"Wages, prices, and employment in a Keynesian long-run" by Marglin

New issue of ROKE with papers by Stephen Marglin, Amit Bhaduri, Esteban Pérez (with your truly) among others. Last issue of the several in honor of the 25 years of the Marglin-Bhaduri papers on profit-led/wage-led growth. From the abstract of Marglin's paper:

The central question this paper addresses is the same one I explored in my joint work with Amit Bhaduri 25 years ago: under what circumstances are high wages good for employment? I extend our 1990 argument in three directions. First, instead of mark-up pricing, I model labor and product markets separately. The labor supply to the capitalist sector of the economy is assumed à la Lewis to be unlimited. Consequently the wage cannot be determined endogenously but is fixed by an extended notion of subsistence based on Smith, Ricardo, and Marx. For tractability the product market is assumed to be perfectly competitive. The second innovation is to show how disequilibrium adjustment resolves the overdetermination inherent in the model. There are three equations – aggregate demand, goods supply, and labor supply – but two unknowns – the labor–capital ratio and the real price (the inverse of the real wage). Consequently equilibrium does not even exist until we define the adjustment process. The third innovation is to distinguish capital deepening from capital widening. This is important because, ceteris paribus, wage-led growth is more likely to stimulate the economy the greater the fraction of investment devoted to capital deepening. A final section of the paper shows that US data on employment and inflation since the 1950s are consistent with the theory developed in this paper.

Marglin's paper is open, and so is Bhaduri's. Full issue here.

The Gospel of Co-operative Capitalism: Acolytes and Apostates

Bill McColloch published a new PERI Working Paper, which deals with the same issues, in a very different way, raised in the last chapter of his dissertation. From the abstract:

The present paper seeks to locate the Bhaduri-Marglin (B-M) model as an historical outcome of the Left's internal disputes over the prospects for social democracy. In better  contextualizing the B-M model as a historical response to the perceived political economic failings of the social compromises upon which the growth of post-War advanced capitalist economies had rested, both the model’s popularity and its potential limitations can more easily be understood. Though the B-M framework has frequently come to be referred to as the neo- or post-Kaleckian growth model, such labels perhaps obscure the model's diverse ancestry. The model constituted an attempt to reconcile seemingly incompatible theoretical perspectives, and to highlight those special conditions that made possible a ‘Golden Age’ of social democracy. Moreover, they sought to show that the conclusions of Keynesian social democrats and of radical Marxists could be viewed as two possible outcomes of the same broadly Keynesian theoretical framework in which investment played a leading role. While this synthesis has fostered a vast literature and useful dialogue, it is argued that it should, nevertheless, be seen as the outcome of a generation Left social scientists that had become deeply skeptical of the possibility of egalitarian redistribution under capitalism, and of the political ambitions of Keynesian and social democratic parties.

The paper explains to some extent the political economy of the rise of the profit-led models within radical economics.

Kaldorian and Sraffian supermultipliers: a clarification

This is a post for those interested in demand-led theories of growth. Not long ago I wrote a post on misconceptions about Sraffian economics. Marc Lavoie sent me a nice email about it, and a recent paper he published in Metroeconomica (subscription required), which comments on a paper I wrote with Esteban Pérez (working paper available here). In his discussion of supermultiplier models, which put the multiplier and the accelerator together to explain -- not fluctuations of the level of output around its normal position -- but the determination of trend or normal output. Lavoie says:

"Other post-Keynesians, also assume that non-capacity creating autonomous expenditures are the driving force, rather than investment. Serrano himself refers to Kaldor (1983, p. 9) to provide support for this reversal of causality. Fazzari et al. (2013) assume that there is some unidentified demand component that grows autonomously, in order to tame Harrodian instability; Godley and Lavoie (2007, ch. 11) and, as already pointed out Allain (2015), rely on autonomous government expenditures. Indeed, there is a large Kaldorian literature that relies on exogenous growth components other than business investment, most particularly the whole literature on Thirlwall 's law with its exogenous exports (McCombie and Thirlwall, 1994), as well as Godley and Cripps (1983), with both government expenditure and export sales."

And in a footonte to that passage he says:

"Thus, adding to the confusion over terminology, Pérez-Caldentey and Vernengo (2013) refer to the Kaldorian tradition when discussing models based on induced investment and non-capacity creating exogenous growth components such as Serrano's Sraffian supermultiplier analysis."

So let me clarify our use of Kaldorian, and also why I believe that it is a mistake to refer to the Sraffian supermultiplier as neo-Kaleckian, even though it does have evidently Kaleckian elements. As I understand the distinction that came to dominate demand-led models of growth, there are basically two* main traditions, one that is referred to as neo-Kaleckian, and one that is referred to as Kaldorian.

