Does Reduced Consumption, and Increased “Saving,” Result in “Capital” Formation?

Matthew Yglesias riffs off my recent post, “Saving” ≠ “Saving Resources,” and there’s been quite a bit of commentary there, plus on Asymptosis and Angry Bear (plus a bit of twitter talk that I can’t figure out how to link to easily and usefully).

There are a dozen things I want to discuss on the topic, but I’d like to address the key belief underpinning much of the commentary (including Matthew’s). In my words:

If you don’t spend all your income, the unspent part is used by others to produce/purchase* “fixed” or “real” or “productive” assets. More money gets spent on investment, and less on consumption.

There are all sorts of problems with this notion, empirical and theoretical (notably the confusion of an accounting identity, “S is identical to I,” with economic incentives). I want to try and cut to the crux, with this:

A. If I transfer $100K from my bank to yours to purchase goods or labor, is there more money to produce/procure productive assets?

B. If I (or all of us) instead transfer $75K, leaving (“saving”) $25K in my bank, is there more money to produce/procure productive assets?

The answer to B is obviously “no.”

I hope not skipping too many steps here, so as to render this incomprehensible, I think Dan Kervick makes the key point in his comment on Matthew’s post:

a significant portion of monetary saving is just used to purchase government bonds

I would add, “directly or indirectly.” And: government bonds are only part of it.

This imparts the crucial understanding of aggregate, inter-sectoral balances that people lose sight of when thinking in terms of personal, individual “saving” of “money.” (Usually, implicitly, people are thinking about an isolated, domestic, private, non-financial sector — U.S. households and non-financial businesses.)

The financial system (including treasury and Fed) is constantly creating new, more, financial assets. New government bonds and currency, in particular, have no direct relationship to real investment. When you (or your bank) buy(s) a newly-issued government bond, you’re not funding/financing/incentivizing real private-sector investment in productive/useful capacity. (Though in one accounting view, you could argue that you’re “funding” government investment.)

So in a very real sense those financial assets (and arguably many [private-though-not-necessarily-“real”-sector] others) “absorb” “money” without creating new productive capacity. (This does not imply “crowding out.” Interest rates are at historic lows, and corporate cash hoards are at historic highs, even while government bond issuance has also been at historic highs.)

Funds flow from the private domestic nonfinancial sector into the the financial and government sectors (in return for an increased stock of IOUs). But absent intentional action (lending by the banks, deficit spending by government), they don’t flow back into the private domestic nonfinancial sector — and even less certainly into investment by that sector.

This is greatly simplified, and there’s much more I’d like to say, but I’m hoping to impart a straighforward (though incomplete) understanding of this view.

Here’s how I see it (this is the most important part of this post):

Production produces surplus. Output > Input.

That aggregate surplus is monetized by trade and a financial system (including treasury and fed), in a stunningly complex process that I won’t detail here. That’s why the quantity of financial assets (“money”) keeps increasing — because the surplus increases the stock of real assets, and the stock of financial assets (loosely) represents the value of those real assets.

If producers can’t sell (trade) their goods — in the process monetizing the value of the surplus created — they don’t produce them (as a successful serial entrepreneur, I’m here to tell you…), and you get less surplus. So less saving. My saving happened because people spent.

Spending causes saving. (Counterintuitive, huh?)

Though I prefer the term “accumulation.” The moral valences associated with “saving” — and the misunderstandings of its technical meaning(s?) in the national accounts — have resulted in no end of economic confusion (and confution*).

And yes: spending — and the production/trade/surplus-creation it spurs — causes monetary saving. The creation of surplus effectively forces the financial system to create new financial assets, so the producers of that surplus (workers and businesses) can monetize that surplus, and store it in their accounts. If the financial system doesn’t effectively monetize workers’ and producers’ surpluses via wages and profits, they have less incentive to work and produce, so a weak economy/slow growth results. Fed governors get replaced, politicians get voted out, and banks lose money or at least lose out to competitors who are willing to monetize the surplus.

