If you’d asked me a week ago what impact means, I’d have said it was about increasing people’s welfare and I would have considered that to be standard economics. Now having read a paper that explains how standard economics measures how businesses affect people’s welfare, I might give a slightly different answer. The problem is that for economists welfare is derived from consumption, and I do not think that investing to satisfy people’s preferences for consumption captures the spirit of impact investing.
“An Economic View of Corporate Social Impact” by Hunt Allcott, Giovanni Montanari, Bora Ozaltun, and Brandon Tan, is a paper that I have been wanting someone to write. Last year I had plans for a project to put impact investing on a firmer economic footing – you can read the concept note here (I am still interested in doing that, please get in touch if you are too). I wanted to know why in theory we might expect some investments to have a higher social impact than others, before we start appealing to ‘externalities’ and ‘frictions’ which are the usual justifications for intervening in markets to influence private investment to the benefit of society.
Allcott and co-authors work with standard economic theory in which firms are maximising profits, consumers are maximising their utilities, and prices and quantities are set in markets. They are deliberately starting with the most basic theoretical building blocks, so they do not include the possibility of positive externalities that could explain why some investments have unusually large social returns (see The economics of development finance for more on that). The only deviations from “econ 101” they incorporate are negative externalities from the oil and gas industry’s carbon emissions, and the possibility that people self-harm by smoking and drinking soda (which, writing from British International Investment, I should call fizzy drinks).
In this theoretical world, the social impact of a business consists of the surplus workers obtain from having a better job than they might otherwise; the surplus consumers obtain from buying something they like more cheaply than they might otherwise; the profits enjoyed by the owners of businesses, and the negative effects of carbon and smoking. They define the impact of an individual business as how the sum of these thing would change if it ceased to exist. This is “enterprise impact”.
They also suggest a second approach, where a business might instead want to claim its share of its market’s impact. This is a response to an issue I have raised before. If a market is reasonably competitive then the marginal impact of any individual firm could be close to zero, as its customers can easily turn to its competitors if it disappears, despite the fact that taken as group the set of competing firms may have a large positive impact on society. It does not seem right that each individual firm should be assigned zero impact as they would be, taking the first approach. The second approach divides up the impact of an entire segment, such as grocery retailers, between the competing grocers.
Their model estimates the change in total social welfare, should a business cease to exist, in the simplest possible way by assuming workers and consumers move to the other firms already in existence – they do not introduce possibilities like new firms entering or try to think through the consequences up and down supply chains. The surpluses that workers and consumers derive from a business can be calculated in these models from the slope of supply and demand curves, which capture how workers and consumers would respond to changes in wages and prices. And this is where the empirical part of the exercise comes in – they select a sample of businesses (such as Walmart and BMW) and ask thousands of customers and workers in the USA how they would respond to price changes. With that information in hand, they can plug in the numbers and calculate which businesses have the greatest impact, measured in dollars (the holy grail of impact measurement).
The final – and crucial – ingredient in the model is that they apply weights to these surpluses, so that poorer people carry more weight than rich. This is very important. It turns out people really love their BMWs and without those weights this method would identify the luxury car brand as an impact leader. It also means that profits are an unimportant part of impact because they go to the already rich.
I am going to discuss three elements of the model – consumer and labour surplus, and weighting. I start from the position that any method for measuring impact is going to quickly lose credibility if it picks out companies that seem obviously wrong to most people. That is pretty unfair on researchers who are trying to do something very difficult and are consciously putting many complications aside. But this is the standard that impact monetisation must meet if it is to be adopted in practice. Any method that omits significant things is going to make errors. Some ESG indices included the fast-fashion company BooHoo because it was creating jobs in low income areas, but omitted the fact staff were underpaid and working in dangerous conditions. That error revealed those indices were not fit for purpose. If methods to place a dollar value on impact are ever going to trusted, they will need to stand up to scrutiny. My intention here is to outline what I think some of the challenges will be. I will start with those weights, because they are so important.
Thus far, to the great frustration of all, nobody has yet produced a method for comparing the impact of different investors doing different things in different places. If the impact investing industry is ever going to arrive at a consensus about how to do that, one of the first things it will need to agree is how impact changes as the incomes of the people your investment is affecting changes. The idea that an additional dollar makes more of a difference to the life of a poor person than to a rich one is widely accepted but quite how much more impact it represents is not agreed. I have previously written about how the Global Innovation Fund (GIF) is one of the few impact investors to propose a specific methodology – it treats all percentage changes as equal. That means as people get better off, you need achieve a larger absolute change in their incomes to have the same impact.
It seems uncontroversial to regard simply increasing someone’s consumption as impact, when that person is living in poverty. That gets more debatable at higher levels of income. There is a macro version of the same question. A development finance institution might reasonably consider an investment that raises national economic growth in a low-income country as impactful, but the idea that making some contribution to economic growth in a rich country should qualify as impact investing seems dubious. Any impact monetisation method is going to need to tackle this issue, and I am very glad to see this paper draw a line in the sand by proposing a specific weighting – which I think corresponds to the GIF approach, although it’s expressed differently.
So, with weights applied, the impact of BMW is reduced. Still, I am not sure that adjustment alone will be accepted by would-be impact investors. I suspect people want something more categorical. If I had to guess, in my household the Nintendo Switch is producing a great deal of consumer surplus and, unlike BMWs, one does not need to be especially wealthy to own one (at least not if you live in a rich country). I don’t think the impetus behind impact investing – the desire to avert environmental catastrophe and cure society’s ills – will be satisfied with an impact measurement method that rewards Nintendo for entertaining my family in London.
