skip to Main Content

Money neutrality – another ideological contrivance by the conservatives

I have noted in recent weeks a periodic reference to long-run neutrality of money. Several readers have written to me to explain this evidently jargon-laden concept that has pervaded mainstream economics for two centuries and has been used throughout that history, in different ways, to justify the case against policy-activism by government in the face of mass unemployment. It is once again being invoked by the deficit terrorists to justify fiscal austerity despite the millions of productive workers who remain unemployed. I have been working on a new book over the last few days which includes some of the theoretical debates that accompany the notion of neutrality. There will also be a chapter in the macroeconomics text book that Randy Wray and I are working on at present on this topic. Essentially, it involves an understanding of what has been called the “classical dichotomy”. It is a highly technical literature and that makes it easy to follow if you are good at mathematical reasoning. It is harder to explain it in words but here goes. I have tried to write this as technically low-brow as I can. The bottom line takeaway – the assertion that money is neutral in the long-run is a nonsensical contrivance that the mainstream invoke to advance their ideological agenda against government intervention. It is theoretically bereft and empirical irrelevant. That conclusion should interest you! But be warned – this is just an introduction to a very complex literature that spans 200 years or so.

In a pamphlet to support Lloyd George in the 1929 British election, J.M. Keynes wrote:

The Conservative belief that there is some law of nature which prevents men from being employed, that it is ‘rash’ to employ men, and that it is financially ‘sound’ to maintain a tenth of the population in idleness is crazily improbable – the sort of thing which no man could believe who had not had his head fuddled with nonsense for years and years.

When I started learning economics, there was high unemployment in most advanced economies (second-half of 1970s). Given the massive waste that unemployment involves I was curious as to how the mainstream economics profession could on the one hand argue that the absence of government intervention would allow the optimal allocation of resources yet it was obvious from history that whenever governments used their fiscal powers to generate spending (aggregate demand) unemployment fell.

It was clear as I learned more that the mainstream theory was preposterous. It justified mass unemployment by claiming it was a voluntary phenomena arising out of free and maximising behaviour by the unemployed who were cast as preferring leisure at that point in time to work (and income). In that sense, high unemployment was “full employment” because it was a chosen state.

That allowed the models they used to indoctrinate students with to assume full employment as a starting point. Yes, there were some economists who could see the business cycle. But this was just a transient deviation around the “long-run” trend. There was a sequence of interrelated theories that claimed this was optimal and inevitable and any efforts by government to divert this “natural” tendency would result in inflation.

Accordingly, we were presented with the classical dichotomy or classical neutrality that said that nominal variables in the economy (money stock, prices) were independent of the real variables (employment, production etc) in the long-run. Extreme versions (rational expectations) later denied any relationship between the nominal and the real at any time! So the short-run was the long-run.

So if you wanted to understand the evolution of real output or employment then you needed no knowledge of the monetary system (or money). The rate of inflation was a purely monetary phenomenon and was analysed separately from the state of production.

The other way of saying this is that money is neutral (with respect to the real economy) and changes in the stock of money did not affect the real economy – only the rate of inflation. This was the Classical position that dominated economic theory until the Great Depression in the 1930s and although Marx had demolished it in the C19th, it took the work of Keynes and Kalecki in the 1930s to really expose its mythical nature.

So the Classical economists considered money to be a “veil” over the real economy that could be lifted without losing any insights. Moreover, if money is neutral in the long-run then governments could not hope to influence the real economy by manipulating the stock of money. So if the government thought the unemployment rate was too high and tried to expand aggregate demand (operating under the assumption that spending equals income equals output) then the classical dichotomy claimed this action would be futile and result in ever-accelerating inflation.

I outlined the labour market theory that underlies the classical dichotomy in this blog – Less income, less work, less income, more work!.

The mainstream analysis divides households (consumers) and firms (producers) and fudges the so-called aggregation problem (adding all these different entities up) by dealing with average agent – representative agent which is constructed as being rational and maximising in all their decision making. Behavioural theory refutes that assumption but the mainstream have rarely consulted psychology. They just assert this type of behaviour irrespective of the evidence.

All these agents are price takers in all markets – that is, there is perfect competition and no firm has any price setting power.

What determines employment? The simple classical model used what was called the Corn model to derive the classical dichotomy. They claimed that to examine household choice, we only need to know: (a) the initial goods endowment (wealth); (b) the terms of exchange between goods; and (c) Tastes and preferences of households. The first two define household income.

They consider two goods: (a) Corn (C); and (b) Leisure (L). The relative price of leisure is number of units of C that you give up to have one unit of L. Giving up L involves offering equal units of labour so relative price of leisure is the real wage rate, w. The relative price of corn is thus 1/w.

So the workers are deemed to have a labour-leisure choice (per day).

In Chapters 2 and 7 of my 2008 book with Joan Muysken – Full Employment abandoned – we consider the Classical labour market which is the basis of the standard Classical view on unemployment.

The Classical/neo-classical model is represented by the following figure:

For those who like equations, the lines in the graph can be written in algebraic form as follows:

Labour demand: Ld = f(w)    f’ < 0

Labour supply: Ls = f(w)    f’ > 0

Equilibrium: Ld = Ls

where w is the real wage which is the ratio of the nominal wage, W and the price level P.

The real wage is considered to be determined in the labour market, that is, exclusively by labour demand and labour supply. Keynes showed that this assumption is clearly false. It is obvious that the nominal wage is determined in the labour market and the real wage is not known until producers set prices in the product market (that is, in the shops etc).

The labour supply (Ls) function, which is based on the idea that the worker has a choice between work (a bad) and leisure (a good), with work being tolerated only to gain income (in the simple model – units of Corn). The relative price mediating this choice (between work and leisure) is the real wage which measures the price of leisure relative to income. That is an extra hour of leisure “costs” the real wage that the worker could have earned in that hour. So as the price of leisure rises the willingness to enjoy it declines.

The worker is conceived of at all times making very complicated calculations – which are described by the mainstream economists as setting the “marginal rate of substitution between consumption and leisure equals to the real wage”. This means that the worker is alleged to have a coherent hour by hour schedule calibrating how much dissatisfaction he/she gets from working and how much satisfaction (utility) he/she gets from not working (enjoying leisure). The real wage is the vehicle to render these two competing uses of time compatible at a work allocation where the worker maximises satisfaction.

What happens when the relative price between C and L changes? The final result looks scientific but is in actual fact a total fudge.

They isolate two separate “effects” of such a real wage change (say a rise): (a) a substitution effect; and (b) an income effect.

So an increase in the real wage (more corn is foregone for an hour of leisure) leads the worker to consume less leisure and to work more. This is the substitution effect. So if the real wage rises, work becomes relatively cheaper (compared to leisure) and the mainstream theory asserts via the so-called law of demand that people demand less of a good when its relative price rises. So real wage up, less leisure, more work.

But there is another effect to consider – the so-called income effect. When the real wage rises, the worker now has more income for a given number of hours of work.

The mainstream theory of normal goods (as opposed to inferior goods – the distinction is just made up largely) tells us that when income rises a consumer will consume more of all normal goods. The opposite is the case for an inferior good.

Leisure is considered to be a normal good as are other consumption goods the worker might buy with the income he/she earns. So as the real wage rises, the income effect suggests that the worker will demand more of all normal goods (because they have higher incomes for a given number of working hours) including leisure. That is the worker will work less and consume more leisure.