I really have to finish this up by citing Dan Becker again, in a response to Pete Petepete at Angry Bear:

As I read your postings, it seems you are moving the discussion toward the chicken or the egg type.

But it’s not just Pete Petepete. We’re all really rehashing the old Say’s Law argument here: does production cause consumption (“demand creates its own supply”), or the reverse? The obvious answer is “Yes. Both.” But I think it’s clear which side I fall on, and I fall on that side because we have a sovereign-currency-issuing government, and a massive financial system which also constantly creates new financial assets. Say’s Law only makes sense if 1. there’s full employment***, and 2. there are no new financial assets to monetize/store/”hoard” the surplus from production and trade.

If all the “so-called” quotation marks in this post are driving you batty, my apologies. So many of the key terms in economics are used so sloppily and in so many ways, I often find it impossible to talk about the subject without constant parenthetical definitions of terms — which definitions themselves often deserve full blog posts. I hope this post will at least encourage my gentle readers to think very carefully about what I (and they) mean when using these terms.

* The produce/purchase distinction is conceptually problematic in itself (and as it’s tallied in the national accounts), as made clear by discussions among Kuznets and company back in the days when they were creating the national accounts; trade is the juncture where real surplus from production is monetized, which drops us into the thorny theoretical thickets of “value,” “capital,” and the mysteries of “money profits.”

** Yes I know that’s not a word. But it should be. Hey: good name for a new blog?!

*** Full employment is another problematic concept. Are there realistically imaginable scenarios — i.e. wage inflation without commensurate price inflation — in which large numbers of permanent non-workers would be coaxed into the work force, increasing employment without changing the percent “unemployed”? Full employment compared to what?

Cross-posted at Angry Bear.






23 responses to “Does Reduced Consumption, and Increased “Saving,” Result in “Capital” Formation?”

  1. Ben Avatar

    “A. If I transfer $100K from my bank to yours to purchase goods or labor, is there more money to produce/procure productive assets?

    B. If I (or all of us) instead transfer $75K, leaving (“saving”) $25K in my bank, is there more money to produce/procure productive assets?

    The answer to B is obviously “no.” ”

    I think that if you change the word money in each sentence to “labor and materials”, then your answer changes, assuming that you’re paying me for something, and not just giving me money for free. Your purchase incentivizes me to not produce productive assets with my time and materials. By purchasing 25% less, you save 25% of my time and materials, which I can use to produce productive assets.

  2. Ben Avatar

    I think Sumner’s argument for taxing consumption is that when people purchase goods and services to consume, they’re essentially forcing the economy to allocate resources–either materials or people’s time–in ways that benefit (usually only) them, and that this is the action we want to disincentivize through taxes. When people save, the economy is able to allocate those same resources in other (potentially more productive) ways.

  3. Asymptosis Avatar


    Again: aggregate vs individual confution.

    If we all “save more” there’s nobody to pay you that $25K to “produce productive assets.” So why would you do so?

  4. Ben Avatar

    I don’t think Sumner wants everyone to save more. He wants people who consume excessively to save more / consume less.

    By buying a latte, we encourage the marginal laborer to become a latte-maker rather than a grocery clerk. If we forego lattes, then that laborer’s time can be used to produce and distribute more groceries instead. The individual goods and services don’t matter. The point is that there’s a finite amount of materials and laborer-time, and they can be devoted to building yachts or to building factories; to operating nightclubs or to operating schools; to advertising departments, or to R&D departments. These tradeoffs involve real resources, and not just nominal cash.

  5. Tom Hickey Avatar


    The economic point is, however, that when there is an output gap and elevated unemployment, there are idle resources available to be put to use. It is neither effective nor efficient to repurpose them in aggregate, although individually well-managed firms are always seeking ways to be more productive.

    When available resource are idle, what is needed is for effective demand to increase in order to send a signal to firms to invest, e.g., by hiring clerks to handle the increased demand at their counters. When an economy is producing all the non-discretionary goods it needs, then it can turn to producing whatever discretionary goods it can support. Only a poorly managed system would produce discretionary items when non-discretionary items are needed but lacking, because effective demand will turn to non-discretionary first — unless inequality of income and wealth distribution influence behavior in aggregate.

