Comments On Older Neoclassical Theory Versus Recessions

As I recently discussed, although empirical recession models have some attractive features -- particularly being rigorous econometric models -- they leave open the question as to why recessions happen. I am commencing my survey of the theoretical aspects of recessions. Since I am not a particular fan of neoclassical ("mainstream") economics, attempting to survey its contributions is arguably difficult. This piece is largely an abridged summary of Robert E. Hall's discussion of this topic in the context of the earlier 1990s recession in the United States. Obviously, the analysis is decades out of date, but it does appear to cover the difficulties that the core of neoclassical approaches have with recessions.

(Once again, this is a very preliminary draft of a section that will make its way into my book on recessions. Although I favour post-Keynesian analysis, I expect that I will have at least one chapter on the neoclassical approach. The issue is to make the treatment as fair as possible; I will let the reader pursue critiques of neoclassical economics elsewhere. I expect that the bulk of the chapter will focus on how mainstream economists have reacted to the Financial Crisis recession, but this Hall article pretty much covers the problems associated with the modelling strategy.)

What is the Neoclassical Approach?

The term neoclassical theory is a term that only an academic can love. For my purposes, I prefer using it rather than mainstream as it is less vague -- what is considered mainstream has changed over time -- and has less connotations associated with it. Its use may not be appreciated by economists in the neoclassical school, who sometimes argue that they are just doing economics, with no qualifier. The problem with that stance is the well-documented way in which they have airbrushed competing schools of though out of existence. (The documentation is provided in the many thousands of pages of complaints by post-Keynesians.) 

The defining characteristic of the neoclassical approach is the centrality of equilibrium analysis. It may be that there is a discussion of multiple equilibria, or out-of-equilibrium states. Nevertheless, the concept of equilibrium is still the core concept being appealed to. The models appear highly mathematical, and are often labelled as dynamic stochastic general equilibrium (DSGE) models.

The key to the concept is that all goods and services (including labour) are traded simultaneously, with prices determined by supply and demand curves. Equilibrium means that all markets settle at prices so that supply and demand are balanced.*

What is supposed to happen is that prices clear at a level that is Pareto optimal: it is not possible to change the outcome without making someone worse off. Resources -- capital and labour -- are used to full economic capacity.

Hall's Article

Robert E. Hall's article "Macro Theory and the Recession of 1990-1991"** offers a non-mathematical assessment of the issues face by the neoclassical models of the day. He argued that there were eight standard explanations for recessions. (I have re-ordered his list.)

1. Changes in the world economy had a negative effect on growth in the United States via exports.
2. Regulatory changes reduced the intermediation services offered by banks.
3. Government purchases of goods and services declined.
4. Tax rates increased.
5. There was a price shock, and stabilisation policy reduced output to limit inflation.
6. Monetary policy switched to a lower target for nominal growth.
7. There was a negative shock to productivity (technology).
8. There was a spontaneous decline in consumption.

I will discuss these in turn.

Real Economy: External Demand, Credit Crunch

The first explanation -- an external demand shock -- was rejected by Hall, as U.S. exports were rising. More generally, blaming the external sector for a slowdown creates a theoretical chicken-and-egg problem: if this is the explanation, why did the external sector have a slowdown? For some open economies, an external slowdown is a highly plausible explanation for recessions, and this would hold regardless of the economic theory. The United States economy in the modern era is not export-driven, and so this possibility seems unlikely. (One may note that there was an oil price spike as a result of the invasion of Kuwait, which was one way the external sector had an effect on the United States.)

The loss of bank intermediation is a more plausible concern; regulators clamped down belatedly on banks after the extent of the Savings and Loan debacle became clear. This would then cause a credit crunch, hitting investment expenditures. Hall rejects the explanation in the context of mainstream theory (of the time) for two reasons. Firstly, if resources were not allocated to investment, the theory suggests that they should have been allocated to consumption. (I would note that this not what post-Keynesian arguments would suggest.) The second problem was that the effects of bank regulation were spread over time, and should not have caused a sharp decline of activity concentrated in a span of a few months.

It should be noted that the issues around credit crunches were taken much more seriously by the mainstream after the Financial Crisis. New classes of models were proposed to incorporate the operation of the financial system. We will need to look at those proposed models to see how well these effects are captured. (At the time of writing, I was not particularly happy with the treatments of credit effects in the papers that I have read; however, I will need to review the literature again.)

Fiscal Policy

Points 3 and 4 refer to fiscal policy. Hall looked at the data, and argued that even though some fiscal tightening occurred ahead of the recession, the magnitude of fiscal policy changes were dwarfed by the loss of output in the recession. That is, if we want to blame the recession on fiscal policy, the multipliers are implausibly high.

Although blaming the recession of 1991 on fiscal policy seems implausible, this will not always be the case. We only need to look at the austerity-driven depressions in the euro periphery to see examples of recessions induced by fiscal policy.

Although interesting, I will not pursue the question of the effects of fiscal policy in this context. Most mainstream theory as well as post-Keynesian theory suggests that fiscal policy has the same directional effect on the economy (raising taxes or cutting government spending both dampen aggregate growth). In both cases, the multiplier from government spending depends on behavioural parameters, and so we could hope to use econometric methods to get a similar operational rule from both approaches. If we take the post-Keynesian thinking that I am most familiar with -- Modern Monetary Theory (MMT) or Hyman Minsky's work -- there are some subtle theoretical differences between the mainstream and these post-Keynesians. In particular, both Minsky and MMT proposed variants of a Job Guarantee program, which deals with the perceived main defect of standard Keynesian fiscal remedies: the lack of proper targeting of spending. Nevertheless, if we take a standard Keynesian fiscal policy change -- such as  an across-the-board tax cut -- there is not a big operational difference between the expected effects of the policy.

The interesting exception of the convergence of analysis is with respect to the free market wing of mainstream economics (the so-called "Freshwater School"). At the extreme, the argument is that the multiplier on government spending is zero. However, it takes some seriously implausible assumptions to get to that result. It seems that the position is rather useful from the perspective of political economy: fiscal policy is ineffective as a means to stimulate growth, so the government should get out of the way of the private sector. The reality that pretty well every government -- including parties nominally in favour of free market economics -- opened the fiscal spigots in response to the Financial Crisis. This is a sign that the belief that the fiscal multiplier is zero has zero respect among policymakers. 

Classical Neo-Classical Arguments

The final four recession mechanisms are the ones that set neo-classical theory apart from post-Keynesian theory. They are largely tied to the equilibrium modelling technique.

Explanation number 5 -- that there was a price shock and then growth was reduced by stabilisation policy -- is rejected by Hall. Other than the small bump in the aggregate price level that was driven by the oil price spike that resulted after the invasion of Kuwait, the trajectory of the price level was stable ahead of the recession. (Inflation rates fell after the recession.) Furthermore, there was no evidence that there was such a policy response, as interest rates were stable. (One task I that I should undertake is to read the Federal Reserve minutes, which are now available.)

At present, it may be somewhat surprising to not be sure about monetary policy goals. The financial press is dominated by a cycle of Fed speakers who continuously natter on about policy objectives and policy mechanisms. By contrast, central bankers in the early 1990s were more circumspect, and they did not even announce what the target for the fed funds rate was. Monetary operations were mired in opacity, and Monetarists were still roaming in wide numbers, arguing that the central bank acts by determining the money supply.*** 

The next, somewhat related point is the possibility that the central bank switched its target for nominal income growth. A few years ago, there was a fad in favour of nominal GDP targeting. This idea has a long history; Hall cites James A. Meade (1978) and James Tobin (1980) as proponents.**** Is it possible that the central bank switched its target for nominal income growth? Hall argues that this is plainly not correct: both output and the inflation rate for the GDP deflator dropped below trend. (Once again, we could look at the minutes of meetings to test this theory.)

The emphasis on central bank desires is a key difference between the neoclassical and post-Keynesian tradition. In an equilibrium model, all markets have to be cleared simultaneously. This includes the markets in future goods and services. (Think of having futures markets for everything.) This clearing of markets implies that all actors are satisfied with the mix of relative prices. Moreover, the central bank is one of those actors -- it trades in the market between money and bonds/bills, and so critically sets the discount rate between the present period and the future. The neoclassical assumption is that plans for consumption are based on that discount rate, and so the central bank has a lever to determine the path for economic activity. By changing its reaction function, it forces future activity to follow a trajectory that it wishes.

Post-Keynesian theory does not give the central bank this power. At most, the path for nominal interest rates is set by the central bank's reaction function (which is a topic of interest to bond market participants). Decisions in the real economy are driven by a mix of factors, and the nominal interest rate is normally not that important. This means that changing the reaction function of the central bank in a model might change the interest rate term structure (if that term structure is modelled), but it will have only limited effects on economic activity variables. This means that a change of the reaction function cannot instantly change the path of economic variables. In particular, a change in the desired path for inflation cannot cause a recession by itself: the central bank would have to raise its target interest rate to a level that it has sufficient effect on the economy to induce a recession.   

Neoclassical Real Economy Arguments

The final two points are real economy changes that are treated somewhat differently than in the post-Keynesian literature.

The first is the possibility of a productivity slowdown. In Real Business Cycle (RBC) models (which were new and exciting at the time), output variations are largely determined by changes to productivity, or technology. This is an extra parameter that is part of the production function, which influences the amount of output. That is, with the same labour hour and capital inputs, increasing productivity will increase output. Hall argues:

The effects of technology shocks have received much attention in the real-business-cycle models of Finn Kydland and Edward Prescott (1982) and their followers. In the real-business-cycle model, an adverse productivity shock has a more than a proportional effect on real output, because employment falls in response to the shock  (via a decline in the real interest rate). However, the multiplier is nowhere near high enough to explain much of the recession of 1990.