What is the net result? No analytical solution is provided. They just assert that the relative price (substitution) effect is stronger than the income effect and so the labour supply curve is upward sloping. It is totally made up result and no robust empirical analysis has supported this assertion. The reason they need to assert this result is because they also assert the demand curve is downward sloping and for an “equilibrium” they need the curves to cross. Simple as that.

The model assumes there are millions of firms and labour is one factor of production and indistinguishable from any other (inanimate) input. The single good is produced (Corn). Technology defines a “production function” which maps how much output is gained by adding successive units of input to a “fixed” stock of capital.

The labour demand (Ld) function is the “derivative” of the production function with respect to labour input (the marginal product). The ad hoc imposition of the so-called law of diminishing returns ensures this derivative is positive but declining as employment is increased.

In English, they say that as extra units of labour are added to a fixed stock of capital they become increasingly unproductive and hence there is diminishing products (returns). Hence, the labour demand function is downward sloping with respect to real wages. The argument is that firms have to pay the real wage (an amount of actual product) to the marginal workers and so they will only hire extra labour if the amount the worker contributes to production is more than the real wage. The assertion of diminishing returns assures the demand curve will be downward sloping – so the firm will only be prepared to hire extra workers if the real wage is reduced.

The following diagram shows the relationship between the labour market (top) and the production function (bottom). Here I am using Y to denote GDP and N to denote aggregate employment, The convex shape of the “production function” defines the assertion that there are diminishing returns. I will come back to this graph in moment.

The production function is a “short-run” relationship based on the fixity of other inputs like capital. In the real world, production processes rarely follow the model presented in the textbook. First, the idea that capital is fixed and production expands purely because extra workers are added seems not a reasonable depiction of reality. A cleaning firm who finds extra contracts do not force extra workers to share a single broom or vacuum cleaner. If there is extra work they add a worker and another broom!

So we have what are called “fixed factor proportions” which is just a fancy phrase meaning that workers and capital are added to production more or less in proportion dictated by the current technology. In that case, each additional worker is more or less as productive as the last. There is very little evidence in the real world to support the ad hoc assertion of the “law of diminishing returns”. It always amused me when I was a student how the mainstream attempted to infuse levity into their inane theories by referring to them as “laws” akin to a physical law in physics.

Once you see through this and realise that they are just assertions that these characters dream up (invent) which stand and fall on their empirical applicability you start seeing the nakedness ugliness of the neo-classical theory.

Anyway, the labour demand hiring rule is based on the assumption that firms always maximise profits and in the output market set prices equal to marginal cost (that is, the last unit of revenue they gain (price) is equal to the cost of producing that last unit of output). They cannot do better than that!

This rule translates into a labour hiring rule by noting that marginal costs (MC) equals the nominal wage (W) divided by the marginal product (MP, the extra unit of output per worker employed). So P = W/MP. Re-arranging this gives W/P = MP which is the famous marginal product hiring rule and says in English that firms will hire up to the point where what it costs them in real terms (the real wage – W/P or w) is exactly equal to what they get from the last unit of labour hired (MP).

Given the assertion of diminishing marginal products the firms know that as the real wage falls (on the vertical axis of the labour market diagram) firms will hire an additional worker (who is less productive at margin than last hired).

If you put the supply and demand ideas together you get the classical labour market which guarantees continuous full employment. With no borrowing or lending (simplifying assumption), then a price adjusting economy (if excess demand, price rises) will always guarantee full employment. If there is unemployment (that is an excess supply of labour) then the real wage will fall and vice versa.

In the corn economy this is a simple proposition. If there are workers wanting a job who do not have one then the corn wage falls and firms react by hiring more workers because it is profitable to do so at the lower rate. The corn wage falls until all those who want to work have a job.

Once money is introduced and the real wage is a combination of the money wage and the price level (which are determined in separate markets) then the analysis becomes more difficult.

The important point is that the classical labour market (which underpins the classical dichotomy) denies that involuntary unemployment can exist.

Why? Answer: Instantaneous price adjustment (real wage) keeps economy at full employment. Can there be generalised over-production? Answer: no! How do we know that the output produced by the full employment labour hiring (and capital usage) will be demanded in the goods market? Answer: Say’s law.

Say’s Law asserted that there was no possibility of general over-production. The trivial statement was that supply creates its own demand. This is a trivial point if we are in a barter economy with no borrowing or lending. More later on that.

To see the point consider the determination of aggregates in the Corn model. Reflect back on the two-panel diagram above. The model is entirely in real terms and there is no “price level” or “money” involved. The real system is totally determined by these two relationships.

If we know the market clearing real wage then the system is solved. We know the real wage(w), total employment (N), real GDP (Y), and real aggregate demand (AD). In the latter case, the AD curve would be identical to aggregate supply (AS) curve because supply creates its own demand or, in other words, the product market always clears.

In the bottom panel, technology sets the position of the production function. Market clearing determines the real wage and the employment level in the top panel. This employment level then determines a level of output via the production function deterministically.

This level of real output is all sold because of the assertion of Say’s law. So the real system is determined with reference to money. We do not have a “price level” in this economy and to explain prices (and inflation) the classical economists then had to introduce money. But the important point is that in dealing with a “money” economy the classical economists still asserted the dichotomy – the real sector was determined independent of the “price” level.

So Classical employment theory thus denies unemployment. The classical dichotomy was built on a denial of mass unemployment. The persistence of mass unemployment thus is extremely problematic for their theory and more practical approaches claimed it was a short-run phenomena due to market forces (that is, price clearing) being thwarted by government policy (for example, minimum wages) or trade unions etc. These fundamental institutions of capitalism were thus cast off as “market imperfections” which would disappear in the “long-run” as competitive forces eroded them.

What happens if we introduce borrowing and lending into the simple Corn model.

Borrowing is constructed as the supply of (private) bonds whereas lending is the buying of bonds. A bond is an asset which is a promise to pay corn at some specified future date. Previously, to provide for future consumption a person had to save corn (not consume current production). Now a person can purchase bonds that will yield future income flows (for consumption). So the assets that are now available in this more complex model are: (a) Bonds; and (b) Corn savings

The impact of this was that saving (S) was now identically equal to investment (I) now for any one individual. One can invest more than one saves and save more than one invests. Further it is no longer true that the only way to invest is to save (and the only way to save is to invest). With 2 assets we have S = I + Lending.

But if there is to be no general overproduction then what is not consumed has to be invested. In the simple Corn model if workers didn’t eat a Corn seed they would automatically plant it (that is, invest it). But with borrowing and lending permitted the situation was more complicated.

The point is that for the product market to clear (that is, aggregate demand to equal aggregate supply) saving (S) still had to equal investment (I). How did the classical economists assert that this situation would be continuously achieved (that is, full employment would be continuously maintained)?

Answer: introduce the theory of loanable funds, which is a aggregate construction of the way financial markets are meant to work in mainstream macroeconomic thinking. The original conception was designed to explain how aggregate demand could never fall short of aggregate supply because interest rate adjustments would always bring investment and saving into equality.

You will find this theory still being peddled (and applied to our financial markets) in most mainstream macroeconomics textbooks today (including Mankiw).

Mankiw assumes that it is reasonable to represent the financial system as the “market for loanable funds” where “all savers go to this market to deposit their savings, and all borrowers go to this market to get their loans. In this market, there is one interest rate, which is both the return to saving and the cost of borrowing.”

This is back in the pre-Keynesian world of the loanable funds doctrine (first developed by Wicksell).

