Keynesian Explanation of Unemployment: Seriously Incomplete

Previous post on  Lecture 8A covers first 17m of video lecture in  Advanced Macro  — explains how different ways of articulating the micro-foundations for Keynesian Macro Concepts leads to dramatically different Macro Outcome.

This post is Lecture 8B — from 17m to 37m of video lecture linked at bottom of post. It attempts to make sense of the Keynesian explanation of unemployment based on insufficient aggregate demand. It concludes that several elements missing from Keynes must be added to get to a satisfactory explanation.

 Friedman’s Methodology  leads to crazy models: In previous post (  Lecture 8A – Microfoundations for Keynesian Economics ), we showed that even small differences in the micro-foundations could lead to very large differences in the macro outcomes. Depending on how we choose micro-foundations, we can get almost any result we like at the macro level. So the question arises: HOW should we construct our micro-foundations? How can we choose among the wide variety of possible micro-structures. This leads to a very serious methodological issue of what models are and how they relate to reality. On this topic, see my detailed discussion in post on Models and Reality. Briefly, the standard POV adopted in neoclassical textbooks is that the only job of models is the provide a match to observations. The inner details of the models can be arbitrary.  However neoclassical economists insist that a good model must have optimizing behavior by all agents, and the equilibrium outcome of the model should match observations. There is no requirement for models to be realistic.  Because realism or lack of it is not relevant,  Friedman’s methodology   states that significant models will have wildly inaccurate descriptive representations. This methodology leads to ridiculous and absurd models, which failed completely in the Global Financial Crisis (see  Quotes Critical of Economics ). In fact, as we have argued elsewhere, maximization does not describe either firm behavior or consumer behavior. Furthermore, behavior of a complex dynamic system cannot be understood by looking at its equilibria. So the methodological tenets of neoclassical economists are fundamentally wrong.

Critical Realism in Models: In this course, we take the POV that models are simplifications which help us to understand a complex reality. Because of this, realism is a requirement – we cannot allow wildly inaccurate descriptions. Realism has to be understood correctly. We want to strip away inessential elements and focus on simplest descriptions which capture the mechanisms that produce the phenomena of interest. The process of stripping away complexity can be said to create descriptive inaccuracy, but this is of a different type from the kind of absurdities created by conventional methodological stance a la Friedman. For example, when  massive empirical evidence proves that human agents do not maximize utilities , than we cannot made utility maximizing behavior as an explanation of any observed phenomena. Note that we may nonetheless assume utility maximization as a convenient placeholder for human behavior if the phenomenon we wish to explain does not depend very much on exact nature of human behavior. That is, if we believe that replacing utility maximization by many other types of behavior would lead to the same results.

The phenomenon we wish to explain is unemployment. Consider a simple fixed proportion production function where 10 LLs own 5 Acres of land each, and One Acre + One Laborer can produce 10 units of corn. Each LL can hire a maximum of 5 laborers, so if there are more than 50 laborers, the model will generate unemployment. Here we have a very simple model which produces unemployment. This model generates understanding – if we do not have sufficient productive capacity – ways to utilize laborers for production, then excess labor beyond maximum productive capacity will remain unemployed. When we come to the Great Depression, it is clear that this explanation will not suffice. This is because the economy did have full employment prior to the GD 29, and so potential there was capacity in the economy to employ all workers. So the problem is to find a model in which there is unemployment when the economy does have the capacity to hire all the labor. For example, suppose that there are 40 laborers in the economy given above. Then it is clear that all 40 CAN find productive employment. There is land to spare, and each laborer can produce 10 units of corn when allowed to work.

In this situation, there really is a puzzle as to how unemployment could emerge. Something must go wrong to prevent an activity which yield benefit to all participants – laborer would earn wages, and landlord would earn profits, and there are no parties which would be hurt by this additional production. If we can solve this puzzle, we may get some understanding of why unemployment occurs in the real world.

