performed a similar study, however they did not include the role of culture and, the advertising level was pointed towards private label products instead of aggregate advertising spending.

Lamey et al. (2007 and 2008) did research containing aggregate advertising spending for private label products. They conducted two researches; one study on aggregate level (2007) and one study for different private label product categories (2008). In their first study (2007), they used annual data in a time span for 20-30 years. In this research, the authors used the Hodrick and Prescott (1977) filter (hereafter: HP filter) to extract the cyclical components and quantified the extent of cyclical sensitivity for private label through the cyclical comovement elasticity. In their first study, the researchers concluded that private label share behaves countercyclical. In other words, when there is economic downturn, the share of private label products increases and vice versa.

(13)

The effect of business cycles in advertising sensitivity for different product classes Page 13 purchases done by consumers, almost one-third of these products are private label products. Although this a high percentage of total sales, still two-third of total sales are not explained by this study. This could lead to a large gap concerning advertising level and the business cycle.

Deleersnyder et al. (2009) also investigated the aggregate advertising spending in comparison with the business cycle. Unlike Lamey et al. (2007 and 2008), which investigated the business cycles on private label sales, Deleersnyder et al. (2009) used aggregate sales on different countries and the impact of culture on advertising spending. They used annual data of 37 countries on a time span of twenty years. Just like Lamey et al. (2007 and 2008), Deleersnyder et al. (2009) used the HP filter to extract the cyclical components in advertising and GDP. They concluded that different cultures have an considerable impact on the co-movement elasticity with GDP. Moreover, it was found that the average co-co-movement is 1.4. This explains, for example, that a 1% increase in GDP results in 1.4% rise in advertising spending and vice versa. Limitations of the studies of Deleersnyder et al. (2009) and Lamey et al. (2007 and 2008) is that first, the used data is annual, while this could be more disaggregated to monthly or weekly data. Then, more detailed business cycles could be captured in the used time span, which could lead to more and specific fluctuations. Secondly, this issue is only addressed for the economy of a country as aggregation level (Deleersnyder et. al 2009) while there also could be differences on a different aggregation level on for example, different product classes or product types. In one study, Lamey (2008) made use of a lower aggregation level (different product categories). However, because there was made use of private label products only, lots of information (only one-third of total purchases) is not taken along into that study.

(14)

(15)

The effect of business cycles in advertising sensitivity for different product classes Page 15 5. HYPOTHESES

It is evident that, when there is an economic decline, the ability and willingness of consumers to buy a product will decline and when there is economic expansion, the willingness and ability will increase (Katona, 1975). The willingness to buy for each product class or product type is different and moreover, each product class has a different need of purpose. Some products will be bought faster and on a more regular basis like milk and bread. These products have a high need of purpose because a consumer cannot live without it and are bought regularly. Some products are bought on a less regular basis, for example a car or a television. These products are considered as necessary however, the need of purpose is less compared to for example, food products.

According to Danaher (1992), it is very common that firms spend a percentage of last year’s sales on advertising. When there is a recession, aggregate consumption will be less compared with periods of economic expansion. Then, firms will tend to decrease their advertising level because of lower sales in previous year and will increase the level of advertising when there is economic expansion. Therefore, it is assumed that when the aggregate consumption is lower compared to a previous year, the advertising level of firms will decline as well.

Balasubramanian and Kumar (1990) and Farris and Buzzel (1979) were the first researchers who linked the advertising level on different market categories however, they did not included business cycles. Farris and Buzzell (1979) conducted a study why advertising and promotional costs vary on different variables. The authors used the advertising and promotional to sales ratio (A&P/S ratio). They also used different markets such as consumer market, capital goods, industrial market and services markets. After empirical evidence, it was concluded that there are variations in the ratio of advertising and promotion to sales on the different markets.

(16)

The effect of business cycles in advertising sensitivity for different product classes Page 16 for a 1% increase in advertising. The product types that were included are pharmaceutical, durables, food, nonfood and services. Three product types that are used in this study of Sethuraman et al. (2011), are also used in the current study. These product types (which are named product classes in this study) are durables, food and services. The authors concluded that durables have the highest advertising elasticity followed by services. Frequently purchased food and nonfood products have the lowest advertising elasticity. This states that when advertising level is high, more consumers will buy that certain product with a high advertising elasticity (durables) compared to product types with a lower advertising elasticity (frequently purchased food and nonfood products).

Lots of research has been conducted about advertising and the state of the economy. This differs from aggregate advertising (Deleersnyder et al., 2009) to different product types (Sethuramen et al., 2011). As concluded by previous research, aggregate advertising decreases when there is an in economic recession (Deleersnyder et al., 2011, Frankenberger and Graham, 2003). This results in the following hypothesis.

H1: Advertising expenditures of brands will decrease when there is an economic recession.

5.1 Food and drinks

Food & drinks are necessary products for everyone and therefore, they have to be bought on a regularly basis. Consumers will buy these products always, also when there is an economic recession. Howard and Seth (1969) state that the purchase cycle for food and drinks is short and purchase happens frequently. Because the purchase cycle for food products is short, consumers can not postpone the consumption of these products. Therefore, it is expected that consumption of these products remains stable. As said before, according to Danahan, it is very common that a certain percentage of previous years’ sales is invested to advertising. Thus, when consumption of food and drinks is stable, it is expected that the advertising expenditures of firms are stable as well. This is true according to the study of Sethuraman et al. (2011) who state that advertising elasticity for food and drinks is the lowest of all product types. Therefore the hypothesis is that the advertising level of food & drinks are not sensitive to GDP and the effect is pro-cyclical.

