Dematerialization Through Services: Evaluating the Evidence

Abstract

Dematerialization through services is a popular proposal for reducing environmental impact. The idea is that by shifting from the production of goods to the provision of services, a society can reduce its material demands. But do societies with a larger service sector actually dematerialize? I test the ‘dematerialization through services’ hypothesis with a focus on fossil fuel consumption and carbon emissions—the primary drivers of climate change. I find no evidence that a service transition leads to carbon dematerialization. Instead, a larger service sector is associated with greater use of fossil fuels and greater carbon emissions per person. This suggests that ‘dematerialization through services’ is not a valid sustainability policy.

Appendices

Appendix

This paper has a supplementary website containing raw and final data as well as code for all analysis: https://osf.io/93fpn/.

A Sources and Methods

US Energy Intensity by Sector (Fig. 1)

Sources and methods for calculating US sectoral energy use, employment, and value added are shown in Table 7. Except for energy consumption in the US service sector, all data is taken as given from the official data. The US Energy Information Agency (EIA) uses energy consumption categories that differ from the standard national income and product accounts categories used by the Bureau of Economic Analysis (BEA). The EIA energy data uses four categories: Industrial use, Commercial use, Transport use, and Residential use.

To allocate transport energy to the service sector, I apply the method developed by Giampietro, Mayumi, and Sorman (2012; 2013). I define service sector energy use ($E_{\text {Service}}$) as commercial energy ($E_{\text {Commercial}}$) plus work-related transport energy ($E_{\text {Work-Related~Transport}}$).

$$\begin{aligned} E_{\text {Service}}= E_{\text {Commercial}} + E_{\text {Work-Related~Transport}} \end{aligned}$$

(4)

work-related transport energy is calculated by subtracting non-work-related energy from transport energy. Non-work-related transport energy is defined as all transport energy minus light-duty vehicle energy consumed for non-work-related trips.

$$\begin{aligned} E_{\text {Work-Related~Transport}} = E_{\text {Transport}} - E_{\text {Light-Duty}} \cdot \frac{VMT_{\text {Non-Work}}}{VMT_{\text {Total}}} \end{aligned}$$

(5)

Here VMT stands for vehicle-miles-traveled. Data for US light-duty vehicle energy use comes from various EIA Annual Energy Outlooks from 2000 to 2018. Vehicle-miles-traveled data comes from the National Household Travel Survey 2009 and 2017.

Table 7 Sources and methods for US sectoral energy use, employment, and value added

US Sectoral Composition and Historical Energy Use (Figs. 7b, d, 8a, 9d)

US sectoral labor composition sources are shown in Table 8. Because the series are not mutually consistent, there is inherent ambiguity in the historical data. To quantify this ambiguity, I use a Monte Carlo technique to randomly splice together the series in different ways. I then use the median of this spliced data.

US historical energy and fossil fuel use data (1800–1945) comes from EIA Annual Review 2009, Table E1. Fossil fuel energy use data begins in 1850. I use an exponential regression to extrapolate trends back to 1800. US energy data from 1949 onward comes from the EIA Annual Energy Review, Table 1.3.

Table 8 Sources for US sectoral labor composition

US industry energy use and non-production employment (Fig. 9e)

US industry energy data comes from EIA Annual Energy Review Table 2.1. Industry employment comes from BEA Tables 6.8A-D (persons engaged in production). I calculate employment of non-production workers in industry is using Bureau of Labor Statistics series CES0600000006 (Production and non-supervisory employees, goods-producing) and series CES0600000001 (All employees, goods-producing). I define non-production workers as the difference between total employment and production employment.