In 1803, Jean-Baptiste Say examined, in his book, the evolution of trade between Great Britain and Brazil. He realized that the only way for Brazil to buy British goods was to supply Brazilian goods to Britain in order to obtain the sterling necessary to demand English goods. Say extended this to a single economy in claiming that a demand for a particular set of goods can only be expressed by an equivalent supply of another set. An easier example can make this clearer: how does a shoemaker get hats? He can either make them himself (which he really has no skill or knowledge to do at low cost) or he can simply make more shoes and exchange them for hats. Supply, therefore, "creates" demand.
Almost all of Classical and most Neoclassical theory, is based, to a lesser or greater degree, on this simple, even tautological, assertion. In a simple shoe-hat world, at any one time, three possible situations may arise: (1) there may be a sufficient amount of shoes and hats to satisfy all demand; (2) there may be too many shoes offered on the market - implying too great a demand for hats; and (3) there may be too many hats and not enough shoes. Situations (2) and (3) are situations when markets have not made precise the allocations. However, the essence of Say's Law is that there can never be too much of both shoes and hats. A shoemaker would not make more shoes if he did not desire more hats. Therefore, Say's Law concludes, general gluts cannot exist.
Now, the Classical economists were not so blind that they did not see great amounts of unemployment, or that they thought markets always worked. But they concluded that these were due to excess supplies and demands of particular commodities and not excess supplies (or gluts) of commodities as a whole. As Ricardo notes, "Mistakes can be made, and commodities not suited to demand may be produced - of these there may be a glut" (Ricardo, 1820: 160) and that "it is at all times the bad adaptations of the commodities produced to the wants of mankind which is the specific evil, and not the abundance of commodities. Demand is only limited by the will and power to purchase." (Ricardo, 1820: 161). John Stuart Mill concurs by noting that in these situations, "production is not excessive, but merely ill-assorted" (Mill, 1848: 380).
Ricardo (1817: Ch.21) and Mill (1848: Ch.14) extended this proposition to savings and investment. If one produces more than one consumes, then the surplus is saved and, by definition of terms, invested. No one would produce in excess of consumption needs if one does not have a desire to either exchange it or invest it. Supply, therefore, is demand. This virtually all the Classical economists held to be an irrefutable truth.
All but, of course, Thomas Malthus (1820, 1827) and also the French economist, J.C.L.Simonde de Sismondi (1819). Malthus and de Sismondi believed general gluts could exist and they gave their reasoning as follows. Income is distributed between workers, entrepreneurs and land-owners receiving wage income, profit and land rents. Let us assume the Classical proposition that all wages are consumed and all profits invested but make no stipulation as to what happens to the rents of land-owners. Presumably, landowners will receive a portion of income, but they may choose to consume it or not to consume it.
Let us assume a Ricardian "corn" world, i.e. where only corn is produced and everyone is paid in corn and the only capital is corn (it goes into its own production as seed). Thus: Y = wL + (1+r)M + tT where w is the wage, L labor employed, r is the profit rate, M is capital employed, t the land rent and T the land employed. Thus, wL is the share of output which goes to workers, (1+r)M is the share of output which goes to capitalists and tT the share of output going to the landowner. Let us then define net output, X = Y - M, i.e. total output of corn minus the corn needed as capital implement. Thus, net output can be regarded as that output that is brought to "market", i.e. supplied. Then:
X = wL + rM + tT
so that we obtain wL as the proportion of net output going to labor, rM the proportion going to capitalists and tT the proportion going to landowners.
Let us now assume that tT = 0, i.e. no landowners exist or, if they do, they receive no rent. Then total net output is distributed so:
X = wL + rM
However, since entrepreneurs do not consume, then aggregate demand is composed entirely
of workers' demand. Now, as long as profits are positive, then workers' income is less
than net output, i.e. wL
pcXc = wLc + rpkMc
pkXk = wLk + rpkMk
such that we have net output for consumer (Xc) and capital (Xk) goods respectively. The other terms follow logically (Lc and Lk are the labor employed by the consumer and capital goods industries respectively, Mc and Mk the capital employed and pc and pk are the prices for consumer and capital goods). To reduce the plethora of subscripts, we can normalize for pc, the price of the consumer good, so that pc = 1, so p = pk/pc is the price of the capital good in terms of the consumer good. Thus we get:
Xc = wLc + rpMc
pXk = wLk + rpMk
Now, rp(Mc + Mk) constitutes the total income by capitalists in the consumer goods and capital goods sectors; assuming capitalists save all their income, then it is also equal to total saving and, by extension, total investment demand. pXk, of course, is the supply of capital goods. Thus, for investment demand (savings) to be equal to the supply of capital goods, then:
rp(Mc + Mk) = pXk
or simply, savings is equal to investment. Rewritten:
rp(Mc + Mk) = wLk + rpMk
rpMc = wLk
substituting into the consumer goods equation:
Xc = wLc + wLk = w(Lc + Lk)
i.e. total demand for consumer goods, which is derived from wage income in both sectors w(Lc + Lk) , will equal total supply of consumer goods, Xc. Thus, there is no general glut. This conforms to the simplest form of Say's Law, as stated by Ricardo (1817) and Mill (1848).