This doctrine was a central part of the so-called classical model where perfectly flexible prices delivered self-adjusting, market-clearing aggregate markets at all times. If consumption fell, then saving would rise and this would not lead to an oversupply of goods because investment (capital goods production) would rise in proportion with saving. So while the composition of output might change (workers would be shifted between the consumption goods sector to the capital goods sector), a full employment equilibrium was always maintained as long as price flexibility was not impeded. The interest rate became the vehicle to mediate saving and investment to ensure that there was never any gluts.

The following diagram shows the market for loanable funds. The current real interest rate that balances supply (saving) and demand (investment) is 5 per cent (the equilibrium rate). The supply of funds comes from those people who have some extra income they want to save and lend out. The demand for funds comes from households and firms who wish to borrow to invest (houses, factories, equipment etc). The interest rate is the price of the loan and the return on savings and thus the supply and demand curves (lines) take the shape they do.

Firms were assumed to invest and households assumed to save. The question was how do these disparate decisions of each group coincide given they have different motives? The assumption is that the average firm is a net borrower (to fund capital) and a net bond issuer while the average household is a net lender (purchaser of bonds).

So what determines consumption and savings? Borrowing and lending are called inter-temporal exchange. What determines saving? Saving is constructed as a function of the real interest rate (which calibrates how many units of corn you have to give up now for future corn). What are the terms that this exchange occurs?

The real interest rate is a relative price (current for future corn) – that is, the relative price of today’s corn income is the number of units of corn that an individual will receive in the future if that individual gives up one unit of today’s corn. It is obviously going to be greater than unity.

The excess over unity is the interest rate. So relative price of today’s corn is 1 + i. The relative price of today’s corn is = 1 + i. Relative price of next period corn is the number of units of corn that the person gives up today to get one unit next period = 1/(1 + i) < 1. What does the saving function look like for a household (net saver)? Here the same sort of decomposition that we saw above in the labour supply decision is invoked. If the interest rate rises, the relative price of today's corn rises and two effects can be decomposed: rises and two effects occur: (a) Substitution effect - saving rises; and (b) Income effect (wealthier) - saving falls. The mainstream just assert that the substitution effect is greater than the income effect and so as the interest rate rises, consumption falls. In terms of the investment function, investment is seen as a negative function of the interest rate. Note that the entire analysis is in real terms with the real interest rate equal to the nominal rate minus the inflation rate. This is because inflation "erodes the value of money" which has different consequences for savers and investors. Mankiw claims that this "market works much like other markets in the economy" and thus argues that (Principles, p. 551):

The adjustment of the interest rate to the equilibrium occurs for the usual reasons. If the interest rate were lower than the equilibrium level, the quantity of loanable funds supplied would be less than the quantity of loanable funds demanded. The resulting shortage … would encourage lenders to raise the interest rate they charge.

The converse then follows if the interest rate is above the equilibrium.


The interest rate coordinates the separate saving and investment decisions and brings the net lenders (households) and net borrowers (firms) together to ensure equilibrium. If for example, S > I, i falls and the market clears. If S < I, i rises and the market clears So Say's Law is maintained and interest rate is a real variable (not a monetary variable). By way of summary, real output is determined by the current state of productivity (technology) and the level of employment that clears the market. It is thus supply constrained only by the available supply of labour and the state of technology. Real wage flexibility maintains full employment. Say's Law is maintained by interest rate flexibility in the loans market. So this analysis to date describes the way the mainstream outlined the determination of the real sector. What determines the price level (the nominal values of these real variables) in the Classical system?

In answering this question we explore the full meaning of the “classical dichotomy”. The principle framework used to explain the price level was the Quantity Theory of Money (QTM).

Quantity Theory of Money was used by Classical economists to explain the price of goods. The value of money = 1/p (the inverse of the price level) is determined by demand for and supply of money. There were two versions of this theory: (a) the flow version – Irving Fisher; and (b) the stock version – Marshall, Pigou and Keynes (so-called Cambridge equation).

The flow version is straightforward and best known by lay audiences and is still used by those who want to draw a connection between the money supply (or its rate of growth) and inflation.

In the Classical model, the price level is determined by: MV = PY

M = stock of money balances; V = velocity of circulation (how many times per period the stock of money turns over in transactions); Y = full employment output and P = price level. PY is nominal GDP.

MV is the flow of $s per period and PY is the flow aggregate demand of $s per period. If you think about the MV = PY equation then from a transactional viewpoint it has to hold. It is an accounting statement. All the transactions (left-hand side) have to equal the value of production (right-hand side). That doesn’t get us very far.

What the Classical economists wanted to assert was that if MV > PY then the change in prices was assumed to be positive, and if MV < PY then the price level would fall. Clearly, a steady-state situation was when MV = PY. To explain the price level (P) the Classical economists had to impose behavioural stipulations on V (fixed) and full employment (Y fixed). Then it became clear that the change in M would directly impact on the change in P (that is, inflation).

So Classical theorists (and monetarists and more modern variants) had to make some assumptions or assertions about the behavioural nature of the variables underlying the accounting identity. So they assumed that V is constant and ground in the habits of commerce – despite the empirical evidence which shows it is highly variable if not erratic.

They also assumed that Y will always be at full employment because they invoke flexible price models and assume market clearing – so they take Y to be fixed. This just asserts the real model derived above which includes the denial of unemployment.

This is a case of blind devotion to theory stopping the economist looking out the window and seeing regular periods when productive resources are anything but fully employed.

But with these assumptions – any child could then conclude that changes in M => directly lead to changes in P because with V assumed fixed the left-hand side is driven by M and if Y is assumed to always be at full employment then the only thing that can give on the right-hand side of the accounting identity is P.

While first-year students struggle to learn this theory and think it is high science – it is almost mindless when you think about it. You should easily be able to see the flaws.

The assertion that the real side of the economy (output and employment) are completely separable from the nominal (money) side and that prices are driven by monetary growth and growth and employment is driven wholly by the supply side (technology and population growth), rests on the assertion that the economy is always at full employment (quite apart from the nonsensical assumption that V is fixed).

Anyone with a brain could tell you that if business firms can respond to higher nominal spending (that is, higher $ demand) – either by increasing production or increasing prices or increasing both – and they cannot increase production any more (because the economy is at full employment) then they must increase prices to ration the demand.

That is a basic presumption of modern monetary theory (MMT). But typically, firms prefer to respond to demand growth in real terms to maintain their market share and thus invest in new capacity if they think spending will grow in the future and vice versa.

The overwhelming evidence is that the macroeconomy quantity adjusts rather than price adjusts to nominal aggregate demand fluctuations when there is excess capacity. Otherwise firms risk losing market share.

So, when the economy is in a state of low capacity utilisation with significant stocks of idle productive resources (of all types) then it is highly unlikely that the firm will respond to a positive demand impulse by putting up prices (above the level that they were before the downturn began). They might stop offering fire sale prices but that is not what we are talking about here.

This should discourage you from automatically linking growth in the monetary base and inflation. There is no link.

Inflation is driven by nominal aggregate demand growth that exceeds the capacity of the economy to respond in real terms – that is, to increase output. There are a myriad of reasons why this situation could emerge and some of these reasons implicate poor government policy decisions.

