In: Journal of Environmental Management, July 2001, Volume 62 Issue 3, pages 271-282.

ENERGY EFFECTS ASSOCIATED WITH E-COMMERCE: A CASE STUDY CONCERNING ONLINE SALES OF PERSONAL COMPUTERS IN THE NETHERLANDS

Lucas Reijnders, Martijn J. Hoogeveen

Open University Netherlands

Postbus 2960, 6401 DL Heerlen, the Netherlands
Fax +31 30 2250519.  Tel +31 6 54 25 64 76

lucas.reijnders@ou.nl martijn.hoogeveen@ou.nl

Abstract

The introduction of e-commerce is changing purchase and distribution patterns dramatically. One of the observed effects is that logistics become more efficient as products are directly shipped from a manufacturer or wholesaler to an end-user. Another effect is that market transparency increases, which has a downward pressure on prices of many products sold via the Internet. This article addresses the energy implications of e-commerce at the micro level. This is done by quantifying the transport related energy savings in the case of a Dutch online computer reseller and by assessing the extra energy expenditure associated with increased buying power of online buyers. It is found that energy use per article sold by the online computer reseller is lower. However, taking into account indirect effects such as increased consumer buying power, there are scenarios that lead to an overall increase in energy use.

1. Introduction

The emergence of the Internet has led to the breakthrough of electronic commerce (e-commerce). We define e-commerce in this context narrowly as the purchasing of goods and services via the Internet. As e-commerce is booming, it seems worthwhile to look at environmental impli­cations of e-commerce. Especially, because it seems reasonable to assume that e-commerce may lead to shortening supply chains and related logistic chains. To explore the possible environmental implications of e-commerce in retail, we concentrate here on energetic implications, as energy implications are relatively well documented.

These energetic implications follow from three aspects of e-commerce applied in retail. Firstly, as distribution patterns associated with e-commerce differ from traditional ways of distribution, the energy use associated with e-commerce based distribution will also be different. Online purchased goods tend to be transported directly from a wholesaler or manufacturer to a customer, avoiding a physical detour via the stock facility of a retailer. Also physical mobility of the customer, necessary for visiting a retailer, is substituted by visiting a web site using a computer. Moreover, an efficient e-tailer (electronic retailer) can operate with a smaller sales staff, and has less space requirements than a traditional retailer. This limits energy consumption for mobility and space heating and cooling. Also the use of paper can be less for an e-tailer than in classical retailing, as a web site replaces printed catalogues.

A second effect of e-commerce, energetically relevant, is its impact on the type of expenditures. A rapidly increasing share of expenditures is realised through e-commerce. If being online influences the purchasing profile of a shopper, this may have implications for the energy intensity of the overall demand, given that the energy intensity of purchased products varies.

Thirdly, e-commerce may be associated with price reductions, and thus increased buying power. The way in which this increased buying power is applied has again energetic implications.

2. Direct energy use associated with e-commerce based PC distribution

 It is claimed that e-commerce strongly improves the energy efficiency of the economy (Romm et al., 1999). However, at micro level there is as yet not much evidence that either supports or refutes this claim. To obtain insight in the energetic implications of e-commerce at micro level, we have studied the case of online sales by computer reseller Take it Now in the Netherlands (http://www.takeitnow.nl/). Online computer sales merit consideration as computer hardware is frequently purchased online, and the computer sector is relatively complicated from a logistic point of view, and thus has considerable potential to benefit from e-commerce. Take it Now is interesting because it seeks to organise its activities as efficient as possible. Take it Now mainly sells personal computers to small and medium sized enterprises (SMEs) and households in the Netherlands. Customers are split 50-50 between SMEs and households. We will concentrate on the latter to be able to make comparisons with available data.

What are the potential energy effects and financial savings associated with computer distribution? To characterise the current energetic impact of computer distribution, we made use of a life cycle analysis published in 1998 concerning a representative generic personal computer (PC) in a European context (Atlantic Consulting and IPU, 1998). This PC includes a control unit with CD-ROM drive, a 15" SVGA colour monitor, a keyboard, a mouse and packaging (Atlantic Consulting and IPU, 1998). Table 1 shows the energy use associated with the pre-use stages of the personal computer life cycle.

