No: Saving Does Not Increase Savings

January 28th, 2014

The misconceptions embodied in this post’s headline sow more confusion in economic discussions than any others.

“Saving” and “Savings” seem like simple concepts, but they’re not. They have many different meanings, and their different usages (often implicit or unconscious) make coherent understanding and discussion impossible — even, often, in writings by those who have otherwise clear understandings of the workings of financial systems.

In short: if you disagree with this post’s headline, you are thinking (perhaps unconsciously) in the “Loanable Funds” model. And the loanable funds model is complete, incoherent bunk. (I’m not even going to bother citing the hundreds of supporting links here, including unequivocal papers from central bank research departments worldwide; Google them.) 

Think this through with me:

Your employer transfers $100K from their bank account to yours to pay you for your work. You’ve saved.

But is there more savings in the banks? More money to lend? Obviously not.

You buy $50K in goods from your employer, transferring the money from your account to theirs. They’ve saved. You’ve dissaved (spent).

But is there more or less savings in the banks? More or less money to lend? Obviously not.

You transfer $50K from your bank to your employer’s, in exchange for $50K in Apple stock or government bonds.

Did you just “save” again? Is there more savings? More money to lend? Obviously not.

When people save up money, they do not add to the stock of monetary savings — the mythical stock of “loanable funds.” Monetary saving does not increase monetary savings.

And since saving up money doesn’t increase the stock (supply) of money, it doesn’t lower the cost (interest rate) of money. On the aggregate level, saving doesn’t “fund” lending. (Banks lend by creating new money ex nihilo if they think they’ll make a profit; that’s what they’re licensed/chartered to do.)

This error and misunderstanding exists partially because there are two ways to “save,” which are widely confuted, conflated, and confused:

Pay people to create real assets — drill presses, houses, ideas, skills, etc. — that you own or have a claim on. Paradoxically, in this case you save by spending.

Increase holdings of financial assets (net of debt) — from dollars to debt securities to Dell stock to CDOs. (It’s much easier to think about this coherently if both dollar bills and deeds are viewed as financial assets: legal constructs designating particular rights or claims on real assets.)

The confution of these two, of course, was all hashed out many decades ago in the Cambridge Capital Controversy — when the MIT/classical Cambridgeans admitted defeat at the hands of the Cambridge Cambridgeans, including acknowledging their central point. (The classicals have proceeded to ignore the whole thing ever since, acting and thinking just as they did before, as if it never happened.)

…the measurement of the “amount of capital” involves adding up quite incomparable physical objects – adding the number of trucks to the number of lasers, for example. That is, just as one cannot add heterogeneous “apples and oranges,” we cannot simply add up simple units of “capital.” As Robinson argued, there is no such thing as “leets,” an inherent element of each capital good that can be added up independent of the prices of those goods.

(This also explains why the the Q in MV=PQ is utterly incoherent: the only unit that can be used to measure Q — dollars — varies with P. It’s like measuring a rubber band using a ruler whose inches vary in length with the changing length of the rubber band.)

People get confused about this partially because “saving” in the national accounts encompasses both real capital created (measured by dollars spent to create it), and net acquisition of financial assets. It’s not what people think it is: a simple sum of everybody’s money saving (income minus expenditures) — not even close.

Here’s how monetary saving happens — aggregate increases in people’s net holdings of financial assets:

People spend money (some of it borrowed from the financial sector), paying people to create real assets. (Purchases of existing real assets can spur creation of new ones — a second-order effect — but the creation’s the thing.)

The market decides that the financial assets that are claims on those assets are worth more than was paid to create the real assets. (Sometimes the market overestimates; this is a problem.)

The people’s debt is unchanged, but their financial assets are worth more. Their net worth has increased. They’ve saved up money. (When the market optimistically bids up financial assets, there’s suddenly, magically, more money.)

Saving (not-spending) money doesn’t increase monetary savings. We saw that obvious reality at the top of this post. Spending (partially enabled by new money creation via bank lending and government deficit spending), coupled with market re-pricing of financial assets, increases monetary savings. This is how the financial system monetizes the accumulated surplus from production.

Since saving money is not-spending, more saving results in there being less savings (relative to the counterfactual of more spending). Or if you must call them this, less loanable funds.

Spending increases savings.

There’s more I’d like to say, but I’ll leave this with a question for my gentle readers:

The IS/LM model seems to be inescapably based on the misconception detailed above — that more saving results in more savings hence, because of supply and demand for loanable funds, lower interest rates.

But: if Krugman’s constantly repeated assertions are correct, that model seems to perform very well.

Why is this true? What am I not understanding?

Cross-posted at Angry Bear.