I really liked the way that Michel Kalecki developed the critique of Say’s law. In 1964 Kalecki (in his ‘Why Economics Is Not an Exact Science’, PP, 9, 62-65) offered this view:

This law asserts, speaking roughly, that all incomes, wages or profits are fully spent on the purchase of goods and services. This is self-evident as afar as expenditure on consumption is concerned while accumulation is treated as being always spent on investment. This doctrine obviously rules out the possibility of general overproduction, total demand is always equal to total supply. The law has certain affinities with the laws of preservation of matter and energy. The difference, however, is that it is definitely wrong. It implies that the value of national income is constant. If, for instance, less is spent on consumption then pro tanto more is spent on investment. But it was always clear that this is not the case since the value national income is subject to abrupt changes … Any divergence from Say’s law was interpreted as a divergence of the actual rate of interest from the equilibrium rate. It took a long time to realize that the idea of a rate of equilibrium was totally misconceived since the actual rate of interest had nothing to do with it. To discard the law of the preservation of purchasing power meant admitting the possibility of general overproduction. In particular investment may be lower than the accumulation corresponding to full utilisation of productive resources since its volume is determined by altogether different factors. This leads to piling up of unsold goods. A situation of this kind cannot, obviously, continue for long and is soon followed by a state of underutilization of resources …

How could the belief in the preservation of purchasing power be maintained for so long? In my view for two basic reasons: the class interests of the capitalists and the apparent corroboration of the law by experience of the individual. A doctrine which ruled out general overproduction made the capitalist system appear capable of full utilization of productive resources and dismissed cyclical fluctuations as insignificant frictions. These apologetics were facilitated by the application to the economy as a whole of the experience of housekeeping where clearly a lesser consumption means higher saving. But whereas the income of the individual is given, the national income is determined in a capitalist system by consumption and investment decisions, a fall in one of these components by no means leading automatically to a rise in the other. Thus individual experience does not correspond to the course of the economy as a whole.

So from that very insightful passage you learn a lot of things about macroeconomics that can be applied to understanding the present situation.

First, there is a strong recognition of the way in which the presence of compositional fallacies deceives the mainstream analysis. What applies at the individual (micro) level does not apply at the aggregate (macro) level. Please read my blog – Fiscal austerity – the newest fallacy of composition – for more discussion on this point.

Second, inherent in Kalecki’s work is a split between the long- and short-run but unlike the mainstream (Say’s law) there is no notion that the long-run is a steady-state attractor (that is, a (natural) point that the macroeconomy gravitates to when imperfections are eliminated). Kalecki’s notion of the long-run bore no insinuation of “equilibrium” (competitive equalisation of rates of profit; realised expectations; full employment).

Rather he sees the long-run as a sequence of short-run situations and the long-run influences are manifest and developed within the short-run. What are these long-run forces? Answer: factors that have an enduring influence on the path the economy takes. So these factors are evident by the sequence of short-run situations and so analytically it is impossible to decompose the long-run from the short-run.

He argued factors such as the institutional characteristics of the nation, market concentration, technological change and the class struggle between workers and capital were such long-run factors.

For example, Kalecki’s income distribution analysis was unique for his day because he eschewed the perfect competitive model as a starting point in analysing firm behaviour and price setting. He considered oligopoly (imperfect market competition) to be the norm (that is, concentrated markets where firms had price-setting power). He then wanted to know what determined the firm’s price setting power. In the mainstream, firms are initially cast as having no price-setting discretion as they are perfectly competitive.

Kalecki also began with the observation that full employment was an exception not the rule so firms typically had spare capacity and thus could expand output as demand increased at more or less constant unit costs. In the competitive model firms are always assumed to be oprating at full capacity and face increasing unit costs and so prices rise when demand rises. The real world fits the Kalecki starting point.

In considering the determinants of the “mark-up” that these empowered firms placed on their unit costs, Kalecki interwove his long-run factors with the short-run factors. It is a great example of the way in which he considered it impossible to really differentiate a separate analytical long-term temporal plane.

He said that the degree of industrial concentration (how many firms in a market etc) was inversely related to price-setting power. This is a “long-run” influence because changes in concentration move slowly. However, the business cycle is interposed on these underlying factors and are considered “short-run” factors. So when the economy turns downwards, the capacity of firms to pass on price rises is strained and so the mark-up may be squeezed. Further, while the class struggle underlies the determination of the output distribution, this is, in itself, mediated by the state of the business cycle – so when times a good, unions are stronger than when times are bad.

So you get a good picture of the way he eschewed the mainstream (neo-classical) long-run from these examples.

Kalecki said:

In fact, the long run trend is but a slowly changing component of a chain of short run situations; it has no independent entity …

Some mainstream economists assert that while price rigidities can render the real economy susceptible to changes in the money supply (positive effect) in the long-run, when all these rigidities are eliminated the classical dichotomy asserts itself.

Clearly, Kalecki didn’t agree with that. The concept of a separable “long-run” in highly dubious.

I particularly like this construction:

In sum, in neoclassical economics the “long run” cannot be a temporal concept. It does not refer to any real world process of evolution or change. Even notional processes are shown to be difficult to conceive. “Long run,” by definition, refers to a situation in which no mistakes are made and all the (theoretically expected) adjustments are achieved. It is not in time, it is not even a secular trend. It is just an aprioristic statement of what the world would look like should the theory be correct (emphasis in the original).

The source is Fernando Carvalho (1984-85) ‘Alternative Analyses of Short and Long Run in Post Keynesian Economics’, Journal of Post Keynesian Economics, 7(2), 214-234. For those who have access to JSTOR the link is HERE.

The transmission mechanism that linked the excess supply of money with the excess demand for goods was as follows.

People who hold more “money” than they desire will will eliminate the excess by (a) Directly buying goods and services; (b) Lending the excess supply of money (shift supply of loans) and lower the interest rate and stimulate investment.

This will create an excess demand for goods and prices rise because the economy is already at full employment (by assumption).

Classical dichotomy and the denial of unemployment

So at the heart of the classical system was the classical dichotomy and the QTM was used to generate a theory of absolute price levels while general equilibrium theory was used to generate a theory of relative price determination for the ‘real’ economy in which money was excluded.

In this regard, Walras’ Law provided the foundation of relative price theory. It is the basic principle linking together the goods market, the labour market, the bonds market and the money market.

Walras’ Law requires that the sum of excess demands and excess supplies overall all markets to be identically equal to 0. Thus for unemployment to exist (which would be an excess supply of labour) then by Walras’ Law there must be an excess demand in another market.

So Walras’ Law extends Say’s Law and includes all markets in the aggregation. They invoked what they called the “homogeneity postulate” which just says that the demands and excess demands in all the markets will not change in response to a change in the absolute price level on its own.

The technical term is that the demand functions are homogenous of degree zero in money prices. Accordingly, if relative prices remain unchanged, the demand for a particular commodity will not change when the overall price level changes. Demand functions are thus dependent only on relative prices.

Once an equilibrium is defined as a situation where in each market is in equilibrium (all excess demands are zero) then Walras’ law shows that equilibrium is defined purely in terms of a set of relative prices because the absolute price level can have no real effects. A proportionate change in all prices will not alter the equilibrium.

Thus the QTM is used to determine the absolute price level while the Walrasian GE model determines relative prices.

But the integration of the two approaches is very problematic although that gets too technical for this audience. Maybe another day I will have more time to weave a narrative around the mathematics!

What follows is that the money value of all planned market purchases (summed) are identically equal to the aggregate money value of all planned market sales. The implications of this for an economy-wide analysis were that equilibrium was considered to generalise across all markets. A single market could not be in disequlibrium alone.