(17)

The effect of business cycles in advertising sensitivity for different product classes Page 17 5.2 Household products

In the study of Sethuraman et al. (2011), it is said that nonfood, nondurable products, which household products are, tend to be low involvement products. According to the same study, the advertising elasticity of nonfood products is lowest compared to other product types. Due to the low advertising elasticity of nonfood products, advertising has not a significant influence on sales of these products. Because household product are bought frequently, this product class is considered as a frequently purchased nonfood product. The usage of household products is on a less regular basis compared to food & drinks however, they are still bought and used on a regular base. For example, toilet blocks and paper towels are present in every household. They are always used however, they have not to be bought as frequent as food and drinks. Thus, when there is economic decline, it is possible that the consumption of consumers is less because of their low involvement and therefore, consumers will postpone the purchase in a recession as long as possible. Sales of household products are thus less stable compared with food and drinks. Because advertising has not a significant influence (i.e. low advertising elasticity) on purchasing behavior and consumers will postpone their consumption of low involvement products, less products will be consumed when there is economic decline followed by less advertising. This results in the next hypothesis:

H3: Advertising expenditures for household products are more pro-cyclical sensitive to business cycle fluctuations compared to the overall mean of all product classes.

5.3 Care products

(18)

The effect of business cycles in advertising sensitivity for different product classes Page 18 H4: Advertising expenditures for care products are less sensitive to business cycle fluctuations compared to the overall mean of all product classes.

5.4 Services and durables

For services and durables, the assumption is that the purchase necessity is the same for both categories; not regular. These are products which are purchased once in several years. This purchase time is also different for any durable product. For example, shoes and televisions are both durable goods. However, it is probable that shoes are bought more frequently than a car, but still, both products are bought infrequently. This is confirmed in the study of Howard and Seth (1969) who state the appliance cycle for durable products is ’lengthy and infrequent’. When there is a recession, financial status could be a restraint for a consumer to buy certain products (Howard and Seth, 1969). As said before, it is evident that when there is economic decline, the ability and consumer willingness of consumer to buy a product will decline and when there is economic expansion, the willingness and ability will increase (Katona, 1975). Thus, most consumers will not tend to buy expensive products, because they have a lack of financial resources and will postpone the purchase of durable goods until the ability to buy these products increases again. This is confirmed by the research of Sethuraman et al. (2011), which state that the advertising elasticity for this category is the highest for durables followed by services. It is assumed that advertising will be cut because firms decrease their advertising level when sales decline (Danaher, 1994). For durables and services, the sales will decline more compared to other product classes because of their high advertising elasticity. Moreover, Deleersnyder et al. (2005) conducted a study on business cycles and consumer durables. They investigated the cyclical volatility for 24 durable products in the United States and concluded that this volatility for these products was on average 2.250. They state that ‘Durables are therefore, in terms of their cyclical volatility, more than four times as sensitive to business-cycle fluctuations than the general economic activity.’ Thus, durable products are very pro-cyclical sensitive to fluctuations in business cycles. This results in the following hypothesis:

H5: Advertising expenditures for durables are very pro-cyclical sensitive to business cycle fluctuations compared to the overall mean of all product classes.

(19)

The effect of business cycles in advertising sensitivity for different product classes Page 19 these product types are erotic lines, recreation parks and car rent. When there is economic recession, the ability of purchasing these products declines. This results in lower consumption of services and thus, a reduced level of advertising. The above mentioned arguments result in the following hypothesis:

H6: Advertising expenditures for services are very pro-cyclical sensitive to business cycle fluctuations compared to the overall mean of all product classes.

5.5 Retail

The last category that will be investigated is retail. Retailers are firms that provide goods and/or services to consumers for personal use (Levy, 1992). This product class is a ‘tricky’ one, because of the fact that some retail stores sell regular sold products with a short purchasing cycle (like food & drinks) and certain retailer stores will sell durable goods with a long purchasing cycle like for example, a car seller. Thus, all kind of products classes are present in current product category and are purchased for personal use. This makes it more difficult to expect a certain effect on business cycles. Bell et al. (2011) conducted a study on the unplanned behavior of consumers when they are in a retail store. They state that managers invest considerable resources inside the store to influence shoppers, because there is the ‘belief’ that most purchases in retail stores are unplanned. Out of store advertising has no significant effect on the unplanned buying behavior of consumers. This effect is the same for economic recession and economic expansion. Because most advertising is out of store and it is believed by retailers it has little effect, it is assumed that the level of advertising is low but stable. Thus, due to the fact different product classes are sold by retail stores and that the advertising level is stable, it is expected that macroeconomic fluctuations have no effect on the advertising level. This leads to the following hypothesis:

(20)

The effect of business cycles in advertising sensitivity for different product classes Page 20 6. DATA

(21)

The effect of business cycles in advertising sensitivity for different product classes Page 21 Table 1: Examples product types and brands

A g g r e g a t i o n L e v e l Product Class

Food & Drinks Househol d

products

Care products

Durables Services Retail

(22)

The effect of business cycles in advertising sensitivity for different product classes Page 22 7. METHODOLOGY

7.1 Data conditions

In order to use the appropriate data and which contains the most and valuable information for this study, it has to satisfy certain conditions. Firstly, of each product type, the top five of brands that advertise the most in euro’s, will be used in this investigation. This because these brands are most valuable for research and provide current study with the most information. Secondly, the top five brands per product category have to advertise at least 10% of the total time span. For the total time span of ten years, it tantamount to 52 weeks of advertising. Brands which advertise less, have too little data (thus, less information) too draw fair conclusions about it and are not added into the analysis. Most brands who are in top five of advertising, mostly advertise more than 10%. Thus, there is little loss of information because most top brands of 76 product types are used in the analysis because little brands are deleted for further research. Next to the satisfactions of the data, some years from certain brands are deleted. This because these brands or products did not exist yet or were not present on the market anymore. For example, digital video cameras were (almost) not on the market in the first year of the used data (1994) and therefore, these brands did and could not advertise which resulted in a deletion of those years. This deletion of data was done on yearly level, which results in information loss of certain brands for one of more years. All brands and how many brands per product category which are used in this study, can be found in appendix 1.