Keynes argued that money is an essential ingredient, and has real effects in the short and long run. He also stated explicitly the sequencing that wages are paid upon hiring laborers, and production takes place later. Let us try to use these two ideas as building blocks for a model where unemployment might occur. Since Keynes used a static one-period framework, we will also work with this same case.

Suppose initially that all laborers are paid PKR 100, which is the going nominal wage rate. All 40 are hired, and production takes place, creating 10 units each for a total of 400 units of corn. This sequencing, where payment is made in nominal terms, before the market for corn opens, has rather startling consequences, which run counter to intuition generated by neoclassical theories of price determination.

Model 1: Suppose a closed economy, and no use for corn other than sale on domestic market. Than all 400 units would be put on the market, while laborers have 4000 PKR as their wage earnings, which is all the money available to buy. So price of corn would be 10 PKR per unit, and every laborer would receive 10 units of corn – Wage equals marginal product of labor. But this is a very strange model because the LLs earn zero profits. Why would they go to the trouble of producing corn, if they earn nothing in the process – this question arises in a Market Economy, where production is done for profits.

Model 2: LLs can consume some of the corn, and put the rest on the market. Suppose each LL consumes 200 units of corn, and puts the remaining 200 units on the market. Now there is a total of 2000 units of corn on the market, while there is 4000 PKR available in wages to purchase the corn. So the price would be 20 PKR per unit of corn, and each laborer would be able to buy 5 units of corn, distinctly less than the Marginal Product of labor, which is 10 units of corn. This is a static equilibrium which can repeat. The landlords do not make any monetary profits, but they get to eat a vast amount of corn.

QUESTION: Why does wage fail to equal marginal product of labor in Model 2? The theory is that there would be competition among the LLs to increase the wage, because each has an extra acre of unutilized land, and if he can attract a laborer, he can make more profits. However, in the initial phase where wages are being paid, competition in nominal wages has no effect on outcome. Regardless of how high the money wage is, the price of corn will be set in exact proportion to get back all of the wage income. This illustrates how wages are set in nominal terms and the real wage is an outcome of the system as a whole, exactly as Keynes said.

Both of these models illustrate one of the basic principles of MMT – A sector cannot make profits unless it receives injections from the outside. All inside transactions cancel in monetary terms.

Model 3: The only way for Landlords to make monetary profits is to sell the surplus corn to some other sector. This could be the government, or it could be an urban sector (or any other sector of domestic economy), or it could be exported to the foreign sector. In all such cases, money which flows into the sector from the outside is the profit to the landlords.   As MMT tells us, the private profits of the agricultural/rural sector exactly offset losses of some other sector. How much can the landlords sell to the urban sector (for example)? According to neoclassical theory, there are technological constraints on the profits of the landlord. Their profit should be equal to the marginal product of Land, the input they supply to the production process. In the fixed proportions production function, the marginal product of 1 Acre of Land is equal to the marginal product of one laborer and both are equal to 10 units of corn. Adding up of the two marginal products to the total product does not hold. In this model, the Landlords can sell arbitrarily large amounts to external sectors, and leave arbitrarily small amounts for the domestic market. Regardless of how much or how little they put on the domestic market, the laborers will spend all their wage-money on purchasing the product.

It seems reasonable to think that landlords will leave subsistence amounts of corn on the domestic market, so as to prevent domestic unrest or revolutions, while marketing the rest for highest profit. Since the amount retained for domestic market is based on political considerations, it need not satisfy equilibrium conditions for profit maximization. That is the domestic price can be less than or higher than the foreign price at which the surplus over subsistence is sold.

