Impact stocks beat UK benchmark on risk-adjusted returns, study finds

A portfolio built from seven publicly listed UK impact companies delivered higher returns, better Sharpe ratios and stronger downside protection than the FTSE All-Share benchmark between June 2017 and June 2022, according to a new empirical study of sustainable investing in the British equity market. The advantage held up under a volatility-targeted framework designed to put both portfolios on an equal risk footing. Yet the study also delivers a sobering caveat: when the return differences were subjected to formal statistical testing, the outperformance could not be confirmed as statistically significant, and the positive excess return, or alpha, recorded in regression analysis remained insignificant at conventional confidence levels. The findings, published open access in the journal Digital Economy and Sustainable Development, therefore paint a nuanced picture in which impact stocks appear to offer meaningful diversification benefits and partial independence from broad market movements, while definitive claims of superior returns remain statistically out of reach for this particular sample and period.

The research, authored by Yang Song, Lin Tang and Shuhui Yang, empirically studies the role of sustainable impact investments in strengthening portfolio resilience and risk-adjusted performance in the UK equity market from 2017 to 2022. Rather than relying on the private or proprietary datasets that have limited transparency in much of the impact investing literature, the authors assembled a dataset of impact companies affiliated with the Social Stock Exchange UK, a recognised listing and licensing body for socially driven and financially sustainable businesses. Although the UKSSE ceased operations in 2019, the study continued to track these firms because they remained publicly traded on the London Stock Exchange and its Alternative Investment Market, maintaining their core social and environmental business models. Member companies of the exchange were required to commit to monitoring and reporting their progress toward relevant United Nations Sustainable Development Goals, giving the sample a defined sustainability character.

The impact portfolio was constructed exclusively from seven selected UK-listed companies accredited by the Social Stock Exchange, using monthly closing prices and market capitalisation data to set weights in proportion to each firm’s market value. This market-capitalisation-weighted approach was chosen to reflect real-world investment conditions. The comparison portfolio, described in the study as the normal portfolio, simply tracked the FTSE All-Share index, a benchmark comprising more than 600 listed companies that captures roughly 98 percent of the UK market’s total capitalisation. Both portfolios covered the same period, from 22 June 2017 to 22 June 2022, and historical stock price data and market capitalisation figures were collected from Yahoo Finance and Refinitiv Eikon. The dataset contained 1,263 daily closing prices and corresponding market capitalisations across the five-year window.

Methodologically, the study applied Markowitz’s modern portfolio theory as its analytical backbone, supplemented by Sharpe ratio analysis, a paired t-test and CAPM regression models. To guard against hindsight bias and to simulate the dynamic conditions of actual portfolio management, the authors adopted a rolling walk-forward back-testing framework following the methodology of Dierckx and colleagues from 2020. The first 24 months of data were used to estimate the relevant parameters, and the estimated weights were then applied to calculate the out-of-sample portfolio return for the following month. The estimation window then rolled forward one month at a time, so that each out-of-sample observation was based only on information that would have been available at that moment. Of the roughly 60 months in the sample, the first 24 served as in-sample estimation and the remaining 36 months provided the out-of-sample testing observations.

A second layer of rigour came from volatility targeting. Because a portfolio can achieve higher returns simply by taking on greater risk, the authors scaled both portfolios’ returns to a constant target of 10 percent annualised volatility before comparing them. Rolling monthly standard deviations of historical returns were used to estimate each portfolio’s realised volatility, and returns were then rescaled in proportion to the ratio of the target volatility to the realised volatility. This adjustment ensures that any measured outperformance reflects genuine differences in efficiency rather than disparities in risk exposure. Raw price data were also transformed into logarithmic returns to standardise measurements across companies with very different price scales and to mitigate the heteroskedasticity common in financial time series, keeping the return distributions statistically stable and comparable over time.

The seven impact companies spanned a diverse range of sectors, including materials, real estate and information technology. The study’s figures highlight how varied their risk profiles were: Catenae Innovation Plc showed the most significant daily fluctuations, with sharp peaks and drops indicating a highly volatile trading pattern, while Assura Plc and Primary Health Properties Plc displayed stable daily returns consistent with their focus on real estate and healthcare infrastructure. ITM Power Plc demonstrated consistently high volatility, especially during the COVID-19 turbulence of early 2020, a period in which nearly all the sampled stocks experienced intense volatility reflecting widespread market uncertainty. The authors note that this diversity of risk characteristics within impact stocks itself underscores the demand for careful diversification when constructing portfolios from such assets.