7.2 Methodology

The methodology of this research contains four steps. Firstly, the appropriate level of the GDP has been calculated. Secondly, the full sample filter of Cristiano Fitzgerald will be used to remove the linear trend of the advertising level per brand. Third, a regression analysis is performed to quantify the co-movement of the product categories and brands to the business cycles. Fourth and last, a weighted Z-test is performed to test for the significance of each product class and product type.

7.2.1 Gross Domestic Product

(23)

The effect of business cycles in advertising sensitivity for different product classes Page 23 Then, the Cristiano and Fitzgerald full sample filter was performed on these values to remove the linear trend of this data. As last, the values of the quarterly GDP (after Cristiano and Fitzgerald full sample filter) were calculated into weekly values. For the value of week one in quarter one, the following calculation has been used;

Fq1 + (Fq2 – Fq1) * (N/13) Where:

Fq 1 = the value after (CF filter) for quarter one. Fq2 = the value for quarter two.

N = the number of week in that quarter. 13 = Number of weeks in a quarter.

For the other weekly values of GDP, the same calculations have been made as the calculation for week 1.

7.2.2 Cristiano and Fitzgerald full sample filter

(24)

The effect of business cycles in advertising sensitivity for different product classes Page 24 -.012 -.008 -.004 .000 .004 .008 .012 -.02 -.01 .00 .01 .02 5 10 15 20 25 30 35 40 Filter GDP Non-cyclical Cycle

Asymmetric (time-varying) Filter

-1.2 -0.8 -0.4 0.0 0.4 0.8 0.0 2.5 5.0 7.5 10.0 12.5 15.0 50 100 150 200 250 300 350 400 450 500 Filter Heineken Non-cyclical Cycle

Asymmetric (time-varying) Filter

Figure 1: Cristiano and Fitzgerald full sample filter GDP

(25)

The effect of business cycles in advertising sensitivity for different product classes Page 25 7.2.3 Linear regression

Third, a linear regression is performed with the log-transformed GDP as independent variable and the log-transformed advertising data per brand or product category as dependent variable. Previous data are the log-transformed values for advertising and GDP after the use of the Christiano and Fitzgerald full sample filter.

The regression formula is as follows:

The outcomes of the regression analysis are threefold; a co-movement for each used brand, one for each product class and a co-movement for each product type. The overall weighted beta for each product class and product category are calculated as follows:

̅ ( )

( )

In this way, the weighted beta for each product class and product type is calculated. Beta is the sign of the comovement elasticity.

7.2.4 Weighted Z-test

Fourth, a weighted Z-test is performed to test the significance of each product category. Each product class consists of several product types. The weighted Z-test tests whether the product class or product type is significant as a whole. First, every probability of each P-value is transformed to a Z-value. The Z-transform test converts the one-tailed P-values, from each of k independent tests into standard normal Z values. Then, the weighted Z-value of each product class or category is calculated (standard normal inverse) as follows:

̅ ( )

(26)

The effect of business cycles in advertising sensitivity for different product classes Page 26 8. RESULTS

8.1 Descriptive statistics

Firstly, some descriptive statistics of the used data will be outlined which are visualized in table 2. The descriptives that are presented and discussed are the average advertising (in €) per product class and the brand which advertises the most and least in each product class. Moreover, the range between the brand which advertises the most and least in advertising is visualized. Last, the average advertising spending during a contraction and during an expansion are outlined in the table. An expansion occurs when GDP is positive from period T – T-1 and a contraction occurs when the GDP is negative in period T – T-1. In this way, a first view can be seen whether a product category advertises more or less compared to the business cycle. Product class Average advertising spending (€) Highest average advertising (€ per brand) Lowest average advertising (€ per brand) Range Avg.adv. Contraction (€) Avg. adv. Expansion (€) Food & drinks 27,992 229,837 (Coca Cola) 1,248 (Van Dobben) 228,589 33,723 29,342 Care products 24,841 82,575 (Pampers) 6,130 (Kotex) 76,446 24,022 24,392 Household products 26,681 104,259 (Ariel) 3,081 (Javaanse Jongens) 101,178 27,150 33,160 Durables 9,848 41,244 (Senseo) 902 (Sikkens) 40,342 11,414 9,973 Services 17,158 79,124 (Disneyland Paris) 1,521 (AVIS) 77,603 19,768 15,516 Retail 27,915 132,710

(Seats & Sofa’s) 647

(Speelboom)

132,063 17,456 39,610

Table 2: Descriptive statistics data

(27)

The effect of business cycles in advertising sensitivity for different product classes Page 27 product classes mentioned earlier. This product class only advertises almost one-third compared to food and drinks; €9,848. Furthermore, the range for food and retail is much higher compared to durables. The range for care products and services is almost the same while the range for household products is somewhat higher (€101,178). Also, at brand level, it can be seen that Coca Cola advertises much more on average (€229,837) as number one of food & drinks compared to other number one in advertising of other product classes like for example Seats & Sofa’s (€132,710). The product classes food and drinks, durables and services have an higher average advertising spending during a contraction compared to average spending with an expansion. The other three product classes; care products, household products and retail spend more on advertising during an expansion than periods with contraction. From these numbers, it looks like that product classes which spend more during contraction have a negative co-movement elasticity with GDP and product classes which spend more during expansions have a pro-cyclical comovement.

8.2 Cyclical sensitivity of product classes

After the discussion of the first statistics of the data set, now the cyclical sensitivity of the different product classes are discussed. In table 3, the weighted comovement elasticities per product class are showed. A positive comovement indicates pro-cyclical movement whereas a negative comovement is indicating a counter cyclical movement.

Product class Weighted comovement P-value Z-value

Food & Drinks -6.09 0.00 3.18

Care Products -6.15 0.20 0.83 Household Products 7.71 0.00 13.24 Durables -7.65 0.35 -0.40 Services 2.94 0.00 18.04 Retail -0.39 0.00 5.43 Overall -2.86 0.00 13.66

Table 3 : Weighted comovement of product classes

(28)

The effect of business cycles in advertising sensitivity for different product classes Page 28 eight. From these three product classes, care products and durables are not significant (P = 0.20 and P = 0.35 respectively) while the other product classes are. Retail also behaves countercyclical, however it is inelastic (M = -0.39, Z = 5.43, P <.001). Household products and services have a pro-cyclical sensitivity. The comovement elasticity of household products is 7.71 (Z = 13.24, P <0.01) whereas the comovement elasticity of services is less elastic (M = 2.94, Z =18.04, P <0.01). The overall elasticity of all investigated product categories is -2.86 (Z = 13.66, P <0.01). Thus, the overall advertising has a countercyclical comovement elasticity. This states that when the GDP grows with 1%, the overall advertising declines with 2.86%.

8.3 Differences in product classes

Now that the weighted comovement elasticities of each product classes is known, the differences inside these classes are outlined. It is explained whether most product types have a pro-cyclical or a countercyclical ((in)elastic) comovement elasticity and whether every brand inside a certain product type shows the same effect. Moreover, some extreme or strange comovement elasticities for some (outlier) brands are discussed. All comovement elasticities of the product types can be found in appendix 2.

8.3.1 Food & drinks

(29)

The effect of business cycles in advertising sensitivity for different product classes Page 29 Figure 4: Filter values GDP and Brand

Figure 3: Filtered values GDP and Slim Fast

(M = -37.22, Z = 14.67, P <0.01). However, there are seven product types in which all brands do have the same type of cyclicality. Soup and sport drinks which have a pro-cyclical effect and edible fats, cereal products, subtropical fruits, frozen snacks and cat food which all have a countercyclical comovement elasticity. A last finding that will be outlined are brands which have a very extreme comovement elasticity from above 100. These brand are Slim Fast (countercyclical), Brand and Karvan Cevitam (pro-cyclical).

Brand and Slim Fast are respectively the brands with the highest pro-cyclical and countercyclical comovement elasticity of all product types in this study which advertise every week in the used time span of ten years. In figure 3 and 4, the filtered values of the GDP and advertising for Slim Fast (fig. 3) and Heineken (fig. 4) are visualized. From these figures, the difference between the most pro-cyclical and most countercyclical brand can be seen very clearly. On the vertical axis, the level of filtered values can be seen and on the horizontal axis the number of weeks are visualized.

-8 -6 -4 -2 0 2 4 1 ₂₉ ₅₇ ₈₅ 11 3 14 1 16 9 19 7 22 5 25 3 28 1 30 9 33 7 36 5 39 3 42 1 44 9 47 7 50 5

Filtered values GDP and Brand

GDP Brand -6 -4 -2 0 2 4 6 8 1 ₃₀ ₅₉ ₈₈ 11 7 14 6 17 5 20 4 23 3 26 2 29 1 32 0 34 9 37 8 40 7 43 6 46 5 49 4

Filtered values GDP and Slim Fast

(30)

The effect of business cycles in advertising sensitivity for different product classes Page 30 8.3.2 Care products

The overall cyclical sensitivity of care products is countercyclical (M = -6.15, Z = 0.83, P = 0.2). This can also be seen out of the number product types which have a countercyclical movement (6 product types) against the number of product types which has a pro-cyclical sensitivity (4 product types). The product type toothpaste has the lowest pro-cyclical comovement (M = 4.21, Z = 8.79, P <0.01) compared to shampoo which has a highest pro-cyclical comovement of 19.52 (Z = 10.23, P <0.01). The product types which have a countercyclical elasticity have a more wider range. Drugs against cold have a comovement elasticity of -1.22 (Z = 0.59, P = 0.72) whereas oral care has the highest sensitive countercyclical comovement (M = -25.61, Z = -15.88, P <0.01). In this product class, there are two product types in which all brands have the same cyclical effect; oral care (countercyclical) and diapers (pro-cyclical). For the other product types, all brands have a different cyclical effect. Two brands in this product class have an extreme value. These are Max Factor with a comovement elasticity of -105.36 (Z = -7.94, P < 0.01) and Impulse Body which has an elasticity of 164.78 (Z = 14.67, P < 0.01)

8.3.3 Household products

(31)

The effect of business cycles in advertising sensitivity for different product classes Page 31 8.3.4 Durables

Durables are the most sensitive to business cycles as can be seen from table 3 (M = -7.65) however, this value is not significant (P = 0.30). Five out of thirteen product types show a pro-cyclical effect to business cycles whereas the other eight product types show a countercyclical effect. Washing machines is the least pro-cyclical of these five product types with a comovement of 0.54 (Z = 5.67, P < 0.01). The most pro-cyclical product type is digital video cameras with a comovement of 40.84 (Z = 1.90, P < 0.01). With the countercyclical product types, the range is somewhat wider compared to the pro-cyclical product types. The lowest product group which acts countercyclical is televisions which are inelastic (M = -2.48, Z = -1.89, P <0.05). The product type with the highest countercyclical comovement elasticity is sport shoes (M = -48.98, Z = -7.94, P < 0.01). In five product types, all brands have the same direction of comovement. The brands of kitchen appliances, kitchen equipment and vacuum cleaner all have a countercyclical comovement elasticity whereas digital cameras and coffee/tea machines have a pro-cyclical comovement elasticity. Puma is the only brand in this product class which has an extreme value of -100.75 (Z= -7.94, P <0.01).

8.3.5 Services

(32)

The effect of business cycles in advertising sensitivity for different product classes Page 32 comovement effect; political parties. As can be seen from erotic lines (mentioned above), the comovement elasticities could be very different inside a certain product type.

