Indices insights: Do investors act on shifts in climate concerns?

In the previous ‘Indices insights’ article,¹ we outlined how the carbon footprint metric is useful for mapping out portfolio-level or entity-level decarbonization pathways. However, we also determined that it is less effective in differentiating between climate leaders and laggards. For this purpose, additional stock-level climate data that captures other dimensions is required.

In this article, we start to address this issue by introducing a novel long-short polluting-minus-clean (PMC) portfolio, similar to the carbon portfolio that is put forward in a recent research paper.² It is based on a portfolio that takes long positions in a basket of stocks that contribute negatively to one or multiple climate-related SDGs (‘polluting’), and short positions in a basket of stocks that contribute positively to climate-related SDGs (‘clean’). As such, the PMC portfolio tracks the difference in returns between the polluting and clean companies.

To assess the insights that can be drawn from the PMC portfolio, we first scrutinized its performance over time in relation to sentiments regarding climate change. Specifically, we tested whether companies that contribute negatively to climate-related SDGs are more negatively impacted by an (unexpected) rise in concerns around climate change compared to those that contribute positively to climate-related SDGs.

As a proxy and robustness test for the level of climate concerns, we referred to the climate policy uncertainty (CPU) and temperature anomaly indices, respectively. The former is based on the volumes of text-based newsflow linked to climate change as this reflects uncertainty around future climate policy, while the latter identifies months with anomalously high or low temperatures.

In Figure 1, Panel A illustrates the monthly changes in the CPU index alongside the cumulative returns of the PMC portfolio. The orange bars denote the months in which there was a relatively significant upward shift in the CPU index, which corresponds to months characterized by rising climate concerns. Similarly, the blue and grey bars reflect the months during which climate concerns decreased significantly and were stable, respectively.

Panel B depicts the average performance of the PMC portfolio across different regimes based on sentiment regarding climate change. In periods of rising, stable and subsiding climate concerns, it delivered an annualized return of -4.1%, 1.3% and 1.4%, respectively. This outcome is in line with our expectations and allows us to conclude that climate laggards tend to be more negatively affected during periods of increased climate policy uncertainty compared to climate leaders.

As a robustness test, we also analyzed the performance of the PMC portfolio in months with anomalous temperatures. We assumed that concerns around climate change increase in periods when the average temperature is anomalously high, and vice versa. The results are shown in Figure 2.

Similar to the previous outcomes, the PMC portfolio underperformed on average in months when average temperatures were anomalously high as it generated an annualized return of -4.8%. By contrast, it delivered a significantly higher annualized return of 4.4% during months with anomalously low average temperatures.

The starting point of our analysis is the S&P Global LargeMidCap universe. Using Robeco’s proprietary SDG framework, we identify stocks that either contribute positively or negatively to climate-related SDGs.

The PMC portfolio is constructed by taking long positions in stocks that have an SDG score of -2 or -3 for either SDG 7 (affordable and clean energy), SDG 11 (sustainable cities and communities) and SDG 13 (climate action), and short positions in stocks that have an SDG score of 2 or 3 on either one of the same SDGs. As SDG scores for individual companies are not available before 2017, we construct the long and short legs for that period by sorting stocks on the RobecoSAM carbon intensity measure using the 30th and 70th percentiles, respectively.³

To analyze the performance of the PMC portfolio across different regimes based on climate concerns, we rely on two measures: CPU and temperature anomaly indices.⁴ The CPU index is constructed by focusing on the textual similarities of authoritative texts on climate change – for example frequently used word combinations in Intergovernmental Panel on Climate Change (IPCC) reports – with articles in the Wall Street Journal in a month. Months with high volumes of news linked to climate change indicate heightened uncertainty around future climate policy, and vice versa.

For instance, climate policy uncertainty peaked in December 2009 when the COP15 event took place and in November 2015 which was the month before the COP21 event when the Paris Agreement was adopted. The month-to-month difference in the CPU index indicates either increases or decreases in climate concerns. The three climate regimes (increasing, stable, decreasing) are defined by using the 25th and 75th percentiles of the month-to-month changes in the CPU index over the full sample period.

Similarly, we split the sample into three climate regimes based on the 25th and 75th percentiles of the detrended temperature anomaly index. This is based on data from the Global Land-Ocean Temperature Index sourced from the National Centers for Environmental Information. A temperature anomaly is defined as a deviation from the 20th century average reference temperature.⁵

In our analysis, we assessed the performance of a novel PMC portfolio across different regimes based on climate concerns. In line with our expectations, we found that it tends to underperform in periods of heightened climate policy uncertainty, and vice versa. Moreover, we performed a robustness test and saw that it also lags in periods of anomalously high average temperatures, and vice versa. As a result, we believe the PMC portfolio can be used practically by investors to help distinguish climate leaders from laggards.