On the other hand, what happens when the weighting is applied globally, where national median annual incomes per capita range from $500 to $50,000? My guess is that almost nothing in a rich country is going to have any significant impact set on that scale, and I doubt the impact investing industry is going to swallow that either (maybe it should!). I suspect the impact investing industry would prefer the ability to claim impact by doing categorically impactful things in rich countries.
This is what inferring consumer surplus from the slope of demand curves implies: “a firm delivers more social impact if it would be more difficult for its consumers and workers to find substitute products and employers if the firm were to exit.” There is something intuitive about the idea that a business has more impact if it supplies something people otherwise lack, rather than being just one more producer of something that is readily available. But I am not sure I like all the implications.
Does a business increase the good it does for society by acquiring its competitors and increasing its market power? That would make it more difficult for consumers to find substitutes and I think increase impact as measured by this method. Economists are increasingly concerned that rising market power is doing society no good (see here and here). Do we really think firms with more market power have more positive impact? One of the businesses with the judged to have greatest impact by this exercise is Walmart, because it serves a lot of low-income people who seem to have few other options. Selling things cheaply to people without much money does sound like impact to me, but the impact of Walmart on local communities in America is contested and this methodology does not have any chance of picking up any potential harmful effects.
I have similar concerns about market power here. Your labour supply curve might be steep because you are a wonderful employer, and your staff would still like to work for you even if you cut wages. But there are worse reasons. Exploitation involves taking advantage of someone in a desperate position. This is a tricky one for impact measurement systems to handle. As the economist Joan Robinson famously said, the only thing worse than being exploited by capitalism is not being exploited by capitalism. A fairly standard approach to capturing the impact of creating a job is to ask the worker how much better off they think they are, thanks to having that job. Creating jobs for people in desperate situations does intuitively seem to be more impactful than adding jobs somewhere they are plentiful.
But I am not sure I like all the implications. Are we really sure that the big employers who pay low wages to people who have no better options – which I think would be measured as conferring a large labour surplus under this methodology – are an unambiguous force for good? I can’t make my mind up about that one. I think under this method a firm that hires people in dire need of a job and is also a good employer will face a steeper labour supply curve and hence be judged to generate more surplus than an employer who is more exploitative, but I think we might want to draw sharper distinctions than that. The role of DFIs and impact investors is to move enterprises towards being good employers (here are the standards BII sets) – ideally an impact measurement methodology would not allow bad employers to claim a positive impact.
It was high time that the economics profession engaged with idea that has gripped the impact investing world — that a dollar value can be placed on the social impact of a business. This paper is a real milestone. Impact monetization is an incredibly ambitious thing to attempt, and this ground-breaking work has shown us a methodology that is powerful and generalisable – which makes it very appealing. But is it a foundation that the impact investing industry ought to build on?
At BII our objective is to improve the lives of people who currently have a very low standard of living. Interpreting that as increasing the welfare they derive from consumption (in the jargon of economics) seems acceptable to me and consistent with the ethos of impact investing. But when I see that same logic applied to a country where most people already have a high standard of living, I am less comfortable.
We know that people benefit from a productive economy that provides a range of goods and services, and because the economy consists of businesses, we know that businesses benefit people. But impact investing means more than “invest in a business”. It might mean “invest in those businesses that confer the greatest benefits”. Under this approach that means invest in those business that have the most market power and serve or employ the lowest-income people
The authors start from the observation that “economics offers a set of useful frameworks for conceptualizing and quantifying some of the objects that many impact investors care about: social welfare, consumer and worker surplus, and externalities.” But mainstream economics is very non-judgemental – people have preferences, and the job of the economy is to satisfy them. I am sure that many companies would love to report big numbers in their “impact accounts” because they sell things that people like, but I think it’s important for the credibility of impact investing that’s not where impact monetisation lands – even after weighting for stakeholder income. Impact investing must be about making the world a better place. I am not arguing that satisfying consumers is a bad thing, but that impact investing should be about something more.
Perhaps with some more additions from the economics toolbox, the approach could get closer to what impact investing claims to be about. I would also not be terribly comfortable with some people deciding they know better than everyone else what’s good for society and defining impact along those lines. Nonetheless I feel like a satisfying method for measuring impact will need to be more judgmental than the slope of a demand curve.
 They are not trying to identify the impact of an investment which would result in a change in the impact of a business, nor are they trying to normalise impact by the dollars of investment that were required to generate it, as one would if wanting to allocate dollars.
 They do not attribute impact in proportion to revenues, instead they propose using a concept called “Shapely Values” which are a solution from game theory to a bargaining problem when people work as a team, but make different contributions and want to assign costs and benefits accordingly.
This does not imply we ought to be sniffy about what poor people spend their money on, when they get a little more of it. One first things people buy when they can afford it is a television (see here for an account of the economist Abhijit Banerjee discussing the importance of pleasure in people’s live). If you live in poverty, having a TV can be a significant improvement to your standard of living. When people are poor, I would argue anything that raises their level of consumption can be counted as impact.
 The Sustainable Development Goals can be seen as an attempt to find consensus and legitimacy, via intergovernmental negotiation and civil society consultations, for a conception of what making the world better means.