"Is Von Neumann Square?" is one of my favorite titles for an article in economics1. This post is about a case in which Von Neumann is more hep2.
Sraffa’s book presents a succession of models in which, after the second chapter, the system of price equations have one degree of freedom. This is usually taken to be a trade-off between wages and the rate of profits. Once the distribution of the surplus product is exogenously specified, prices are determined.
Some, such as Michael Mandler and Paul Samuelson, have criticized Sraffian economics on the basis that this number of degrees of freedom is arbitrary. Cases can arise in which the system of price equations has either more or less than one degree of freedom. This post illustrates a case in which more than one degree of freedom exists.
2.0 The Example
2.1 Technology and Quantity Flows
Consider a very simple economy in which laborers produce corn from seed corn on lands of definite types. Two types of land are available. Assume that this economy has 100 acres of land of each type available. Two Constant-Returns-to-Scale processes are known for producing corn. As shown in Table 1, each process requires inputs of a single type of land, as well as labor and seed corn. The technology is such that the order in which types of land will be rented can be read off directly from the technology. As I have previously pointed out, this is not a general property in long period models analyzing rent. I think this special case property, however, is not what drives the existence of possibly more than one degree of freedom.
|Labor||1 person-year||1 person-year|
|Type I Land||1 acre||0 acre|
|Type II Land||0 acre||1 acre|
|Corn||1/5 bushels||1/4 bushels|
|Outputs||1 bushel corn||1 bushel corn|
Under the assumptions, anywhere from zero to 200 bushels of corn can be produced as gross output in this economy. Assume that the gross output of this economy is 100 bushels of corn. Then cost-minimizing firms will cultivate all of type I land, and all of type II land will lie fallow.
2.2 The Price System
For stationary-state prices, no process can earn pure economic prices. This condition imposes the following inequalities:
(1/5)(1 + r) + ρ1 + w ≥ 1
(1/4)(1 + r) + ρ2 + w ≥ 1
- w is the wage (bushels per person-year), paid at the end of the year
- r is the rate of profits
- ρ1 is the rent (bushels per acre) on type I land, paid at the end of the year
- ρ2 is the rent (bushels per acre) on type II land, paid at the end of the year
Land of a given type can be modeled, in an alternative specification of the technology, as jointly produced at the end of the period from the inputs of labor, seed corn, and that type of land3. As long as less than 200 bushels of corn are produced, at least one type of land will pay no rent:
ρ1 ρ2 = 0
2.2.1 First Special Case
Consider what would happen if the gross output was infinitesimally less. Both types of land would be in excess supply. The rent on both would be zero:
ρ1 = ρ2 = 0The solution in this case is:
0 ≤ r ≤ 4
w = (1/5)(4 - r)Only type I land is cultivated. The number of processes in use is equal to the number of produced commodities, that is produced goods with a positive price. The system of price equations has one degree of freedom.
(1/5)(1 + r) + ρ1 + w = 1
(1/4)(1 + r) + ρ2 + w > 1
2.2.2 Second Special Case
Consider, however, what would happen if the gross output was infinitesimally more. Both types of land would be cultivated. Type I land would not be able to produce all the output quantity needed for the requirements for use, and it would have a positive rent:
ρ1 > 0Type II land would be in excess supply, and it would have a rent of zero.
ρ2 = 0The price system becomes:
(1/5)(1 + r) + ρ1 + w = 1
(1/4)(1 + r) + ρ2 + w = 1The solution is:
0 ≤ r ≤ 3
w = (1/4)(3 - r)
ρ1 = (1/20)(1 + r)Two produced commodities with positive prices exist: corn and the first type of land. And two processes are activated.
2.2.3 The Case With Two Degrees of Freedom
But type II land does not need to be cultivated in the case under consideration. Thus, the costs of cultivating type II land can exceed the revenues, and the rent on Type I land is not determined by the price equations. Only the first equation in the system of equations for prices need obtain.
(1/5)(1 + r) + ρ1 + w = 1The second process is still characterized by an inequality:
(1/4)(1 + r) + ρ2 + w ≥ 1This system has the solution:
0 ≤ r ≤ 4
0 ≤ ρ1 ≤ (1/20)(1 + r)
ρ2 = 0
w = (1/5)(4 - 5ρ1 - r)Figure 1 illustrates one projection of this solution into two dimensions. The lines closer to the origin are drawn for a higher rent on the first type of land.
|Figure 1: Variation in the Wage-Rate of Profits Frontier with Rent|
I am loath to argue that the extra degree of freedom in this example is negligible since it arises only for a knife-edge. If the quantity produced is a hair larger or a hair smaller, the input-output matrices for commodities with positive prices are square. But in a larger model, the quantity produced is a choice variable. I also don't see why Sraffian models must not have more than one degree of freedom.
1 If I’ve actually read this article, it must have been in a reprint in some collection.
2 Some posts take me a while to write. I began this one the day after Arthur Laurents died.
3 I don't here show the derivation of rent from such a model.
- Christian Bidard (1986) "Is Von Neumann Square?" Journal of Economics, V. 46: pp. 407-419.
- Michael Mandler (20xx) "Sraffian Economics (new developments)" New Palgrave, 2nd edition.