  1. January 28th, 2014 at 13:40 | #1

    “Pay people to create real assets — drill presses, houses, ideas, skills, etc. — that you own or have a claim on. Paradoxically, in this case you save by spending.”

    This is close (due to your attempt to link S with I), but wrong, I think.

    Say you pay John $100 for a factory. If we take the expenditure approach to national income accounting, we are measuring expenditure on goods produced in a given sector. For your sector, you are spending $100 on an investment good, but you are importing it from another sector (John). Thus spending on goods produced by your sector is 0: Y = C + I + G + X – M (leaving out the govt sector) = $100 (I) – $100 (M)= 0. Likewise, saving (S) in your sector, which is S = Y – C = $0 – $0= $0. You haven’t saved by spending. Rather, you are swapping money for an investment good and your net wealth stays the same.

    For John, on the other hand, he’s earned income of $100 by exporting $100 of goods to John’s sector. Y = X = $100. S for his sector is S = Y – C = $100 – $0 = $100. John has saved by producing an investment good, adding $100 to his wealth.

    If we aggregate to national Y, I, and S, where national means both you and John, then the sum of Y is 0 + 100 = $100. This makes sense, since $100 of goods were produced. The sum of S is 0 + $100, which makes sense, since the goods that were produced were investment goods, and national S = national I in this scenario.

  2. January 28th, 2014 at 13:56 | #2

    So rather, you contribute to national saving S by spending on and acquiring an investment good (I). But you yourself are not saving (S) as an individual economic unit. The counterparty is the one saving (S), but by the same token, that counterparty is not spending on or acquiring an investment good (I).

    So there’s a mirror image thing going on here, where even though S = I, the S and I don’t point to the same thing in a sense. JKH is talking about this here:

    “Suppose private sector saving in the current period consists entirely of household saving of S, with the result being an addition to household net worth and a corresponding bank deposit. Suppose also that a corporation has borrowed an amount equal to S to make a new investment I during the same period. Under these assumptions, S = I. But it is obvious that the substance of I (the material substance whose value is recorded on the corporate balance sheet) is different than the substance of S (a net worth increase whose value is measured on the household balance sheet and which equals an amount held in bank deposit form).”

  3. January 28th, 2014 at 14:05 | #3

    By the way, in all of this, I’m using national income account definitions of saving (S). That definition is different from what you’re trying to get at in this post, which is accumulation/decumulation of financial assets. But I commented specifically on the part of your post where you revisited national income accounting.


  4. January 28th, 2014 at 15:07 | #4

    @ATR: “Say you pay John $100 for a factory.”

    That’s buying a pre-existing asset, not paying John to build a factory. It’s just an asset swap, not saving.

    I was quite careful to say “Pay people to create real assets”

    That purchase may spur the creation of real assets in a second-order effect, but it’s not in itself creating an asset.

    The former is not saving. The latter is in a very real sense.

    But even that doesn’t turn into monetary savings until you sell the asset for more than you paid for it, or until your claim upon the factory (a financial asset) it is re-priced by the markets and that increase is recorded in your books.

  5. January 28th, 2014 at 15:47 | #5


    I’m not sure if we’re talking past each other. I meant that you pay John in exchange for his creation of a factory. You are purchasing a newly created asset for $100. Yes, I agree that is an asset swap. You give away $100 worth of financial assets (cash) in exchange for $100 worth of real goods (factory). But you said you saved by spending on the factory: “Paradoxically, in this case you save by spending.” That’s incorrect. You did not save by spending, in terms of national income account saving S. You did an asset swap. Your saving S was zero. John’s saving S was $100. National saving S was $100, and it’s equal to the factory investment I $100.

  6. January 28th, 2014 at 15:51 | #6

    (Agree none of this has to do with monetary accumulation/decumulation. That needs to be kept separate from national income account saving S.) @ATR

  7. January 28th, 2014 at 16:45 | #7

    “if Krugman’s constantly repeated assertions are correct, that model seems to perform very well.
    Why is this true?’

    ISLM assumes that the interest rate is set in the market by demand for loanable funds, even if the rate is set exogenously by the central bank as in the US. Then, the thinking seems to go, the cb mostly ratifies the rate an efficient market would otherwise set on it own ‚ e.g., the UK uses Libor, which the major market makers set daily, rather than a rate set by the BoE.

    In this view the interest rate “naturally” increases pro-cyclically with increasing demand for funds and decrease counter-cyclically with decreasing demand for funds, whether this is done exogenously by the cb as in the US or endogenously by the banks as in the UK. Thus, the interest rate is demand driven and savers increase the interest demanded as demand for funds rises.

