Or: It’s The Velocity, Stupid.
I got quite a bit of blowback on my recent post suggesting that economists don’t understand accounting.
In response I give you Exhibit A: the almost-ubiquitous notion that more saving increases the supply of “loanable funds” — hence that more saving causes or at least allows more investment. (The absolute classic fallacy of the S=I accounting identity.)
On casual consideration, it seems like it would be right, right? You spend less than your income, so you have more money (stuffed in your mattress?), and you can lend it out.
Or more likely: you “put money in the bank” – deposit more than you withdraw — so the bank has more money; it can lend more.
It’s A Wonderful Life.
Here’s Mankiw in his textbook, saying exactly that:
Saving is the supply of loanable funds — households lend their saving to investors or deposit their saving in a bank that then loans the funds out.
1. A little careful consideration shows that this casual consideration is logically incoherent — just plain wrong, by accounting identity.
2. Economists are not supposed to be thinking, giving their sage advice, or corrupting our youth based on casual consideration.
Think about it:
Meantime, you don’t spend $25,000. You “save” it. The money sits there in your checking account. If the action of spending — transferring money from one account to another — doesn’t change the total stock, how could not transferring money do so? Your bank still has the money, which it can lend out. Other banks still don’t, and can’t.
It may help to think about this as if there was only one bank. (Which is not so far off. Bank deposits all consolidate back to accounts at the Fed.) Every person and business has an account. All the spending/transfers (or non-transfers, a.k.a. “saving”) just shift deposits between accounts, with no change in the (single) bank’s total deposits.
So the saving/spending mix has no effect on the stock of loanable funds. Shifting (or not shifting) those stocks around has no aggregate effect on the total stock.
But what about the flow — new loans from banks? Again: no.
Here’s a behavioral, rather than accounting-based assertion — not a controversial one, I think: In any period, banks in aggregate lend more — “print” more new money and deposit it in people’s/businesses’ accounts — because they think they can make money doing it at current interest rates. They think that for one primary reason: they are confidently optimistic about future prosperity — borrowers’ future income streams. If they’re less confidently optimistic they lend less, or ask for higher interest rates — which has the same effect: less lending.
Likewise borrowers: they borrow because they think future conditions will be good, and they’ll be able to service their loans at the asking rate out of strong income streams (and/including rising financial asset values).
Likewise spenders: they spend (that new) money because they think it will yield good returns from investment, and/or because they think they can consume today and be able to earn more money to pay for it (repay the loans) in the future.
So how does the saving/spending mix affect those expectations? Another behavioral assertion: Those expectations are set, to a great degree, by current conditions, because they’re the best predictor we’ve got. It’s difficult at best to predict future “shocks” that will change those conditions. Or as the Eight Ball says: “The future is … unclear.” Life is uncertain.
So how does a higher proportion of saving to spending affect current conditions?
It makes them worse. GDP is spending. Less spending (as a proportion of either income or wealth) means less economic activity. Less velocity. Less transactions. Less surplus from trade. Lower GDP. People, businesses, and banks, borrowers and lenders, are less prosperous, and less optimistic. So banks lend less, borrowers borrow less, and (in a potential downward spiral) spenders spend less.
Takeaway: An increased saving/spending proportion has no effect on the stock of loanable funds (it can’t), and it has only a second-order, expectation-driven behavioral effect on flows — it decreases them.
You really have to wonder sometimes where economists get this stuff that they put in their textbooks.
Nick Rowe attempted to save this conceptual situation recently in a comment posted hereabouts (emphasis mine):
Suppose there’s an increased demand for financial assets by households (a rightward shift in the demand curve). Will that increased demand lead to an increased quantity of investment by firms and an increased quantity of financial assets sold to households (a movement along a supply curve)? It may do. That depends on the model. It’s a behavioural question, not an accounting question.
His questions in the middle, and the last statement, are completely on the money. But his explanation begins right in the midst of the conceptual confusion, putting the modeling cart before the behavioral horse. The behavior doesn’t “depend on the model”; the model’s accuracy and usefulness depends on its assumed human response to incentives and constraints.
Or perhaps, rather, he’s climbed aboard the wrong behavioral horse — one that is wandering off rather aimlessly.
The “desire to save” is a conceptual representation, a mini-model, if you will, of one aspect of the economic situation. I’m suggesting that that construct is outside of, peripheral and irrelevant to, the behavioral chain of cause and effect.
People might want to save more/spend less in aggregate for various reasons:
• Times are tough — GDP and employment are weak — and they’re worried about future ability to consumption.
• Times are good, and they’re satisfying all their consumption desires.
• Rich people have a larger proportion of income and wealth, and their lower marginal propensity to consume drags down aggregate spending, relative to income and wealth.
Or some other scenario. (As Keynes said — not looking up the exact quote here – all economic activity is driven by the desire to consume.)
In the second scenario banks will want to lend more — but not because people and businesses (want to) save more. If that were true, banks would also want to lend more in the first scenario — which is completely contrary to what actually happens. (The results in the third scenario seem uncertain.)
Here’s a syllogism to make this widespread confusion clearer:
Mankiw’s conceptual confusion is inevitable, and arises from two causes:
1. He’s starting with snaky (and conceptually confused) assumptions about the sources of human behavior, and:
2. New but related subject: He’s trying to think about flows (and get tender young minds to think about flows) using static, of-an-instant models like the standard S/D and IS-LM diagrams. The problem, when you’re trying to think about “supply,” is that a flow can’t exist in an instant (only stocks can); it’s a meaningless, impossible concept. And since stocks in our discussion here are unaffected by saving, he’s in a pickle, cause it’s all about flows. (And no: “comparative-static” methods don’t solve the conceptual confusion; they arguably only contribute to it because they impart the illusion of time and dynamism.)
The only way (that I know of) to model “flow supply” in a conceptually coherent way – or even think or talk about it really, which is mental modeling — is using a dynamic simulation model. Of late I’ve been quite taken with the power grid as a good metaphor for a dynamic model of the economy — one that I’ll expand and expound upon in a future post.
For now I’ll leave you with this: Clower/Burshaw on the difference between “stock supply” and “flow supply,” or peruse the literature here. Nick also talks quite a bit about the largely forgotten old 70s notion of “nominal” (roughly: “potential”) supply and demand — though mainly regarding money, not real goods.
I almost never see any consideration of these seemingly crucial concepts in economic discussions — much less cogent analysis, or incorporation of said concepts.
Which leads me to ask a question of economists:
* I won’t even touch here on the widespread misconception among economists regarding bank lending, except to say that in practice bank lending is not constrained by deposits — banks lend most of their deposits then lend (much) more based on their excess capital (times X) — which, thus, is their effective constraint on lending. Not deposits.
Cross-posted at Asymptosis.