How might throwing landowners into the picture affect this? Well, rewriting:
Xc = wLc + rpMc+ tTc
pXk = wLk + rpMk + tTk
Let us suppose, again, that rp(Mc + Mk) represents total investment demand. Then, the savings-investment identity requires that:
rp(Mc + Mk) = wLk + rpMk + tTk
rpMc = wLk + tTk
so, plugging back in the consumer goods sector:
Xc = wLc + wLk + tTc + tTk
= w(Lc + Lk) + t(Tc + Tk)
total worker income w(Lc + Lk), and landowner income t(Tc + Tk) will exhaust the output of consumer goods. This is where Thomas Malthus's (1820, 1827) argument attempts to part way. What he argued is that if a part of landowners' total income t(Tc + Tk) is not consumed, then total consumption will not be enough to fulfill the total supply of consumer goods. In other words, if only a portion q of landowner income is consumed, so that qt(Tc + Tk) is total landowner consumption then:
Xc > w(Lc + Lk) + qt(Tc + Tk)
In other words, there will be a general glut (excess supply of consumer goods) even though the investment-savings identity still holds.
This may seem reasonable until we ask whither the unconsumed portions of rent income? Presumably saved - and hence, invested (Malthus did not challenge the S = I identity). If q of rent income goes for the consumption, then (1-q) of it is saved. Conseqently, landlord savings must be (1-q)t(Tc + Tk) and total economy-saving becomes rp(Mc + Mk) + (1-q)t(Tc + Tk) and the savings-investment identity is:
rp(Mc + Mk) + (1-q)t(Tc + Tk) = wLk + rpMk + tTk
rpMc = wLk + tTk - (1-q)t(Tc + Tk)
Plugging this back into the consumer goods equation:
Xc = wLc + wLk + tTk - (1-q)t(Tc + Tk) + tTc
Xc = w(Lc + Lk) + qt(Tc + Tk)
since qt(Tc + Tk) is defined as landlord consumption, then this condition states that total consumption demand will be equal to total supply of consumer goods. Thus, S = I and there is no excess supply of goods. There is no general glut, Say's Law is reinstated. The Malthus-Sismondi arguments, then, seem to fail to hold.
However, things might change if we add a dynamic twist to this simplistic model. As Thomas Sowell (1972: p.32) notes, Ricardo erroneously transformed all of Malthus's and Sismondi's arguments, as we just have, into long-run comparative statics when they were, in fact, arguing in short-run dynamic terms.
Malthus and Sismondi recognized that investment increases output capacity. Namely, investment in any period increases output capacity in that period which increases the supply of consumer goods by itself. Let us suppose, then, that Xc is consumer good output at period t and Xc + dXc is consumer good output a period ahead. Now, assuming a Robertsonian lag, so consumption at period t+1 is derived from t period's wage and rent income which are in turn derived from that period's output by the standard Ricardian equality Xc = w(Lc + Lk) + qt(Tc + Tk). However, this result obtains because we assumed that investment occurred at time period t. Therefore, capacity has increased and period t+1 output is Xc + dXc so that obviously:
Xc + dXc > w(Lc + Lk) + qt(Tc + Tk)
the present supply of consumer goods exceeds the demand for consumption (which is derived from last period's income). This is the general glut Malthus was talking about. In this situation, unless consumption increases exogenously in period t+1 to meet the extra output, then either prices for consumer goods will fall or the glut will persist (since next period t+2 demand is based on t+1 income).
If nothing else changes, then the glut persists. In this case, entrepreneurs will have little choice but to reduce investment. However, reducing investment demand at period t+1 will reduce labor incomes and rent incomes in the capital goods sector, so wLk and tTk fall. However, this implies that total consumption will fall further so that, at time period t+2, the difference will be the same or even more pronounced. Consumption can lag persistently behind investment. Thus we get secular stagnation.
One way out of this would be to increase consumption exogenously. Malthus argued for q to be jacked up at time t+1 so that landlords consume more. This would lower investment immediately since landlord savings would be depleted. This would, admittedly, lower worker consumption in the capital goods sector - but it would be replaced by higher landlord consumption. Thus, consumption could be brought up to meet output, mopping up dXc, while, at the same, the growth of capacity is reduced by lower savings. The conclusion Malthus' reached, then, is that as a policy, q should be increased concurrently to increases in output capacity so that the possibility of a glut be preempted. With landlords increasing their consumption autonomously as output increases, then everything will be reigned in. Hence Malthus's benevolence towards the landlord class. By pandering to landlords, he argued, they might just be induced to provide the extra consumption to avert secular stagnation.