8.3.6 Retail

The last product class, retail, is the least elastic of all product classes with an overall elasticity of -0.39 (Z = 85.43, P <.01). Here, the product types which are pro-cyclical have an elasticity around five (toy stores: M = 4.18, Z = 10.23, P <0.01 and furniture stores: M = 5.80, Z = 6.64, P <0.01). Garden centers and computer companies both behave countercyclical with both an elasticity of -2.06 (Z = -3.29, P <0.01) and -6.36 (Z = 2.26, P <0.05) respectively whereas transport stores are more elastic with a beta of -11.18 (Z = -4.93, P <0.01). Retail is the only product class in which the brands in every product type do not have the same effect on business cycle fluctuations. Some brands have a pro-cyclical comovement elasticity whereas other brands in the same product type have a countercyclical effect. Most extreme value in the product class is -107.82 (Z = -1.99, P <0.01) for the stores of Computer Correct.

8.4 Hypotheses

The comovement elasticities show that there are large differences between the six product classes. Hypothesis 1 states that firms decrease their advertising level when there is an economic recession. Thus, it predicts an overall positive comovement elasticity to GDP. However, the overall weighted comovement elasticity shows a countercyclical pattern (M = -2.86) and therefore, H1 is rejected. Moreover, there is not found support for H2 and H4 neither.

H2 posits that food and drinks are less sensitive to business cycle fluctuations compared to the

overall mean while it has a comovement elasticity of -6.09. Thus, the outcome is that food and drinks are very sensitive and more sensitive to the overall mean. Drinks alone are less sensitive compared to food, but still its comovement elasticity is above 1 concluding that this is elastic and sensitive as well. Also, care products are very sensitive to the state of the economy (M = -6.15) while H4 states that care product are less sensitive resulting in a

rejection of hypothesis four. A drawback of this conclusion is that the comovement elasticity is not significant. The hypotheses for household products (H3) is accepted. H3 states that the

(33)

pro-The effect of business cycles in advertising sensitivity for different product classes Page 33 cyclical as expected. Although H5 is rejected, the value for it is not significant. H6 posits that

services are very pro-cyclical sensitive to business cycle fluctuations compared to the overall mean. This hypothesis is accepted, because the services are indeed pro-cyclical and it is more sensitive (M = 2.94) than the overall sensitivity (M = -2.86). The final hypothesis (H7) states

that retail is less sensitive to the state of the economy compared to the overall mean. This hypothesis is accepted due to the fact that retail has the lowest comovement elasticity and it is lower than the overall mean. Thus, this product class is indeed less sensitive.

In the figure below, the difference between the overall mean and all product classes are visualized.

Figure 5: Comparison overall mean and product classes

8.5 Reliability of regression analysis 8.5.1 GDP

In order to measure reliability of the GDP, we also did a regression analysis on 10 brands which were seen as very sensitive to GDP. Base year 2010 is used instead of base year 2012 of the GDP (less recent calculations of monetary value). The result is that the base year of 2012 is reliable due to the fact that the comovement elasticity was almost the same for GDP of the 2010 and 2012. Moreover, the overall correlation between both filtered GDP of both years was somewhat perfectly correlated (0.998) which indicates that GDP of both years is almost the same.

-6,09 -6,15 7,71 -7,65 2,94 -0,39 -2,86 -2,86 -2,86 -2,86 -2,86 -2,86 -10 -8 -6 -4 -2 0 2 4 6 8 10

Food & drinks Care products Household products Durables Services Retail

Comparison overall mean and product classes

(34)

The effect of business cycles in advertising sensitivity for different product classes Page 34 8.5.2 Extreme values

In order to check whether if the results for each product class are robust, all extreme values (Beta around 100 of -100) are deleted. This resulted in a deletion of 13 brands and results in the following values visualized in table 4.

Table 4 : Weighted comovement of product classes after deletion extreme values

All comovement elasticities still have the same effect and are less sensitive to business cycles except for care products and durables. This is caused by the deletion of some brands with an extreme comovement elasticity. For example, the beta for food and drinks was -6.09 whereas it has a comovement elasticity of -6.04 now. The values for care products and durables changed most (M = -6.15 to M = -7.46 and M = -7.65 to M = -6.06 respectively). For these product classes, the significance level decreases as well. It can be stated that all product classes are robust except for the product classes care products and durables. This because the change in comovement elasticities can be considered as high compared to the other product classes.

8.5.3 Lag series

The hypothesis mentioned in this study consist of the extent to which different product classes are sensitive to GDP. A reason for this formulation lies on the fact that firms let depend their advertising level on the sales level of previous period (Danaher, 1992). Therefore, there is also done a regression analysis of lagged GDP and advertising level. In this analysis, the GDP of one year before the advertising level are taken along. Thus, for the advertising level of week 1 in year 1995, the GDP level of week 1 in year 1994 is used. On this way, it can be investigated what the effect is of a lagged series and if this is correct. The results vary very much compared to series where there has not been made use of lagged values and also differs Product classes Weighted comovement P-value Z-value

Food & Drinks -6.04 0.00 2.73

(35)

(36)

The effect of business cycles in advertising sensitivity for different product classes Page 36 9. CONCLUSION

9.1 Conclusion

Mostly, business cycles do affect the advertising spending of most firms. When there is economic downturn, advertising spending goes down and vice versa (Frankenberger and Graham, 2003). Most of this research is done on yearly GDP data and for the industry as a whole. However, this research focusses on weekly data of advertising spending and on a lower aggregation level on what impact the economy has; brand level. Moreover, the influence of GDP on different product types like soda and candy are estimated and also, several product classes like durables and services are.

Firstly, some previous research about the effect of business cycles on the industry and other variables were explained. This was followed by the hypothesis of how sensitive different product classes are with fluctuations in business cycles. Then, the advertising level and GDP were filtered with the full sample filter of Cristiano and Fitzgerald. With this filtered values, a regression analysis was performed with GDP as independent variable and advertising of a brand as a dependent variable. Furthermore, a weighted Z test was performed in order to get a weighted value of all product types into a certain product class or product type. The results are different compared with previous research. Finally, some reliability tests consisting of GDP and lag series were performed.