Real Business Cycle models are not particularly popular within the mainstream, they have largely been eclipsed by New Keynesian variants that have price stickiness ("the Calvo fairy") which allows for slightly more realistic outcomes. Nevertheless, they are only a tweaked version of the original RBC framework,. and productivity shocks are a feature of the models. If we expanded Hall's analysis, it is hard to see productivity shocks being a plausible explanation for the rapid downturn in activity during recessions.

The final explanation is a drop in consumption behaviour. Hall states:

There is remarkably little disagreement among schools of macro theory on the principles governing consumption. The life-cycle-permanent-income model of Modigliani and Friedman informs essentially all [emphasis mine - BR] research on consumption: the contentious issues are mainly how much of an adjustment needs to be made for the fact that households cannot borrow against future labor earnings.

(I do not want to delve into the debates about consumption functions, but I think it is safe to say that post-Keynesians have strong reservations about the life-cycle-permanent-income model. It is rather revealing that a published paper will say that all research subscribes to that theory. It is another example how mainstream economists airbrushed heterodox economics out of existence.)

Household consumption obviously fell during the 1990-1 recession, Hall suggests that it explains more than half the decline in output. His analysis, based on regressions, suggests that a shortfall of $18 billion (in inflation-adjusted 1987 dollars) may have come from consumption shifts, which is a "a very small fraction" of the $250 billion (1987 dollars) deviation of output from trend.

To a certain extent, a fall in household consumption is almost a circular explanation of recession: the only way that we can have a recession without declining consumption is to have a rather spectacular drop in exports, investment, or government consumption. For an economy that is not export-led (such as the United States) and in the absence of a sharp government spending cutback, such an outcome would be very implausible in a post-Keynesian model.

The difficulty that older neoclassical models have with explaining consumption shifts is straightforward. Consumption is planned over the agent's lifetimes (which may be infinite...), and gives an optimal allocation with future consumption discounted at a smooth rate. Since behavioural parameters are supposed to be stable, there is no reason for today's optimal infinite horizon consumption plan to be much different than yesterday's plan. New information will shift plans, but what kind of new information? Why is this information change bunched into a short-lived recessionary period, and then typically unwound within a few months?

More recent models have added various bells and whistles to the story (households that are liquidity-constrained, etc.), these are being bolted on to a framework that has a hard time generating recession-like outcomes.

Concluding Remarks

Hall's statements about the limitations of neoclassical models is very familiar to anyone who has slogged through economic squabbling in recent years.

I conclude that established models are unhelpful in understanding this recession, and probably most of its predecessors. There was no outside force that concentrated its effects over a few months in the late summer and fall of 1990, nor was there a coincidence of forces concentrated during that period. Rather, there seems to have been a cascading of negative responses during that time, perhaps set off by Iraq's invasion of Kuwait and the oil-price spike of August 1990. Consumers responded to the negative forces as they would to a permanent decrease in their resources. [Comments on auto purchases, interest rates, investment cut for brevity - BR.] Little of this falls into the type of behavior predicted by neoclassical models.

If we look at the pre-2008 neoclassical models, they performed well during the so-called Great Moderation. If they have any predictive value, that value is demonstrated during the fairly steady growth of an expansion. The assumption of optimal smoothing over infinite horizons is hard to square up with a rapid decline of activity in a matter of months.

Explanations that the price level and nominal GDP follow the whims of monetary policymakers are perhaps plausible when we look at the track record of inflation-targeting central banks largely hitting their target since the early 1990s. (We need to ignore the case of the Bank of Japan missing its inflation target repeatedly, but that is typically explained away by the suggestion that the Bank of Japan doesn't really mean it when they declare they want 2% inflation.) However, that theory is highly question when activity is collapsing when inflation is well-contained, which implies that nobody wants that outcome.

The extreme difficulty that neoclassical models have with recessions explains why I have narrowed my scope to that topic for my book on business cycles. The reality is that the neoclassicals have churned out a deluge of highly-mathematical papers that are neither interesting nor useful. If I kept my scope too broad, the effort required to do an adequate survey of mainstream thought explodes. However, by narrowing the scope to recessions, the chapter on mainstream theory will legitimately be short. The bulk of the chapter will likely be a survey of how the mainstream attempted to get a better grip on reality in response to the Financial Crisis.

Footnotes:

* The mathematical definitions used by neoclassical economists are rather vague, as I discussed in other articles.

**  Macro Theory and the Recession of 1990-1991, Robert E. Hall,  The American Economic Review Vol. 83, No. 2, Papers and Proceedings of the Hundred and Fifth Annual Meeting of the American Economic Association (May, 1993), pp. 275-279

*** They don't. See chapters 10 and 12 of Abolish Money (From Economics)!

**** I believe the novelty in the new approach is that there is a level target for nominal GDP. Although this may be novel theoretically, it would be unfeasible in practice. Missing the target would imply massive required acceleration or deceleration in growth, which would be politically unsustainable.

(c) Brian Romanchuk 2019