The Irrelevance of Deposit Creation for Prices and Allocation: Comments on Selgin

by Arash Molavi Vasséi

In a previous post, Andreas refers to George Selgin’s recent discussion of the place of fractional reserve banking in the Austrian Business Cycle Theory (ABCT). There, Selgin takes a swipe at the monetary pillar of the ABCT. According to the Austrian model, fractional reserve banking is inclined to create money out of “thin air” and, therewith, admits investment spending in excess of “voluntary saving”. This imbalance, allegedly induced by a decline in reserve ratios, is reflected in a Wicksellian interest rate gap, which is supposed to impact prices and the allocation in a systematic way (the real pillar of the ABCT). Selgin argues that fractional reserve banking does not account for the Austrian business cycle, and Andreas expresses sympathy for this view.

Basically, Selgin’s argument relies on the money multiplier and Laurence Laughlin’s Law of Reflux (when he invokes “deposit destruction”). I agree with the general direction of the argument, but go further and approach the problem from a completely different angle: under the assumptions of the ABCT, which includes competitive and frictionless banking, the creation of inside money cannot account for the business cycle, Austrian or otherwise. I go even further: given competitive and frictionless banking, inside money does not affect relative prices and allocations; it does not induce a real-balance or hot-potato effect, even if it circulates as a means of payment; control over inside money does not solve the indeterminacy problem, despite its ‘moneyness’.

If inside money has macroeconomic implications, this is because of non-monetary frictions or banking regulation. Of course, this is the view expressed by James Tobin, Fischer Black, and Eugene Fama. I know that Andreas and Mr. Selgin do not share this view. I would like to know why. Subsequently, I list some arguments to be picked apart. Please, prove me wrong!

Banks, like all intermediaries, transform liabilities that debtors are able to issue into liabilities creditors are willing to hold. Whenever an ultimate debtor is able to issue new liabilities, this is because somewhere on this planet some ultimate creditors are willing to increase their net financial positions accordingly. These ultimate creditors are not necessarily the ones that accept the additional risk associated with the new liability. By risk stripping and adjustments in gross financial positions, this risk can be channeled to third parties.

Deposits, the characteristic liabilities of the commercial banking sector, are the result of such an asset transformation. Risky, non-tradeable credit portfolios satisfying low-reputation borrowers are transformed into interest-bearing senior liabilities that find their way into the portfolios of households, firms and governments as (1) safe stores of value, and (2) are used to transfer wealth. More precisely, bank accounts grant access to a reputation-based central ledger technology (CLT), which banks cooperate to provide, and which is required to run through a public ledger operated by the central bank. Economic agents can use their balances to transfer wealth (cum current income), which is recorded as a debit, in exchange for commodities or securities. A transfer received is accordingly recorded as a credit. In this sense, deposits are ‘liquid wealth’, which for many translates into ‘moneyness’.

This fact accounts for a widespread misunderstanding. Specifically, there exists a long-lived belief that deposits and cash should trade as perfect substitutes (e.g., Mises’s “money substitutes”‘), probably due to the convertibility services associated with the banking sector (i.e., the promise to redeem on demand at par value). This belief implies that, in equilibrium, deposits do not return-dominate cash, which links them to the hot-potato effect. The correct belief, I think, is that deposits should be rather treated as close substitutes for short-term senior debt like T-bills, irrespective of their ‘moneyness’ or convertibility. Let me explain why:

  1. Given unregulated banking, deposits are free of risk if all (diversifiable) credit risk is allocated to shareholders and to those who invest in a bank’s ‘non-core’ liabilities. The ability of banks to create riskless liabilities is, therefore, nothing special but relies on Modigliani-Miller mechanisms that, in principle, apply to all balance sheets (or partially fail to apply in case of differential taxation, asymmetric information, etc.). Risk stripping is basically how to do it, at least under conditions that admit ‘linear pricing’. In principle, all financial intermediaries could provide some form of access to the CLT, in which case deposits would be claims to just some other types of managed portfolios.
  2. Deposits issued by different unregulated banks are made homogeneous by a sufficiently large equity buffer, i.e., by one that shields deposits from capital gains and losses ‘almost surely’ (with probability one), say 30, 40, or 50%. The $-denominated value of individual accounts is only affected by executed transactions and the accumulation of interest at the risk-free rate, adjusted for a competitive fee to cover the resource costs of the CLT. In principle, however, the CLT could operate on risky deposits, that is, on standing facilities that do admit capital gains and losses. In this case, deposits are heterogeneous with respect to their risk-return profiles. We would witness differential risk premia, which is what preserves the ‘moneyness’ of deposits since differential risk premia establish indifference at the margins of choice.
  3. In the free banking case, the deposit market is regulated by competitive interest rates. There is no dispute over the fact that deposits differ from other short-term senior debt in that they also provide access to the CLT. All else being equal, this ‘specialty’ of bank debt constitutes a comparative advantage and accounts for some extra demand. Yet moneyness, though necessary, is not sufficient to account for a convenience yield. In addition, aggregate deposits must be in suboptimal low supply. Otherwise, the use of the CLT via deposits involves zero opportunity costs. The substitutionality of deposits and currency decreases in relative deposit supply, and the deposit yield converges to the risk-free rate (adjusted for fees).
  4. Is there any reason to believe that competitive and frictionless banking fails to produce an efficient supply of deposits? No, because a convenience yield would constitute an arbitrage opportunity, at least under the standard assumption of zero marginal resource costs of deposit production. Free entry and optimization imply no-arbitrage, so deposits become arbitrary close substitutes to T-bills and the like. The supply of deposits is not determined, but it is bounded from above by the total demand for safe assets. By increasing the overall supply of deposits, the banking sector crowds out the demand for other types of “information-insensitive” assets without any implications for equilibrium prices and allocations. Since the deposits market is equilibrated by interest rates, there is no hot-potato effect.
  5. To survive the market process under uncertainty, all economic agents build up a liquidity buffer to synchronize the inflow and outflow of funds at the cost of decreasing average portfolio returns. The optimal liquidity buffer of unregulated banks and other firms that issue redeemable liabilities is just larger than the optimal buffer of firms that don’t. Higher prudence is a difference of degree, not of kind. Convertibility services grant the possibility to exit the reputation-based CLT in favor of “memoryless” and, thus, anonymous P2P transactions (e.g., due to libertarian or criminal preferences for anonymity; a collapse of bank reputation; etc). This makes a commercial bank’s senior liabilities run-prone, accounting for multiple equilibria. In contrast to Selgin, I am happy to live in a world where central banks select for the good equilibrium at zero costs (as lenders of last resort). In the good equilibrium, there is no difference in kind between the liquidity management of banks and that of other financial intermediaries.


5 thoughts on “The Irrelevance of Deposit Creation for Prices and Allocation: Comments on Selgin

  1. Dear Arash,

    I don’t quite share the assumptions you are making.

    In ABCT (with the central bank), banks extend loans beyond the natural level because the central bank is accommodating. Banks lack the knowledge to judge the state of the economy and trust in price signals skewed by the central bank. They provide loans to businesses that will fail as interest rates rise. Banks lack information and are unable to sort out good from bad projects at a time due to false signals.

  2. Useful analogy: A checking account dollar is a call option on a green paper dollar, with a strike of zero and no expiration date. Just as the issuance of call options does not affect the price of the base security, the issuance of checking account dollars does not affect the price of the green paper dollar.

  3. I haven’t time, for now, to consider and respond to the arguments here at length. I do however wish to correct the claim that “Basically, Selgin’s argument relies on the money multiplier and Laurence Laughlin’s Law of Reflux (when he invokes “deposit destruction”).”

    Well, although my argument presumes a “money multiplier,” that multiplier is nothing more than an implication of fractional reserves, where the broad money stock necessarily tends to be some multiple of the monetary base. I never claim any simple mechanical relationship. On the contrary: I am concerned about the possibility of endogenous changes in the multiplier’s value. Someone might of course tell stories about money supply without mentioning the fact that M = mB where m>1. But they could hardly pretend to have avoided “relying” on the multiplier’s presence.

    Finally, while it has never been constant, the multiplier is no mere will-o-the-wisp: it reflects banks’ choices, conditional on regulation, of desired reserve ratios, as well as the public’s relative demand for currency. Throughout the postwar period the U.S. multiplier varied within fairly narrow limits, even allowing for changes in reserve requirements etc., until October 2008, after which it collapsed, with bank reserve ratios increasing almost passively in response to growth in B. But far from throwing a wrench in the multiplier concept, that development merely illustrates the powerful yet fully predictable consequence of the Fed’s new policy of paying relatively generous interest on bank reserves. Pundits who missed this went around declaring that the multiplier was a fallacy etc., when in fact that had an omitted variable in their multiplier models.

    As for Laurence Laughlin and the “law of the reflux,” the last of which is actually more appropriately associated with John Fullarton, my thinking rests on neither. Laughlin and Fullarton both believed in the real bills doctrine; and the “law of the reflux” with which that doctrine is often associated explains how excessive credit creation is checked by … bank borrowers repaying their loans! I’ve long been a staunch critic of all of these ideas. (See for example my JITE article on”The Analytical Framework of the Real Bills Doctrine.”) My own views on the regulation of credit in a competitive banking system are fully set out in The Theory of Free Banking.

    None of this pretends to address your other points. I hope I can find time later to respond to at least some of them.