This result was used to examine what happens when there is an excess supply of laboru in one market (that is, a disequilibrium). If the nth market is in disequilibrium then there must be an offsetting disequilibrium in at least one other market. This is Walras’ Law.

If there is excess supply in one market then there must corresponding to this be positive excess demand in at least one other market. Important to notice that the excess demands and supplies are measured as differences between planned (or notional) demands and supplies and not necessarily actual demands and supplies).

What is the relevance of this? Answer: Walras’ Law is crucial for understanding different constructions of how unemployment occurs in the mainstream approach and provides a more sophisticated version of the simple “workers are lazy”.

A Walrasian would say that either all markets are in equilibrium, or more than one is in disequilibrium, but we can’t have a situation where only one market is in a state of excess supply.

This is in conflict with Keynes’ own claim that excess supply in the labour market can be an equilibrium and thus a persistent state of affairs in a (free) market economy.

So for a Walrasian, involuntary unemployment cannot persist in a market economy with flexible wages and prices. The excess demand in the product market (accompanying the excess supply in the labour market) will drive prices up until excess demands are exhausted. The excess labour supply will push money wages down. Together, the real wage falls and this removes the excess supply of labour.

So unemployment cannot be a persistent equilibrium that the economy tends to.

So an excess supply of labour (unemployment) must be accompanied by an offsetting excess demand elsewhere (say for commodities). According to this story, the unemployed workers must have been intending to buy something with the wages they hoped to earn.

So the question then is why the bosses won’t hire the workers given that the demands for commodities exactly offsets the supply of labour.

Clue: the Keynesian re-appraisal which built on insights provided earlier by Marx and then Keynes (as well as others).

Macroeconomics emerged in the 1930s out of the failure of mainstream economics to conceptualise economy-wide problems – in particular, the problem of mass unemployment. Marx had already worked this out many years earlier.

In Theories of Surplus Value – Chapter XVII entitled Ricardo’s Theory of Accumulation and a Critique of it, Marx confronted Ricardo’s contention that there could never be a general glut in the market (which in modern terminology relates to a deficiency of aggregate demand).

Marx summarised Ricardo’s claim as follows:

more of a particular commodity may be produced than can be consumed of it; but this cannot apply to all commodities at the same time. Because the needs, which the commodities satisfy, have no limits and all these needs are not satisfied at the same time. On the contrary. The fulfilment of one need makes another, so to speak, latent. Thus nothing is required, but the means to satisfy these wants, and these means can only be provided through an increase in production. Hence no general overproduction is possible.

This was the logic behind Say’s Law which posited that “supply brings forth its own demand” and so macroeconomic crises (and unemployment) could not occur. It is also the logic behind the loanable funds doctrine which is still the basis of the claim that rising budget deficits lead to rising interest rates.

In developing his own explanation of how economic crises occur as a result of a failure of aggregate demand, Marx said:

Could there be a more childish argument? … In periods of over-production, a large part of the nation (especially the working class) is less well provided than ever with corn, shoes etc., not to speak of wine and furniture. If over-production could only occur when all the members of a nation had satisfied even their most urgent needs, there could never, in the history of bourgeois society up to now, have been a state of general over-production or even of partial over-production. When, for instance, the market is glutted by shoes or calicoes or wines or colonial products, does this perhaps mean that four-sixths of the nation have more than satisfied their needs in shoes, calicoes etc.? What after all has over-production to do with absolute needs? It is only concerned with demand that is backed by ability to pay. It is not a question of absolute over-production – over-production as such in relation to the absolute need or the desire to possess commodities. In this sense there is neither partial nor general over-production; and the one is not opposed to the other.

In the wider context of the development of economic theory, the development of Marx’s arguments in TSV were prescient indeed. The idea that capacity to pay was the key to understanding crises underpinned the notion of effective demand that appeared in the writings of Keynes and his contemporaries some 70 odd years later. It is the basis of the famous attack against the emerging monetarism of Friedman and others by Robert Clower and Axel Leijonhfuvd in the 1960s where they noted the difference between notional and effective demands and supplies. Marx had worked all that out 100 years earlier.

The more recent chapters in the debate constituted the Keynesian reappraisal.

In the 1950s, particularly, there was a major debate proceeding concerning the microfoundations of macroeconomics. As noted above, the neoclassical paradigm hardly considered macroeconomics to be a separate conceptual structure to microeconomics. The emergence of Keynesian economics in the 1930s had coincided with the emergence of macroeconomics as a separate line of enquiry.

The break with neoclassical thinking came with the failure of markets to resolve the persistently high unemployment during the 1930s. The debate in the ensuing years were largely about the existence of involuntary unemployment. The 1930s experience suggested that Say’s Law, which was the macroeconomic component and closure of the neoclassical system based on the optimising behaviour of individuals, did not hold.

The neoclassical economists continued to assert that unemployment was voluntary and optimal but that some factors not previously included in the model prevented Say’s Law from working. Keynes, following Marx and Kalecki, adopted the distinctly anti-orthodox approach and refuted the basis of Say’s Law entirely.

Marx was the first to really understand the notion of effective demand – in his distinction between a notional demand for a good (a desire) and an effective demand (one that is backed with cash). This distinction, of-course, was the basis of Keynes’ work and later debates in the 1960s where Clower and Axel Leijonhufvud demolished mainstream attempts to undermine the contribution of Keynes by advancing a sophisticated monetary understanding of the General Theory.

Bob Clower and Axel Leijonhufvud separately attacked the relevance of Walras Law for situations in which there is involuntary unemployment.

The theoretical push to reassert Say’s Law by neoclassical economists was severely dented by the work of Clower (1965) and Leijonhufvud (1968). They demonstrated, in different ways, how neoclassical models of optimising behaviour were flawed when applied to macroeconomic issues like mass unemployment.

They agreed that it is correct to say that in aggregate an excess of planned supply in one market must be matched by an excess of planned demand in at least one other market. But, they argued that Walras’ Law measures excess demands and supplies as differences between planned (notional) demands and supplies. These are not actual (effective) demands and supplies.

Keynes maintained that there can be conditions under which excess demands (or supplies) will not be effectively communicated in the market. So although certain prices (including wages) are at disequilibrium levels, no process of bidding them away from these inappropriate levels will get started. In particular, the excess demand for commodities is not effective market failure. It persists because “the market signals, pre-supposed in general equilibrium analysis are not transmitted” (Leijonhufvud, 1981: 70).

Clower (1965) showed that an excess supply in the labour market (unemployment) was not usually accompanied by an excess demand elsewhere in the economy, especially in the product market. Excess demands are expressed in money terms. How could an unemployed worker (who had notional or latent product demands) signal to an employer (a seller in the product market) their demand intentions?

Leijonhufvud (1968) added the idea that in disequlibrium price adjustment is sluggish relative to quantity adjustment. Leijonhufvud interpreted Keynes’s concept of equilibrium as being actually better considered to be a persistent disequilibrium. Accordingly, involuntary unemployment arises because the labour market is not in equilibrium and there is no way that the unemployed workers can signal that they would buy more goods and services if they were employed.

Any particular firm cannot assume their revenue will rise if they put a worker on even though revenue in general will clearly rise (because there will be higher incomes and higher demand). The market signalling process thus breaks down and the economy stagnates.


This blog aimed to introduce you clearly to the notion of the classical dichotomy which is based on the assertion that the real economy can be analysed completely separately from an analysis of price level determination. The sinister side of the dichotomy is that it underpins a case for zero policy intervention because it assumes full employment.