Table 1: Primary energy use in MJ (MegaJoules) for pre-use stages of personal compu­ters (including packaging) Source: Atlantic Consulting and IPU (1998).       

We recalculated the direct energy use for mobility associated with distribution, using data collected at the Eidgenossische Technische Hochschule Zürich (Gruppe Energie-Stoffe-Umwelt ETH, 1996). The value of 13 MJ in table 1 seems a satisfactory value for lorry and van based transport linking factory, wholesaler and retail outlet. We calculated typical values for these combined elements of distribution ranging from 12.8 to 16.9 MJ. However, this value does not take into account energy use associated with the consumers' round trip by car to the retail outlet to pick up the computer. A typical roundtrip of 10 km leads to an average energy use of 29.3 MJ (Gruppe Energie-Stoffe-Umwelt ETH, 1996).

Energy use concerning the distribution stage is lowered at Take it Now, when compared with traditional ways of distribution, as the roundtrip of the consumer is eliminated. Furthermore, purchased PCs are shipped directly by the wholesaler to the consumer, eliminating transport to and from the retailer. On the other hand, computer based contacts with, and use of information of, Take it Now requires extra energy in contrast to traditional sales. For a buyer of a personal computer, energy requirements related with Internet use are expected to be at least a factor 10 less than the energy requirements associated with personal transport that Internet use replaces (Romm et al., 1999). We estimate that e-commerce distribution, as practised by Take it Now, reduces energy require­ments with roughly 30-40 MJ in comparison with more traditional distribution systems.  

3. Energy effect of PC expenditures

As the rise of the Internet is reported to have a stimulating effect on PC sales, it is not only relevant to focus on pre-use energy effects but also on energy use in the use stage. Especially, as most energy use during the life cycle of a PC, occurs in the use stage (see Table 2).

If one looks at the price of a PC, which is typically in the range of fl. 2000 to fl. 4000, its energy intensity is roughly between 1 and 2 MJ/guilder. This compares favoura­bly with the average energy intensity of household expenditures in the Netherlands, which ranges between 6.1 and 7.8 MJ/guilder, depending on the nature of the household (sin­gle/multi-person, old/y­oung), with standard deviations in the range of 1.0 to 1.5 MJ/guilder (Schneider, 1994). These energy intensities have been calculated on the basis of life cycle energy analysis of the products and services concerned.

Table 2: Primary energy use (in MJ) associated with the total life cycle of a personal computer and its use stage. Source: Atlantic Consulting and IPU (1998)

With an energy price of f 0.025/MJ, the life cycle energy inten­sity of a personal computer will be roughly 3-6 MJ/guil­der. This still compares favourably compared to the average energy intensity of household expenditure (6.1-7.8 MJ/guilder).

One may therefore conclude that when the direct effects of e-commerce based distribution are analysed the energy use and its associated environmental impact are reduced in comparison with more tradi­tional distribution systems. Moreover, the buyer pays via an e-tailer on the average a 10% lower price, whereas energy efficiency of computer related spending compares favou­rably with average expenditu­res. This seems to be a win-win situation.  

4. Limitations of distribution related energy analysis

There are a number of limitations with regard to above-given analysis related to PC distribution. First, non-distribution related direct effects of e-commerce are not quantified, as no benchmark data were available. For example, Take it Now operates no physical shop, employs less sales and logistic staff, and spreads no printed promotion materials. These differences strongly contribute to the efficiency gain that is translated into lower purchase prices. Secondly, energy intensities are only a partial reflec­tion of the overall environmental impact of PC usage. In the lifecycle of PCs hazardous chemical­s are used, that are relatively burdensome to the environment (Atlantic Consulting and IPU, 1998). So, one should be aware that the overall environmental perfor­mance of PCs, compared to other products, may be less favourable than energy intensity data suggest. On the other hand, the unfavourable impact of hazardous substances associated with the PC life cycle is undergoing a downward trend due to better environmental design, application of cleaner technology, and better reuse of wastes (Hoffman, 1997)(Graedel, 1998). Furthermore there is as yet no generally accepted weighing system for different environmental impacts of products (Reijnders, 1998).

Finally indirect effects, such as the energy effects associates with increased buying power following from lower purchase prices remain to be considered.  

