The Liquidity Approach to Financial Crises1
The Liquidity Approach to Financial Crises1
Abstract and Keywords
The chapter focuses sharply on liquid assets, and is the heart of the book. It distinguishes between intrinsic and extrinsic liquidity, centering on the latter. Extrinsic liquidity may break down on the spur of the moment and generate socially costly Liquidity Crunch. A substantive part of the chapter is devoted to discussing relative resilience of liquid assets, and focuses on Keynes's Price Theory of Money, the resilience of the US dollar, the weakness of bonds denominated in that currency, and of currencies of emerging-market economies. The chapter claims that recent financial crises can realistically be modeled as old-fashioned bank runs, and that assets' liquidity may be a function of policy. Special attention is paid to a phenomenon called Liquidity Deflation, which helps to rationalize Liquidity Trap as a consequence of loss of money liquidity rather than on the conventional explanation based on the infinite interest elasticity of money demand.
… the use of money is enough in itself to make a free market system potentially unstable; and that the higher the degree of development, or sophistication, that it exhibits the greater does the danger of instability become . … It is a “psychological” instability, not a mechanical, which is in question: so it cannot be remedied by the application of a formula, as so many, both then and in later days, have been tempted to suppose.
As I noted in the previous chapter, “liquidity” or “means of exchange” problems are factors that cannot be ignored in a reasonable narrative of financial crises. In this chapter I take a more theoretical tack and explore the potential explanatory power of the view that puts liquidity at the center of these crisis episodes, a view that I will call the Liquidity Approach (to financial crisis). The Liquidity Approach focuses on the primary triggers and possible magnifiers of financial crises. I argue that the Liquidity Approach can offer a rationale for central and puzzling phenomena that cannot be easily explained by mainstream macro models. An example to which I devote some attention is the striking resilience of the output price of fiat or paper money (“cash” or “currency,” for short) in developed market economies. DMs were at the (p.24) epicenter of the subprime crisis. Nevertheless, the demand for DM cash (particularly the US dollar and dollar-denominated assets fully guaranteed by the US government, e.g., Treasury bills; “dollar” for short) has exhibited a sizable increase (Liquidity Trap). This puzzle is exacerbated by the simultaneous occurrence of a large price meltdown in other DM liquid assets (liquidity crunch).
I start with a definition of liquidity and its sharp demise (liquidity crunch), and continue with a discussion of the strength of DM currencies during crisis, a fact that stands in sharp contrast with the weakness of emerging market currencies under similar circumstances (see chapter 6). The centerpiece of this section will be the Price Theory of Money proposed by Keynes (1936). I argue that liquidity crunch can give rise to credit Sudden Stop, and a coordinated fall in aggregate demand with dire consequences for output and employment. Making these points does not require abandoning the assumption of rational expectations, or assuming animal spirits that coordinate gloomy views about the marginal efficiency of investment by CEOs, or trigger massive thriftiness attacks on the part of consumers.
A lesson of the chapter is that discussion of liquidity issues reveals aspects of modern economies that are radically different from the conventional view and show the relevance of central banks and financial regulators for preventing and managing financial crises.2 However, the discussion will also make it evident that liquidity opens up a variety of questions for which we do not have clear answers.
Liquidity and Its Demise: Liquidity Crunch
“Liquidity” is a slippery concept, and is all the more so given the lack of attention it has received from economists since (p.25) World War II. In the conventional model, liquidity is represented by something called “money,” and a central proposition is that “money is a veil.” This view, taken to an extreme, is tantamount to saying that money is essential for the determination of the price level, and nothing else. Under this lens, monetary policy has no significant effect on output or employment (i.e., money is “neutral,” in the macroeconomist’s lingo).3 Of course, standard Keynesian models give a nonneutral role to money, but this is achieved by assuming some sort of nominal price/wage rigidity, not by delving deeply into the nature of money or liquidity. Thus, aside from the distinction between outside and inside money to incorporate in the analysis commercial banks as producers of inside money (e.g., sight deposits), conventional macroeconomics has had little to contribute to the liquidity discussion.
The subprime crisis and the type of phenomena highlighted in the previous chapter have given a strong impetus to the liquidity literature. A dominant view that emerges from that literature is that liquidity is an essential characteristic of means of exchange, and that its effectiveness in that role is an increasing function of the ability of making means of exchange equivalent to a basket of highly tradable goods (i.e., items that directly enter utility or production functions), at highly predictable relative prices and involving relatively low search costs. Models differ as to the specifics, but a term that is becoming increasingly popular is “pledgeability.” The word is especially central in models that highlight the credit market (e.g., Kiyotaki and Moore 1997; Holmström and Tirole 1998, 2011). Pledgeable is the share of a good/asset that can be easily transformed into a basket of highly tradable goods.4 In models with a few goods (which hide the imprecision of the expression “basket of goods”), (p.26) pledgeability offers respectable foundations for the liquidity concept. By “respectable,” I mean foundations that facilitate inserting the model into the general equilibrium Arrow–Debreu model—expanded to account for asymmetric information or agency considerations—without having to model objects like “money” that have always looked alien or redundant in that kind of framework.
However, the appeal of those models to the professional economist is obtained at a cost, since it ignores assets that are deemed highly liquid in practice but have no intrinsic value and no predefined claim on goods. A classic example is fiat or paper money. To discuss this issue, it is useful to extend the discussion and define two types of pledgeabilities, namely intrinsic and extrinsic. Intrinsic pledgeability is the concept mentioned above—the ability of transforming a given good or asset into a basket of highly tradable goods. On the other hand, extrinsic pledgeability is the ability to quickly get real goods on the expectation that someone else will accept the liquid asset in question in exchange for a basket of goods. In other words, an intrinsically liquid asset contains real goods in its core, while extrinsically liquid assets have none. To transform the latter into goods, there is at least one more step that has to be taken, involving at least another agent that is willing to accept the extrinsically liquid asset for real goods. As indicated, fiat money belongs to that group.