It was found that the overall comovement elasticity in this study is -2.86 which is countercyclical and elastic. This also states that aggregate advertising is more sensitive than the economy. In previous research, Deleersnyder et. al (2009) found a pro-cyclical comovement elasticity of 1.91. The cyclical sensitivity of both studies are elastic however, the effect is contrary. Although the values of both studies are very contrary, the cause could be that Deleersnyder et. al used annual data while weekly data is used in current study. In the current study, much more ‘points’ are used; 520 weeks compared with 20 annual points of Deleersnyder et al.

(37)

The effect of business cycles in advertising sensitivity for different product classes Page 37 of 2’. Also, Sethuraman et al. (2011), found a pro-cyclical comovement elasticity. Thus, the value for durables in current study differs from comovement elasticities and the effect is contrary compared with previous research. The estimated value for durables is not significant though.

The finding for food and drinks is also different than expected, because the comovement elasticity for food and drinks is countercyclical elastic instead of less sensitive and pro-cyclical. Thus, the advertising expenditures rises very much if the GDP rises by one percent and vice versa. It was supposed that advertising would be stable due to low advertising elasticity for food (Sethuraman et al. (2011).

Care products differ also from the hypothesis although it is not significant. The hypothesis posits that care products are less sensitive to business cycles, however they are countercyclical and sensitive. It is less contrary than the two other product classes discusses above. It could be possible that producers of care product are afraid to a loss of sales during recessions and raise advertising level in the hope of increasing sales again.

Household products are indeed more sensitive than the other product classes. A supposed, consumers will postpone the purchase of household products as long as possible when there is economic downturn. Then, advertising will decrease as well which leads to a pro-cyclical comovement elasticity (M = 7.71).

The hypothesis for retail and services are both accepted. For retail, it was expected that the advertising level was stable due to the fact that advertising does not have a significant effect on sales and different product types are represented in this product class. This is confirmed by the comovement elasticity of retail which is -0.39 which is inelastic.

Most services which are taken along in this study, represent more luxury products. This could be the reason why the effect is elastic and pro-cyclical. People tend to buy less products of this product class when there is economic downturn. Firms react on the reaction of the customer and decide not to invest in advertising, because turnover is under pressure.

(38)

The effect of business cycles in advertising sensitivity for different product classes Page 38 which consists of weekly data in current study instead of annual data which has been used on previous studies. This difference of weekly and annual data could consist of more precision for weekly data compared to annual data due to the higher level of degrees of freedom (Ganjaberndee, 1972). Moreover, in a study of Dekimpe and Hanssens (2000), they state the weekly or monthly data results in a more precise estimation in post-promotional dips and competitive reaction patterns. This precise estimation is harder to detect with temporally aggregated data like annual data. The above mentioned arguments could be reasons for the counterintuitive results compared with previous studies, because there is made use of a different temporal interval compared with previous studies.

9.2 Managerial implications

Contrary to previous research, firms do not cut back their advertising level when there is an economic recession. Deleersnyder et. al (2009) conclude that companies which had a pro-cyclical comovement, that these companies ‘perceive advertising as a short-term elastic expense. ‘ From this study it can be said that most firms perceive advertising as a long-term investment for itself, because business cycles have a countercyclical effect on most product classes. However, there are still some differences between the used product classes. Services and household products are the two product classes who do have a pro-cyclical comovement elasticities. For these product classes, advertising is apparently still not perceived as an investment. From previous research, it is found that firms which implement a counter cyclical advertising strategy, their long term firm outperforms firm which implement a pro-cyclical advertising strategy. Thus, it would be better for product types and brands which have a pro-cyclical elasticity, to invest in advertising when there is economic recession to increase firm performance. Moreover, a benchmark with other product types can be easily conducted. If a certain product type has a countercyclical effect and the performance of this product types is well, it could be easier for a pro-cyclical brand of product type to decide to follow the same strategy.

9.3 Limitations and further research

(39)

The effect of business cycles in advertising sensitivity for different product classes Page 39 for a product class. Although the used product types are very extensive and widespread, it could be that important product types are not used in current study. Furthermore, only brands that advertise the complete time span could be taken along in a next study in order to lose as little information and thus, to include as much as information as possible.

(40)

The effect of business cycles in advertising sensitivity for different product classes Page 40 10. REFERENCES

Ashley, R., Granger, W.J, and Schmalensee, R., (1980). Advertising and Aggregate Consumption: An Analysis of Causality. Econometrica, 48 (5): 1149–67.

Balasubramanian, S.K. and Kumar, V. (1990). Analyzing Variations in Advertising and Promotional Expenditures: Key Correlates in Consumer, Industrial, and Service Markets. Journal of Marketing. Vol. 54 (2): 57-68.

Bell, D.R., Corsten, D. and Knox,G., (2011). Journal of Marketing. Vol. 75 (1): 31-45. Danaher, P.J. and Rust R.T. (1994). Determining the Optimal Level of Media Spending. Journal of Advertising Research. 34: 28–34.

Deleersnyder, B., Dekimpe M.G., Sarvary M., and Parker P.M.,(2004). Weathering Tight Economic Times: The Sales Evolution of Consumer Durables over the Business Cycle. Quantitative Marketing and Economics. 2: 347–83.

Deleersnyder, B., Dekimpe, M.G., Steenkamp, J.E.M., and Leeflang, P. (2009). The Role of National Culture in Advertising’s Sensitivity to Business Cycles: An Investigation Across Continents. Journal of Marketing Research. 46 (October): 623–636.

Dekimpe, M.G. and Hanssens, D.M. (2000). Time series models in marketing: past, present and future. International journal of research in marketing. Vol. 17 (2): 183-193.

Didow, N.M., and Franke, G.R., (1984). Measurement Issues in Time-Series Research: Reliability and Validity Assessment in Modeling the Macroeconomic Effects of Advertising. Journal of Marketing Research. 21 (February): 12–19.