    The problem with this is that loanable funds presumes scarcity and scarcity should increase with demand for funds. But that is opposite to what actually happens in an endogenous system, when banks can created deposits through loan extension and seek funding afterward, not only from deposits but also from none markets and the cb. So the supply of loanable funds is variable, expanding and contracting with loan demand.

    I think that Krugman claims that this really doesn’t matter, since interest rate changes behave as if loanable funds were in place, except at the lower bound where the cb does not set the rate negative. At this point, fiscal policy is required, as ISLM predicts it should be in the absence of negative rates.

    Of course, I may be misunderstanding Krugman’s position, but that is the way I read it, since he assumes both loanable funds in assuming ISLM and he also assumes endogenous money, which seems to contradict ISLM. Maybe he will make this clearer some time, or if I am wrong, perhaps someone will set me straight on his view.

    The general Post Keynesian view, again as I understand it, is that the central bank simply sets the interest rate based on inflation targeting using a rule in order to macro-manage through the price of funds since it does not control quantity exogenously. LT yields are a projection of expectations of future Fed rate changes, and the Fed can control the yield curve anyway using either price to set an exact rate or less precisely by quantity through taking bonds off the market using QE. So no need to bring in ISLM on an as-if basis since the assumptions on which it is based do not apply in the current system.

    According to Krugman PKE converges with the ISLM view that zero bound but these views diverge above the zero bound. So he sees as an opportunity to test the ISLM view against the PKE view when rates rise again.

    BTW, in the loanable funds model, S (saving) = I (domestic investment) + NFI (net foreign investment). Thus saving (S) is the account record of investment (I + NFI) in an open economy. See Wikipedia/loanable funds.

    All the talk about people putting money in the bank ostensibly to funding investment has nothing to do with either S = I (aggregates) in the two sector macro model of an economy or in the loanable funds model. It’s a confusion of the ordinary language meaning of “saving” and the technical meaning in macro.

    In addition, the economic meaning of “saving” is different from financial saving in a portfolio, the nominal value of which is constantly fluctuating with market prices and real purchasing power, just as “real” (different from real v. nominal) investment as spending on capital goods is different from financial investment as portfolio items. This moment to moment fluctuation of portfolio value and “paper” net worth is meaningless wrt to actual income and saving, since gains and losses are not booked until realized. Pace the wealth effect. (Then there is nominal-real adjustment.)

    “Saving” as it is used the macroeconomic sense is also different from the ordinary language meaning of individuals’ “saving” as money in the bank plus cash under the mattress, or pennies in the piggy bank. Because children are taught the meaning of saving as pennies in the piggy bank, this concept of saving persists as a connotation of “saving” later on, resulting in confusion where terms are not used equivocally between ordinary language and technical usage.

    Important terms like “saving” have different meanings in different contexts. Failure to keep the meaning straight in a context leads to ambiguity, and ambiguity leads to confusion. I think we are dealing with at least three contexts here, the ISLM context, the PKE context, and the ordinary language contexts, and probably the portfolio context, too. Lots of balls in the air all called “saving.”

  8. January 28th, 2014 at 16:54 | #8


    Actually, I’d avoid use of the term ‘asset swap’ here and just stick to the GDP accounting. Also, my understanding of GDP accounting is that the factory takes on value that flows into GDP (and thus measures like Y, I, S, etc.) when it is purchased as a final good on the market.

  9. January 28th, 2014 at 16:58 | #9

    This is what I was going to post originally, before I spotted the above error – which I think I am correct on.

    As Tom’s said: it’s important to be very cautious about linking national income account variables, such as saving (S), to concepts such money accumulation/decumulation – even though in our everyday life, we associate the word “saving” with changes in money balances. Here I agree with you. The former was ultimately designed to measure the production of goods and services in a nation, and the concept of saving (S) reflects this.

    Some easy ways to see this:

    An investment (I) in the form of a factory is defined to simultaneously create national saving (S), but this says nothing about national money balances that may or may not have been accumulated or decumulated in the process.

    More generally, one could imagine a closed economy producing a significant amount of GDP, but the variables for national Y, C, I, and S tell us little to nothing about the existing stock of money or credit created that facilitated the transactions leading to that GDP. Obviously, some amount of money must have existed, but we don’t know how much from those variables. It could have been any mix and quantity of net cash balances and credit. But that stuff is in the background and not captured by Y (and all its sub-component variables).

    The overall point being that “saving (S)” in the national income accounts is very much a separate concept from the everyday concept ‘money saving.’

    As for your questions regarding IS/LM and loanable funds, it’s a future project of mine. It does seem, however, that carrying the confusion of S vs. money saving into that model could lead to a whole host of misunderstandings. That’s just a precautious hunch I’m carrying into the analysis, though.