The first tradition developed from Bob Rowthorn's expansion of Joan Robinson's 1960s model. And because Joan Robinson was influenced by Kalecki, and  Rowthorn, a Marxist author, was seen as Kaleckian, the name stuck. The original model, one must note was wage-led. And causality basically determined whether the authors was Keynesian or Marxist, with Ed Nell famously referring to one author that suggested that causality went from income distribution to growth as Jean Baptiste Marglin. At any rate, Marxist and Keynesian closures, to use the term popularized in this context by Lance Taylor, were special cases of the neo-Kaleckian model. Later developments introduced changes in the independent investment function which allowed for a profit-led closure.

As I noted before, the term Kaleckian is a bit of a misnomer. The current version of the model allows for a profit-led closure, which is not clearly in Kalecki, and, besides its derived from Joan Robinson's model. The Kaleckian feature is that often it is assumed that workers do not save, and capitalists do not consume, for simplification, a classical political economy type of assumption really.**

The genesis of supermultiplier models is more convoluted. On the one hand, the combination of multiplier and accelerator was used to discuss economic cycles, not growth, including by Hicks, who first discussed the idea of the supermultiplier. By the late 1960s, Kaldor moved away from the differential savings or neo-Keynesian growth models (sometimes referred to as Kaldor-Pasinetti or Cambridge growth model), and adopted the supermultiplier model, formalized by Thirlwall in the 1970s. The model assumed as a simplification that exports were the only autonomous component of demand. In accordance with the accelerator, investment was seen as derived demand. That is the main difference with the so-called Neo-Kaleckian models, namely: there is no independent investment function.***

The idea of the supermultiplier was later, in the 1980s and 1990s, developed by Bortis and Serrano,**** both authors sharing a Sraffian perspective. In these versions, autonomous spending was not restricted to exports, and government spending was also relevant. The term Sraffian or classical-Keynesian has been used to describe these models. In essence, they are Kaldorian models, since investment is derived demand, as much as in Thirlwall's model. In this sense, even though the Kaldorian models a la Thirlwall are a special case of the Sraffian supermultiplier, as discussed here in my debate with Jaime Ros (in Spanish), and by definition more general than the export-led growth model, they came later, and can be seen as a development within this tradition.

So certainly the intention is not to create confusion. In my view, models with an independent investment function are broadly speaking neo-Kaleckian, while models in which investment is derived demand are Kaldorian. And there are differences between models within those broadly defined traditions.

* All taxonomies are somewhat arbitrary and one might see some sub-divisions from the two main branches discussed here as standing in the same footing, for example, some might argue for an explicitly Marxist tradition.

** Goodwin predator-prey models, which have become quite fashionable, can be seen as a variation of these neo-Kaleckian models.

*** I think these Cambridge models have been completely abandoned since the 1960s, and that is the reason why I don't have three types of models in my taxonomy. They are a historical curiosity, associated to a response to the Harrod instability problem, at a time when full employment seemed like a stylized fact in advanced capitalist economies. For a clear explanation of the implications of the different model closures see the paper by Franklin Serrano and Fabio Freitas here.

**** The Sraffian versions of the supermultiplier model also assume differential savings by workers and capitalists, as many other classical political economy inspired models, and in that sense have Kaleckain features. But they are not neo-Kaleckian, since there is no independent investment function.

On Marx and Keynes, the NAIRU and Say's Law, again

Many of the entries in this blog directly or indirectly deal with the relation between the old classical authors of the surplus approach and the more radical authors that followed after the Keynesian Revolution (see here, for example).  David Fields has pointed out this post by Michael Roberts on Marx and Keynes, from a Marxist point of view (after a debate with someone he refers to as left post-Keynesian).

This is not a bad post at all. The interpretation of Keynes given by the post-Keynesian author (in Roberts' description) is not the best, but it is one of the possible interpretations for sure. The unknown postie also correctly notes that:

"Keynes and Marx were united in their critique of Say’s law (that supply creates its own demand) as a common starting point for theorising about the possibility of insufficient effective demand and the realisation problem."

Which is important to emphasize, since some Marxists tend to have an attachment to Say's Law. In this respect the postie [again in Robert's description of what he said] is actually unfair to Marx and Ricardo, suggesting that:

"But he [Marx] did (like the other classical political economists) expected the economy to always tend towards full capacity utilisation even if he (like Ricardo) theorised about technological unemployment in an economy operating at full capacity utilisation."

Actually there is no indication that Marx or Ricardo suggested full capacity utilization. Ricardo's version of Say's Law does not require full employment, and there is no mechanism to bring investment to the level of full employment savings. Investment equals savings by definition, and it might be at a level with significant levels of unemployment and spare capacity. Reductions in the real wage or the rate of interest would not bring about full utilization of resources. That's a concept that appears only with Marginalism.