Any unemployment that is observed must be transient (policy interfering with market forces) or voluntary (leisure preferred over work). According to this theoretical structure, in the “long-run” money and inflation are proportional (and causal from money to prices) and government intervention designed to lower the unemployment rate is purely inflationary.

The framework fails theoretically and has no empirical content.

That is enough for today!

This Post Has 96 Comments

  1. There is a time series in the UK statistics called ‘Inactive – wants a job’ (LFM2). The last recorded figure at time of writing is 2.379 million people. People who answer a survey with ‘want a job’ are by definition not voluntarily idle. So that is prima facie evidence that the theory is the proverbial crock.

    The ‘official’ unemployment number (which doesn’t include those people) is 2.448 million people (MGSC). I graphed up the unemployment rate against the ‘excess of job demand over supply’ –

    This doesn’t include underemployment. This is just joblessness.

    In the UK there are 4.8 million people chasing 466 thousand vacancies. That’s a 10 to 1 ratio.

    Which is more likely – there are severe structural supply side issues even after thirty years of trying to sort them out, or the market simply isn’t creating enough jobs?

    I know which one lex parsimoniae goes for.

  2. It has been my experience that in any religion such as mainstream economics that when up against reality something esoteric (laws, theories, math) is used to justify the results. My question is; if I don’t have a job because I am too lazy, then how is cutting wages going to make me more inclined to go to work? In the real world that would make me even more lazy. If the people on the business news could only hear the things they say.

  3. “The overwhelming evidence is that the macroeconomy quantity adjusts rather than price adjusts to nominal aggregate demand fluctuations when there is excess capacity. Otherwise firms risk losing market share.”

    Is that evidence listed anywhere? Is there a published consolidation paper anywhere that makes this assertion? (If not can you get a PhD student to draft one up!)

    For me that empirical evidence is key to the entire debate. The whole neo-classical argument appears to be based upon suggest any central action will cause an immediate price adjustment upwards.

  4. Today IS the long run of 30yrs ago and the short run of last year. We are ALWAYS in the long AND short runs. Its all relative to when the change started.

    What a worthless concept, totally meaningless.

  5. We’ll all be going back to a local, market economy anyway. “Market” as in the original sense of the word – farmers bringing their goods to market, people buying and selling their wares etc. You see, once oil runs out we’ll have to.

  6. Some employers dream of having no unions, no welfare and no minimum wage. They pay economists to justify this dream. In the long run, they are willing to settle for something less.

  7. Neil: So is it the case that for a traditional economist “the long run” is just another way of saying “in our dreams”?

    In the sciences a technical term without an operational definition is essentially meaningless. Terms used loosely often have some ambiguous meaning because of the way they are used colloquially.

    For example, it is a truism of commonsense that inflation increases “in the long run” because, e.g., a loaf of bread costs more today that it did say fifty years ago. But as we know, defining “inflation, ” unemployment”, etc., operationally is rather difficult and somewhat arbitrary, since the indexes are constructed selectively and the selection process is often controversial.

    Is there even an attempt to define “long run” and “short run” operationally? Or is it by rule of thumb?

  8. “There is very little evidence in the real world to support the ad hoc assertion of the “law of diminishing returns”.

    At any point in time, maximum real output is finite, right? How would you model those constraints without some form of diminishing return assumption?

  9. Brilliant, thanks a lot!

    While first-year students struggle to learn this theory and think it is high science – it is almost mindless when you think about it. You should easily be able to see the flaws.

    Some students fall for the mysticism of holy equations conjured up by great white men, others just fall asleep. I’m proud to say I belonged to the latter category in my days of neo-liberal indoctrination. A distinct gut feeling (aka chundering in your country, I believe) and my infallible moral compass (aka luck) kept me from ever acting upon what I’d ‘learnt’. Ten years on, my mental wounds have slowly healed and seeing professional affirmation for what I always felt were highly questionable foundations of economic theory encourages me to relearn some of what I originally signed up for, paid for but sadly never received.

    Expert textpert choking smokers,
    Don’t you think the joker laughs at you?
    See how they smile like pigs in a sty,
    See how they snied.
    I’m crying.

    I Am The Walras (Lenin & McCarthy)

  10. In my view, the “veil of a barter” concept of money is another manifestation of the fascination some economists feel with respect to physics… (perhaps to feel deserving the social credit given to physicists…) without due regard the unsurmountable differences between the fields of study.

    The idea of the “veil of barter” has a curious analogy with the physical principle that the value of a measure is the ratio to an arbitrary chosen unit. Physical laws do not change if one changes the measurement units. Unfortunately, money is not a veil: it is the medium through which different things can be made “measurable economic variables”.

    Some decades ago Norbert Wiener observed that defining some “economic variables” as “measurable variables” does not stand to the measurement requirements of physical variables. Consider Domestic Product, DP, for instance. It is made of a myriad of different things in different quantities. If such a thing is to be “measured” than it must be “measured” through a giant list or vector of things. It can never be measured by a one only number.

    Of course, if one multiplies each element in the list by its selling price and adds up all prices one gets $GDP. If one calculates the average price weighted by quantity numbers one gets a number that one can call the price level, $P. Then, one can *define* DP to be $GDP/$P.

    Of course if all prices are scaled by the same factor, $GDP and $P will be scaled and DP will have the same value. This does not make money a veil, because such idea of “scaling a measure” is very easy to put at work in a physics laboratory, but has no operational correspondence in economic reality.

  11. @Elizabeth

    Ireland now has a political crisis and almost certainly a General Election in the New Year. This uncertainty, when the markets think about what the outcomes might be, could be very damaging to Europe. Ireland may well be back to barter and other forms of exchange simply for the people to survive; the Euro abandoned.

    I feel the financial future of Europe now depends entirely on the attitude of Germany. How much financial support is possible? I just cannot see their economic powerhouse bailing out all the countries that may come under market attack and thence get into difficulties. The reassurance that has been given in Ireland’s case won’t be forthcoming. As there is no plan B for the Euro, there can only be chaos economically [and probably worse]. A real own-goal for mainstream economic theory. Result – money, as we know it now, will have no value, unemployment will rise horrendously. I really do hope my feelings of foreboding as a result of recent events are totally wrong and I can be classed as economically illiterate, scaremonger and an idiot.

    I wish they would listen to Bill!

  12. @pebird

    I agree the Keynes quote is snappier, but I thought the Robinson quote was more appropriate to Greg’s comment.

  13. PG:

    In my view, the “veil of a barter” concept of money is another manifestation of the fascination some economists feel with respect to physics…

    I think it’s much more than this. Really what is going on is two separate accounting systems.

    There is a “goods” accounting, and a “financial” accounting, and both are used inconsistently in order to derive results, as goods flow in the opposite direction direction of money. You cannot just divide by the price level in order to convert from one system to another. They are completely different systems.

    In the Goods Accounting system, to produce real output is to “earn” income. To consume real output is to spend the income received.

    Given that output consists of consumption goods (which are consumed) and investment goods (which are not consumed), then by definition, the difference between income and expenditure is “savings”, which is investment.

    It’s a theorem.

    Similarly, a government tax in the goods accounting system consists of output consumed by the government and not by the private sector. Government also produces output which adds to “real income”. The difference between the output that government consumes and the output that Government creates is the “real” deficit.

    That is how they “prove” that taxation is a real harm. This how they “prove” Say’s Law holds, and this is how they prove that deficits crowd out the private sector. After all, the government is consuming more than it is adding in real terms. This must come as a loss, right?