  10. January 28th, 2014 at 17:21 | #10


    Great paragraph:

    “As Tom’s said: it’s important to be very cautious about linking national income account variables, such as saving (S), to concepts such money accumulation/decumulation – even though in our everyday life, we associate the word “saving” with changes in money balances. Here I agree with you. The former was ultimately designed to measure the production of goods and services in a nation, and the concept of saving (S) reflects this.”

    Kuznets was trying to depict actual production, and real assets. As he said they are the true “wealth of the nation.” But of course there’s no such thing a “unit” of production, or of real assets. Sticky wicket.

  11. January 28th, 2014 at 17:31 | #11

    @Tom Hickey “I think that Krugman claims that … interest rate changes behave as if loanable funds were in place”

    I’d been thinking this. But the empirics don’t support it.

  12. January 28th, 2014 at 17:34 | #12

    Everyone: go look at how the International Macroeconomic Accounts (based on the internation System of National Accounts) tables are structured and labeled.

    The word “investment” (in the NIPA sense) does not appear anywhere. Ditto “saving” in the NIPA sense — the thing that equals “investment.” There’s only “net saving,” which I’m pretty sure means something different here than it does in the NIPAs. Still tracking that down…

  13. TRF
    January 29th, 2014 at 21:18 | #13

    I’m confused about your statement: “Banks lend by creating new money ex nihilo if they think they’ll make a profit; that’s what they’re licensed/chartered to do.”

    How does a bank create new money ‘out of thin air’ by lending? My understanding is that a bank has a balance sheet like everyone else, and that assets (loans) must be funded by liabilities (customer deposits, debt issuance) and equity (shareholder contributions). Before a bank can make a new loan (increase assets) it needs to ‘get’ the money from somewhere – either from new customer deposits or by borrowing in wholesale markets (increase liabilities). The only entity able to create new money is the Fed?

  14. January 29th, 2014 at 22:13 | #14


    Banks loan when there is demand by creditworthy customers and then fund their loans afterward. That’s the business of the Asset & Liability Management Department. Loan officers don’t consult with ALM before they make loans to see whether funding is available or if the bank has excess reserves at the cb. They let ALM know of loans coming down the pike that will need to be funded and ALM takes care of that iaw the bank’s policy wrt maturity transformation.

  15. Asymptosis
    January 30th, 2014 at 12:27 | #15

    @TRF “How does a bank create new money ‘out of thin air’ by lending? My understanding is that a bank has a balance sheet”

    You walk in and give them an IOU — an asset for them. They give you cash (bank deposits), a liability for them.

    Their balance sheet expands.

  16. TRF
    January 30th, 2014 at 20:40 | #16

    Sorry I think I was getting mixed up between central bank money (MB) and commercial bank money (M2?) I’ll admit now that I’m in the deep end and way over my head on this site!

    I think what you are describing is the manufacturing of M2 money (financial assets/liabilities) out of MB money (cash)? i.e the money multiplier. But doesn’t this still require some MB and funding – it cannot be done ‘ex nihilo’.

    As Tom points out above, a bank can make loans only while it has excess reserves at its CB. But just like everyone else it has to settle the draw-down of those loans with cash from its CB reserve accounts. If it starts running low on its CB reserve accounts it needs to top them up by sourcing funding in the form of deposits, savings accounts or debt issuance. A savings account / debt issue is just a bank converting a financial liability (an IOU to the saver/purchaser) into MB cash.

    So it seems to me that a bank has no more ability to ‘create’ M2 money than you or I – we are all constrained by the same limits of needing ‘cash’ in the form of CB reserves/liquidity.

    Just like a bank I can agree to lend you $100k (‘create $100k of money?) but if I only have $10k cash in my bank account then this ‘money’ has only been created in my imagination. However, if I can convince Tom to ‘save’ $90k with me by giving him a certificate of IOU in exchange for getting him to transfer $90k into my account, and I then transfer $100k to you, then I have ‘created’ $100k of M2 balance sheet money (MB is exactly the same). But it certainly wasn’t out of thin air. I had to convince Tom to ‘save’ with me so I could fund the loan to you. I’ve also left myself with no liquidity which will be a problem when I go to buy my groceries.

    So, I guess the point I’m trying to make is that while I agree that in general ‘the financial system’ creates financial assets/liabilities out of MB ‘money’, it’s not ‘ex nihilo’. It takes a lot of work and liquidity management.