When Roberts comes to his argument, it starts with a discussion of Keynes' political position (saving Capitalism) rather than the analytical elements of his theory. Then he discusses the limits to Keynesian policies, which he narrowly defines as fiscal activism (note that this could have been the position of an Old or New Keynesian, defending fiscal activism on the basis of market imperfections). Roberts misses the radical component in Keynes analysis, which implies that market economies are NOT efficient in the allocation of resources, and that long-term equilibrium could imply unemployment (yes long-term equilibrium unemployment, not disequilibrium, and not short-term unemployment is Keynes' central proposition).

Further, Keynes' Principle of Effective Demand requires to be fully operational the abandonment of the neoclassical (Marginalist) theory of distribution, since variations of the rate of interest (or the real wage) do not lead to full utilization of capacity, and are compatible with a coherent recovery of the old and forgotten theories of the classical authors and Marx. So, complementing the postie, Keynes and Marx are united, not only the rejection of Say's Law, but also on the need for an alternative to the Marginalist theory of distribution, one in which conflict is necessary.

Keynes didn't get to this point, but understood the need of getting rid of the natural rate of interest concept. And that leads to an abandonment of the neoclassical theory of distribution per force.

Then Roberts gets to his main point. He says:

"Keynes says the crisis comes about through a lack of ‘effective demand’, namely an unaccountable fall in investment and consumption and this causes profits and wages to fall. Marx says: let’s start with profits. If profits fall, then capitalists would stop investing, lay off workers and wages would drop and consumption would fall."

This is a variation of the traditional Marxist theory of the business cycle. There are multiple versions, from Andrew Glyn profit squeeze to Goodwin predator prey (which by the way suggests full utilization of resources in his model in Marxo-Marginalist fashion), and many others. All rely, like Roberts above on the notion that the rate of profit (or the profit margin in Marglin and Bhaduri) drives investment.

In this case, high employment generates wage inflation which can increase the wage share of workers in output; but this will, in turn, reduce the profits of capitalists and thus reduce future investment and output, leading to a recession. The recession, in turn, reduces labor demand and employment and consequently leads to lower wage inflation or even deflation and reduces the wage share of workers. But as workers wage share declines, then profits increase and, with them, investment and a new boom starts.

There are a few problems with this view, the profit-led view of investment, discussed here before (and here). As mentioned before, from a theoretical point of view, it is difficult to justify why a firm would invest, even if the rate of profit is high, if demand is not growing. And, in reverse, why would a firm not invest, and increase its capacity, if demand is growing even if the rate of profit is relatively low. Wouldn't it make sense to keep pace with demand, rather than let the competition take advantage of expanding demand? From an empirical point of view (as noted here, and here) the accelerator principle rules the roost.

Finally, note that all of the profit-led based theories of cycles and growth are fundamentally dynamic Say's Law stories, in which savings, determined by profits determine the behavior of investment, and the system fluctuates around a level of output in which worker's bargaining power does not lead to inflation, or in other words stable inflation. That is the so-called Marxist version of the NAIRU (see my previous post on the late Andrew Glynn on the subject; for a relatively recent Marxist defense of the NAIRU see Pollin here; subscription required). As I noted before (and here) there are reasons to be skeptical about the Phillips curve and the NAIRU, in all their versions.

PS: Comment by Franklin Serrano, that I paste in full:

"In Marx there is indeed no indication of a tendency to full capacity utilization nor full employment of labour. But the author is right about Ricardo. Ricardo assumed Say's law which means that all that "demand is only limited by production", this DOES imply that investment is equal to and determined by full capacity savings as in Ricardo it is only capital (full capacity) not labor that determines potential output. Ricardo did NOT contemplate "spare capacity"  and yes his technological unemployment was based on the fully capacity output not generating enough jobs for the economy to get to full employment. You are right that in Ricardo investment does not adapt to full employment savings. But it does adapt if arbitrarily to full capacity saving."

Same issue brought up by PGB below. I conflated full capacity with full employment. Mind you, I suspect that this is what the post-Keynesian author referred in the post was doing, and hence my mistake.

Stephen Marglin on Heterodox Economics

I have lots problems with Marglin's views, not just on heterodox economics, but on the mainstream too. Marginalism (neoclassical economics), the supply and demand story that emerged in the 1870s, is not simply about efficiency. It is about a certain type of efficiency (factors of production are fully utilized). There are different views of efficiency, and certainly that was the case for the authors of the surplus approach (Smith/Ricardo/Marx), for whom efficiency was essentially about capital accumulation, i.e. the wealth of nations. Also, the problem is not that the mainstream ignores income distribution, but that it provides an untenable position on income distribution (to each according to its productive). In contrast, for the surplus approach authors income distribution depends on class conflict.