Ricardo (1820) translated this argument in long-run comparative statics terms - and showed, indeed, that the argument is incorrect as a long run argument. If the above inequality is true in the long run, then payments to factors in the consumer goods sector exceeds demand for consumer goods:
wLc + rpMc + tTc > w(Lc + Lk) + qt(Tc + Tk)
rpMc + tTc > wLk + qtTc + qtTk
But since, by definition S = I, so rpMc = wLk + tTk - (1-q)t(Tc + Tk), then this inequality must be rewritten:
wLk + tTk - (1-q)t(Tc + Tk) + tTc > wLk + qtTc + qtTk
qt(Tc + Tk) > qt(Tc + Tk)
which is a contradiction. Now, Malthus (1820: p.38) believed savings equals investment at all times, then was he wrong? Not really. After all, as we know, in the short run, the term on the left is determined at time t+1 whereas that on the right is determined at time t. Since output is higher at t+1 (from higher capacity) then this is perfectly sensible.
But in the long run, the equality will have to emerge. The economic reasoning for this, as explained by Ricardo, Mill and others, was that any overproduction in the consumer goods sector necessarily implies that profits of entrepreneurs will have to fall (since they are not capable of selling). This will itself reduce saving and thus lead to a fall in investment - reducing capacity accordingly. Alternatively, the price of consumer goods would fall, therefore raising p (recall that p = pk/pc) so that pXk rises and so too does wage and rent incomes from payments in the capital goods sector. This will raise consumption accordingly.
As we see, then, the Malthus-Sismondi story is only really applicable to short-run disequilibria. In the long run, it simply cannot be true. Now, to defend his position, Malthus might have well picked up one of Keynes's famous sayings about the short-run lasting a generation or that "in the long run we are all dead" - and Malthus did indeed come up with something similar:
"The question of a glut is exclusively whether it may be general, as well as particular, and not whether it may be permanent as well as temporary...[The] tendency, in the natural course of things, to cure a glut or scarcity, is no more a proof that such evils have never existed, than the tendency of the healing processes of nature to cure some disorders without assistace from man, is a proof that such disorders never existed."
(T.R. Malthus, 1827: p.62-3)
Such indicates, then, that the protagonists were really speaking at cross purposes. Ricardo was concerned almost exclusively with long run possibilities of "secular stagnation" whereas Malthus seems to be more concerned with the mere possibility of general gluts existing in the short-run and, of course, that being a specific, practical problem that had to be dealt with. Note that this implies that Malthus succeeded in breaking what Becker and Baumol (1952) call "Say's Identity" (that there can be no general gluts at any time) albeit not "Say's Equality" (that in the long run there can be no general gluts).
That this is true has been accepted by most economists. However, what has not been decided was whether Malthus's argument was really about short run frictions (as argued by Sowell (1972)) or rather that he was actually trying to prove, but failed to prove, the existence of long-run secular stagnation (as argued by Patinkin (1956) and Blaug (1962)).
Nevertheless, Malthus (1820) had some clear policy propositions to cure the short-run problem. His most famous was that landlords' consumption be increased to "fill" the glut in demand. Since they do not produce anything, nothing is added to output but their very unproductiveness is actually welcome since it maintains demand for all goods while, at the same time, reduces investment. But should landlords hold back consumption, then we have a general glut. Malthus's identification of the landlord class as beneficial stands in stark contrast to Ricardo's (1817) view of them as parasitical.
Karl Marx's(1910) critique of Malthus started from a position of agreement. Marx's idea of capitalist production, however, is characterized by his concentration on the division of labor and his notion that goods are produced for sale and not for consumption or exchange. In other words, goods are produced simply for the intention of transforming output into money to purchase other commodities. The possibility of a lack of effective demand, therefore, is held only in the possibility that there might be a time lag between the sale of a commodity (the acquisition of money) and the purchase of another (its disbursement). This possibility, also originally crafted by Sismondi (1819), endorsed the idea that the circularity of transactions was not always complete and immediate. If money is held, Marx contended, even if for a little while, there is a breakdown in the exchange process and a general glut can occur.
Marx, like Malthus, also accepted the savings-investment identity but reached a different conclusion. Since investment is part of aggregate demand, circulation does continue in time even if money is held. The drive for accumulation, Marx concluded, will continue unhindered and thus a crisis of the sort Malthus described can never happen and if it did, would be practically inconsequential. What can happen, as Ricardo originally claimed, is that a single good may be oversupplied causing a very temporary and small adjustment of proportions which might seem as a general glut but in fact is not.
Thus, all the Classical economists, except for poor Malthus and Sismondi, therefore were generally in agreement over the validity of Say's Law, at least in the long run. They all also agree on the identity of savings and investment as well as the separability of output and price theory.
Resources on the General Glut Controversy