Unfortunately, this post comes a little early, because I was hoping to review some of the more recent literature on the zero lower bound (ZLB) first. But, Simon Wren-Lewis’ recent piece on New Keynezian ZLB models pretty much covers a point I’ve been thinking about, and it prompts my skepticism of fiscal policy. In this post I want to summarize why fiscal stimulus is popular at the ZLB, and I want to provide some thoughts on why we should think twice about it.
A resurgent theme over these past years is that fiscal policy is really just monetary policy, with the added role of industrial policy (see, for example, Lars Christensen’s “There is No Such Thing as Fiscal Policy“). This viewpoint isn’t new; old monetarist Clark Warburton, for example, made this point in the 1940s. The question this leads us to is, given the inefficiencies of industrial policy (central planning), why opt for fiscal policy at all?
Most economists see the trade cycle as being at least aggravated by a shortage of money, and the traditional response is to increase the money supply (or, alternatively, increase velocity). But, given existing monetary institutions, how does the central bank increase the money supply? The transmission mechanism most economists have in mind is to use the central bank to target a lower interest rate, with the hope of inducing a greater volume of lending. The central bank can also target some inflation rate, the idea being to tax the demand for money, but, still, to actually cause this inflation the central bank has increase the supply of money. It can do so in a number of ways, but especially open market purchases (exchanging cash for some asset, usually a treasury bill). The exchange takes place with a bank, and the bank can use the more liquid liquid asset (cash) to support more lending.
At the zero lower bound, this traditional transmission mechanism may not work. The definition of the zero lower bound is provided by Hicks (1937), “If the costs of holding money can be neglected, it will always be profitable to hold money rather than lend it out, if the rate of interest is not greater than zero” (pp. 154–155). To get a better idea of what this means, assume that holding money is costless and that money earns a zero percent return. If the loanable funds rate of interest is zero, holding cash becomes a perfect substitute. Holding cash actually becomes preferable, because of the risk premium attached to any loan and the liquidity premium attached to preferring more liquidity over greater liquidity, all else equal. Keynes (1936) argues, following the logic of the liquidity premium, that the lower bound is actually closer to somewhere near two percent. But, if we consider that, in an environment where inflation is positive, the rate of return on cash is negative, the rate of interest on alternative assets to induce a substitution can fall. (Which is one reason why some economists, see Greg Hill and Miles Kimball, support measures that essentially act as a tax on money.)
If the rate of interest falls to, or below, the ZLB there will be a “liquidity trap,” where banks prefer to hold cash, or other safe assets (e.g. treasuries), over making loans. If banks aren’t lending, the traditional transmission mechanism for monetary policy no longer works, and we have to think of alternatives. This is where fiscal policy comes in. A shortage of money implies that there are exchanges that should be taking place, but aren’t. If a central bank can’t eliminate this shortage, then you can conduct these exchanges through collective action. This is a long-winded way of saying that fiscal policy helps maintain nominal income (demand).
A great example of an economist who makes this connection clearly, if not explicitly, is Richard Koo, in The Holy Grail of Macroeconomics. Koo is mainly interested in explaining two events: the Great Depression and Japan’s “lost decade.” His theory is that during an asset price boom firms accumulate these, and when the boom ends these asset prices pop and lose value. Subsequently, firms have an asymmetry in their balance sheet and are forced to increase their demand for money to pay down their debt. Koo’s narrative focuses on the non-financial firm, such that the demand for credit is low — this helps to explain low interest rates. But, in the U.S., it was the financial firms who suffered a balance sheet recession (which, for credit, is supply-side relevant, so we need a different explanation for interest rates). Because of this rise in the demand for money, nominal income falls and there’s a recession. Koo advocates fiscal spending for the sake of maintaining this nominal income: he’s using fiscal policy in lieu of effective monetary policy (Koo believes that there’s no effective monetary policy during a balance sheet recession).
But, the fact that fiscal policy includes industrial policy makes it very different from straightforward monetary stimulus, and this fact should make economists very wary. First, we need to define what an ideal policy would look like, the simplest definition being a neutral policy. A neutral policy is one that mimics the allocation of goods that would exist in a perfectly competitive, information superabundant, frictionless world of barter — an allocation that is symmetric with the web of society’s members’ value scales. The best way to interpret this is to start in a world of pure barter. Barter is imperfect, because information-related transaction costs may indirect exchange difficult. To improve our ability to exchange one another, society developed certain goods (namely, money), organizations, and institutions. Assume that a free banking system would be able to provide a perfect counter-cyclical monetary policy; sans free banking, then, an ideal monetary would mimic free banking’s result.
