“Thinking in Systems” and Economics

Thinking in Systems by Donella Meadows uses many economics-related examples. Even though Meadows’ background is as an environmental scientist, she clearly knows quite a bit about economics.

In this post, I outline the key economic examples Meadows refers to in the book and offer up some of my thoughts on her conclusions.

The Examples

Markets and market failures

In a section on bounded rationality, Meadows makes several well-trodden arguments against standard economic assumptions about human rationality:

Economic theory as derived from Adam Smith assumes first that homo economicus acts with perfect optimality on complete information, and second that when many of the species homo economicus do that, their actions add up to the best possible outcome for everybody. Neither of these assumptions stands up long against the evidence.
— Donella Meadows in Thinking in Systems

This is somewhat of a caricature of classical or mainstream economic thought. Adam Smith was well aware that cases of market failure existed and never claimed humans were perfectly rational. In fact, he wrote a whole book (The Theory of Moral Sentiments, written before The Wealth of Nations) about emotions, morality and other complexities of human behaviour.

To be fair to Meadows, “free market economics” was far more prevalent when she was writing in the early 1990s than it is now. I also expect there were many free market proponents (especially politicians) espousing a bastardised, “perfect rationality” version that they claimed was based on Smith’s or Milton Friedman’s work, even if the economists themselves held more nuanced views.

Meadows also sensibly recognises that markets and price signals have some benefits. As a result, her solution isn’t a blanket call for government to step in and regulate everything (c.f. Evicted by Matthew Desmond):

[Markets] can indeed by marvels of self-correction, as prices vary to moderate supply and demand and keep them in balance. … Prices that reflect full costs will tell consumers how much they can actually afford and will reward efficient producers.
— Donella Meadows in Thinking in Systems

Thankfully, Meadows is balanced enough to recognise that government is also fallible. She observes that both companies and governments have frequently used subsidies, taxes etc to fiddle with prices in a way that weakens (rather than strengthens) the feedback power of price signals.

Instead, her answer is simply that we need to better understand systems before trying to reform and restructure them:

A free market does allow producers and consumers, who have the best information about production opportunities and consumption choices, to make fairly uninhibited and locally rational decisions. But those decisions can’t, by themselves, correct the overall system’s tendency to create monopolies and undesirable side effects (externalities), to discriminate against the poor, or to overshoot its sustainable carrying capacity.

To paraphrase a common prayer: God grant us the serenity to exercise our bounded rationality freely in the systems that are structured appropriately, the courage to restructure the systems that aren’t, and the wisdom to know the difference!
— Donella Meadows in Thinking in Systems

That I can totally get on board with.

Limits to growth in a finite environment

Meadows argues that the goal of a corporation is to grow and eventually engulf everything — just like a cancer. But she also points out this is actually the goal of every living population, and it’s only bad when there isn’t a balancing feedback loop to keep it in check.

Meadows is probably one of the most prominent proponents of this argument. She co-authored a report called Limits to Growth back in 1972, which later turned into a book.

My thoughts

I find the limits to growth argument pretty convincing if describing GDP rather than GDP per capita. Meadows doesn’t clarify what she’s talking about in Thinking in Systems, but I think it’s the former. She says we need to slow down both economic and population growth in order to solve many social and environmental problems, which suggests she’s focusing on GDP.

Evidence more recently shows a negative correlation between GDP per capita and population growth. As countries get richer, fertility rates have generally fallen (see e.g. The New New Zealand). Slowing population growth could help ease some of the pressure on the environment. So maybe the answer to many social and environmental problems is to focus on increasing GDP per capita, not GDP.

Of course how you go about increasing GDP per capita also matters. For example, some argue that increasing GDP per capita is good because richer countries can better invest in technologies that reduce their environmental footprint. Yet wealth doesn’t necessarily translate to a lower environmental footprint. The US is one of the richest countries in the world, and its environmental footprint per capita is arguably the worst. If environmental externalities are not properly priced in, there’s less of an incentive for countries to invest in green technologies.

I’m not sure what evidence existed about the relationship between economic growth and population growth when Meadows was writing. But her broad argument that slowing down a reinforcing loop is usually more effective than creating a balancing loop seems plausible. Slowing down when we don’t really understand the system also seems generally prudent. (I feel this way about Artificial Intelligence, but that’s another story.) I also believe that it’s usually dangerous to try to maximise any one measure because:

  • most measures are imperfect — i.e. they are proxies for things we care about, rather than things we actually care about; and
  • most things come with trade-offs — even if a measure is something we directly care about, it’s unlikely to be the only thing we care about. Diminishing returns will usually mean that, past a certain point, the benefits of further pushing that one measure are outweighed by the costs of neglecting other things in life.