5. Indirect effects on energy use

E-commerce leads in the case of Take it Now to on average 10% discounts on PC prices. The price discounts are not associated with more efficient distribution (i.e., transportation). On the contrary, e-commerce distribution leads in the case of Take it Now to a small increase of distribution costs for the e-tailer, that is passed on to the consumer. This, because delivery to the home is provided for.  

The most important business variable that changes with e-commerce is that the sales operation can be managed with far less personnel: fewer personnel in sales, logistics and purchase departments. Reduction in personnel costs is the main business variable accountable for the increased operational efficiency and discounted PC prices. If we assume that personnel not hired by Take it Now is, in the fast-growing Dutch economy, not sitting at home but working elsewhere (and actually producing additional economic growth), what then would be the indirect effect associated with more efficient retail operations due to e-commerce?

In this case, the mentioned price discounts will lead, ceteris paribus, to more buying po­wer for consumers, and in practice to spen­ding most of the savings associated with lower e-commerce pri­ces. Just to explore the importance of this indirect effect we will concentrate on its impact in terms of energy use by households. The methodology that we follow is in line with other studies on indirect or rebound effects associated with energy saving at the micro level (Schipper and Grubb, 2000). These studies suggest that rebound effects usually eat up 10-40% of initial energy savings (Schipper, 2000).

The price reduction of 10% for PCs sold by Take it Now leads to an extra buying power of 200 - 400 guilders per purchased PC. Let us assume that this sum will be spent integrally, and will not be set apart on a savings account. As savings, as a rule, are also economically recycled, involving the use of energy, this is unlikely to lead to major errors in estimating the impact on energy use.

It is known that marginal budget shares of expenditures on specific products and services change with increasing buying power of persons and fami­lies. Empirical evidence shows, for instance, that in post World War II industrial coun­tries, increased expenditures lead to declining marginal budget shares for food and clothing, whereas the marginal budget share for other expenditu­res did increase (Pollak and Wales, 2000) (De Combrugghe et al., 1997) (Van Driel et al., 1997). Different types of expenditure lead to different energy intensities per guilder spent (Van Driel et al., 1997). For instance, clo­thing typically has an energy intensity below 4 MJ/guilder, whereas personal care and do-it-yourself products pro­ducts have energy efficiencies that tend to range between 4 and 8 MJ/guil­der and the energy effi­ciency of food varies from 2 to 20 MJ/g­uil­der (Schmidt and Postma, 1999). Again these energy intensities have been calculated on he basis of life cycle energy analysis of the products concerned.

Empirical evi­dence shows that additional buying power tends to be used for expenditures requiring less energy than average expenditures (Schneider, 1994) (Blok and Vringer, 1995). Calculation on the basis of empirical data pertinent to Dutch consumers in the early nineteen nineties (Blok and Vringer, 1995) sug­gests that the energy elasticity is 0.8, i.e., that on the average a 1% increase in expenditures will lead to a 0.8% increase in the direct and indirect use of energy. This may correspond with energy intensities associated with additional buying power between 4.8 and 6.2 MJ/guilder, with a standard deviation of 1.5 MJ/guilder.

Taking into account this energy elasticity of expenditures, the extra energy use following from financial gains of customers associa­ted with e-commerce of PCs is 960 - 2480 MJ. Taking into account this energy elasticity of expenditures the extra energy use following from financial gains of customers associa­ted with e-commerce of PCs (960 - 2480 MJ) far outweighs the energy gains in the field of distribution (30-40 MJ).

However, there may be an objection against the use of an energy elasticity of 0.8. Energy elasticity of expenditure may change over time. In fact calculated energy elasticities from 1960 onwards vary between 0.6 and 0.9 (Blok and Vringer, 1995). Moreover, as pointed out in the Introduction, e-commerce may itself be an agent of change. There is as yet no data available that allows us to assess a possible change in energy elasticity of expenditures since the early nineteen nineties. There are, however, data on sales via Internet, and these may give an indication of energy use associated with the product mix obtained via e-commerce. 

Table 3 lists the major categories of products sold on the Internet in the Netherlands between March 1998 and March 1999 (Multiscope.nl, 1999) with estimates of the energy intensity following from direct and indirect energy use associated with these products.