    And all of this seems plausible unless you think about it a little.

    To an individual, it may seem that their consumption opportunities are reduced when they pay taxes, but we do not pay taxes by sending goods to the government, but with money. The private sector as a whole does not consume less if everyone pays taxes — not unless you are assuming Keynesian nominal effects, which they aren’t.

    But how many seconds of contemplation are needed to see this fallacy?

    So unless you are Keynesian, then you don’t even see the financial transactions such as income taxes or deficit spending. It could well be that the government is running enormous financial deficits, but the goods deficits are small, or are even goods surpluses.

    It is at this point that they switch to using the financial accounting system in the public discourse, and in their inner discourse, to argue that

    * financial deficits crowd out the private sector
    * financial taxes are a dead weight loss
    * financial profits need to be encouraged because they fund real investment

    And underlying this is an assumption that the “financial” accounting system is the same as the real accounting system, if you only divide by the price level. And they have “proved” the corresponding analogues in the “goods” accounting system, so that they must be true in the financial system as well.

    Of course, if it was equivalent, then no one would care about the financial deficit, or taxation, or borrowing. They would only care about government consumption of output and government provision of output.

    That would be if the model was internally consistent. There are “theorems”, but these theorems need to be interpreted at the fundamental level, rather than just criticized for being too simplistic. And the heterodox people will lose the argument if they just keep arguing that more complicated models will fundamentally reverse the conclusions of the theorems. In most cases they won’t. You have to tackle the theorems to see what they are actually saying and what they are actually proving, and then point how it has nothing to do with what the economists are concluding.

  14. NKlein and Pebird

    Both fine quotes. Much better than most of us here could have said it at the time.

  15. Who are you RSJ?

    With your eloquence and level of understanding. In the public realm, you could achieve something great for the common good.

  16. RSJ:

    “At any point in time, maximum real output is finite, right? How would you model those constraints without some form of diminishing return assumption?”

    By saying that there is a finite amount of workers with only 18 waking hours a day each. When you run out of unemployed people, you hit your maximum real output. No diminishing returns required.

  17. RIght, except that we do not hold to the labor theory of value.

    The resource constraints are not just on labor, and even for labor, as you work everyone 18 hours a day, experiences diminishing marginal returns for that 18th hour and the last set of workers that you pulled into the workforce.

    I think you would hit diminishing returns after about 9 hours a day, if not before that.

  18. “Similarly, a government tax in the goods accounting system consists of output consumed by the government and not by the private sector. Government also produces output which adds to “real income”. The difference between the output that government consumes and the output that Government creates is the “real” deficit.”

    This appears to be backwards.

    Government expenditure at the margin is a financial deficit and a real surplus. Real resources are transferred from the private sector to the public sector.

    Taxation at the margin is a financial surplus and a real deficit. Real resources are transferred (back) from the government sector to the private sector.

    Where there is taxation without corresponding expenditure, taxation becomes a financial claim of the government on the private sector. Such a claim macro net is an ultimate claim on real investment. Taxation transfers that real resource from the government sector to the private sector (as “paid for”). That is a real surplus to the private sector.

  19. “The resource constraints are not just on labor, and even for labor, as you work everyone 18 hours a day, experiences diminishing marginal returns for that 18th hour and the last set of workers that you pulled into the workforce. ”

    Nah – the preference for leisure will kick in ;0 Just joking – most workers would not agree to work 18 hour days. A gulag, on the other hand . . .

  20. It seems to me that the corn model is too simplistic to be taken seriously. It could be the basis for conjecture, at best.
    Were there more realistic models presented in the intervening years by mainstream economists?

  21. Anon,

    Government expenditure at the margin is a financial deficit and a real surplus. Real resources are transferred from the private sector to the public sector.

    Here, the point of view is of the private sector, not the government. For the standard view, the only real resources relevant are consumption, or a stream of consumption.

    To the individual, when they receive a check from Treasury, they are in effect receiving consumption goods. But from the point of the private sector as a whole, that is just a shift of consumption to the person receiving the check from the private sector person who would have otherwise consumed. It does not matter if the check is “paid for” by taxing someone else, or by increasing the money stock, or by borrowing. In either case, someone else in the private sector funds the additional consumption. Total output and consumption will not change. The market will still clear, just at different prices. By taxing or mailing checks out, what you are doing is re-shuffling the endowments, but you are not increasing the total endowment of the private sector as a whole.

    Again, this is the mainstream view, and it assumes full employment.

    The only way that government can transfer real resources to the private sector is by the provision of public goods.

    And the only way that government can subtract real consumption from the private sector is by consuming output directly in the course of providing the public goods.

    The net of these two activities is the “real” deficit.

    Here, I am assuming that government does not own private sector capital. Perhaps in Australia, the government buys stock or bonds of private firms when it runs a surplus and then sells the stock back to the private sector when there is a deficit. But this is idiosyncratic, and not significant in scale. It’s also not something that most economists would support.

    What Bill and others would point out, is that when the economy is below full output, then you do not get crowding out when the government consumes output. And they would also argue that nominal effects are important, so that by mailing a check to everyone, you can get the private sector to produce more and to consume more. Therefore there is no real deficit, but a real surplus. Those complaining about the deficit are suffering from money illusion, as are those complaining about taxes being too high, etc.

    But I don’t want to get into that discussion now, I just wanted to point out what the theorems e.g. Barro (1974) are really saying. They are using a different accounting system in which their version of “tax” is really government consumption of output, their version of “borrowing” is equivalent to a tax (hence the “crowding out” argument). And their version of “repayment” is really the provision of public goods. You need to properly interpret the theorems, and separate the math from the public policy arguments, rather than attacking the theorems when you disagree with the policy.

  22. RSJ:

    I agree, you certainly will get diminishing results per worker. But if all workers had an 8-hour day, you would not get diminishing returns *per worker added*, which is a huge difference, and critical to the classical construction of full employment.

    Also, you asked how you could model finite real output with *no* diminishing returns. I did so. Criticizing my example for containing no diminishing returns, like you specifically requested, seems somewhat unfair.

  23. Grigory,

    You would certainly get diminishing returns per worker added, as you scraped the bottom of the barrel in terms of the labor force.

    “Also, you asked how you could model finite real output with *no* diminishing returns.”

    No, that’s not the problem in terms of modeling, and that’s not what I was asking.

    The problem is how to determine what the actual output level will be, given that we know a priori that actual output is constrained.

    The best way to see this is using your labor example.

    As you make people work longer and longer hours, and as you pull more and more people into the labor force, there must be some sort of rising cost, so that the level of “full” employment and full output is less than having the entire population work 18 hours per day. You could achieve more output with child labor and 18 hour workdays, but you decide not to, because the costs of doing so exceed the benefits.

    But for working less than 8 hours, the benefits outweigh the costs. At 8 hours, the benefits are equal to the costs.

    So even in your example, there are rising marginal costs and declining marginal benefits, at least near the region of “full” output (and therefore full employment).

    There is no escaping this basic premise.

    Now, whether a market allocation will actually realize the optimal allocation is a separate question. Bill has reviewed the arguments that the discrepancy between effective demand and notional demand will lead to a market failure. Unemployed workers have no way of signaling that they would consume, if they would only have a job, and so there is no intrinsic mechanism to kickstart the production process. We can say that externalities, and specifically values (e.g. the disreputable nature of child labor) are not incorporated in the price vector.