  17. January 30th, 2014 at 21:47 | #17


    “As Tom points out above, a bank can make loans only while it has excess reserves at its CB”

    That’s a misreading of what I said. Excess reserves are not required prior to making loans for the simple reason that if the bank come up short, the central bank acting a the lender of last resort will make up the shortfall automatically and charge the bank the penalty rate. Banks wish to avoid the penalty rate so when loan officers indicate loans are in the pipeline, the appropriate staff arrange for adequate reserve balances to meet the reserve requirement by borrowing in the interbank or money markets. Banks don’t wait for to have excess reserves on hand to make loans.

    BTW, there is no money multiplier as an ex ante factor. It is an ex post accounting residual. The quantity theory based on the money multiplier is wrong. It’s got things reversed.

  18. TRF
    January 31st, 2014 at 01:18 | #18

    Sorry @Tom – didn’t mean to put words in your mouth!

    I’ve just found a whole discussion over at which you seem to have also contributed to. I’ll spend some time reading through the 238 replies to see if I can get a clearer understanding. From a brief look its refreshing to see I’m not the only one struggling with this…

  19. Asymptosis
    January 31st, 2014 at 06:36 | #19


    That NEP piece is a great place to start. Here’s another writeup over at Winterspeak:

    I so completely understand your confusion. I’ve felt like Jacob wrestling with the angel on this for years. None of the discussions I’ve read have fully satisfied me. Follow some Related Links in this post and you’ll see some of that wrestling.

    So much to say but I’ll cut to the chase: The endless (fascinating) discussion by dozens, hundreds of incredibly sharp and knowledgeable people proves to me that economists, people in general, don’t know what money or capital is.

    Or more precisely: they have no agreed-upon and conceptually tractable definition of these terms. As terms of art, technical definitions, distinct from their various vernacular usages.

    I’m going to just throw out some of my assertions/conclusions in somewhat random order. Just give these a try in your thinking, see if they work for you. They’ve really helped me sort out all the arguments I read.

    • A dollar bill is not money (in this technical, nonvernacular, conceptually tractable set of definitions). It’s an embodiment of money — exchange value, claims, legal rights. As is a treasury bond or a deed to a house. They’re all financial assets, and they all embody money.

    • All financial assets are embodiments of money, that cannot be consumed (to derive human utility), and require (almost) no real inputs to production. No use value. They only embody exchange value. This is what distinguishes financial assets from real assets.

    • Financial assets (so-called “financial capital”) are not capital. (Another vernacular problem.) Only real assets are capital. And financial assets are not “inputs to production.” Financial assets are (roughly) claims on real capital. This is why “saving” gets so confusing — it tends to conflate the real and financial. (Understandable: if you “own” a factory, that’s part of your savings, right? But if you think of that ownership as a financial asset, a claim or particular set of rights to that factory, which financial asset has a dollar value determined by the market, you’ve distinguished, conceptually, between the asset itself and the claim on that asset.)

    • When 1. more financial assets are issued, or 2. when the market bids up the price of financial assets, there’s more money. Both bank lending and government deficit spending contribute to both of these effects.

    • When the market bids up the price of financial assets (claims on real assets), capital gains result. But in the national accounts, capital gains aren’t considered to be “income.” So it gets confusing, trying to figure out how wealth increases without it flowing from income. See the 3-step process described in this post.

    • M2 etc. are categorizing attempts to resolve an unresolvable conceptual problem — confusing certain types of financial assets (all of which embody money) with the vernacular term “money” that has always meant coins, currency, etc. in the vernacular. If it’s easily exchangeable for real goods (or reliably 1:1 exchangeable for other financial assets that can be exchanged for real goods) — if it’s “liquid” in this sense — a financial asset is considered “money.” Otherwise not.

    • Both banks and govt create money (embodied in financial assets) out of thin air, using different but related financial mechanisms. But when River A and River B flow into an ocean, is it useful or possible to then distinguish between River A water and River B water in the ocean?

    • Most real capital — the huge bulk of which consists of humans’ abilities to produce future output — is not monetized/financialized/”capitalized”. An exception, by example, is the huge runup in student loans — resulting claims on those people’s abilities to produce in the future (which claims are then packaged into securities/financial assets).

    Sorry, kind of random but I hope you find it useful. This setup is the only way I’ve been able to understand all the confused and contentious thinking about money and capital that I’ve been reading with fascination and perplexity for years, from every corner of the ideological and economic spectrum.

  20. February 2nd, 2014 at 09:37 | #20

    Don’t tell Steve Keen, but I’m actually convinced that the endogenous money view and the loanable funds view are equivalent above the zero lower bound, and maybe below it — dual theories, like the many varieties of string theory. It makes no difference if banks lend first and then come up with the necessary reserves — they are still constrained by the prevailing interest rate, because they expect they will have to borrow to get reserves and what rate they can borrow at affects which loans they are willing to make. So in one view the central bank exogenously sets the size of M0 and the interest rate is determined endogenously by the supply and demand for savings, whereas in the other view the central bank exogenously sets the interest rate and then the money supply is determined endogenously by how much banks are willing to lend at that rate, with the CB acting to fix the interest rate by acting as lender of last resort.