  4. @ Andreas

    Yes, my argument is based on strong assumptions: bank clients have sufficiently many outside options and there are no frictions. Yet, these are not my assumptions, but the assumptions underlying the ABCT, including Wicksell’s “pure credit system” on which the ABCT is built. If you adhere to the ABCT, these are your assumptions. 😉

    Under these assumptions, so my argument, deposit creation has no relevance for prices and allocation (no Cantillon or Ricardo effects; not even the monetarist/old-Keynesian real-balance effect). So I reject the ABCT for its lack of internal consistency (the real pillar is problematic, too).

    In your comment, you restate basic elements of the ABCT. I totally agree with your exposition of the Austrian view, although I don’t share it. Again, what I argue is that these elements do not follow from the (extremely weak) restrictions that Austrians usually impose on the banking sector.

    Outside this model, my point of view is this: to the extent we saw “excessive credit” leading to the GFC, this is more because of (1) overfitted internal risk models (explaining the miserable out-of-sample forecasts), and (2) banking regulation that allowed banks to economize on equity by using such overfitted models. Seeing “the” interest rate, even at it’s “natural” level, is not sufficient to rule out excessive credit and the like.

    In my mental model, economic agents are subjective expected utility maximizers, which implies that they are Bayesian learners. Even in a world where “money” and “natural” rates never diverge, learning the truth is non-trivial: Bayesian learners, including bank managers, consistently update their priors but might eventually fail to learn the properties of the price and return processes they care about. All this has nothing to do with central banks, which are too often blamed (or hailed) for outcomes they do not control.

  5. @ George Selgin

    The compression of a complex argument into a single sentence invites misunderstandings. I had an entire paragraph describing your argument, but unfortunately replaced it by that admittedly misleading sentence. I caused you the inconvenience of writing a clarification, and I am sorry for that. No, you do not assume behavioral constants. Yes, you do have proper “microfoundations” to explain variations in reserve ratios.

    What I should have said is that your argument is based on the long-lived belief that M=mB is an equation (or identity) of strategic importance in the analysis of banking and the banking sector’s macroeconomic implications, usually disregarding the liability structure (M&M-theorems). Let me use the example in your post:

    Base money B is a gold stock, but the public prefers to hold “bank-created money” M. I assume that banks issue no notes, so M measures deposits. The reserve ratio is defined by equation (1) of your post, i.e., M=R(1/r), where R is the level of reserves. Given that the public doesn’t demand gold, “the supply of bank reserves is always equal to the supply of basic money”, i.e. R=B, so (1′) M=B(1/r) with 1/r=m.

    At least since Keynes (in the Tract) and Fisher used this logic to integrate fractional reserve banking into the quantity theory (QT), equation (1′) (or 1) is chosen to determine M. Let’s lump (1′) into the quantity equation, like Keynes and Fisher did … B(1/r)V=PY … and gone are the deposits, which reveals that they don’t have a live of their own: they are fully explained by B and r (optimized or not).

    For the sake of the argument, let us also assume that Y is fixed at the full-employment level. Then, and increase in B or a decrease in r increases the price level … the hot-potato effect. Specifically, a decrease in r implies credit expansion financed by money out of thin air, deposits. The excess supply of such transaction money implies an excess demand for commodities, driving up the price level to ensure market clearing. Both, the young Keynes and Fisher argued that changes in r induce business-cycle effects. Keynes used momentum effects to show that the QT-prediction may not hold (the Keynes-Cagan effect).

    This is the view I critizise. And I think this is a fair description of the model outlined in your post. Here is the first of your two questions: “To what extent have historical money-fueled booms been associated, not with growth in the supply of either commodity money or central-bank supplied bank reserves, but with declining banking system reserve ratios?” …. clearly, I have identified the two drivers of deposits in my example.

    This view treats deposits like yieldless cash, thereby artifically fixing prices (zero yield). Unsurprisingly, if one presumes perfect substitutionality, the model predicts that deposits behave like base money: they induce a real-balance effect …. simply because the interest rate is arbitrarily fixed.

    With respect to the Law of Reflux: OK, if it isn’t the Law of Reflux, I don’t understand the argument:

    “… a deposit made at one bank involves a corresponding withdrawal of funds from another bank, as when the deposited sum takes the form of a check. In that case, the process of deposit expansion that Block and Garschina describe will have as its counterpart a like process of deposit destruction, where the ultimate result (assuming the simple case in which all banks maintain the same, given reserve ratio) is an unchanged total money stock, with the only actual change consisting of a change in the distribution of bank deposits among the various banks.”

    The view you critizise assumes a sectoral credit expansion financed by new inside money. If there is no compensating loan repayment (or price level adjustment as the currency school would insist), how can an interbank transfer of deposit be of help? Some non-bank must increase its net financial position (the ultimate creditor), or nominal income must adjust.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s