By Brett Fiebiger, Ph.D
A liquidity trap may be defined as a situation in which conventional monetary policies have become impotent, because nominal interest rates are at or near zero: injecting monetary base into the economy has no effect, because base and bonds are viewed by the private sector as perfect substitutes … [W]hatever the specifics of the situation, a liquidity trap is always the product of a credibility problem: the public believes that current monetary expansion will not be sustained [Emphasis added]. – P. Krugman, ‘It’s Baaack: Japan’s Slump and the Return of the Liquidity Trap’, 1998, pp. 141, 166.
Look, please, at my Brookings Paper on the liquidity trap … Look, we have a framework here that has been a stunning success in practice. Why won’t you guys admit it. – P. Krugman, ‘Been There, Done That’, 24 July 2013.
Paul Krugman often refers his NYT blog readers to his Brookings paper on Japan for details on why the balance sheet substitution channel of monetary policy is relatively impotent in a liquidity trap. The gist of the argument is that central bank open market (OM) operations become ineffective at raising prices and/or output when base money and bonds are ‘perfect substitutes’. The framework and predictions of the Brookings paper are arguably less robust than touted.
Consider first an obvious objection that monetary authorities need not confine their dealings to short-term bills (as advised by Keynes (1936) in Chapter 15). Since the crisis began the US Fed has added over $1 trillion of T-securities to its books and about the same in longer dated MB-securities. Back in 2007 short-term T-bills comprised close to one-third of the Fed’s holdings of T-securities, now, the amount is zero. Banks have been receiving 0.25% on reserves; still, it is not the case that the Fed has just been swapping near-zero yielding debts for near-zero yielding reserves.
While long rates have fallen, they are not near-zero, especially private debts (Table 1). Interest rates on bank credit products are also well above that on reserves. Even if banks could lend out reserves to nonbank agents (instead of only to other banks in the federal funds market at a rate set by the central bank) the reason why banks are holding large amounts of excess reserves cannot be because there is no ‘opportunity cost’ to lending: many interest rates are quite a bit above zero.
At a basic level of scrutiny the ‘lack-of-an-opportunity-cost’ explanation falls down. Krugman (1998) did not consider that the central bank could buy securities with different maturities and risk profiles as in a quantitative easing (QE). With QE it cannot be argued within an exogenous money framework that OM operations are ineffectual because private agents are holding base money just for the store of wealth function (due to ‘lack-of-an-opportunity-cost’ / ‘perfect substitutability’).
Eggertsson and Woodford (2003) did consider QE and stipulated an irrelevance proposition for all OM operations (for rather dense reasons). Bernanke, Reinhart and Sack (2004) in turn objected to the contention that shaping interest rate expectations is the only effective policy tool. The pre-crisis New Keynesian liquidity trap literature was fragmented with some of the key contributors either ambivalent or sceptical about a role for fiscal policy: a view now untenable.
The Brookings paper advice was for monetary authorities to commit to sustaining a high level of base money. The US Federal Reserve has yet to adopt that advice as part of a deliberate attempt to induce economic recovery via “expected inflation” (though expanding and maintaining base money is one aspect of Abenomics). Instead, ever since the US Fed’s balance sheet expanded in late 2008, officials have sought to dismiss fears based on the simple idea that “an increase in outside money—the monetary base—must raise prices (Krugman, 1998, p. 141).” The Brookings paper is still thought useful to counter the dubious views of the monetarists-in-mind like, Alan Meltzer and John Taylor, who fear that the Fed’s base money expansion portends runaway inflation.
Krugman’s (1998) analysis is nonetheless monetarist-in-spirit. It argued that the orthodox theory of money neutrality (and exogenous money supply) may not always work; unless the central bank can “credibly promise to be irresponsible”, and convince everyone that a high level of base money will be sustained into the future thus “creating” the theorised “natural” solution of “expected inflation” in the current period. Amongst the assumptions were a stable money demand function, expressed in the monetarist equation of exchange, such that a given volume increase in base money induced a simple multiplicative increase in broad money and income (at least outside of the liquidity trap).
The monetary solution in the Brookings paper resembled monetarism more than the neo-Keynesian version of the liquidity trap (which for all it defects at least supposed a fiscal solution was needed). Increase base money, signal that it will be sustained, and then the “money multiplier” will get going regardless of the demand-side or any complicated financial disturbances like a debt-deflation:
One often hears, for example, that the real problem is that Japan’s banks are troubled, and hence that the Bank of Japan cannot increase monetary aggregates; but outside money is supposed to raise prices regardless of the details of the transmission mechanism. Aside from the bad loans, one also often hears that corporations have too much debt … All of this may be true and may depress the economy for any given monetary base, but it does not explain why increases in the monetary base should fail to raise prices, or output, or both. Recall that the neutrality of money is not a conditional proposition; it does not depend on banks being in good financial shape, or the service sector being competitive, or corporations not taking on too much debt (Krugman, 1998, pp. 141-2).
The details of the transmission mechanism of monetary policy do matter. The supply of bank money is determined primarily by the endogenous demand for bank credit from private agents within the limits imposed by capital adequacy and bankers’ judgements of creditworthiness / prudent leverage. One should not expect monetary policy will work if: (a) banks have concerns about capital adequacy; (b) asset prices have collapsed reducing creditworthiness; or, (c) there is a lack of demand for credit.
To suppose that monetary policy—which is essentially about influencing the terms on private debts (with a greater and more direct influence at the short end of the term structure of interest rates than at the long end) and thereby indirectly the quantity of credit and level of economic activity—should work regardless of any detailed consideration of the balance sheet quality of heterogeneous private agents and the burden of debt thereupon is profoundly counterintuitive.
At year-end 2007 the ratio of US base money to bank deposits was around 1:10. If the Fed had faith in the “money multiplier” it should have created say about $0.4 trillion in excess reserves and waited for the “money multiplier” to do the rest. That bank excess reserves at around $1.95 trillion are not “money multiplying” into a quantity of new bank deposits somewhere in the vicinity of $19.5 trillion should be taken as an unambiguous empirical testing of a theory; with the result, that of failure.
The exogenous money camp is divided about why the “money multiplier” is stalled. Some argue that banks have been impeding the recovery by hoarding reserves and making a mint from doing so at a 0.25% return minus about the same paid on deposit liabilities. The US Fed’s policy of paying interest on reserves is blamed for removing the incentive for banks to lend reserves. Another rationalisation is the lack-of-an-opportunity-cost which is unconvincing given the well-above-zero lending rates in Table 1 (plus QE) and also because banks cannot lend out reserves to nonbank agents.
There is no reason to fear some sort of “money-multiplying” of excess reserves into trillions of new bank deposits or some pre-1900s currency-transacted spending spree (Krugman, 14 January 2013). Any talk about the “money multiplier” being stalled is a diversion. The theory is faulty as it supposes that banks just wait around for the Fed to increase reserves, and then automatically expand the supply of broad money; as if capital adequacy, creditworthiness and credit demand do not matter.
The Brookings Paper Monetary Framework and Monetary Predictions
Cullen Roche (15 August 2013) has drawn attention to the counterintuitive ‘banks lend out reserves’ aspect of Krugman’s (1998, p. 185) monetary framework as per this passage:
Banks, however, need hold only a fraction π of their deposits in reserves and will hold no more than necessary; they lend the rest out (which is how consumers get the money for the deposits). So bank deposits will be a multiple 1/π of the monetary base; the velocity of base will be 1, that of deposits, π.
Krugman (1998, p. 184) envisaged a sequence where “Consumers come into existence and receive the money supply M*” with M* meaning here outside money. That is a variant of the Friedmanite “helicopter drop”, with outside money just falling from the sky, and then OM operations can further affect the supply. In the Brookings paper bank deposits were “created” after, and in proportion to, consumers depositing currency. That contrasts with Krugman’s (24 August 2013) new acceptance of “commercial banks… can make a loan simply by crediting the borrower with new deposits…”
Krugman (16 August 2013) also has a new take on “banks lend out reserves”. JKH (18 August 2013) explains why “lend reserves” is not an apt shorthand for describing the funding of bank activities; especially, the main lending portfolio. There is another counterintuitive aspect in arguing that banks lending reserves is “how consumers get the money for the deposits”. Aren’t bank deposits… money? Krugman’s (1998) paper contains several models. For succinctness I will skip to the introduction of financial intermediaries (“banks”):
To do this [i.e. put financial intermediaries and monetary aggregates into the basic model], return to a one-good endowment economy, but now suppose that at the beginning of each period a three-step process takes place, as follows: (1) individuals trade currency for bonds in a capital market and are also able to make deposits at a class of banks, (2) individuals discover whether they derive utility from consuming in the current period, (3) those who do want to consume withdraw the necessary cash from their bank accounts (Krugman, 1998, p. 156).
Why would the individuals settle bond trading in currency instead of using bank money? Why would the consumers, who want to consume, need to make currency withdrawals? EFTPOS has been used in the US since the 1980s, credit cards since the 1950s, and cheques for over a century.
It is curious that Krugman’s (1998) “banks” issue deposits, but not money, which functions as a medium of exchange and unit of account. [Eggertsson and Krugman (2012) do no better as their model has no “banks” and does not require an explicit money supply]. It would be surprising if a framework informed by the logic of a currency-transacted economy and exogenous money supply did yield correct predictions. Krugman’s (1998, p. 157) predictions for money stocks were stated as:
Applying what one of my colleagues calls the principle of insignificant reason, one may surmise that an increase in monetary base will lead to substitution in all three directions. This means that under liquidity trap conditions, such a base expansion will (1) expand a broad aggregate slightly, but only because the public holds more currency; (2) actually reduce deposits, because some of that currency substitutes for deposits; and (3) reduce bank credit even more, because banks will add to reserves.
Figure 1 indicates substantial disparity. The growth in reserves after the collapse of Lehman Brothers was voluminous; yet, a surging demand for currency did not take place. And, bank deposits have grown by a couple of trillion since late 2008, rather than fallen or remained stagnant.
The dotted lines in Figure 2 underline what Krugman’s (1998) predictions for currency and deposits should have looked like (or approximately so) after the US Fed hit the zero nominal lower bound on short-term interest rates (said sometimes by New Keynesians to be the liquidity trap) in late 2008.
Nor is it clear that banks are indifferent to holding base money vis-à-vis bonds. The Brookings paper prediction was for a rise in reserves matched by a like fall in bank credit; specifically, the bonds sold in OM operations. The data does not show that (see Figure 3). Bank credit did decline (notably loans) but not bank holdings of Treasury and Agency securities. If the Fed merely swapped reserves for bonds with banks in its QE asset acquisitions and the balance sheet substitution channel was stalled, then, banks should be left with reserves and diminished holdings of Treasury and Agency securities.
Judging by these remarks Krugman (11 April 2013) may dispute aspects of the above data analysis:
So, at this point America and Japan (and core Europe) are all in liquidity traps … Incidentally, this isn’t just a hypothetical: there has been a surge in currency holding, although a lot of it is $100 bills held overseas: [Figure showing US currency outside banks as a percentage of GDP].
Krugman (14 August 2013) again refers to a surge in public holdings of currency though this time he did not mention the overseas drain (over two-fifths of US currency is held abroad). A figure of the ratio of total currency in circulation to GDP was provided (with the axis zoomed right in) followed by these remarks: “Again, this is exactly what liquidity-trap models, like my 1998 paper, predicted.”
It is debatable that US residents have taken to currency as a prudent portfolio choice. From the start of 2008 to 2013Q1 the amount of currency outside banks held domestically by US residents grew from 3.3% of GDP to 3.9% (Figure 4); and, by $155bn in dollar terms. There certainly are reasons to be unconvinced that the paltry trickling in currency holdings can explain key aspects of the crisis.
In another post Krugman (19 September 2011) pointed instead to “surging bank deposits” as proof of the liquidity trap. Can a surge either in currency or in bank deposits be taken as vindicating the predictions of liquidity trap economics? It is worth underlining what was actually predicted:
If an economy is truly in a liquidity trap, failure of broad monetary aggregates to expand is not a sign of insufficiently expansionary monetary policy: the central bank may simply be unable to achieve such an expansion because additional base is either added to bank reserves or held by the public in place of bank deposits. However, this inability to expand broad money does not mean that the essential problem lies in the banking system; it is to be expected even if the banks are in perfectly fine shape [Emphasis added] (Krugman, 1998, p. 158).