The headline descriptive results favoured the impact portfolio. Comparing the 60-month raw data, the impact portfolio produced an annualised return of about 2.36 percent, against a loss of roughly 0.29 percent for the FTSE All-Share. That outperformance came alongside higher volatility, 19.47 percent versus 13.99 percent, and a deeper maximum drawdown of minus 35.10 percent versus minus 26.94 percent. Even so, the Sharpe ratio of the impact portfolio stood at 0.121, exceeding the benchmark’s minus 0.021 and indicating a modest improvement in return efficiency. A cumulative return analysis, assuming an initial investment of £1, showed the impact portfolio ending the period with a higher value than the benchmark, with a growth curve that recovered more strongly after downturns, a pattern the authors interpret as adaptive resilience.

Under the volatility-adjusted framework, in which both portfolios were normalised to the 10 percent annualised volatility target, the gap widened considerably. The impact portfolio generated an annualised return of 5.84 percent, compared with minus 3.69 percent for the FTSE-based portfolio, while the Sharpe ratios were 0.42 and minus 0.311 respectively. The impact portfolio also showed greater financial resilience on this equalised-risk basis, with a smaller maximum drawdown of minus 18.18 percent against minus 22.18 percent for the normal portfolio. The authors argue that these results indicate impact investment strategies can provide superior performance under equalised risk conditions, though the statistical significance tests discussed below temper how firmly that conclusion can be stated.

Indeed, the formal significance testing did not confirm the outperformance. Although the impact portfolio’s average monthly return exceeded the benchmark’s in both raw and volatility-adjusted analyses, with annualised differences of roughly 2.65 percent and 9.53 percent respectively, the paired t-tests returned p-values of 0.77914 on the raw comparison and 0.34044 on the volatility-adjusted comparison. Both failed to reject the null hypothesis that the true mean difference between the portfolios’ monthly returns equals zero at the 5 percent significance level. In other words, despite the outstanding mean performance observed, the excess return cannot be confirmed as statistically different from zero at conventional confidence levels for this sample and period.

The CAPM regressions told a similar story with an important twist. Both regressions generated positive Jensen’s alpha, suggesting the impact portfolio produced returns above those predicted by its market exposure, but the alphas were not statistically significant. For raw returns, the estimated alpha was 0.0033 monthly, or 3.99 percent annualised, with a p-value of 0.6398; for volatility-adjusted returns, the alpha was 0.0062 monthly, or 7.4 percent annualised, with a p-value of 0.3756. The beta coefficients were 0.38 for raw returns and 0.19 for adjusted returns, indicating lower sensitivity to market movements and reduced systematic risk. The low R-squared values, 0.0727 and 0.0247 respectively, imply the FTSE benchmark explains only a small fraction of the impact portfolio’s return variation, pointing to a greater reliance on idiosyncratic or impact-specific factors. The authors suggest asset managers and institutional investors could integrate impact-oriented assets into mainstream portfolios to align financial performance with sustainability goals, while future research applying multi-factor models such as Fama-French or Carhart, and wider geographic and temporal data, would help confirm whether these benefits hold more generally.

Subject of Research: Risk-adjusted performance and diversification benefits of impact investment portfolios in the UK equity market

Article Title: Do impact investments enhance portfolio resilience? Evidence from UK sustainable firms

Article References: Song, Y., Tang, L., & Yang, S. (2025). Do impact investments enhance portfolio resilience? Evidence from UK sustainable firms. Digital Economy and Sustainable Development, 3(1), Article 25. https://doi.org/10.1007/s44265-025-00074-5

Image Credits: AI Generated

DOI: 10.1007/s44265-025-00074-5

Keywords: impact investing, portfolio resilience, UK equity market, FTSE All-Share, Sharpe ratio, CAPM regression, volatility targeting, Social Stock Exchange, sustainable finance, diversification, out-of-sample backtesting, Jensen's alpha