So far, there is no unemployment in this model, and unemployment cannot exist unless some amount of produce remains unsold. If this happens, it would give a signal to the landlords that over-production has taken place, and they would make decision to reduce production. But how can produce remain unsold? Let us consider the Keynesian idea that the labors SAVE some portion of their wages for use later. This would reduce the effective demand and perhaps lead to surplus production which could not be sold. To see how this works, suppose the LLs sell 200 units of corn to external sector at some price p, making profits of 200p. The remaining 200 units of corn is put on the domestic market for sale, there is a total of 2000 units for sale, while the laborers have 4000 PKR in total income. If they spend all of their income on purchasing corn, the price of corn would be 20 PKR per unit and every laborer would get 5 units of corn.

Now suppose that the laborers decided to save half of their income for potential contingencies in the future. Saving 2000 PKR leaves them with 2000 PKR to purchase corn. When 2000 units are available for sale, the price will fall from 20 PKR/unit to 10 PKR/unit and all of the corn will be sold. Since all the corn is sold, there is no reason for the Landlords to limit production in the next round, and this situation can repeat.

If there is price rigidity, and the price of corn cannot fall from PKR 20/unit, THEN we have the possibility of surplus. Only 1000 units of corn can now be sold on the domestic market and 1000 units of corn remains as surplus. If this can be sold to some external sector – government, urban sector, foreign sector – then again there is no serious problem. But if there is some capacity limit on outside sales, then the surplus cannot be sold. In this case, the LandLords get the message that what was produced could not be sold, and so they should cut back on production. This can produce unemployment in the next period.

Lessons from this model/analysis

In one-period, static model, shortfall in aggregate demand can only come from fixed prices. This is not true in DYNAMIC Multi-period models, but these require more sophisticated analysis, which Keynes did not make. SAVINGS by laborers DOES create disequilibrium: Money profits to landlords become NEGATIVE. They do not recover in revenue what they paid in wages. However, this should result in declines in nominal wages. What happens depends on disequilibrium dynamics.

The final conclusion from all of these considerations is that the mechanism of insufficiency in aggregate demand which was used by Keynes to explain unemployment has not been sufficiently articulated. Many factors were left unspecified, all of which matter to the outcome. A well-articulated Model must specify the technology of production: Cobb-Douglass, IRTS, DRTS, Leontief. These all make a difference. In addition, market structure and sequencing must be specified. Do we have competition, monopoly, monopsony, posted prices or oral double auction. Who produces, and who markets the product. The Information Structure is also important – the Sequencing of Decisions and information flows, who know what and when?. Relative Bargaining Power, and Social Norms govern the outcomes. Of Vital importance is disequilibrium outcomes and how they are resolved. The Class Structure – Landlords, Laborers, Others, also matters – even in our simple models, consumer behavior of laborers is essentially different from that of the Landlords for structural reasons – Landlords do not consume out of wages they earn. Similarly, if the economy is driven by profits, then some other sector must be involved, because profits can only be earned if money is injected into the system. So a well-specified model must have multiple sectors, such as Domestic Other, External Sector, and Government.

We abandon Keynesian analysis of unemployment, noting that the explanation it provides is seriously complete, and must be augmented by additional factors. In particular, fixed prices does provide an explanation, but this is clearly a short run phenomena and to get a serious explanation along these lines requires consideration of disequilibrium dynamics. Subsequent analyses by many authors show the importance of debt, and in particular leveraged debt, which plays a key role in producing major financial crises. To introduce these ideas, we consider a different mechanism – Fisher’s theory of Debt-Deflation – which was also proposed as an explanation of unemployment and the Great Depression, but did not receive much attention in the mainstream. In recent times, elements of this theory have been picked up by Minsky, Koo, Mian and Sufi and others.

This post (Lecture 8B) covers the 20 minute segments from 17m to 37m of the lecture 8. The first 17m were covered in previous post on  Lecture 8A: Microfoundations Matter 

The remaining portion of the lecture explains Fisher’s Debt-Deflation Theory.

Advertisements

Leave a Reply

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

WordPress.com Logo

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

Google+ photo

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

Twitter picture

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

Facebook photo

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

Connecting to %s

%d bloggers like this: