Does a value-added smile curve imply an environmental frown curve?

A distinction can be made between two waves of globalization. The first wave took place in the second half of the 20th century, triggered by falling transport costs. This reduction made it increasingly profitable to separate production from consumption locations, resulting in widespread outsourcing to low-cost countries. The second wave of globalization resulted from the advances of the ICT, which enabled further global segmentation of the production stages. As a result, production was not only geographically distributed, but also became increasingly fragmented within the individual production sites.

The “Smile Curve” hypothesis originally proposed by Acer founder and managing director Stan Shih states that the value-added shares in the pre-production (e.g. R&D, design) and post-production phases (e.g. marketing, sales) tend to be higher compared to the standardized manufacturing processes, which leads to a U-shaped relationship along the production steps. While the distribution of value creation along the value chains is well documented, the distribution of emissions embodied in exports is less researched. Since China's accession to the WTO in the early 2000s and subsequent production shifts, not only has production and income have skyrocketed, but emissions have also increased drastically. However, this increase in emissions is not limited to China alone. OECD studies have shown that non-OECD countries, which are usually net exporters of emissions, have overtaken OECD countries in both production and consumption-related emissions in recent years.

This article focuses on the relationships between production and the emission shifts caused by the emergence of GVCs in the last quarter of a century. We use the smile curve approach originally designed to map the distribution of production and value creation in GVCs and expand it to the measurement of emissions. Our special focus is on relocating activities with low added value of high emission intensity. In addition, we examine how the patterns of value creation and emissions vary along the value chains between countries, sectors and over time.

Our central hypothesis is that the majority of industrial value chains have a U-shaped value-added curve and an inverse U-shaped emission curve. Our goal is to develop a unified concept and provide an interactive tool to study the interactions between these two curves.

Main Results

The concept of the 'Smile Curve' has proven to be an efficient tool for analyzing the value-added distribution in global value chains (GVCS). This paper extends this concept with the introduction of “Environmental Loss Curves” that map embedded CO2 emissions along GVCs. We use detailed OECD input-output data and analyze the emissions and value-added contributions of 45 industries in 66 countries. We can aggregate the results for economic groups (EU27, OECD, G7) and certain industrial sectors over time.

Our methodological framework for estimating value and emission curves shows different patterns in the environmental profiles compared to conventional value-added distributions. While the value creation usually follows a smile curve pattern, we find the first indications of an inverse U-form in terms of emissions in the GVCs, although this pattern is less pronounced. In addition, the classic “Smile Curve” has changed over time, which is primarily due to the successful integration of emerging countries like China into global value chains.

The increasing fragmentation of the GVCs and the involvement of emerging markets in the last three decades has two remarkable effects: a flattening of the value-added profile and an intensification of emissions in early production phases. This emission pattern can be attributed to the outsourcing of intermediate production to emerging emission intensities, especially in energy generation.

Our study has three limitations to consider. First, due to data restrictions, we were unable to include functional specializations in our framework, as has been done in previous studies. Secondly, the TECO2 database of OECD emissions from chemical processes in the production stages excludes, which could significantly affect emissions patterns in the various value chains. Thirdly, the OECD's input-output database covers a large number of countries, but does not contain data on low-income countries, which could affect our results. Databases with a larger country coverage, such as B. Eora, however, suffer from a lack of granularity in terms of industries and have been criticized for their poor accuracy in other studies. These limitations offer promising opportunities for future research to improve our understanding of both the economic and environmental dimensions of global value chains.