Growing inequality because of the “rich get richer” reinforcing loop

Another example discussed in the book which is similar to the “Limits to growth” argument is the “rich get richer” system archetype (Meadows calls it “success to the successful”). This archetype exists when the winners of a competition get an advantage that helps them compete in the future.

In the context of growing inequality, this may look like: poor children getting worse education; rich people getting better prices because they can buy in bulk or for the long-term; and compound interest in general. The “rich get richer” phenomenon can also occur at the corporate level, as larger firms benefit from things like economies of scale or market power.

A problem that arises is that, if the winning takes place in a limited environment (meaning the winner’s gains are at the losers’ expense), the losers will eventually be bankrupted, starved or forced out.

Meadows suggests several ways to arrest the reinforcing loop:

  • Diversifying. In the ecology context, a species may learn or evolve to exploit new resources. In the economy, smaller firms might find niches to compete in, away from bigger firms. However, diversification is not always possible if the winning species/firm has enough power to crush any offshoots.
  • Balancing loops. For example, many sports and games provide handicaps for weaker players to keep things interesting. Antitrust laws are supposed to act as a balancing loop to keep any firm from taking over completely. However, one “reward” of winning enough power may also include the power to weaken antitrust laws.
  • Periodically levelling the playing field. Examples include progressive taxation, charity, public welfare, and inheritance taxes.

My thoughts

I thought Meadows’ assessment of the problem was generally accurate. While she didn’t go into a lot of detail on the solutions, I found her categorisation helpful.

One balancing loop she didn’t mention was that there are also costs to being big and having lots of power. For an organisation, this could mean less agility, weaker information flows, and more delays. For an individual, see Jeffrey Pfeffer’s explanation of the costs of power. This balancing loop is in-built and, depending on its strength, may lessen the need for interventions.

Problems with common economic measures and analysis

Meadows points out that economists and economic news tend to focus on events and flows because that’s where interesting variations crop up. A commentator might say that stocks went up because interest rates fell, but won’t explain further. Econometric analysis goes one level deeper to look at a system’s behaviour and statistical trends over time.

However, Meadows argues that this is still insufficient. A statistical link between flows that previously held up will not keep holding if something in the system’s structure changes. So, although behaviour-based econometric models are pretty good at near-term predictions, they’re quite bad at predicting longer-term performance. And they’re terrible at telling us how to improve the economy.

Similarly, Meadows points out some limitations around gross national product (GNP), which is a measure of a nation’s income (a flow). Because she argues we should care more about a nation’s total capital stock than its flows, she thinks GNP is overrated. Moreover, GNP doesn’t capture all flows since it doesn’t accurately account for declines in a stock (e.g. from wars).

Meadows sensibly acknowledges that “there is every reason to want a thriving economy”. She just points out that governments sometimes do dumb things in mindless pursuit of GNP.

My thoughts

Meadows’ critiques seem valid to me, but I should clarify I know nothing about econometric analysis. As far as economic news goes, Meadows’ points support keeping a low-news diet (an idea which I find appealing and has gained traction more recently).

One quibble I have is that she could have been more balanced on the GNP point. As noted above, there are problems with every measure and with single-minded pursuits in general. But, for all their flaws, GNP and its close cousin gross domestic product (GDP) are better than nothing. A 2016 study found GDP per capita to be an excellent indicator of welfare across a broad range of countries. The correlation between GDP per capita and the broader welfare measure was a whopping 0.98! (There are caveats — e.g. for any given country, the difference between the two measures can be high, and the welfare measures themselves are imperfect.) People relying on GNP — or any measure — just need to be aware of its limitations.

Interest rates and discounting the future

The most surprising economics-related view Meadows expressed was this:

One of the worst ideas humanity ever had was the interest rate, which led to the further ideas of payback periods and discount rates, all of which provide a rational, quantitative excuse for ignoring the long term.
— Donella Meadows in Thinking in Systems

My thoughts

This is a somewhat radical view, and calling the interest rate one of humanity’s worst ideas is surely an exaggeration. But I think Meadows raises an interesting point.

There’s nothing “objective” or “true” about the idea that the future is worth less than the present. It’s only true because the market prices it as such. The interest rate just reflects the fact that most people are myopic. (It also reflects inflation and risk, but we can ignore those for present purposes.) And because you’re trading with other people, that leads you to be somewhat myopic too.