From the start of 2008 through to 2013Q1 the volume of deposits issued by US commercial banks grew by $2,493bn (and from 52.9% of GDP to 62.0% of GDP) while the volume of bank credit grew by only $1,101bn. How did the US Fed manage to expand broad money while banks were reluctant to increase lending; and, why didn’t that expansion produce a timely and robust recovery? I’ll have a go at that in another post. A few brief remarks are required here on the broad money expansion.
A noted above, Krugman (19 September 2011) has shifted from surging bank deposits is proof of the liquidity trap, to the “money supply broadly defined hasn’t taken off (Krugman, 14 August 2013)” amidst surging currency growth vindicates the Brookings paper liquidity trap predictions. Certainly, bank deposits did not “take off” as expected if the “money multiplier” was descriptive, but that theory is not a useful guide to anything. When considering that the growth during 2008-2013Q1 in US commercial bank deposits was 1,608% greater than the growth in US resident currency holdings it is difficult to sustain that it is the latter money which was surging.
Crisis lessons: credible base money expansions or fiscal policy relevance?
It seems to be part of the folk wisdom in macroeconomics that this is in fact how the Great Depression came to an end: the massive one-time fiscal jolt from the war pushed the economy into a more favorable equilibrium … My point is that… the view that a one-time fiscal stimulus can produce sustained recovery, does not actually appear to fit the story line too well … If the central bank can credibly promise to be irresponsible… it can bootstrap the economy out of the trap. – P. Krugman (1998, pp. 159-161).
The “natural” solution to the liquidity trap as argued first by Krugman (1998) and reiterated by Eggertsson and Krugman (2012, pp. 11, 20-1) is “expected inflation”. One would think that the “natural” solution to a big crisis is for policymakers to, first put a floor under aggregate demand via fiscal policy, and then to create credible expectations of growth in aggregate demand going forward. The key policymaking task is not to “credibly promise to be irresponsible” with monetary policy but to “credibly promise to be responsible” with fiscal policy, that is, actively manage demand to create the conditions and confidence conducive to a private sector growth dynamic taking hold.
The central bank can play a constructive and responsible role (e.g. lender of last resort activities and actions to lower long interest rates); nonetheless, fiscal policy is the relevant countercyclical tool. Before Krugman (1997) rediscovered and modified the concept of the liquidity trap, he was arguing monetary policy was nearly-omnipotent, and that the Fed Chairman had a magical wand for the unemployment rate in the form of the federal funds rate. He was not alone in thinking so and it is important to not personalise a critique that applies to the “mainstream economics profession”.
When Krugman (1998) stumbled on the liquidity trap (and I write stumble because it happened as a by-product of him endeavouring to discredit the said “vulgar Keynesians” who viewed ideas like the thrift paradox as relevant) the argument was then made that a monetary solution alone could work. Support was also given to Romer’s (1992) non-fiscal policy explanation of how the United States exited the Great Depression. The paper thus did not prepare policymakers for the fiscal solution needed in the Lesser Depression. In the build-up to the current crisis Krugman (2003) warned of a fiscal crisis in the United States and also a creditor-led ‘dollar crash’ (July 2007): neither occurred.
Since the crisis began Krugman has pushed fiscal reflation and deserves some kudos for doing so. The claim however that there is a liquidity trap framework which has been vindicated by the crisis (e.g. Krugman, 14 December 2011; 24 July 2013; 14 August 2013; 16 August 2013; 27 August 2013) infers a level of continuity and coherence that is not there. None of the key Minskyan insights on endogenous financial instability, private debt, banking and the relevance of fiscal policy made it into the pre-crisis New Keynesian liquidity traps (and it is also debatable how well so post-crisis). Nor is it evident that the Brookings paper predictions played out in the United States’ Lesser Depression:
- Banks reserves did rise but not because of a ‘lack-of-an-opportunity-cost-to-lending-reserves’.
- Bank credit did fall for awhile but not because banks just swapped reserves for bonds (indeed, bank holdings of securities of the type the Fed bought in QE increased rather than declined).
- Domestic currency holdings hardly grew at all (and not unusually so) which was not predicted.
- Bank deposits did expand by a non-trivial amount (and usually so given the developments in bank credit) which was not predicted.