Farris, P.W. and Buzzell, R.D., (1979). Why advertising and promotional costs vary: some cross-sectional analyses. Journal of Marketing. Vol. 43 (4): 112-122.

Frankenberger, K.D. and Graham, R.C., (2003). Should Firms Increase Advertising Expenditures During Recessions?. Marketing Science Institute Report No. 03.

Ganjaberndee, S. (1972). The effects of temporal aggregation on a model of the Australian monetary sector. Economic Record. Vol. 48 (124): 527-535.

Graham, R.C. and Frankenberger, K.D., (2011). The Earnings Effects of Marketing Communication Expenditures During Recessions. Journal of Advertising. Vol. 40 (2): 5-24. Heerde van, H.J., Leeflang P.S.H., Wittink D.R., (2001). Semiparametric analysis to estimate the deal effect curve. Journal of Marketing Research. 38 (May): 197–215.

Heerde van, H.J., Leeflang P.S.H., Wittink D.R.,(2004). Decomposing the sales promotion bump with store data. Marketing Science. 23 (3): 317–334.

(41)

The effect of business cycles in advertising sensitivity for different product classes Page 41 Hodrick, R.J. and Prescott E.C., (1997). Postwar U.S. Business Cycles: An Empirical Investigation. Journal of Money, Credit and Banking. 29 (February): 1-16.

Katona, G., (1975). Psychological Economics, New York: Elsevier Scientific Publishing Co. Kijewski, V. (1982). Media Advertising When Your Market Is in a Recession. Cahners Advertising Research Report, Strategic Planning Institute.

Lamey, L., Deleersnyder, B., Dekimpe M.G., and Steenkamp, J.E.M., (2008). How to Mitigate Private-Label Success in Recessions? A Cross Category Investigation. Working Paper, Catbolic University of Leuven, Department of Marketing and Organization Studies. Lamey, L., Deleersnyder, B., Dekimpe M.G., and Steenkamp, J.E.M., (2007). How Business Cycles Contribute to Private-Label Success: Evidence from the United States and Europe. Journal of Marketing. 71 (January): 1–15.

Levy, M., and Weitz, B., 1992. Retailing Management, Homewood IL, Richard D. Irwin. Sethuraman, R., Tellis, G.J. and Briesch, R.A., (2011). How well does advertising Work? Generalizations from meta-analysis of brand advertising elasticities. Journal of Marketing Research. Vol. 48 (3): 457-471.

Srinivasan, R., Lilien, G.L., and Sridhar, S., (2011). Should Firms Spend More on Research and Development and Advertising During Recessions?. Journal of Marketing. Vol.75 (3): 49-65.

Srinivasan, R., Rangaswamy, A. and Lilien, G.L., (2005). Turning Adversary into Advantage: does Proactive Marketing During a Recession Pay Off? International Journal of Research in Marketing. 22: 109-125.

Steenkamp, J.E.M., Nijs, V.R., Hanssens, D.M., Dekimpe, M.G., (2005). Competitive Reactions to Advertising and Promotion Attacks. Marketing Science. Vol. 24 (1): 35-54.

Tellis, G.J. and Tellis, K., (2009). Research on Advertising in a Recession A Critical Review and Synthesis. Journal of Advertising Research. Vol. 49 (3): 304-327.

Tikoo, S. and Ebrahim, A., (2010). Financial Markets and Marketing: The Tradeoff between R&D and Advertising During an Economic Downturn. Journal of Advertising Research. Vol. 50 (1): 50-56.

White, J.B. and Miles, M.P., (1996). The Financial Implications of Advertising as an Investment. Journal of Advertising Research. 36 (July–August): 43–52.

(42)

(43)

The effect of business cycles in advertising sensitivity for different product classes Page 43 11. APPENDICES

11.1 Appendix 1: Brands used in study Product type (# of brands): Brands

Sport drinks (4) Extran, Aquarius, Karvan Cevitam Sportdrank, AA Drink

Soda Coca Cola, Fanta, Pepsi, Seven Up, Sisi

Beer Heineken, Amstel, Grolsch, Bavaria, Brand

Drop (4) Venco, Klene, Autodrop, Oldtimers

Chewing Gum Stimorol, Sportlife, Freedent , Xylifresh, V6 kauwgom

Coffee Douwe Egberts, Idee Koffie, Kanis & Gunnik, Max Havelaar, Van

Nelle

Pretzels Croky, Pringles, Smiths, Duyvis, Klijn Noten

Cheese Frico, Kaas Collectief, Maaslander, Milner, Slankie

Candy Fruittella, Haribo, M&M’s, Red Band, Werther’s snoep

Fruit juice Appelsientje, Coolbest, Hero, Roosvicee, Taksi

Whisky Jameson, Johnnie Walker, The Famous Grouse, William Lawson,

Ballantines

Biscuits Bolletje koeken, Liga Biscuits, Lu Koekjes, Sultana, Verkade

Candybars Cote D’or, Kitkat, Mars, Snickers, Twix

Cat food (4) Kitekat, Gourmet, Sheba, Whiskas

Mealseducer complete warm

Heinz, Honig, Knorr, Grand Italië, Conimex Mealseducer components

cold (4)

Calve, Heinz Sauzen, Knorr Sauzen, Remia Sauzen

Dietetic Pool Plus, Profiel, Slimfast, Voldafar, Slankvitaal

Mixdranken premixen (3) Bacardi, Passoa, Smirnoff

Gingerbread (3) Kapitein Koek, Peijnenburg, Wieger Ketellapper

Potatoes (3) Celavita, Aviko, Mccain

Subtropcial fruit (3) Appels collectief, Chiquita, Max Havelaar Fruit

Soup Dry (3) Honig, Knorr, Unox

Meals Conimex, Honig, Knorr, Struik, Unox

Breakfast products (4) Brinta, Hero Fruitontbijt, Kellogg’s, Quaker

Milk (4) Campina, Domo, Melk Collectief, Melkunie

Ice impulse(3) Ola/Cornetto, Ola/Solero, Ola/Magnum

Edible fats (4) Becel, Blue Band, Croma, Diamant

Vlees Vis gevogelte snacks diepvries (3)