Table 3: Spending on major product categories by Dutch customers in the period March 1998-March 1999 and energy intensities in MJ/guilder. Sources: Schmidt and Postma (1999), Multiscope (1999), this article  

The energy intensities shown in table 3 suggest that e-commerce purchase profiles are unlikely to have a dramatic effect on the energy elasticity of household expenditures.    

4. Discussion

An important question is whether our conclusions with regard to direct and indirect effects of e-commerce are only pertinent to e-commerce as applied by computer reseller Take it Now, or may be generalised to other applications of e-commerce.

In this context, one should note that the overall energy savings of e-commerce involving the distribution of computers might be relatively small if compared with other e-commerce operations. Furthermore, it is important to note that in the case of Take it Now financial savings related to more efficient operations have a far greater energetic impact than the improved efficiency in physical distribution.

The energy balance may be quite different in case that traditional distribution related energy expenditures are relatively higher, or absolute price reductions are lower or even absent.

In this context, one can think of products with a far lower monetary value per kilogram, or situations in which the roundtrip of customers to retail outlets is much longer than is typical in the Netherlands.

It can also be argued that the representativeness of the strong price reductions of Take it Now is limited. Research into vendor characteristics in the United States (GTRC, 1998) suggests that low prices are only somewhat important to 55% of online buyers. Other factors, such as reliability, seem to be much more important. Similarly, on the Dutch market reliability is a primary concern with lower prices being second (Multiscope.nl, 1999). In this respect there is also a difference between market segments: the PC market is far more price sensitive than many other market segments. Furthermore, while Internet consumers tend to expect a price bonus when purchasing online from newcomers in the online market, as reliability cannot yet be used as a convincing sales argument. Take it Now belongs to the category of newcomers. In other market segments, and in a more established market, price differences with traditional channels will become smaller.

However we feel that as increased business efficiency that is characteristic for e-commerce in general leads to economic growth one should not be optimistic as to the overall environmental effects of e-commerce.

References:

Atlantic Consulting & IPU (March 1998). EU Ecolabels for Computers. Brussels: Atlantic Consulting and IPU.

Blok, K. & Vringer, K. (July 1995). Energie-intensiteit van levens­stijlen. Utrecht, The Netherlands: Vakgroep Natuurwetenschap en Samenleving, University of Utrecht.

 De Combrugghe, D., Palm F.C., & Urbain J.P. (1997). Statistical demand functions for food in the USA and the Netherlands. Journal of Applied Econometrics, 12, 615-645.

Graedel, T.E. (1998). Life Cycle Assessment in the Service Industries. Journal of Industrial Ecology, 2, 57-70.

GTRC (April 1998) GVU's WWW User Survey. GA: Georgia Tech Research Corporation.

Gruppe Energie-Stoffe-Umwelt ETH (1996). Oekoinventare von Energiesystemen. Zürich, Switzerland: ETH.

Hoffman, W.F. (1997). Recent Advances in Design for the Environment at Motorola, Journal of Industrial Ecology, 1, 131-140.

Multiscope.nl (September 1999), Den Bosch, The Netherlands: Multiscope.

Pollak, R.A., & Wales, T.J. (1992). Demand System Specification and Estimation. New York: Oxford University Press.

Reijnders, L. (1998). The Factor X Debate. Journal of Industrial Ecology, 2, 13-22.

Romm, J., Rosenfeld, A., Hermann, S. (December 1999). The Internet Economy and Global Warming. Washington D.C: Cool Companies.

Schipper, L., & Grubb, M. (2000). On the rebound? Feedback between energy intensities and energy uses in IEA countries. Energy Policy, 28, 367-388.

Schmidt, T., & Postma, A.D. (1999). Minder energiegebruik door een andere leefstijl? Den Haag, The Netherlands: Dutch Ministry VROM.

Schneider,H.C. (March 1994). Op zoek naar energie-extensieve leefstij­len, bestedingspatronen en energiebeslag van Nederlandse huishoudens. Rotterdam, The Netherlands: CEA.

Van Driel, H., Nadall, V., & Zeelenberg, K. (1997). The demand for food in the US and the Netherlands. A systems approach with the CBS model. Journal of Applied Econometrics, 12, 509-532.

Schipper, L. (2000). On the rebound: the interaction of energy, efficiency, energy use and economic activity. An introduction. Energy Policy, 28, 351-353.

© 1995-2002 Martijn Hoogeveen