    The answer in an economy with government is not for the private sector to reduce consumption to save more, but for government to deficit spend to meet the saving desire by providing net financial assets to non-govt in aggregate to offset demand leakage to saving and thereby support full employment.

    The way to optimize capital formation along with resource use is to ensure adequate effective demand to drive investment, and that means that when non-govt desires to save instead of consume to the degree it creates drag on effective demand, then govt needs to step in fiscally to offset the demand leakage.

  6. […] Roth (1, 2), Scott Sumner (1, 2, 3), Bill Woolsey, and Matt Yglesias have been debating questions of saving […]

  7. Asymptosis Avatar

    Totally agreed on all that.

    That’s unaffected by the key point: everybody (in the private domestic nonfinancial sector) leaving $25K in their accounts, instead of transferring it to each other’s accounts, doesn’t increase the amount of money available for, or devoted to, investment spending.

    QTC, really, because each of those transfers constitutes trade in real goods, and causes production of real goods (both consumption and investment goods), which produces the surplus from which savings derive.

    If those transactions don’t happen, the production and trade don’t happen, so the surplus isn’t produced, so the saving doesn’t happen.

    And that not-spending doesn’t (quite obviously) increase the amount of spending overall. Reduced spending does not cause increased spending. At least in the NIPA expenditure approach, spending is GDP.

  8. Ben Avatar

    So I think the argument you’re responding to is basically as follows:

    (1) When people make consumption purchases, they exchange money for goods and services, and then consume/destroy those goods and services.
    (2) When people make investments, they exchange money for a financial instrument, possibly tied to the creation of an investment good.
    (3) By making fewer consumption purchases, we necessarily create more investment goods.

    And I think you’re arguing against point 2, by saying that most investments are not tied to investment goods.

    But suppose we reframe the argument as follows:

    (1a) When people make consumption purchases, they incentivize workers to produce goods and services that are quickly consumed/destroyed.
    (2a) When people make investments, they do not incentivize workers to produce goods and services that are quickly consumed/destroyed.
    (3a) By making fewer consumption purchases, we free up resources to produce goods and services that are not quickly consumed/destroyed.

    I’m not sure which, if any, of these points you object to.

  9. Tom Hickey Avatar


    A tradeoff between the funding used for consumption and the funding saving makes available for investment only holds with a fixed money stock. In an endogenous money system, it’s not either/or but both/and. When consumption is increasing, firms borrow “new money” (loans create deposits) and the money stock expands to accomodate the increasing propensity to consume with new investment. That increase in investment is owned by households with equity in firms, so saving increases correspondingly.

  10. Ben Avatar

    @Tom Hickey

    I don’t think the flow of money is relevant to the point that Scott Sumner is making, and which I’m defending. What we’re concerned about is the behavior or real resources–materials, and laborers’ time. The point Sumner makes is that when I order a latte and drink it, I reduce the stock of real resources, because the coffee and the barista’s time are gone. If instead I leave my cash in a checking account, or buy T-bills, or invest in a hedge fund or startup, none of these actions reduce the stock of real resources.

  11. Asymptosis Avatar

    @Ben “When people make investments, they exchange money for a financial instrument,”

    Couple of problems here. One semantic but important.

    In national accounts, “investment” means spending to create/buy *newly produced* “fixed assets” that will last beyond the current period. Structures (residential and non), Equipment, and software. Plus inventory. (Notably, it does not include investments in “unfixed” capital like education, research, and health care.)

    “Investment” in the vernacular means you buying a share of apple stock from me. This is not investment in the national accounts; it’s an asset swap — cash (bank deposits) for share of stock, or an existing house.

    “possibly tied to the creation of an investment good.”

    That “possibly” is the very crux of the question we’re asking here, right? If you trade your bank deposits (ultimately, Fed reserves) for a newly issued government bond (or just leave the money in your bank account), how many (presumed) steps does it take before somebody manufactures a drill press?

    “By making fewer consumption purchases, we free up resources”

    In a service economy, not necessarily so unless we’re at “full” employment.