Here’s an analogy: the market prices chicken at $10 per kg and lobster at $100 per kg. That doesn’t mean lobster is “objectively” better than chicken — it just reflects the relative scarcity of lobster. But even if I personally prefer chicken, I wouldn’t trade lobster for an equivalent amount of chicken because I know the market values lobster more. Similarly, you’d rationally demand a premium for deferred returns if you know the market offers one, even if you personally don’t discount the future.

The Effective Altruism movement has also discussed whether to discount the welfare of future generations (see my summary of an 80k Podcast Interview with Will MacAskill). Where I personally land is — from an objective and rational standpoint, no. But we are humans, not fully objective and rational beings, so we’ll naturally discount, just like we might prioritise our families and friends over strangers.

Conclusion

I was generally impressed by the economics points Meadows made in Thinking in Systems, given economics was not her background. While I had a few minor quibbles, overall I think she raised relevant systems-related issues with balance and prompted me to think about some of these issues in some new ways.

I’m also giving Meadows extra leeway because the book was published posthumously, years after she unexpectedly died. So she didn’t have the luxury of updating her work for more recent developments in either the evidence or her thinking.

If you enjoyed this post, you may also like:

6 thoughts on ““Thinking in Systems” and Economics

  1. I’ve got three comments.

    1. The limits to growth argument is interesting. First, let’s assume it’s true that GDP growth cannot go on for ever.

    The question I always struggle with is: what is the implication of this right now? Is it that we should stop GDP growth? Why is this true now, but not true in 1974? Or was it also true then? What about 1924 or 1874? I think we’d stopped GDP growth when Meadows was originally writing, millions of people (billions?) would be living below bare minimum poverty lines (https://ourworldindata.org/grapher/the-share-and-number-of-people-living-in-extreme-poverty).

    But also, there is the possibility of economic growth without resource use. I’ve seen this used as dematerialisation. An example would be switching from printing CDs to delivering music through streaming.

    2. Rich get richer

    It seems very important to me whether the rich get richer and the poor get poorer, or the rich get richer and the poor get (less) richer. If I am in a rich country and my wealth/income grows at 5% p.a, but in a poor country under our current model, it grows at 3% p.a, then inequality will grow but absolute poverty will fall.

    A different regime where no one grows at all seems far worse to me.

    3. I am enjoying the Pinker book, Enlightenment Now, and even though he doesn’t directly address systems-thinking, in a way he has a criticism of a particular type of systems-thinking. Here’s the quote:

    “During the high-crime decades, most experts counseled that nothing could be done about violent crime. It was woven into the fabric of a violent American society, they said, and could not be controlled without solving the root causes of racism, poverty, and inequality. This version of historical pessimism may be called root-causism: the pseudo-profound idea that every social ill is a symptom of some deep moral sickness and can never be mitigated by simplistic treatments which fail to cure the gangrene at the core. The problem with rootcausism is not that real world problems are simple but the opposite: they are more complex than a typical root cause theory allows, especially when the theory is based on moralizing rather than data. So complex, in fact, that treating the symptoms may be the best way of dealing with the problem, because it does not require omniscience about the intricate tissue of actual causes.

    Indeed, by seeing what really does reduce the symptoms, one can test hypotheses about the causes, rather than just assuming them to be true.”

    1. Hey Phil – thanks for commenting. I think you raise very valid points.

      1. Limits to growth
      You’re right that “When?” is a really difficult question. But just because we can’t pinpoint when the tipping point is, doesn’t mean such a point doesn’t exist.

      While I generally believe that economic growth brings many benefits, I do think that past some level of material affluence, the marginal benefits of extra GDP will be outweighed by the marginal costs (which are harder to measure). I am not saying we were at that point in 1974 or even now, because there are still hundreds of millions of people in extreme poverty.

      If we ever progress to the stage where pretty much everyone has their basic needs met — and then some, I think we’ll have passed that tipping point. Yet it’s also possible we’ll pass the tipping point before reaching such an ideal. And I can’t be sure we haven’t passed that point already.

      You may have heard the Edward Wilson quote: “The real problem of humanity is that we have Paleolithic emotions, medieval institutions and godlike technology.” It’s somewhat hyperbolic, but I do worry that our pursuit of economic growth and technological progress (which tend to go hand-in-hand) can sometimes be reckless. I referred to AI in my original post but nuclear weapons and catastrophic climate change are also candidates for concern. Like, if the Cuban Missile Crisis caused a nuclear war back in 1962, maybe the point where we should’ve stopped pursuing growth was sometime before then …. So some survivorship bias here.