Beckers, Mora, Van Dobben

Inlegkruisjes (3) Alldays, Libresse, Kotex

Diapers (3) Pampers, Huggies, Libero

Toothpaste Colgate, Aquafresh, Paradontax, Sensodyne, Prodent

Deodorant Impulse bodyspray, Impulse deo spray, Nivea, 8x4 , Rexona

Shampoo Andrelon, Garnier, Head & Shoulders, L’oreal/Elseve, Pantene

Lipstick L’oreal, Lancome, Max Factor, Maybelline, Rimmel

Cold and cough preparations

A Vogel, Otrivin, Vicks, VSM, Bisolvon

Pain killers Aleve, Advil, Panadol, Nurofen, Saridon

Oral care (4) Kukident, Listerine, Steradent, Aquafresh

Bath/Shower products Badedas, Fa, Kneipp, Nivea, Sanex

Shag (4) Drum, Jakobs, Javaanse Jongens, Van Nelle

Batteries (3) Duracell, Panasonic, Philips

Soucepan (3) BK, Sola, Tefal

Newspapers Algemeen Dagblad, De Telegraaf, De Volkskrant, NRC

(44)

Air freshener Airwick, Ambi-Pur, Brise, Febreze, Robijn

Detergent (cotton) Ariel, Corol Reus, Omo, Persil, Robijn

Detergent wash Ariel, Dash, Omo, Persil, Sunil

Hygiene wipes (3) Cif, Glassex, Glorix

Toilet blocks (3) Ambi Pur, Bref, Glorix

Paper towels (3) Bounty , Edet , Lotus

Kitchen Appliances (4) Atag, Miele, Siemens, Zanussi Digital video camera (3) JVC, Panasonic, Sony

Televisions JVC, Panasonic, Philips, Samsung, Sony

Vacuum cleaner (2) Miele , Rowenta

Sport shoes Adidas, Asics, Nike, Reebok, Puma

Paint (3) Flexa, Histor, Sikkens

Brown goods B & O, JVC, Philips, Samsung, Sony

Color printers (4) Canon, Epson, Hewlett Packard, Xerox

Washing machine (3) AEG, Miele, Zanussi

Audio combination (4) B&O audio, Bose, JVC, Sony Coffee/Tea machine (3) Krups , Nespresso, Senseo

Dishwashers (2) AEG, Miele

Kitchen equipment (4) Atag, Miele, Siemens, Zanussi

Pay services ABN Amro, Beanet, ING Bank, Postbank, Rabobank

Erotic lines (4) 09 ITS Lyn, Pek 0906-lijnen, Rodi Media , Telemade

Recreation parks De Efteling, Disneyland Parijs, Duinrell, Walibi World, Warner Bros Movie World

Mortgage service ABN Amro, Fortis Bank, Postbank, Rabobank, SNS Bank

Universities Open Universiteit Nederland, TU Delft, Universiteit Leiden,

Universiteit van Amsterdam, Vrije Universiteit Opleiding

Political Parties (4) CDA, PvdA, SP, VVD

Car rent (4) Avis, Budget Rent-A-Car, Europcar, Hertz

Funeral services (4) Dela, Monuta, NUVA, Yarden

Car insurances ANWB, Centraal Beheer Achmea, FBTO, Ohra, Polis Direct

Travel insurances ANWB, FBTO, GWK, Postbank, Rabobank

Retail transport Bovag Collectief, Focwa, Halfords, Kwik Fit, Profile Tyrecenter

Retail computer Computer Company, Correct, Dynabyte, Vobis, Mycom

Retail garden centers (3) Intratuin, Europatuin, Groenrijk

Retail toys Bart Smit, Intertoys, Top 1 Toys, Toys ‘R Us, Speelboom

(45)

The effect of business cycles in advertising sensitivity for different product classes Page 45 11.2 Appendix 2: Co-movements of all product types

Product class Weigthed

comovement Z-Value P-value Voeding Sports Drinks 20.63 15.45 0.00 Soda 10.46 11.05 0.00 Beer 15.33 16.98 0.00 Drop -9.38 -1.64 0.05 Chewing Gum -5.15 -3.01 0.00 Coffee -0.72 6.54 0.00 Pretzels -23.44 -6.76 0.00 Cheese (Holland) -30.97 -8.95 0.00 Candy -3.88 0.90 0.18 Fruit juice -23.49 -5.48 0.00 Whisky 4.67 3.65 0.00 Biscuits -29.69 -7.65 0.00 Candybars -3.81 -2.23 0.01 Catfood (wet) -19.70 -12.30 0.00 Mealseducer complete/warm 2.28 7.95 0.00

Mealseducer components cold -1.90 -1.63 0.05

Dietetic -11.79 1.48 0.07

Mixed drinks/ Premixes -34.43 -7.27 0.00

Gingerbread -29.32 -10.83 0.00 Potatoes -2.60 1.99 0.02 Subtropical Fruit -6.19 -1.19 0.12 Soup dry 42.43 25.41 0.00 Meals -3.48 6.16 0.00 Breakfast Products -11.25 1.03 0.15 Milk -20.93 -2.93 0.00 Ice (impulse) -1.16 3.32 0.00

Edible fats (liquid) -38.86 -13.17 0.00

Frozen snacks -8.40 -0.53 0.30 Care products Cross inlays 11.10 4.82 0.00 Diapers -39.66 0.15 0.44 Toothpaste 4.21 8.79 0.00 Deodorant 14.13 3.00 0.00 Shampoo 19.52 10.23 0.00 Lipstick -10.08 -1.89 0.03

Cold and Cough preparations -1.22 0.59 0.28

The advertising sensitivity of different product classes to business cycles: an empirical investigation