With collective policy in general, you don’t need industrial policy to make it non-neutral. For example, if a central bank induces an excess supply of money, the result is capital consumption (whether by increasing consumption or malinvestment). But, industrial policy renders all these other considerations almost superfluous. It essentially replaces the market as the means of allocation, giving up the market’s various institutions that help maximize the efficiency of the allocational process. A few big examples,
- Profit and loss: The real world isn’t one of profitless competitive equilibrium, but a trial-and-error process that includes a lot of miscalculation and human error. Prices have an ex post disciplining role, where movements in prices determine, in turn, firms’ revenue. In a dynamic market, the disciplining process of profit and loss works continuously to reallocate capital to those who perform well, and take it away from those who perform poorly. Most government programs aren’t supposed to make a profit, because otherwise they would be provided for by the market — these are cases of market failure. As a result, profit and loss’ disciplining constraint doesn’t bind collective action as much as it does private action;
- Competition: I borrow this argument from Friedman and Kraus (2011) (a book on the financial crisis that everyone should read; pp. 138–140). The market rewards heterogeneity. In markets, there are a large number of different investment strategies, even within the same industries. Where there’s competition and a variety of different “trials,” the trial-and-error process operates more smoothly, and the disciplining process of profit and loss helps push those who do relatively poorly to copy those who do relatively well. Friedman and Kraus use regulation, a somewhat different focus than mine, to show where government doesn’t have the same benefit. A regulation is usually applied industry-wide, and if it’s a bad regulation it will affect the entire industry. We lose the advantage of there being heterogeneous approaches, from which we can pick the best one. This post is especially relevant when you consider that humans are fallible beings, and so any any set of allocational institutions should work towards minimizing the impact of our shortcomings;
- Production stage: One of the most important topics in economics is the allocation of inputs, and it matters whether an input goes towards the production of a consumers’ good or if it goes towards the production of some other input (e.g. like steel being used in the construction of a ship). Hayek (1939) argues that government policy that stimulates consumption runs the risk of changing revenue streams in such a way that it forces people who were skilled laborers in some early stage of production to seek work in the final stage of consumption. If there is a skills mismatch, there is a period of friction where this person is essentially unemployable. (This is one of Hayek’s arguments that people who invoke Joan Robinson don’t seem to know.) But, Hayek’s argument isn’t the only one. A policy that stimulates consumption must necessarily reduce the amount of inputs being used to produce capital goods, and if consumption is stimulated too much, you can consume capital to the point of reducing the capital stock to a point below where it originally was at the beginning of the boom (this is an extreme example, as well, though).
Most economists recognize these arguments, or at least acknowledge that we should rely on markets more than we should rely on government. At the zero lower bound, though, one could make a public goods argument for fiscal stimulus. But, I just don’t find the public goods argument to be very strong.
If there is a macroeconomic market failure (given certain institutional constraints), like a shortage of money or too high a rate of interest, and fiscal policy is the only solution, the costs of fiscal policy have to be less than the costs of the market failure. This is difficult to measure, and as the spending policy becomes less and less local, and larger, the probability of misallocation increases (one implication, and a question, is whether a Federal stimulus that allocated directly to local governments, who did then the specific investing, would be a superior stimulus). Misallocation matters; it’s not about increasing employment and output for its own sake, but to increase the general volume of trade to return society to some level of wealth. (This is why Peter Boettke continues to stress that economics is still a coordination problem, even at the zero lower bound.)
The net costs (or net benefits) of fiscal policy also has to have to be less than (or greater than) alternative policies. And we know that there are alternative policies, even if we may not agree with all of them. For example, in his influential 1998 paper on Japan’s liquidity trap, Paul Krugman advocates a high inflation target. The idea is to influence peoples’ expectations, inducing a rise in the flow of money (remember, inflation acts as a tax on a zero return asset). Scott Sumner has also been consistently arguing for a monetary policy that is completely independent of movement in short-term interest rates. He looks to target NGDP expectations, and one example of unconventional monetary policy that he looks to is the 1933 re-evaluation of the price of gold. (I don’t find Sumner’s arguments convincing, because he doesn’t talk about the transmission mechanism as much as he should.) The point is that we know that there is an idea that monetary policy can work even at the zero lower bound. My skepticism is so much that I’d claim that any stimulus without industrial policy is better than stimulus with industrial policy (of comparable sizes).
A one paragraph summation of this long post: New Keynesian models justify fiscal expenditure at the zero lower bound, because of the breakdown of the conventional monetary transmission mechanism. But, to justify fiscal spending you also have to commit to an institutional comparison, and it’s hard to see, a priori, how fiscal policy is superior to alternatives which don’t include a non-market allocation of goods (of course, one can argue that certain monetary policies are very non-neutral, as Jeffrey Hummel does in “Ben Bernanke versus Milton Friedman“). Government suffers from institutional deficiencies that make it comparably less efficient than markets, so any policy that includes the highest degree of market participation — balanced with the goal of reducing the market failure as much as possible — ought to be favored. If short-term interest rates don’t have to be important for monetary policy to work, the ZLB argument for fiscal stimulus loses a lot of strength.