      Your point about the implications also brings up the question of “How?” — which is just as difficult as “When?”. Meadows doesn’t talk about this much (in Thinking in Systems, anyway) but I’m not sure the answer is to try and *stop* economic growth per se. It may just be to deprioritise it — reducing R&D incentives, etc. But both the nuclear weapons and AI examples show there’ll often an incentive to outdo competitors.

      On your second point about economic growth without resource use — I don’t have any objection to that. In fact, it seems to be what the “Limits of Growth” report was arguing for — changing the quality of growth rather than stopping growth altogether.

      2. Rich get richer
      Meadows makes clear that a key assumption behind her conclusion that this system is unsustainable if the winners’ gains are at the losers’ expense (i.e. zero-sum). What you’re describing — where the rich get richer and the poor get slightly less richer — is a different type of system (win-win).

      Real life I think contains a mix of both zero-sum and win-win games. So sometimes the rich do gain at the poor’s expense, but other times everyone wins. I think that if we didn’t have at least some of the measures that we do to arrest the reinforcing loop (e.g. progressive taxes, public welfare, antitrust laws), there’d be more zero-sum than win-win games.

      I generally agree with you in that I care more about reducing absolute poverty than relative poverty, but I wouldn’t dismiss relative poverty entirely, either.

      3. Symptoms vs root cause

      I haven’t read Enlightenment Now but your quote sounds consistent with what Meadows advocates — humility, testing, being flexible and adapting, etc. I agree that requiring “omniscience” to intervene is far too high a bar and would mean we’d never end up doing anything.

  2. Thanks for thoughtful reply.

    On 1, I’d have to read the report proper, but your earlier quote from Meadows saying we would concentrating on growth in the wrong direction (I assume she was saying we should shrink), and the wikipedia entry you link to do seem to suggest they are arguing against growth overall, not just its quality. The modern degrowth movement seems to be an offshoot of the Club of Rome so I think it might really be that they want economic activity to shrink overall.

    On 3 I was thinking of those who argue that the systems-thinking approach to (e.g.) housing is to “decommodify” it and treat it as a human right instead of a financial asset. I don’t think that would work at all. (It might solve some problems but would create worse ones).

    Now, arguably that’s just bad systems thinking and does seem inconsistent with what Meadows advises. But the quote made me think that systems thinking might attract (among other more genuinely thoughtful people) some people who like to make grand pronouncements about paradigm shifts, but without doing the hard thinking of how everything would actually work in their new paradigm.

    1. On 1 – when I first read Conclusion #2 in the Wiki article I thought it was advocating for sustainable growth but, on a closer look, I think you’re right as it doesn’t mention “growth” – just a sustainable equilibrium.

      I wonder if part of the problem here is that we’re using different definitions of what we mean by “growth” – the “getting bigger/consuming more” kind of growth or the “increasing efficiency/technological progress” kind. As noted in my original post, Meadows didn’t distinguish between GDP and GDP per capita (which is most unhelpful), but I think she’s largely talking about “getting bigger” as that makes the most sense in the context of a “finite environment”. Whereas the Noah Smith blog you linked to seems to be defending largely the “increasing efficiency” type of growth, though he’s not entirely clear about this, either.

      I too am guilty of switching back and forth between the two types of growth. I’m pretty comfortable with the idea that the “getting bigger” kind of growth is not something we should be aiming for, which is why I’m not too concerned about Meadows’ statements. I’m more ambivalent about the “increasing efficiency and technological progress” kind of growth – I generally think this type of growth is good but that it also comes with risks and shouldn’t be pursued at all costs.

      On your last point – yeah, unfortunately I think every good idea or way of thinking also has the potential to be simplified and bastardised.

  3. I follow Noah Smith on twitter, and he’s now taken to saying “growth is the same output with fewer resources” to antagonise degrowth people.

    But I think his definition of growth isn’t right. Growth is also more output with more resources. His definition is really about efficiency/productivity.

    Perhaps your position isn’t necessarily about growth, but about particular technologies?

    For example, if solar cells and batteries keep improving, then eventually energy may become too cheap to meter. I love that idea, and I think if it were genuinely carbon-free (or almost carbon free) then any thinking person really should, right? But I also have some concerns about AI (even though at the same time think it is quite great and brilliant), and would worry about particular technologies.

    It makes more sense to me to worry about particular concerns like climate change, or AI extinction, rather than *growth* per se.

    1. Agreed. I think arguing about “growth” in the abstract without clarifying exactly what is meant leads to people talking past each other. Same with arguing about “technology” in the abstract since, as you point out, some technologies are more dangerous than others.

      Pushing to stop all “growth” in general is, imo, just as wrongheaded as trying to pushing for all growth.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.