    Again, competition between different industries, trades, and professions is not the same as the whole economy.

  12. Tom Hickey Avatar

    Ben :
    @Tom Hickey
    I don’t think the flow of money is relevant to the point that Scott Sumner is making, and which I’m defending. What we’re concerned about is the behavior or real resources–materials, and laborers’ time. The point Sumner makes is that when I order a latte and drink it, I reduce the stock of real resources, because the coffee and the barista’s time are gone. If instead I leave my cash in a checking account, or buy T-bills, or invest in a hedge fund or startup, none of these actions reduce the stock of real resources.

    What is the point in reducing the use of available resources unless they are in short supply and need to be conserved for future use. “Saving” real resources makes no sense in a capitalistic economy based on growth. Fewer consumption purchases don’t free up resources, they idle resources. Some think that saving causes investment, when the truth is that investment creates income and the residual of income not consumed is saved. Investment results in saving, not the other way around.

    As far as firms and industries go, capital flows to where it can optimize return relative to risk. It not a matter of some people deciding to reduce consumption at one firm or industry and this being transferred to another. As consumption falls, profit falls, and capital switches to more profitable uses — some firms close and some industries shrink, and other firms start up and other industries grow. Schumpeter called this process “creative destruction” and “innovation.” It’s going on all the time. If this process were perfectly flexible it would be perfectly efficient but it is not. Significant capital is wasted in the process of moving an economy forward through creative destruction and innovation.

    There is no causal connection running from saving to investment, because in an endogenous banking system, banks don’t loan out savings. If someone comes to a bank for a loan to increase investment, a bank will make a loan to anyone with a reasonable plan who is creditworthy and is willing to pay the going rate of interest rather than let the business go to a competitor. The bank doesn’t look at its deposit position or its reserve position, although it does check its capital position. If it has requisite capital then it will make the loan and borrow reserves as needed to clear the deposit. The bank generally seek to attract deposits for funding since they are the least expensive source of funds. Having deposits on hand is essentially irrelevant to credit extension, which is based on demand of creditworthy customers and a profitable spread. The rest, like asset-liability management, comes later, after the loan is made.

    At the macro level, the paradox of thrift shows that if many save instead of few, that is, propensity to consume drops relative to income, this sends a signal to the owners of firms that effective demand is decreasing, so they should think about reducing staff. No signal is sent to another firm to increase staff so the result is that there idle available resources, including unemployed or underemployed workers. As saving propensity increases there is more money sitting in bank accounts, money market accounts, and other financial assets that make up portfolios that is offset on the investment side as unplanned inventory. As we are seeing, businesses are not borrowing at historically low rates as they “should” according to the loanable funds doctrine and ISLM model, where low rates encourage borrowing. This results in economic contraction until the inventory is worked through and effective demand signals the need to produce more. Unless govt injects enough to offset the increased saving desire, including external saving desire due to a trade deficit. In order to return to full employment, the govt fiscal deficit has to offset the non-govt surplus at optimal capacity. Otherwise the economy is wasting resources by idling them due to demand leakage to saving.

  13. Ben Avatar

    “What is the point in reducing the use of available resources unless they are in short supply and need to be conserved for future use.”

    The available resources are in short supply. Always. There are never enough innovators, entrepreneurs, inventors, teachers, researchers, artists, healers, philosophers, etc. These are examples of service equivalents of productive goods, in that they generate long-run return, rather than just being consumed and forgotten. There are only so many people who can do these things, and right now many of them are instead producing, distributing, and marketing consumption goods. Which is fine, if that’s what we want. There’s nothing wrong with Gucci selling people suits if that’s what they want. But there’s also no denying that if everyone stopped buying Gucci, then the people working in Gucci’s marketing department would need to find new jobs, and some of them would become entrepreneurs or philosophers.

  14. Tom Hickey Avatar


    “The available resources are in short supply. Always.”

    That’s just daft. Looking at national unemployment in the US and UK and also globally, you could fool me on that. Looks to me like a vast pool of idle resources waiting to be put to work.