    It’s not of any practical consequence whether the CB is changing the market interest rate by fixing the supply of money or whether they are changing the endogenous money supply by fixing the interest rate. If the models are properly understood they give the same results. And the CB doesn’t care about manipulating either the size of the money supply or the interest rate — they are both tools for manipulating inflation and unemployment, which is what they ultimately care about.

    Loanable funds breaks down at the ZLB, but ISLM explains why — because the market interest rate would be negative, which can’t happen, so the relationship between M0 and M1 breaks down. Here the endogenous money view gives a clearer explanation of what the size of M1 will be. But ISLM is a simpler model for understanding how the CB should behave above the ZLB.

    As for why ISLM works at all, keep in mind that it is only necessary for investment to respond to the interest rate for it to work. Also, the concepts of saving and investment used in GDP accounting aren’t the best versions of these concepts for the ISLM model; the relevant quantities are the flow of money into the financial world from those who are net savers, and the flow of money into the real world to those who are net borrowers. It doesn’t much matter whether lowering interest rates causes businesses to take out loans to buy factories or whether it causes consumers to run up higher credit card debts to eat out or take a vacation — the “investment” label isn’t a helpful one, we’re really talking about any new debt.

  21. February 2nd, 2014 at 10:45 | #21

    @Eric L

    Eric, I think that is Krugman’s view. He says the test is whether ISLM and the endogenous view are equivalent in result above the zero bound.

    Changing the interest rate works through the spread that banks charge and recent research shows that the rate is inverse to investment and not directly proportional as ISLM assumes. That is to say, lower rates are associated with lower aggregate demand and higher rates with higher aggregate demand. Demand for money increases as aggregate demand increases and decreases with its decline. Moreover, increasing interest rates are inflationary in that they increase Treasury injections, while decreasing interest rates decrease Treasury injections. QE acts as a tax by transferring the interest on the tsys purchased from the private sector to government.

    • February 2nd, 2014 at 12:21 | #22

      Eric: great comment and thanks.

      Maybe not replying directly to you, but you inspired the following, which I’m dropping in here to see what folks say before posting, in my ongoing effort to avoid looking like an uninformed ignoramus. Thoughts?

      Why IS-LM Works: It’s the Spending, not the Saving

      In my last post I asked a question: the IS-LM model is based on a completely false notion — that more saving increases the stock of savings, or loanable funds — but it nevertheless seems to do a good job of predicting macroeconomic outcomes. Why?

      Eric L in the comments gave a very thoughtful response (read it; his blog is here), which may have dropped the scales from my eyes. As thus:

      IS-LM works because it’s depicting real economic relationships, telling a real economic story — but not the story people think (“desired saving” interacting with “desired investment”).

      What it’s showing, instead: the interaction between desired spending and desired borrowing.

      When people want to spend more of their income (side-effect: save less), they achieve those desires partially by borrowing more. That increased borrowing (desire) drives up interest rates. Pretty simple story. When they want to spend less (side effect: save more), there’s less demand for loans; interest rates drop.

      IS-LM is cast as a supply story — more saving creates more loanable funds. (It doesn’t, so the model’s assumptions seem incoherent.) But it’s really a much more straightforward demand story: more (desired) spending causes more borrowing.

      • February 2nd, 2014 at 12:24 | #23

        Oh: and the spending could be on investment or consumption. That’s a separate issue.

  22. February 2nd, 2014 at 13:26 | #24


    “a completely false notion — that more saving increases the stock of savings, or loanable funds”

    Whether this is completely false depends on whether you treat the central bank behavior as endogenous or exogenous. Note that endogenous money makes the CB partially endogenous by assuming things like as a rule the central bank will expand the money supply as necessary to meet demand at its interest rate. If you further model it as adjusting its target rate according to a Taylor rule to keep demand high enough for full employment, then an increase in desired saving will in fact increase loanable funds, at least above the zero lower bound, because the central bank will expand the money supply to keep demand up.

    “the interaction between desired spending and desired borrowing”

    I don’t think this is quite right. Borrowing is just a subset of spending. The interaction is between borrowing and saving, though we have to be careful here because if we’re making borrowing broader than just investment then we have to choose a definition of saving that matches, which we can, by not subtracting consumer debt from saving.