From the point of view of an individual, the difference between intrinsic or extrinsic liquidity is not major. However, in models where utility and production functions are the essential ingredients of human welfare, extrinsic liquidity appears as a flimsier form of liquidity than intrinsic liquidity. Extrinsic liquidity is a concept alien to Robinson Crusoe, for example. Extrinsic liquidity is liable to vanish by the simple (p.27) fact that individuals suddenly refuse to surrender goods in exchange. This type of shock can take place even though there is no major shock to utility or production functions—no earthquake that destroys real assets, for instance. For the man in the street (including many economists in the immediate aftermath of the shock), the shock would look as mysterious as a thunderbolt for the primitive man. These shocks characterize most liquidity crunch episodes in recent financial crises (see chapter 1). The following discussion will focus on the economics of extrinsic liquidity and liquidity crunch episodes.
A liquidity crunch can be triggered by factors that are not mere illusions or delusions. Consider the familiar example in which there is only one commercial bank.5 The bank is funded by demand deposits denominated in terms of cash, purchases illiquid assets (i.e., makes loans that are hard to exchange for cash on call), and holds a precautionary stock of money to cover deposit withdrawals. Under normal circumstances, bank deposits are close substitutes of cash, which leads banks to feel safe lending a large share of deposits and holding a small amount of cash in their coffers for unusual deposit withdrawals. This situation I call “the good equilibrium”; it characterizes capitalist economies in “tranquil,’ non-crisis times. However, one cannot discount the existence of a “bad equilibrium” in which depositors expect that a large number of fellow depositors will try to take their cash out of the bank. In that situation, depositors have incentives to do the same because they know that the bank coffers are pretty slim. If every depositor attempts to get her money out of the bank at the same time, the bank cannot honor the contract to exchange deposits for cash on call—and goes bankrupt. Bank runs impair the liquidity of bank deposits suddenly, causing liquidity crunch.
(p.28) Bank runs were the main motivation for the establishment of central banks, which, as the subprime crisis has demonstrated, help stave off harmful bank runs. But, especially in DMs, only a small share of the stock of extrinsically liquid assets is protected by a central bank. Thus it would be fair to say that liquidity crunch is an endemic disease in economies where extrinsically liquid assets play an important role. The associated crisis, it should be noted, is not incompatible with rational expectations and therefore could be rationalized in terms of mainstream theory (see chapter 3).
A Resilient Dollar—The Price Theory of Money to the Rescue!
The subprime crisis has shown that not all extrinsically liquid assets are equally fragile. The dollar, for example, has exhibited remarkable resilience during the subprime crisis. This is very puzzling because dollar bills have no intrinsic value and the Fed makes no commitment to exchange dollar bills for goods and services.6 In fact the dollar’s “fundamentals” are weaker than bank deposits because the latter are collateralized with claims on real private sector assets, whereas dollar bills have no collateral.7 So, why was the dollar so resilient during the subprime crisis?
To answer this question, it is convenient, first, to discuss the reasons why the relative price of cash in terms of goods and services is positive.8 The microeconomic literature has explored different reasons for this to be the case, such as legal tender or the obligation to settle tax liabilities with cash. However, as argued by Calvo (2012 a), these explanations are not persuasive, especially in a multi-currency world. An explanation that has been largely ignored by the profession and that I find very appealing—although admittedly (p.29) incomplete—was proposed in Keynes’s General Theory. He writes: “the fact that contracts are fixed, and wages are usually somewhat stable in terms of money, unquestionably plays a large part in attracting to money so high a liquidity-premium” (1936, chapter 17; italics are mine). By “liquidity premium” Keynes means the difference between the market’s output value of money and its intrinsic value (virtually zero).
I discuss Keynes’s conjecture in Calvo (2012 a) and call it the Price Theory of Money (PTM). Intuitively, the PTM asserts that money’s output value cannot be nil in equilibrium because economic agents set several key nominal prices/wages in advance, thereby offering a ground-up output guarantee to cash.9 It is extraordinary that to this day this kind of guarantee that holds in all capitalist economies has largely been overlooked by the micro-and macro-literature. I suspect the omission stems from the lack of a deep discussion of liquidity issues in macroeconomics and also the obliteration of price/wage stickiness issues from microeconomics.
The PTM offers a partial rationale for the resilience of the dollar relative to other liquid assets. Most prices and wages (“prices,” for short) in the United States, for instance, are set in dollars and show some degree of temporary stickiness, validating the assumptions of the PTM. This puts the dollar in a privileged position relative to other extrinsically liquid assets, like asset-backed securities (ABS). There is no commitment—comparable to that enjoyed by cash—to exchange ABS for goods and services on short notice. This does not imply that ABS and other financial assets have no real backstop. Market makers do that to some extent. But the stickiness of market makers’ prices is short-lived compared to labor contracts, and during financial crises there are several instances in which market makers have shut down, (p.30) paralyzing security market transactions and creating a situation in which market prices disappeared into “thin air” (see Mehrling 2011; Financial Times 2014). This situation debilitates the real anchor enjoyed by liquid assets other than cash. It helps rationalize a scenario in which, say, ABS prices collapse while the demand for cash rises and liquidity trap takes place (for a formal model displaying this feature, see chapter 5).
Finally, the output guarantee for cash becomes even stronger/more credible if prices are set in an uncoordinated fashion, and there is no significant bunching of price revisions, a realistic situation that the literature calls staggered prices. To see this, notice that if, contrariwise, all prices were set at the same time, price setters could all simultaneously change the unit of account in which their prices are quoted—from dollars to bitcoin, for example—keeping intact relative prices. The latter does not hold under staggered pricing, which generates disincentives for marginal price setters to pull out of the pack and set their prices in a different unit of account.
The PTM implies that even fiat money involves some output pledgeability. However, fiat money’s liquidity is fundamentally extrinsic, since in most cases there is no formal or legal commitment to exchange fiat money for goods. Its acceptability as a means of exchange may be enhanced by the existence of staggered prices but, as a rule, fiat currencies are just worthless pieces of paper/metal (relative to what they can command in the market)—a far cry from, say, a gold watch used as collateral in a pawnshop whose market value normally exceeds, by a wide margin, that of the corresponding collateralized credit. The lack of predetermined output pledge may detract from fiat money’s liquidity. However, there is an important offsetting factor. An advantage of fiat (p.31) money over other assets with pledgeable goods is that, unless the pledge is expressed in terms of money—which would bring us back to the PTM—those pledges are specified in terms of a narrow set of goods (a house, a gold watch, a laptop, etc.), while fiat money offers command on all goods whose prices are predetermined in terms of the corresponding unit of account. This gives an additional reason why a liquidity trap can occur during a liquidity crunch episode. Moreover it helps explain why a reserve currency like the dollar, a global unit of account, can be subject to a liquidity trap while EM currencies undergo liquidity crunch or are victims of capital flight. This will be discussed next.
The resilience of the dollar during the subprime crisis, abstracted from the fact that there are many different fiat monies in circulation. The latter opens up the possibility that the degree of output anchoring differs across currencies. Actually the US dollar holds a prominent position because the US economy is very large, US domestic prices are set in dollars, and even import prices are remarkably insensitive to swings in the dollar exchange rates (see Dornbusch 1987; Gopinath et al. 2010). Moreover the dollar is a worldwide unit of account. Commodity prices are set in dollars, and until the Lehman crisis, oil prices were somewhat sticky in dollar terms, thanks to OPEC’s quantitative controls on oil supply. In fact the dollar has been a key unit of account in global financial transactions for an extended period of time, and dollar-denominated financial markets dominate those of other major currencies. To wit, the eurodollar market has no parallel in a US–euro or US–sterling market. Nevertheless, the dollar preeminence is not guaranteed going forward (p.32) because there are elements of self-fulfilling prophesy behind the resilience of the US dollar. This is important to keep in mind in assessing stimulus packages based on making dollar holding less attractive, such as through higher dollar inflation or negative nominal rate of interest on money holdings (e.g., Gesell’s Stamped Money proposal; see Keynes 1936, ch. 23).
Emerging market fiat monies represent an interesting contrast. In many EMs, for instance, the dollar or euro are units of account for domestic transactions, especially for big-ticket items like real estate and luxury cars. Moreover the pass-through to domestic prices from changes in the exchange rate is large relative to DMs for EMs with a history of high inflation (Choudhri and Hakura 2001; Burstein and Gopinath 2014). Thus the real anchor of EM fiat monies—their PTM component, if you will—is likely to be significantly weaker than that of the dollar. EM policy makers are keenly aware of this and are adamant to let their currencies fluctuate—a phenomenon studied in Calvo and Reinhart (2000, 2002), and labeled “fear of floating.” Under the PTM perspective, fear of floating is equivalent to hardening the real anchor of EM cash by pegging to fiat monies that are perceived as being firmly anchored onto the real sector (which I call “safe monies”), like the dollar. This is not cost free because it induces EMs to accumulate costly international reserves, namely a stock of safe monies and liquid assets backed up by their corresponding central banks. Thus fear of floating shows that EMs have implicitly embraced a system akin to that of Bretton Woods, in which pegged but adjustable parities are the rule. Many EMs accomplish this unilaterally, without relying on international financial institutions (IFIs).10 Consequently, the stock of international reserves in the hands of EMs has skyrocketed (e.g., see Obstfeld et al. 2010), and (p.33) this, in a way, has further reinforced the real anchor of DM currencies—particularly the dollar, given that it still has the lion’s share in EM international reserves.11
Liquidity Crunch and Sudden Stops: Collateral Damage
A liquidity crunch episode, like the one that took place around the Lehman crisis, can have severe effects on credit flows. Large and largely unexpected contractions of credit flows are called Sudden Stops (SS). SS have been shown to cause major disruptions in the real sector.12 SS can hit an economy as a whole, as was the case in EM financial crises, or just a subsector, as in the US real estate market in the wake of the Lehman crisis episode (in which case term-mismatch between assets and liabilities may wreak chaos, even if contracts are denominated in the same unit of account; see below).
Well-anchored fiat monies can oil the wheels of the credit market. The credit market deals with intertemporal transactions in which, say, today one agent supplies a piece of bread to another in exchange for a promise from the buyer to deliver a (usually larger) piece of bread tomorrow. A simple way to carry out this transaction is for the buyer to issue an IOU. However, assessing the credibility of the IOU calls for substantial information on the part of the seller. A way around this problem is for the buyer to pay with money that is well anchored in terms of output. Other liquid assets are also regularly utilized for intertemporal transactions. But, if PTM considerations or a lender of last resort do not protect the output value of those assets, they could be subject to liquidity crunch. It follows that a liquidity crunch episode can have severe effects on credit flows. Moreover, as discussed above, during an SS/liquidity crunch episode not all (p.34) liquid assets are hit alike, and may have disparate impact on collateral values.
This is a critical issue that deserves greater attention. Consider the case of mortgage-backed securities (MBS) that played an important role in facilitating housing market credit, but suffered a major liquidity setback during the Lehman crisis. Arguably, the shock increased intermediation costs in the real estate market, which reduced the demand for housing and provoked a meltdown in real estate prices. Developed market central banks reacted by increasing the supply of base money (i.e., cash in the hands of the public and banks), a different type of liquid asset than MBS. This had a positive effect on the CPI (the consumer price index) but did not prevent a massive fall in house prices. Thus, even though cash and MBS are two forms of liquid assets, central banks were unable to offset the effects of liquidity crunch on MBS. Chapter 3 will discuss this in terms of a formal model.
The point above is worth stressing because the conventional wisdom prior to the subprime crisis took to heart Friedman and Schwartz’s (1963) conjecture that the 1930s Great Depression would have being a regular recession had the Fed increased money supply and stopped producer and consumer prices from falling precipitously. Fisher (1933), in particular, discussed a mechanism that helps explain why the massive price deflation that took place in that episode caused major financial disruption. He argued that price deflation could give rise to unplanned overindebtedness, given that most debts were denominated in dollar terms and were non-state contingent (a phenomenon that he labeled “debt deflation”). I do not disagree with the basic intuition. However, the previous paragraph suggests that monetary expansion might not have been powerful enough to prevent the fall of real estate relative prices during the subprime crisis, (p.35) ven if debt deflation were prevented.13 Furthermore, since US mortgage contracts are set in nominal terms, preventing a large capital loss for mortgage debtors would have called for consumer or producer prices to increase by around 30 percent on impact! Implementing this was very difficult in view of the liquidity trap phenomenon. Besides, if successful, this policy would have been equivalent to a major devaluation of the dollar’s real anchor which, as pointed out above, could have debilitated the safety of the dollar unit of account—the corner stone of the global financial system—possibly exacerbating recession and causing severe political backlash.
Interestingly, a phenomenon akin to debt deflation occurred in EM financial crises, even though these economies were free from liquidity trap (if anything, they suffered from capital flight) and, on average, they exhibited sharp inflation spikes (see chapter 6). Debt deflation–like phenomena happened because a large share of EM debt (not exclusively external) was denominated in terms of foreign exchange—another consequence of emerging markets’ perceived weak real anchors (see Eichengreen et al. 2005). Thus large devaluation resulted in a sizable increase in external debt as a share of GDP. To avoid debt deflation under these circumstances would therefore call for tightening monetary policy to prevent large currency devaluation, exacerbating the slump!
Even if overindebtedness is avoided by resorting to aggressive monetary/fiscal policy—inclusive of debt forgiveness and rescheduling—liquidity crunch could still have a deleterious effect on the credit channel. For example, it may be hard for house credit to bounce back to the situation prior to crisis, unless the market for MBS is thoroughly repaired, which is not automatically granted by protecting house debtors. Absent full repair of extrinsically liquid assets, a (p.36) liquidity crunch may generate a saving’s glut, given that the increase in savers’ expenditure is unlikely to be offset by the forced decrease in dissavers’ expenditure (due to the liquidity shock). Saving’s glut is more likely to arise if the episode takes most agents by surprise, giving little time for saving flows to find alternative outlets. Moreover a Keynesian impasse, where output is the adjustment variable, cannot be discounted if the central bank succeeds in staving off price deflation.14 However, recession would be less acute for economies that can generate a surge in exports. This mechanism operated in EMs (thanks to weak currencies!) and was instrumental for quick recovery (see chapter 6). But it is worth noting that EM export expansion took place during the Great Moderation period in DMs (the destination of a large share of EM exports) and that output recovery was nevertheless far from instantaneous. Moreover there is some evidence pointing out to a negative impact of financial crisis on output trend (see Cerra and Saxena 2008).15 This suggests that even in cases in which demand expansion can be quickly triggered from abroad, liquidity considerations could have long-lasting detrimental effects. (These issues are further discussed in chapter 6.)
Liquidity crunch of financial assets can also have an impact on the relative price of factors of production. As noted above, access to credit is facilitated, and attendant interest rates lowered, by being able to post liquid collateral assets. However, at least two types of collaterals can be distinguished: intrinsic and extrinsic.16 Intrinsic collateral is the collateral inherent in the goods that are purchased on credit. For example, a laptop that can be repossessed in case of default exhibits an intrinsic collateral linked to its price in the secondary market. However, financing the costs associated with a search for new employees has hardly any intrinsic collateral. Arguably, (p.37) projects that are more extrinsic-collateral intensive are likely to be hit harder by liquidity crunch on financial assets. This conjecture is explored in Calvo et al. (2012) and in chapter 6 under the assumption that capital accumulation projects are less extrinsic-collateral intensive than projects related to the labor market. An implication is that liquidity crunch hits the labor market by either increasing the rate of unemployment or depressing real wages relative to pre-crisis levels, even if output recovers its pre-crisis level.
Collateral distortions can also impinge on growth. Reis (2013), for instance, argues that Portugal’s dismal growth since the beginning of the twenty-first century can be partly explained by the existence of collateral constraints and other financial frictions that apparently helped channel the increase in external credit associated with the adoption of the euro to relatively unproductive sectors. His model is very interesting because it can rationalize a case in which the negative growth effects of capital inflows are so strong that, contrary to most capital inflow episodes in EMs, even a large surge of capital inflows fails to give rise to consumption boom.
Policy Challenges under Liquidity Deflation, and Other Forms of Liquidity Mischief
A natural implication of the PTM is that there may be cases in which an increase in money supply may fail to generate higher liquidity, limiting the effectiveness of monetary policy during periods of liquidity shortage. Below I discuss this topic under the title “liquidity deflation,” and argue that it offers a novel explanation for liquidity trap. The section continues with a discussion of shadow banking, and the effects of low international interest rates. These topics have “liquidity” as a common thread and are linked to dismal outcomes. (p.38) Hence, paraphrasing Friedman (1992), this chapter’s examples can be viewed as different manifestations of liquidity mischief.
Liquidity depends on market characteristics. The PTM, for instance, stresses the role of sticky prices cum demand-determined output as a factor behind the liquidity of fiat money and helps explain why there are cases in which fiat money’s output price does not suffer a major collapse. However, the output backup is limited; it depends not only on sticky prices but also on suppliers’ readiness to expand output as demand goes up. Suppose, for the sake of the argument, that money’s output backup has an upper bound, Q. Therefore, if real monetary balances exceed Q, the excess has no output backup and would be subject to liquidity crunch. This is an important observation because it shows that under liquidity crunch conditions, price level deflation may not help increase real liquidity, even though it increases the stock of real monetary balances. Thus, under severe liquidity shortage conditions, Pigou’s real-balance effect may stop working, even though nominal prices are perfectly downward flexible. Likewise Pigou’s effect may not work if money supply increases when real monetary balances have reached Q, even though nominal prices are upward inflexible. In both examples money’s liquidity per unit of money supply falls if real monetary balances are pushed beyond Q. Mainstream models in which money’s liquidity is taken for granted and is invariant to its real value miss this feature. I will call this effect liquidity deflation (LD) because the concept is a close relative of debt deflation (see Fischer 1933, and discussion below).
LD is a relatively novel concept, and I am afraid it runs the risk of being misunderstood. A banking example may (p.39) help intuition. Consider the T-account for a banking system with deposits on the liability side (which I will identify with money) and land on the asset side. The nominal value of land is assumed to be proportional to the price level. Under these conditions a fall in the price level increases real monetary balances (i.e., the real value of bank deposits), but it provokes a capital loss for the banking system because land prices fall. Thus the increase in real monetary balances does not necessarily translate into higher liquidity, since the fall in the price level may increase the chances of a bank run, making deposits less safe. Notice, incidentally, that LD in this example could be modeled as a pecuniary externality for individual banks and households because, in an atomistic highly competitive environment, LD would be invariant to changes in the demand for money of individual agents. In the example, LD is only triggered by changes in the aggregate stock of real monetary balances.
The relevance of LD can be illustrated by reference to the equilibrium condition in the money market (i.e., the LM curve). Consider the following equilibrium equation:
where M, P, i, and Y stand for money, the price level, nominal interest rate, and output, respectively. Moreover me stands for real money balance, M/P at equilibrium. Function L is the demand for liquidity, and Z stands for the new money liquidity component. Except for Z, the equation above is a standard money market equilibrium condition in textbook models.
Given that, at equilibrium, me = M/P, across equilibrium points the equation above can be expressed as follows:
(p.40) To capture the LD effect, I assume that the new money liquidity component declines with the stock of equilibrium real monetary balance. Formally, I assume Z ′< 0. Hence the left-hand side of this equation has an ambiguous sign with respect to real monetary balances. The case stressed in the verbal discussion above corresponds to a situation in which the positive and negative effects with respect to M/P offset each other. This is an interesting special case because it would be observationally equivalent to liquidity trap as usually depicted in textbook models. In textbook models, liquidity trap corresponds to a situation in which the demand for money becomes infinitely elastic with respect to the interest rate i. If the left-hand side of the equation is invariant with respect to M/P, pumping in more money does not change the supply of liquidity, and therefore equilibrium will hold with constant i.
The LD phenomenon is not only applicable to money. It also applies to all types of bonds denominated in terms of cash that are not fully indexed to the price level. The rise in the price of US Treasury bills, for example, may find increasing difficulty in generating greater overall liquidity because, as suggested by LD, their market value depends more heavily on extrinsic liquidity considerations than on standard ‘fundamentals’, including the factors associated with the PTM. This is important because it shows that liquidity shortage cannot necessarily be alleviated by a fall in the price level or the rise in the price of ultra-safe assets (i.e., assets whose payoff is fully guaranteed in terms of cash). This observation sheds further light on the current literature on safe asset shortage, and casts some doubts on the effectiveness of trying to relieve such shortage by expanding the supply of public debt in reserve-currency economies (e.g., see Caballero and Farhi 2015). In fact LD helps rationalize (p.41) the point of view that is becoming increasingly popular these days—given the tepid recovery of the global economy after the subprime crisis—that QE is perilously losing its “punch.”
Interestingly, while QE may lose its punch, the central bank interest rate may still keep its own. I will denote the central bank interest rate by im, which, in the present context, I will identify as the interest rate on money.17 Hence the opportunity cost of holding money boils down to i − im, and the equation above becomes
Here, even though the left-hand side of the equation is invariant with respect to M/P, changing the central bank’s interest rate, im, can still change the position of the LM curve (i.e., the set of (Y, i) that clear the money market) and have an impact on output and employment. The effect will likely be magnified if the private sector perceives that the central bank does not realize that in a liquidity crisis, with the above-mentioned characteristics, interest rate policy could be significantly more effective than control of monetary aggregates.
Clearly, there are strong parallels between LD and Irving Fisher’s debt deflation. Debt deflation arises when bonds or debt are denominated in terms of money/cash and prices fall precipitously, resulting in a sizable increase in the stock of real debt (i.e., debt in terms of output) – a situation that reached its climax during the Great Depression (see Fischer (1933)). During the Great Depression wholesale prices fell by about 30 percent, which catapulted the real or output value of debt to levels that wreaked havoc in the financial sector by generating unplanned overindebtedness, and (p.42) consequently tended to paralyze credit market flows (a phenomenon akin to Sudden Stop). LD is a related phenomenon because our discussion above was also motivated by price deflation, but instead of zeroing in on solvency, as in Fischer (1933), it focused on liquidity. The question of solvency is not absent because, as prices fall, the output backup of money becomes smaller relative to the stock of real monetary balances. Hence the value of money becomes more dependent on its extrinsic liquidity, which may put a limit on the ability of price deflation to generate higher liquidity.
Shadow Banking and International Liquidity Issues
Liquid assets come in different forms and shapes. Cash is the ultimate liquid asset but, as a rule, its nominal (cash) rate of return is nil. Banks are allowed to pay interest on bank deposits, making them strong competitors vis-à-vis cash. However, bank deposits are not safe assets for large clients, like pension funds, because deposit insurance is low relative to their deposits. This poses a problem for banks. Large depositors are keen on holding low-risk deposits because those funds are employed as means of exchange. Absent safe liquid assets that offer attractive rates of return, the lack of deposit insurance would thus call for banks to pile up the asset side of their balance sheets with cash or public debt instruments like Treasury bills.18 This situation hardly allows banks to offer deals that attract large depositors. Actually large depositors could be better off bypassing banks altogether, and purchasing Treasury bills on their own. This situation became even more challenging for regular banks in the early 2000s, as the Federal Funds rate fell sharply and stayed low for a considerable period of time in response to the dotcom crisis. Arguably, these conditions (p.43) gave incentives for the creation of Shadow Banks. These banks were able to collateralize their liabilities by investing in ABS.19 By design (i.e., securitization), ABS are safer than its individual components. Moreover, investing in the upper tranches of these assets, that is, standing first in line for repayment in case of default, can enhance their safety. In addition large deposits are in the form of overnight repo agreements, which further increases the sense of safety since banks would not necessarily go bankrupt if they are subject to a bank run. Under those circumstances the bank still honors the deposit contract by handing over to depositors the assets involved in the repo agreement. There were well-run institutions in place that ensured a low-cost mechanism for the resolution of these claims. Hence repos helped lower bankruptcy costs and ensure quick settlement if shadow banks ran into payments’ difficulties, enhancing the liquidity of deposits in shadow banks. Moreover the stability of such systems has been further enhanced in depositors’ eyes by the fact that these banks invest in assets that are actively traded in the bond market, and this could have led to the belief that their prices were unlikely to fall in response to problems in individual banks. Another factor pushing in the same direction was the very surge of shadow banking, giving rise to what I like to call “inverse bank run” (see below), a momentum effect. Finally, the apparent safety of shadow banks could have been reinforced by the popular belief that in case of financial turmoil the Fed would prevent a meltdown of ABS prices (a situation commonly known as the “Fed put”), an instance of “moral hazard.” The latter was likely enhanced by the fact that rating agencies got paid by the very ABS issuing financial institutions (e.g., investment banks), another instance of ‘moral hazard’. All of these considerations support the view that financial innovation (p.44) and regulatory agencies policy were major factors behind the significant growth of Shadow Banks.
Impressive as shadow-banking architecture looked prior to crisis, though, these arrangements were not free from a “bad equilibrium.” Without a lender of last resort, shadow banks are unable to shield depositors from the consequences of a systemic shock in which most of them simultaneously try to withdraw their deposits. This is so despite the fact that, as noted above, these banks need not go bankrupt. They comply with the contract by simply handing over ABS instead of cash. But depositors end up bearing a capital loss as they try to liquidate their ABS holdings. True, on impact wealth redistribution across big players takes place. This needs not have dire output and employment consequences. However, the turmoil involving shadow banks may generate distrust about these institutions, resulting in a credit Sudden Stop (i.e., a large and largely unanticipated fall in credit flows). At this juncture Wall Street hits Main Street, causing deleterious outcomes in output and employment. This is in line with events during the 2008 Lehman crisis (see Gorton and Metrick 2012). Chapter 3 will show a simple model bearing out some of these phenomena.
Financial engineering is not the only mechanism for creating liquidity. Higher asset turnover could be enough, as is well documented in the specialized literature (see Foucault et al. 2013). For example, a surge of capital inflows toward EMs could increase EM bonds’ liquidity by enlarging their markets, especially during the first stages of a capital-inflow episode in which, perhaps due to heterogeneous investors’ expectations, investors do not jump in all at the same time. Under these circumstances, a phenomenon that I call “inverse bank run” (see Calvo 2014a) may take place. To illustrate this phenomenon, suppose new investors (p.45) enter the EM in a staggered fashion, the best-informed go first, and so on. Thus, before reaching the end of the line, each new entrant knows that there will likely be investors coming after her who are ready to bid for the securities she holds, enhancing the EM bonds’ extrinsic liquidity. Better informed agents will invest in EM securities, not only on account of those securities’ intrinsic rates of return (including expected capital gains) but because the queue behind them makes them to (rationally) expect that it will be easier to liquidate these assets in case informed investors must make, say, an unanticipated cash payment. This increases the incentives of the better informed to purchase the new EM bonds, giving rise to what might look like a rush of depositors eager to cash in their deposits during a bank run—except that in this instance “depositors” are eager to deposit, not withdraw cash from the bank (hence the qualifier adjective “inverse”).
Furthermore capital inflows may be stimulated by sizable stocks of international reserves held at the central bank since, as pointed out above, it is reasonable to expect that the latter increases the output anchor of domestic debt instruments. This could conceivably lead to a complex liquidity cycle. For example, a central bank faced with capital inflows triggered by external factors (e.g., QE in DMs) may accumulate international reserves to shield the economy from a potentially larger capital outflow episode. The latter enhances the liquidity of the EM liabilities, further increasing capital inflows—and inducing the central bank to accumulate more reserves, and so on. This may converge to an inefficient equilibrium, which is not necessarily free from Sudden Stop of credit flows – because international reserves are implicitly held to protect the stock of bank liabilities.
The Price Theory of Money: A Rallying Cry against “Money Causes Prices?
The PTM and extensions discussed here may sound like a rallying cry against the conventional view that “money causes prices” or, recalling Friedman’s dictum, “inflation is always and everywhere a monetary phenomenon.” But this could not be farthest from the truth. PTM is not saying, “Friedman is dead!” PTM is a theory that helps to explain why cash can survive even under circumstances in which other important liquid assets, such as bank deposits in the 1930s or ABS recently, suffer a major shakeup. The resilience of cash is traced to the fact that a critical number of wages and prices are largely predetermined (sticky) in the short run. This assumption was reasonable at the time of the General Theory, because dollar/sterling inflation was not a major issue. The same holds true nowadays in DMs. However, this does not imply that inflation is no longer a relevant consideration. Woodford’s (2003) textbook, for instance, shows models displaying staggered prices in which a more lax monetary policy can give rise to higher inflation, in line with Friedman’s dictum.
However, inflation may debilitate cash’s output anchor. This phenomenon becomes evident during high-inflation episodes. A common market response to high inflation is shortening the duration of price quotations or linking the latter to, say, the exchange rate or some average price level index like the CPI. In the limit, indexation may largely remove cash’s role as a mean of exchange (and hence its extrinsic liquidity) by, for example, replacing the local currency by the dollar, a common phenomenon in EMs, and called Currency Substitution (e.g., see Calvo 1996, pt. III).20
(p.47) Nevertheless, high inflation does not eliminate the relevance of the PTM. Even during the record-breaking interwar European hyperinflation episodes, domestic monies kept circulating until they were removed by monetary reform. The realm of these rapidly depreciating units of accounts shrunk but wages continued to be defrayed in terms domestic money. Wages were frequently updated, but some degree of price stickiness prevailed, which might have given incentives to quote prices in terms of domestic money and put in motion the output backup mechanism associated with the PTM.
A Digression on the Keynes/Minsky Discussion of Liquidity Issues
The claim that liquidity is a fundamental factor behind deep financial crises can be traced back to the insightful discussion of liquidity issues in Keynes (1936). Actually the present chapter stands on two golden pillars in the General Theory: the Price Theory of Money (Keynes 1936, ch.17) and the Beauty Contest (Keynes (1936, ch. 12), the latter being an example of equilibrium multiplicity. As argued, these two pillars help rationalize the existence of a relatively robust unit of account, while some assets denominated in that unit of account suffer liquidity crunch. Curiously, though, neither the General Theory, nor Minsky (2008a, b), who followed on Keynes’s footsteps, employ the two pillars to develop a simple and appealing first-approximation rationalization of deep financial crisis and liquidity trap. Keynes (1936), for instance, discusses the Beauty Contest in the context of longterm expectations, and although there are some references to equilibrium multiplicity in chapters 17 and 19, the latter is not given a central role there. Instead, Keynes/Minsky discussions wandered into complex territory involving (p.48) informational imperfections and pecuniary externalities. This leads me to wonder if analytical complexity plus the long tranquil macroeconomic environment that prevailed after WWII in developed economies conspired to throw liquidity issues into the trash basket, despite Minsky’s later strenuous efforts to refloat them. And, as a result, nourished the twentieth-century naive view (with the benefit of hindsight) that, say, automatic stabilizers and prudent monetary policy are all one needs to defeat the business cycle, particularly deep recession.
A Semantic Note
To some extent, liquidity crunch does involve some sort of “liquidity illusion”—but not the “money illusion” that so worried our ancestors (e.g., see Fisher 2012). Money illusion refers to wrongly assessing the real value of money or assets denominated in money, as a consequence of employing a wrong price level as a deflator, namely, by selecting the wrong denominator. In contrast, liquidity illusion is an error in assessing the real value of assets but, in this instance, the error lies in not taking into account potential liquidity crunch, implying that the error is in evaluating the resilience of the numerator.
Summary and Implications
1. The liquidity approach helps to rationalize phenomena that would look highly disconcerting if liquidity considerations are not brought to bear. A prominent example is the resilience of the US dollar relative to other liquid assets during the subprime crisis, while other liquid dollar-denominated assets suffered a phenomenal liquidity crunch. Building on Keynes’s Price Theory of Money, I claim (p.49) that this has much to do with the dollar’s prominent role as a global unit of account, linked to nominal stickiness in labor and product markets. Another example is global Sudden Stops. Explanations of the latter that do not rely on liquidity considerations would likely call for a radical revision of the dominant macroeconomic paradigm.
2. A liquidity crunch has effects on relative prices. The nature of these effects depends on which assets suffer the crunch.
3. A natural implication of the PTM is that there may be cases in which an increase in money supply may fail to generate higher liquidity, limiting the effectiveness of monetary policy during periods of liquidity shortage.
4. Moreover, under liquidity deflation, the Pigou effect may stop working. Thus full employment may not be achievable by (1) price deflation, even absent Fisher (1933) debt deflation phenomenon; or (2) expanding money supply, even though prices are sticky.
5. Financial stability is no guarantee against liquidity crunch. The stability of a given arrangement could lead to the false conclusion that risks of financial turmoil are over. This and the previous chapter have argued that extrinsic liquidity may be subject to self-fulfilling shocks that not only harm financial instruments but also end up hurting the real economy. This is not necessarily a consequence of nonrational expectations. Misinformation is enough. The latter is particularly plausible when there is room for multiple equilibriums and the latter are associated with new markets or untested instruments. From this perspective, one could somewhat paradoxically conclude that financial instability may help shield the economy from future liquidity crunch episodes—since crisis teaches something that was not well known before it occurred!
7. Capital inflows can increase the liquidity of EM liabilities (inverse bank run), thereby stimulating further capital inflows. These flows, however, are highly sensitive to a rise in DM policy interest rates. EM monies have weak real anchors. Instead of liquidity trap, they are likely to experience sharp currency devaluation during liquidity crunch. Moreover the exchange rate could become highly volatile even under normal conditions.
8. The liquidity approach does not offer easy solutions for policy makers but should open their eyes to risks that were largely ignored by mainstream macroeconomics until the subprime crisis—and are unlikely to go away in the near future.
This chapter has focused on understanding financial crises. But one should not dismiss the positive role of a well output-anchored fiat money or equivalent asset in terms of which prices and wages are quoted. A lack of this kind of instrument might seriously impair the functioning of trade and credit markets. Thus, in a way, liquidity is both a blessing and a curse of modern capitalist economies.
(1.) I am deeply indebted to Sara Calvo, Enrique Mendoza, and Carmen Reinhart for their insightful and comprehensive comments. However, I am solely responsible for errors and opinions.
(2.) Hicks’s quotation at the beginning of this chapter, though, shows that our predecessors’ best minds were keenly aware of the possibly serious fragilities underlying a monetary economy. Unfortunately, these issues have proved to be too subtle for textbook writers, who have left their readers blissfully ignorant about major liquidity disruptions until the Lehman episode.
(3.) Models where money is just a veil over the real sector offer a deep insight about the nature of “money.” However, as will become apparent in the ensuing discussion, this insight applies to a special kind of money under special circumstances, and may have blindsided economists about the vulnerabilities associated with ‘money’ in actual monetary economies.
(4.) “Basket of goods” is an avowedly imprecise expression. It is intended to convey the notion that the pledgeable share of an asset can be readily employed to acquire a wide variety of goods or assets.
(6.) This would not hold true if the central bank pegged the exchange rate to another currency, a usual practice in EMs (see Calvo and Reinhart 2002). But, even in that case, the central bank’s obligation is typically to exchange domestic currency for foreign currency, which does not take domestic currency outside the extrinsically liquid assets’ circle.
(7.) This statement may be questioned because the Fed could resort to the Treasury for additional resources. However, this is doubtful in the case of the US because public debt and fiscal deficit are large, especially if one takes into account the present discounted value of unfunded liabilities (e.g., Medicare).
(8.) Actually Hahn (1965) shows that, in standard general equilibrium models, barter equilibriums, where money’s output value is nil, cannot be ruled out. This poses a fundamental challenge to monetary theory, which macroeconomists have thoroughly disregarded (see Calvo 2012a).
(p.204) (9.) One is tempted to quip that perhaps “In God We Trust,” written on the back of US dollar bills, should be modified into “In God and Sticky Prices We Trust.”
(10.) This is partly due to the fact that despite the existence of new instruments to palliate the consequences of liquidity crunch (e.g., the IMF Flexible Credit Line), the rules adopted by the IFIs are unclear, and EM policy makers are afraid to fall prey of the IFI officer in charge.
(11.) The strength of DM currencies has permitted DMs to utilize monetary policy as a fiscal revenue instrument. For example, the Fed has accumulated large amounts of “toxic assets,” while the European Central Bank is starting to buy public debt from peripheral members of the euro zone, whose “ability to pay” is highly debatable.
(12.) The role of credit in exacerbating the impact of financial crisis on the real sector has been recently been emphasized by Mendoza (2010), Brunnermeier and Schnabel (2015), and Jordà et al. (2015).
(13.) Notice that in the United States, price level deflation was staved off, but it could be argued that debt deflation was not prevented because monetary policy was not strong enough to keep nominal real estate prices from collapsing.
(14.) This outcome can easily be shown in the context of the IS/LM model with predetermined prices.
(16.) These concepts are akin to intrinsic and extrinsic liquidity discussed above, but they are not identical. I hope this causes no confusion. I keep this terminology because we used it in an already published paper (see Calvo, Coricelli, and Ottonello 2013; see also chapter 6).
(17.) This way of formalizing central bank interest rates was originally suggested in Calvo and Végh (1995). The present version would strictly hold if, say, Treasury bills are the only form of money and the central bank sets its interest rate, here denoted im. This will be further discussed in chapter 3.
(19.) A question that I do not think has yet been fully answered is why, relative to the subprime, the dot-com crisis had minor and short-lived deleterious effects. My conjecture is that shadow banks were less important in 2001 than in 2007. However, the dot-com crisis could have set the ground for the 2008 blowout by inducing the Fed to implement low-interest rates for a (p.205) considerable period of time, which might have led to a strong surge of shadow banking (see Calvo 2012b and chapter 6 below). Another conjecture is that the real estate boom was financed by credit while the dot-com bubble was mostly equity financed; see Turner (2015).
(20.) During the hyperinflation episode in Argentina in 1989, price indexation and high volatility were rampant. As a result many stores closed and posted intriguing notices reading “closed for lack of prices.” They should have said “lack of externally liquid assets.” Even the dollar failed as a unit of account!