Does superior social performance reduce the cost of capital – the return investors require for investing in a company? That’s a key question because, if so, then CEOs have strong incentives to improve their social performance. We’d expect the answer to be “yes”. Higher social performance could improve a company’s financial performance or make it less risky, or socially-conscious investors willingly accept a lower return in a responsible company. But, we need to rigorously test it.
Most papers do so by studying the cost of equity. But this is very difficult, because you never actually observe the cost of equity – only the stock price. Now it’s true that the stock price implies a cost of equity – if it’s £10 today and is expected to be £11 next year, then investors must be requiring a 10% return to hold the stock. But the problem is that we have no idea what investors expect the stock price to be next year. The simple solution is to wait and see what happens – if the stock price ends up being £11 next year, then we might be tempted to conclude that investors expected it to be £11, and thus expected a 10% return.
The big problem is that investors don’t always get what they expect. The above method confuses realised returns with expected returns. The realised return is what investors actually get. That’s unambiguous – it’s 10% in the above example. But, investors might not have expected a 10% return. It may be that they expected only a 7% return, and the stock performed better than expected. For example, one of my own studies finds that companies with high employee satisfaction do 2.3-3.8%/year better than what investors expect.
This ambiguity gives ESG (Environmental, Social, and Governance) advocates – me included – tremendous flexibility to interpret the results how we want. If we find that high-ESG companies deliver higher returns than the market, we claim that ESG is “good” as it led to the company performing better than expected. If we find that high-ESG companies deliver lower returns that the market, we can still claim that ESG is “good” as it led to investors requiring lower returns to hold the stock. In the first case, we interpret the high realised returns as high unexpected returns; in the second case, we interpret the low realised returns as low expected returns. Since we never observe expected returns, we can claim whatever suits us. (Thus, in my paper, I had to do several other studies to show that the high returns were unexpected).
A better approach is thus to study bonds. With bonds, you can directly observe the cost of debt – it’s the interest rate on the bond. Moreover, responsible investors might be even more likely to sacrifice financial returns for a responsible company’s bonds rather than its stocks. If an investor company buys a responsible stock, another investor has to sell. No new money flows into the company. Since the investor creates limited social value by buying the stock, it may be unwilling to sacrifice financial return. But, companies issue new bonds all the time. If an investor buys a new bond, it’s buying it from the company. New capital flows into the firm, which allows it to do more social good – so the bondholder might indeed be willing to give up return.
A new paper by Michael Halling, Jin Yu, and Josef Zechner studies the link between a company’s ES scores and yields on its newly-issued bonds (the G, i.e. governance, is less relevant for bonds as they are often secured). They find that better overall ES performance leads to lower bond yields – even after controlling for other factors that drive bond yields. This seems to be a resounding victory for ESG advocates.
But, as in many things in responsible business, the results are far more nuanced than often claimed. When decomposing overall ES performance into its individual components, it’s the product-related dimensions of ES scores that lowers the cost of debt. Scores for the environment, community, and human rights – which get much more attention by the media – are actually linked to a higher cost of debt, although the results are statistically insignificant. Thus, it seems that the “actively do good” aspects of ES matter more than the “do no harm”.
Moreover, the characteristics that matter vary over time and across industries. Starting with the former, employee relations scores are insignificant during recessions. But in booms, good employee relations do reduce bond yields. This may be because, in booms, workers are scarce and strong employee relations help recruit and retain the best workers. Moving to the latter, environmental scores are linked to lower bond yields in the agriculture, forestry, fishing, and mining industries. This makes sense, because these industries are particularly exposed to environmental risks. In the language of Chapter 3 of the book, environmental factors are especially material in these industries. But in the transportation, communication, and trade sectors, high community scores actually lead to a higher cost of debt, perhaps because they signal that the company is focusing on immaterial issues.
These results are striking. The effect of ESG on the cost of capital is much less unambiguous than commonly claimed by ESG advocates. Moreover, it highlights the dangers with lumping the different components of ESG (or even ES) together. Some factors matter; others don’t. Moreover, the factors that matter vary from industry to industry depending on materiality. This is consistent with the evidence in Chapter 4 of the book for equities. Stocks with high ESG scores across the board don’t beat the market. But the ones that outperform on material issues, and scale back on immaterial issues, do.