    Also, while I think that thinking about borrowing instead of investment makes the model more complete and accurate, I don’t think it’s a particularly consequential change — ISLM would work fine if investment were the only kind of spending that was sensitive to interest rates, and it works if you name any kind of spending that’s sensitive to interest rates “investment”.

  23. February 2nd, 2014 at 13:39 | #25

    Tom Hickey:

    “the rate is inverse to investment and not directly proportional as ISLM assumes. That is to say, lower rates are associated with lower aggregate demand and higher rates with higher aggregate demand.”

    Whether that contradicts ISLM depends on the direction of causality. It is expected that the CB responds to low aggregate demand by lowering rates. However ISLM models what we would expect in a given economic situation if the CB chose its rate arbitrarily, ignoring its normal mandate to control aggregate demand.

    “Moreover, increasing interest rates are inflationary in that they increase Treasury injections, while decreasing interest rates decrease Treasury injections.”

    Why would that be inflationary? There are more treasuries and fewer dollars, so more people are putting off consumption until later.

    “QE acts as a tax by transferring the interest on the tsys purchased from the private sector to government.”

    A voluntary tax?

  24. February 2nd, 2014 at 14:30 | #26

    @Eric L

    “Whether that contradicts ISLM depends on the direction of causality.”

    Causality is the foundation of good explanation rather than correlation. That is, some transmission mechanism must account for the change and it can’t be a “fairy” (Krugman) either. Post Keynesians claim that ISLM has the causality backwards, just as the money multiplier has the causality backwards, and the fact that it either correlates is due to ex post residuals rather than ex ante causal factors. This is basic kerfuffle with Krugman. Post Keynesians deny Friedman’s “as-if” basis of explanation based on unrealistic constructs as long as the correlation works. But the correlation doesn’t work with ISLM at the zero bound. According to endogenous money view in contrast to monetarism is that changes in nominal aggregate demand are the fundamental causal factor rather than interest rates affecting the demand for borrowing versus saving. Investment responds to changes in aggregate demand rather than changes in the interest rate. The transmission mechanism is demand signals rather than price of money signals in the PKE endogenous money view.

    “Why would that be inflationary? There are more treasuries and fewer dollars, so more people are putting off consumption until later.”

    Tsys are only issued when the government deficit spends as a politically required offset in the US. When the government deficit spends, net non-government financial assets in aggregate increase, since tsys and reserves are both NFA. When the Fed credits bank accounts at the direction of Treasury the rb used to purchase tsys remain the same and the amount of NFA increases in the amount of the tsys issued into non-government. ST tsys are effectively cash, and even LT tsys are highly liquid and don’t reduce the ability to spend, so don’t decrease nominal aggregate demand as would savings made inaccessible to the market. The tsys head by non-government also increase NFA though the interest payments, which potentially increase nominal AD. The role is tsys is not to control price inflation but appreciation of risky assets by providing safe assets for the desired level of aggregate saving. When the desired level of aggregate saving decreases, then the cb conducts OMO it is not paying IOR to change the relative composition of NFA between tsys and rb in order to provide liquidity for increased credit demand while maintaining its target.

    “A voluntary tax?”

    Voluntary in that the interest is forgone voluntarily by the sellers of tsys. This results in lower NFA in aggregate than would have been available otherwise and potentially reduces nominal AD in the economy.

  25. February 2nd, 2014 at 18:09 | #27

    Eric, you might want to look at Steve Keen’s latest paper, Modeling Financial Instability, which he contrasts loanable funds and PKE endogenous money.

    In this paper I attempt to conclusively determine whether aggregate private debt and banks matter in macroeconomics by putting the two rival models of lending—Loanable Funds and Endogenous Money—on a common footing. Using the dynamic Open Source monetary modeling program Minsky, I firstly put the New Keynesian model of banking in Eggertsson & Krugman 2012b into a strictly monetary model and I show that, if the structure of lending in this model accurately characterizes actual lending, then the Neoclassical perspective that aggregate debt is unimportant, and that banks can safely be ignored in macroeconomics, is correct. I then modify this model to match the Post Keynesian perspective on the structure of lending, and show that in this structure, changes in the aggregate level of private debt have a direct impact upon aggregate demand, and banks therefore play a crucial role in macroeconomics.

  26. TRF
    February 3rd, 2014 at 03:45 | #28


    @Asymptosis – thanks for those pointers. It has really helped with my understanding. Just one thing I would like to check:

    Would it be correct to say that when a bank makes a loan it’s effectively creating money by converting an illiquid financial asset (the IOU from borrower) into a liquid financial asset (deposit account = money)?

    And the Bank’s ALM/funding department is then responsible for ‘converting’ liquid money (deposits) into illiquid financial liabilities (savings accounts, CDs, issuance of bank debt, covered bonds, etc)?

    In other words, a banks lending creates liquid ‘money’ in the banking system, but a banks funding activities ‘removes’ money from the system? i.e. when a customer with $100k cash in a deposit account buys a bond issued by the bank, $100k of ‘money’ is effectively removed from the banking system (and replaced with $100k of bonds).

  27. February 3rd, 2014 at 09:25 | #29

    Those who extend credit lend against collateral. The lender makes the collateral liquid by lending against its recoverable value after a haircut as a margin of safety. Even unsecured loans are lending against collateral since the lender can recover in case of default through either bankruptcy or a judgment against the borrower’s assets. Interest rates are much higher for unsecured credit than secured in order to cover losses.

  28. February 4th, 2014 at 00:49 | #30

    @Tom Hickey

    Steve Keen’s models aren’t particularly relevant here, as he was examining the impact of private debt in models that do not have a central bank, whereas we are discussing the effect of the central bank.

    That said, having seen previous presentations of his models I also don’t think they make the point he thinks they make. There are two problems I have with his models. Basically there are ways his models differ from each other in which they shouldn’t have to differ. One is his patient agents behave like his impatient agents in his “Krugman” model without banks, whereas in his model with them banks are also consumers, but they are more patient — they spend their wealth more slowly. The second problem is that his banks consider their outstanding loans as part of their wealth, but his patient agents consider wealth they have leant to be gone until it is paid back, and so lending suppresses their consumption. These differences aren’t inherent to the models we’re comparing. The point demonstrated by these models as I see it are that people don’t consume less because they lend; they lend because they have less pressing consumption needs. Also, people don’t consider themselves to be less wealthy when they trade cash for a bond or other financial asset. These are valuable insights but they have nothing to do with the difference between loanable funds and endogenous money; these ideas can be applied in either framework.

  29. February 4th, 2014 at 09:18 | #31

    @Eric L

    What I think Steve’s model suggests is the direction of causality is demand-driven with banks meeting demand for loans regardless of deposits, as the endogenous story goes, rather than interest rate driven, as the monetarist narrative runs. Conceivably, they could correlate with each other at least most of the time, but only by being mirror images in the sense that the ex ante factor is demand and the ex post factor is an accounting residual.

    The money multiplier works under certain circumstances that are also the normal circumstances, i.e., when the Fed is setting price (the interest rate) and controlling quantitate of rb with OMO to keep the rate on target. But the effect is not causal, the textbook account goes, but rather is merely an accounting residual that reflects the reserve requirement, making it seem that banks are dependent on the RR or are stimulated to loan up to the RR, when the the causality actually runs the other way, with banks loaning and the Fed as LLR standing by to ensure that the RR is met at the penalty rate if need be. Now we see with the Fed paying IOR the money multiplier is meaningless and all the people betting against tsys expecting inflation because of it were caught short.

    I think that the issue is confusing correlation with causation, and at certain points, the correlation breaks down. Since this happens at the margin, those betting on causation that get it the wrong way lose at key turning points. More importantly from the theoretical POV, explanation is supposed to be causal, based on actual exa ante transmission mechanism rather than ex post residual effects.

    BTW, I am not claiming that SK’s model is correct in all respects. Nick Edmonds and Ramanan have criticized it, for example.

  30. Detroit Dan
    February 15th, 2014 at 11:22 | #32

    I’ve concluded that IS/LM, like Market Monetarism, is a subject that can get a discussion off track. I wrote a post of my own on this subject: Economic Discourse: Focus on the Problem, Not the Model.

    Otherwise, I like your post on the macroeconomic meaning of saving. Without getting into the IS/LM nonsense, I think the concept of saving, at the national level, can be expressed simply as follows:

    Saving = Net Real Investment + Fiscal Deficit + Trade Surplus

    Saving as not-spending can be expressed simply as:

    Saving = Income – Consumption – Taxes

    but one person’s consumption is another’s income, so reducing consumption does not increase saving.

  31. February 15th, 2014 at 11:55 | #33

    @Detroit Dan

    “one person’s consumption is another’s income, so reducing consumption does not increase saving.”

    Right, when consumption desire falls relative to saving desire, then income from consumption falls and saving/investment equivalence increases. This “saving” shows up on firms books as unplanned (unwanted) inventory, which sends a signal to firms to lower prices decrease or quantity of production. Firms using administered (fixed) rather than flexible pricing generally choose to adjust quantity rather than price, so this means cutting back production by laying off workers, which further reduces income, which further reduces consumption and the economy contracts. The majority of firms dealing in final goods in contemporary economies used administered (fixed) pricing.

Comments are closed.