  15. Tom Hickey Avatar


    By the way, scarcity of all resources is a fundamental assumption of neoclassical economics. This only holds at full capacity-employment as constant, and neoclassical economics presumes this to be the equilibrium state.

    Old Keynesianism (General Theory) and Post Keynesianism deny the neoclassical assumptions and believe to have disproved them. (Samuelson and New Keynesianism are built on neoclassical economics.)

    So we are likely not going to resolve this here, since the differences are ideological, depending on opposing models, rather than factual, i.e., based on differences regard factual data in the same model.

    Models differ based on how well they explain and predict, and they are evaluated on results. If two models yield the same result, then the simpler one is preferred. but if a simpler model does not yield as good results, then it is presumed to be simplistic.

    The neoclassical presumption of scarcity of resources and equilibrium at full capacity-employment implies that all unemployment is voluntary, which is critics hold is patently not the case in situations such as we are experiencing today.

    Presuming equilibrium neoclassicals economists have to bring in an external shock disrupting the system and then claim that left to itself, the system will right itself most efficiently. Government should therefore not intervene with stimulus, which just takes resources away from the private sector and slows the recovery. Providing unemployment relief is therefore inefficient.

    Post Keynesians like Wynne Godely, Steve Keen, and Michael Hudson predicted the crisis arising from financial causes that would lead to debt deflation and reduce effective demand. Post Keynesians not only foresaw the crisis but also explained its causes, while neoclassical economists admit that their model did not.

    Moreover, the Post Keynesian propose how to fix the issues using the same model that predicted the crisis. The neoclassical economists are trying to fix the issues using the same model that did not foresee it.

    Go figure.

  16. Ben Avatar

    @Tom Hickey

    Sure, you can punt and ascribe our disagreement to ideological differences, or you can address the thrust of my argument:

    If everyone stops shopping at Gucci, and Gucci has to lay off its entire workforces, what do those employees then do? It seems like the answer consistent with your position is, “they all do nothing”.

    If instead your answer is, “some of them become unemployed, some of them find similar jobs for other firms, and some of them find completely different jobs”, then you’re conceding that by reducing consumption of Gucci, we free up real resources to do other (potentially more productive) things.

  17. Tom Hickey Avatar


    That might be true if there are more productive things to do. For example, most top level grads in physics and math go into finance where their ability is directed to rent-seeking rather than production. Closing down the big banks would free them up to return to productive work.

    But the point is whether there is productive work. The workers that are being let go in the US now, if they can find a job it is at a lower level. What is expanding in not productive work but menial work. In fact the job qualifications for menial jobs are increasing due to the increased numbers of skilled workers that can only find unskilled work, if they can find a job at all.

    There is no plausible transmission mechanism for a worker put out of work at one place and find an equally productive job elsewhere in the economy. In fact, with global labor being fungible, that job is likely to end up in another country because there being a surplus of labor globally, capital is flowing to low wage locales.

    Your argument may hold in a simple model but not the actual world, especially at the macro level. Taking your Gucci worker as the representative worker, as neoclassical economics is wont to do, leads to fallacies of composition, e.g., one employed worker finds equally productive work so that they all do. They all don’t, unless the economy is operating at full capacity and there are positions available due to transitional unemployment. And even then not all do, since transitional UE is about 2% and the so-called natural rate of unemployment wrt setting monetary policy is several points higher than that, creating a permanent buffer stock of unemployed that controls inflation by suppressing labor bargaining power.

  18. Ben Avatar

    Tom Hickey :
    Taking your Gucci worker as the representative worker, as neoclassical economics is wont to do, leads to fallacies of composition, e.g., one employed worker finds equally productive work so that they all do. They all don’t…

    It depends how you measure productivity. If consumption goods and services (expensive shirts, yachts, massages) are inherently not as productive as productive goods and services (factories, education, R&D), since they are destroyed rather than being used for further creation, then there are plenty of ways for laid off employees of Gucci, ISA Yachts, or Massage Envy to find a more productive job, even if it pays less. Anytime a salesperson leaves to become a kindergarten teacher or lab assistant, that’s a big win for productivity.