From my point of view, Marglin is very close to presenting a post-modernist view of critiques of the mainstream, that is, several alternatives to the mainstream are acceptable, as the much as the mainstream is. This is Deidre McCloskey's view that economics is about the rules of conversation, rhetoric. I'm more of a realist, I think there is an external reality independent of my views about it, and that there are, besides logical aspects, facts that allow us to discern between theories that are not correct (the mainstream) and theories that have not been proved wrong (classical-Keynesian). For my view on heterodoxy read this.

PS: Marglin's syllabus here. Thanks to Pedro Américo for the tip.

More on income distribution and growth (wonskish)

Kaldor and Kalecki

A few years back Sam Bowles presented a paper (Kudunomics: Property rights for the information-based economy) at the University of Utah. At dinner he reaffirmed his conviction that Arrow-Debreu General Equilibrium (GE) is compatible with different kinds of behavior and can be a force for progressive economics. Conventional marginalist theory suggests that income distribution is the result of relative scarcities, and, as a result, real wages should equal the marginal product of labor, i.e. labor productivity. When asked how he squares the belief in GE with the fact that wages in the US do NOT follow productivity since the 1970s, Bowles seemed puzzled. And the relation of income distribution and growth remains puzzling for the mainstream and its sycophants.

In the heterodox camp, the discussion has been centered, for the most part, between the so-called Kaleckian and Kaldorian models. First, I should note that from a history of ideas point, the Kaleckian name is a misnomer. Kalecki’s models where about the interaction of multiplier and accelerator, with shocks and lags, to produce fluctuations. In the various forms of his accelerator equation Kalecki included a trend, producing fluctuations around a trend. The so-called Kaleckian models derive from Harrod and Joan Robinson’s attempts to extent Keynes’ Principle of Effective Demand (PED) to the long run.

The PED says that an increase in investment is matched by an exact increase in savings, and that the level of income is the main adjusting variable (rather than the interest rate as in the Loanable Theory of Funds). The Kaleckian models basically normalize the IS identity by the capital stock, assume (in the extreme case) that the propensity to save out of wages is zero, and a propensity to save out of profits (s) between zero and one, and in Keynesian fashion have investment determine savings. The difference with the short-run story is that now accumulation (investment-to-capital ratio) determines income distribution (the rate of profits), a result often referred to as the Cambridge equation.

The various incarnations of the Kaleckian models are defined by the way the investment function is specified. For example, in the influential paper by Bhaduri and Marglin (B-M) (subscription required) they argue that investment and savings are functions of the profit share (h) and capacity utilization (z). In other words:

I(h, z) = shz

Solving for z and deriving with respect to h we have:

dz/dh = (Ih – sz)/(sh – Iz)

Where Ih is the response of investment to profitability and Iz to capacity utilization. Assuming stability, that is, that savings respond to profitability more than investment to capacity utilization and the denominator is positive, the sign of the equation depends on the numerator. If investment is strongly responsive to profitability (Ih > sz), then the system is profit-led (exhilarationist in B-M terms). If not we have the wage-led (stagnationist) regime.

As I suggested in my previous post, there are some theoretical problems with the type of model used to argue that the US economy is profit-led, besides the empirical ones alluded before. The independent investment function suggests that capacity utilization affects capital formation, if capacity is low there is more investment, and vice versa when z is high. In other words, firms would try to adjust capacity to demand. If that is the case you would expect that a normal relation between capacity and demand would be established in the long run (in the neoclassical view demand adjusts to capacity; that’s Say’s Law), which could be seen as the relatively stable output-to-capital ratio over the whole period for the US, in my previous post.

If that is the case, investment is determined by the adjustment of capacity to exogenous demand in order to reach the normal capacity utilization, and it is essentially derived demand (the accelerator principle). It is not instrumental in determining the normal level of capacity utilization, which must be determined by the exogenous components of demand. This is the basis of the supermultiplier models, first developed by Hicks, and then by Nicholas Kaldor, and referred to as Kaldorian in the heterodox literature (for more on that see this paper).

That is the essential difference between the Kaleckian and Kaldorian models, whether investment is partially autonomous and determined by profitability or it is derived demand. Of course income distribution in Kaldorian models might have ambiguous effects on growth, but firms would not investment more if profits went up, if there is no increase in demand. In this sense, worsening income distribution might lead to higher growth if demand keeps going for some reason (say more private debt stimulates consumption; or stimulates the consumption of a higher income group). But in general profit-led growth that stimulates investment, as in the M-B framework seems hard to explain from a theoretical point of view. Hence, the confusion it generates empirically (e.g. in the case of the US the notion that a debt-led consumption boom is a profit-led story).

PS: The typical Kaldorian model is based on Thirlwall's work, but the book by Bortis and Serrano's dissertation (or his paper; subscription required) are essential readings.