Recent Posts

Wednesday, February 3, 2016

Money Hoarding Versus Saving, And Economic Growth

One of the ongoing arguments in political economy that has followed on from Keynes is the debate over the "Paradox of Thrift." This paradox could be loosely summarised as: an attempt to increasing savings by households will lower incomes, and will actually reduce their financial resources. Free market supporters tend to reject this logic, instead arguing that increased savings increases investment, boosting growth (as discussed here). Professor Nick Rowe argues in a recent paper (paper link; comments link) that the problem is not "thrift," rather the desire to "hoard" monetary assets. This debate is not just of theoretical interest; we need to understand the effects of increasing savings in order to gauge what the effect of increasing pension contributions would have on the economy (for example).

For a further background on this debate, please see the article by Ramanan: "Nick ROKE, Thrift, and Hoarding." He discusses the concepts further, and includes a comment by JKH which indicates that Keynes was aware of the distinction between hoarding and saving.

I have left some comments on the Worthwhile Canadian Initiative site, to which Professor Rowe responded. At the time of writing (which is actually a couple of days before publication), I was still unclear about the distinction between "thrift" and "hoarding" that he draws. Therefore, I will not address his argument here, although I have a general discussion of the demand for money within classical (and some post-Keynesian) models.

I am not attempting to explain the mechanics of the Paradox of Thrift here, nor attempt to justify that it is an effect that we see in the real world. I dislike the traditional way of explaining the concept, rather I would prefer to see it appear within a simulation of an economic model. The text Monetary Economics by Godley and Lavoie demonstrate this effect on models of varying complexity (such as in Section 10.7.6).

Why This Matters

I am in the process of following up an earlier article on universal pension systems, and one of the potential solutions for inadequate pension provision is to increase the universal state pension (in Canada, the Canada Pension Plan). We could imagine a rather large augmentation of the plan which results in household savings immediately increasing by 3% of GDP,* which would be invested in non-monetary assets.
  • A traditional Keynesian response (that I would agree with) is that this forced increase in financial saving would hit final demand. There is no reason to believe that fixed investment would rise by 3% of GDP to balance this  and would resemble a tax hike of 3% of GDP. This would probably be sufficient to drive the economy into a deep recession (assuming that it is not in one already)
  • A doctrinaire believer in Say's Law might argue will be increased investment to match the increased saving, and there would be no immediate negative effect on growth (and future growth will be higher due to the increased investment). I am unsure who in academia (if anyone) would take this line, but it would be easy to find market economists who would agree with this assessment.
From the point of view of a policymaker, we need to know which side is correct -- if we did expand the CPP programme, would it have to be accompanied by a loosening of fiscal policy to offset the demand destruction? (Of course, if we are uncertain about the effects, we could slowly ramp up contributions, and attempt to offset weakness as it becomes apparent.)

The incoherence of the two responses is not exactly a selling point for the state of modern economics. The Paradox of Thrift (or lack thereof) should be an empirical statement for which we have something resembling a definitive answer.

Demand For Money Versus Demand For Financial Assets

The difficulty I initially had with Nick Rowe's description of the distinction between "hoarding" and "thrift" is that I cannot see how "hoarding money" matters. I follow a fairly straightforward methodology for thinking about this topic, which is compatible with financial theory and practice, mainstream "classical" economics, and at least some post-Keynesian approaches.

Money is just a financial asset, and the decision-making process for determining holdings can be thought of as follows. (A equilibrium model is more complicated in that all decisions are simultaneous, rather than sequential steps, but in practice we can largely view the outcome as being sequential steps.)
  1. A household decides upon its aggregate saving rate ("propensity to save") based on the expected return on assets, and other factors. (Retirement income needs, utility optimisation, heuristics in Keynesian models.)
  2. The portfolio allocation is set.  In finance, this might depend upon the trade-off between perceived risk and the risk premia on offer. In a simplified economic model where the only financial asset are Treasury bills, this will depend upon the Treasury bill rate (which is the policy rate of the model).
In an optimising framework, it makes no sense to hold money that pays no interest when interest rates are positive.** Therefore, various modifications to models were introduced to induce non-zero money holdings. Broadly speaking, the desired holding of money would be zero, it was just forced upward via those modifications. However, the switch towards money holdings (using parameters that are roughly reflective of the real world) are not large enough to greatly change the expected rate of return on the aggregate portfolio (including money). Since the aggregate expected return will not move much, reasonable changes in money demand will have limited effect on the optimal solution.***

Was The Financial Crisis The Result Of Increased Money Hoarding?

In the abstract to his article, Nick Rowe wrote:
I argue that Keynes missed seeing the importance of the distinction between saving in the form of money (‘hoarding’) and saving in all other forms (‘thrift’). It is excessive hoarding, not excessive thrift, that causes recessions [emphasis mine - BR] and the failure of Say’s law. 
In my view, the demand for money has almost nothing to do with recessions. However, the difference in view largely reflects modelling preferences. Professor Rowe prefers to use models where the demand for money is a critical driving force, whereas I prefer models where money is just another financial asset, and demand for money is largely ornamental. (In fact, my preferred models absorb money into the supply of Treasury bills.) His statement reflects his model assumptions; my explanation of recessions would reflect my preferred assumptions.

However, I think that explanations such as an increased demand for money, or "a shortage of safe assets" is not a good verbal description of the impetus behind the Financial Crisis. (The Financial Crisis is the best candidate to explain recessions as a result of some monetary imbalance; other recessions occurred without disordered money markets.)
  • For households, I would argue that the cutting off of credit was far more important than any desire to raise cash balances. As for the financial crisis, it was centred in the international financial sector, and what was happening in the real economy stopped being important once the unravelling occurred. The deterioration of confidence within the financial sector was much more rapid than slow-moving real economy defaults.
  • Investors applying leverage were forced to exit positions at very unsatisfactory price levels. Technically, de-leveraging is a form of forced savings, but it cannot be interpreted as a voluntary desire to increase transaction balances.
  • Non-levered investors discovered that their portfolios were full of securities that were actually toxic garbage. They were essentially forced to clean up those portfolios.
  • Money market investors discovered that private short-term securities were actually non-money good long-dated corporate credits. They were forced via investment mandates to exit those positions and replace them with "safe assets." Although this sounds like a "shortage of safe assets," this shortage did not exist until they discovered that the private securities were not actually "money."
  • However, an "increased demand for money" does match up with the behaviour of issuers of short-term debt. There was a desire to "get liquid" in order to shore up the perceptions about their solvency.
In summary, the "demand for money/safe assets" is just a description of the symptoms; it is like diagnosing the "cause" of a recession as being a fall in employment. What set up the Financial Crisis was the excessive use of leverage. Reading Hyman Minsky offers much more insight into that process than pondering the demand for a medium of exchange.


* It would be easy to calculate how a change in CPP policy would affect the inflows into the plan. However, it is unclear how much this would affect the aggregate savings of the household sector, as some people may decide to lower their other savings by some amount.

** The DSGE framework does predict that nobody would hold bonds with a negative yield, as people would allocate towards money. This is one of the few falsifiable predictions that can be made by the DSGE framework, and it was of course falsified.

*** For simplicity, assume that the demand for money function is such that an increased holding of money is offset (exactly) by higher expected return on non-monetary assets (that is, the rate of interest goes up) so that the aggregate rate of return on all financial assets is unchanged. If we have an optimal solution S, we could construct S* such that S* has the same aggregate saving and consumption path, and only the money/Treasury bill holdings are different. It is straightforward to demonstrate that S* would be an optimal solution; a formal proof would depend upon the structure of the model.

(c) Brian Romanchuk 2015


  1. "A traditional Keynesian response (that I would agree with) is that this forced increase in financial saving would hit final demand."

    How is it a forced increase in financial savings if you are just investing in non-monetary assets? Surely that's just a transfer from workers to whoever gets the money at the end of the portfolio shuffling process - generally those with pensions in payment.

    If you increase savings to a pension fund, then you have more money chasing the same set of assets and the price of the assets goes up to compensate until the money leaks out either to monetary savings, debt repayments or spending and taxation.

    1. I'll assume we are talking about a purchase of equities by the CPP (pension fund), and the seller was a household (otherwise we have to start worrying about the balance sheet of financial intermediaries, which are ultimately owned by households).

      The effect on final demand depends upon what the seller does. (This came up in the discussion I had in the comments section on Worthwhile Canadian Initiative.)

      If the seller of the asset was a household which goes out and buys groceries with the money, the propensity to consume out of income for the household sector in aggregate would be unchanged. (That is, total consumer spending is unchanged, while incomes are largely unchanged, other than second order capital gain effects.)

      But if the seller takes the money, buys a T-bill, and the central bank neutralises the operation (sells a T-Bill and absorbs money), the propensity to consume out of income for the household sector in aggregate would fall, which implies an increase in inventories, which would trigger a scaling back in business sector activity, etc. (The neutralisation by the central bank means that the net effect on the aggregate household balance sheet is an increase in T-Bill holdings, plus a potential change in value of equities.)

      The precise mechanism depends upon the model, but if you looked at the simulation of a SFC model, there would be a drop in activity (in the short term).

      I believe that the most likely outcome will be something that is closer to the "saving" case than the "spending on groceries" case.

      If we had a solid grasp of empirical relationships in economics, we should be able to estimate the sensitivity of other variables to such a change.

  2. Joan Robinson was sceptical about about the idea hoarding, as she made clear in “The Concept of Hoarding” (Economic Journal, June 1938). In that piece, she notes that, for Keynes, the notions of “savings lying idle in the banks, and of banks’ loans as a supplement to current saving are purely mythical conceptions.”

    1. Thanks. I am not hugely familiar with Joan Robinson's work; I will have to track down some texts some time.

  3. Arghh! I just finished writing a loooong comment but pushed "sign-out" instead of "publish". Kids are hungry, I guess, and they can't take it when I'm on the laptop. Anyway, I'll try and summarize what I wrote later.

    Nice post, though!

  4. The paradox of thrift as discussed in the video under this link states that collective effort to save more in a recession paradoxically causes a deeper recession:

    Actually the balance sheet problems are more complex because the banks and other financial intermediaries are experiencing reduction or contraction of their balance sheets at the same time that their customers are taking up less debt and seeking to save more. In order to create more financial savings the financial intermediaries must expand balance sheets by making loans on the asset side and issuing more liabilities. But the collective effort to net save makes financial intermediaries sell financial assets in a "fire sale" and cancel liabilities. So the collective effort to save more destroys financial savings in a pure market economy with financial intermediaries. This is called a "panic."

  5. Begin with any NEW loan.

    A loan can come from two places:

    1. Any entity unwilling to spend (but willing to lend).

    2. A bank (including central banks).

    We can safely assume that a NEW loan will initiate spending that would not otherwise occur IN THE NEAR TERM.

    Where does NEW MONEY come from? Only central banks can lend without first saving. Everyone else must work to earn money which can subsequently be placed on loan.

    From that descriptive base, I feel safe in discussing "hoarding". If a person (or a large number of entities) decided to spend less money, we could have a condition described as a "hoarding of money". I don't like the term "hoarding", thinking it is a politically charged description for the condition of "a personal or collective decision to spend less money".

    Now we can think about recessions. Can a recession be caused by a collective decision to spend less? Of course it can. And a boom can be caused by a collective decision to spend more.

    Can a recession be caused by a governmental decision to BORROW LESS future money? Well, we first must know if the PAST BORROWED MONEY was being used for consumer spending (including infrastructure enhancement) or monetary asset accumulation. If the past borrowed money was used for monetary asset accumulation, a decision to borrow less would have little effect on consumer spending (but could have a large effect on pension accumulation).

    I think I am mostly agreeing with you but I am emphasizing a closely prescribed assumption base.

    Thanks for yet another thoughtful post.

    1. A new loan implies that someone is willing to increase their leverage; they emit a liability to the bank, and get a bank deposit (money) in return.This is an increase in balance sheets; and then the borrower can spend the money, generally increasing someone's income. This will show up as a form of dis-saving.

      Saving by the household sector interrupts the circular cash flow within the private sector. A business pays wages, and if they are spent, they return as sales. If they are not returned to business as spending, business sector income falls.

      The context of the decision is quite different, and shows up differently in sectoral cash flows.

  6. This comment has been removed by a blog administrator.

  7. This comment has been removed by a blog administrator.

  8. This comment has been removed by a blog administrator.

  9. This comment has been removed by a blog administrator.


Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.