  19. Tom Hickey Avatar


    There are several things operative here.

    First is the neoclassical assumption that consumption goods capital and workers are exchangeable with capital goods capital and workers. Keynesians deny that that assumption. When consumption declines and consumption goods firms shed workers and pare back production, capital and workers do not flow seamless to investment in capital goods.

    This is assumption is based on the representative workers, where sales people become teachers or lab assistants. Big assumption. What we are seeing is that workers are becoming underemployed and descending into jobs of lower skill level rather than finding jobs of comparable skill in the field in which they are trained.

    The neoclassical answer is that the issue is retraining workers for available jobs, which implies that the representative worker assumption is incorrect.

    The Keynesian solution is running a full employment budget to maintain the effective demand need to run the economy at optimal capacity. Capital will naturally flow in aggregate to the highest return, and that is where the investment go, and job offers will increase in that sector. It’s unnecessary to decrease consumption to feed investment, and workers will go to the work that pays highest for their skill level. It is unnecessary to free up the Gucci worker. If the Gucci workers can make more at producing capital goods with their skills, then that will be the workers’ choice if there are openings. Closing the Gucci store is not going to cause new investment in capital goods. It’s more likely going to result in cut backs at Gucci suppliers, and up the line toward higher level capital goods and resource production. Will these be repurposed to “more productive uses.” What’s the rationale they will be. Getting rid of one frivolity doesn’t mean that other frivolities won’t replace it.

    Since we have an endogenous money creation system, the money supply expands and contracts as needed to provide funding for investment. Decreasing consumption to increase saving in order to increase investment in capital goods is irrelevant in this type of economy. Neoclassical economics is not modeling the actual world because neoclassical economists are thinking in terms of a barter economy in which money is a veil rather than as a factor with it own dynamics that depend on the type of monetary system, financial system, and fiscal and monetary policy.

  20. Ben Avatar

    @Tom Hickey

    I’m all for enacting policy to create full employment, and if can produce a plan to get there, i.e. a series of testable experiments, then I’m all ears.

    I’m not clear as to why shifting from income to consumption tax would hinder that plan, and I maintain that even after achieving full employment, there’s good reason to shift workers from producing consumption goods and services to producing productive goods and services. Whether at 15%, 2%, or 0% unemployment, I’m still going to want the Gucci marketing analyst to find a new job.

  21. Tom Hickey Avatar


    A consumption tax increases the cost of consumption goods, whether across the board like a sales tax, or else in a target way like a luxury tax or other excise taxes. But taxing consumption just makes consumption goods more expensive reducing sales either across the board, or else in different sectors. Increasing the sales doesn’t increase private sector saving by reducing consumption; it just increases govt revenue by the amount of the tax.

    Taxing luxuries does affect the luxuries taxed, as the effect of the 1990 luxury tax in the in the US showed. Yacht makers lost business as the wealthy shifted consumption to other luxuries that were not taxed or at least not taxed as onerously. The capital goods made available by yacht makers going out of business were not easily repurposed, and there is no evidence of which I am aware that workers went into more productive jobs. 7600 jobs were lost in the yacht industry and a total of 75,000, when suppliers are included. There was no revenue gain since yacht sales transferred overseas where US buyers paid no tax. The downside of the luxury tax was found to outweigh the objective of increasing revenue and most provisions were repealed in 1993. Been there done that.

  22. Ben Avatar

    I believe Sumner’s proposal is to a revenue neutral substitute of implementing a progressive consumption and reducing other taxes. I don’t know of any historical attempts to do so.

    I can’t tell you what those 7600 people did, but I strongly suspect that if you accept my earlier definition of productivity, their productivity increased.

  23. Tom Hickey Avatar

    Ben :
    I can’t tell you what those 7600 people did, but I strongly suspect that if you accept my earlier definition of productivity, their productivity increased.

    7600 people in the yacht industry and 75000 related when suppliers are included. That’s a lot of people unemployed. Their productivity all increased due to repurposing? That’s a huge assumption, given the unemployment rate at the time.