    We can generally point out that in a market with n goods and barter, there are n(n-1)/2 prices, but in a monetary market with n goods, there are only n prices. Therefore there is an information loss in a monetary economy, and this could also prevent the market allocation from correctly equalizing the total costs and benefits, because certain options aren’t visible to the market economy (e.g. specifically, the option of selling a good to an unemployed worker, and employing him to produce said good).


    But before we can even study how to compensate for the market failure, we need some model of what the optimal allocation level would be. We can’t even begin to discuss this issue intelligently without first pre-supposing rising marginal costs and declining marginal benefits, at least surrounding the region where full output is reached.

  24. In the interest of full disclosure, I was also previously vehemently arguing against rising marginal costs, and against the information loss due to barter on Nick Rowe’s site, until I thought more about it, and realized I was totally wrong.

    There have to be rising marginal costs and declining marginal benefits, the only issue is that these costs and benefits are not accurately reflected in price, which is how the market mediates these trade offs. And from a more technical view, you would need complete futures markets in all commodities as well, as well as future knowledge of all future prices.

    There are just massive informational problems here that make an efficient market allocation implausible, even before you introduce uncertainty. Once you introduce randomness, then you have the distinction between expected utility and actual utility, so that there may be an ex-ante efficient allocation in terms of probability weightings before the outcome is known, but the same allocation, once the uncertainty is resolved, may not be efficient ex-post.

    The whole issue is very fascinating, but denying the basic tenets of marginal costs and benefits is a lazy response to challenging the policy prescriptions of the free marketeers; as long as the heterodox branch keeps doing this, they aren’t going to be taken very seriously, and for good reason.

  25. RSJ:

    as future knowledge of all future prices.
    Would not that impossible requirement make any attempt at modeling rather futile ?

  26. RSJ – you’re describing a full employment scenario where workers hours are increased. You’d simply add additional workers prior to getting to that state. So, yes, at full employment marginal costs would rise. And that would be a source of inflationary pressure.

    As for those marginal workers – ever see that TV show about the worst jobs? Those workers usually don’t have a great deal of education, but they sure do seem to get a lot done efficiently. So, perhaps, “marginal worker” aren’t as marginal as the neo-liberals want us to believe. In fact, maybe they’re legitimate people and not marginal at all!

  27. “So, yes, at full employment marginal costs would rise. And that would be a source of inflationary pressure.”

    But this statement is a bit non-sensical, right. “Full employment” is that state where marginal costs have risen so high that you do not want to produce anymore. Unless you pre-suppose that marginal costs are rising, you never get to full employment in the first place. It’s always better to work more and more hours and use more and more resources and there is no limit, unless this limit is constrained by rising marginal costs and diminishing marginal benefits.

    In terms of whether low paid workers are “marginal”, that is of course a conflation of terms. At some point, you want to become serious, and produce testable hypothesis that are internally consistent, rather than rely on discursive complaints that some workers are underpaid. Obviously some workers are underpaid and others are overpaid. If you can define which workers are underpaid and under what circumstances this occurs, then you can propose a tax policy solution, or some other solution to transfer income to the groups that are consistently underpaid. But making the case for that requires a bit more than a discursive essay or argument via homonym.

  28. “Would not that impossible requirement make any attempt at modeling rather futile ?”

    No, the result is that ex-ante optimal allocations are not always optimal ex-post allocations unless you have strong assumptions about knowledge of future prices as well strong assumptions about common beliefs.

    That’s a result that modeling gives you.

    Such results are a good thing, not a bad thing.

    For example, they explain why social insurance is more efficient than having each person buy their own insurance.

  29. >>RSJ says:
    >>Tuesday, November 23, 2010 at 2:19
    >>”There is very little evidence in the real world to support the ad hoc assertion of the “law of diminishing returns”.
    >>At any point in time, maximum real output is finite, right? How would you model those constraints without some form of >>diminishing return assumption?

    Have a re-read of your question. That’s what I was answering. I gave you an example of a model with finite real output and no diminishing returns. It was oversimplified, obviously, but it showed how.

    >>”Also, you asked how you could model finite real output with *no* diminishing returns.”
    >>No, that’s not the problem in terms of modeling, and that’s not what I was asking.

    Seriously, just scroll up and look at your first post again. You were fairly clear.

    All your examples are to do with diminishing returns from a single worker. You want to maximize productivity for each worker; of course that means letting them sleep and take bathroom breaks. But why does every Nth worker have to be less productive than the (N-1)th worker if they work in the same conditions with the same equipment?

  30. The Carvalho paper is reprinted in Alternatives to Economic Orthodoxy: A Reader in Political Economy , so most of it is freely available from google books here. Was going to suggest it myself until I saw Bill did.

  31. Bill,

    I have to sit through hours of this stuff every week and I have to say that I have never come across anyone who mistakes modelling assumptions for reality. No one is forcing you to use a tool in a particular way. (Who was it who fed Marx into a DSGE model–Romer? I forget). There’s just so much here that I don’t know where to start to dispute the details. But in any case, the characterisation of your fellow economists as religious zealots is flat out false, in my experience. No one actually thinks like that.

  32. Everyone else,

    “Some students fall for the mysticism of holy equations conjured up by great white men, others just fall asleep. I’m proud to say I belonged to the latter category in my days of neo-liberal indoctrination.”

    Not to pick on Oliver, but I think this is significant. Did someone mention parsimony upthread?

    If you ever actually study economics (and not sleep through it!), you will get a much better understanding of the limitations of the models. And the orthodox bogey men that get so little love round here are in general much more cognisant of the limitations of their models for that very reason. If you doubt this, you obviously haven’t given it much thought.

    In general, if you don’t understand a concept, there is more than one possibility. It’s possible, of course, that it is in fact meaningless; but it’s also possible that you just plain don’t understand it–possibly because you’ve dismissed it out of hand without spending any time trying to figure it out. I don’t know if there is any reliable non-subjective rule of thumb for this. Are you amazingly smart? Is your name Grisha Perelman? Perhaps a bit of humility, then.

    “Ben Bernanke doesn’t know what he’s talking about”. I have lost count of the number of times that I have seen this in the comments section. Just mad, on a 9/11 troofer scale…

  33. Dear vimothy (at 2010/11/27 at 6:45)

    You say:

    I have to sit through hours of this stuff every week and I have to say that I have never come across anyone who mistakes modelling assumptions for reality.

    What stuff? In what context?

    You go on:

    No one is forcing you to use a tool in a particular way.

    It is a way of thinking. The tools are one thing but the method of enquiry (equilibrium – long-run neutrality) is another and suppresses imagination and alternatives.

    You then said:

    There’s just so much here that I don’t know where to start to dispute the details.

    That is always the way to avoid debate. Like – Oh there is some much wrong with your analysis Bill – and you are really so stupid – I just give up!


    But in any case, the characterisation of your fellow economists as religious zealots is flat out false, in my experience. No one actually thinks like that.

    Professions have in-built devices to safeguard the paradigm. After a career in the profession of 30 odd years where I hold the top academic rank etc I can tell you I understand the way my profession thinks and acts. I can tell you in chapter and verse the subtle and not so subtle ways that thought is controlled and research is channelled. No-one would admit to be a religious zealot but when the empirical world bears no relationship to the theoretical models and you start talking about unemployment and you don’t even have a labour market in your theoretical model etc then it becomes a “matter of faith”. Joan Robinson said long ago that mainstream economics was a branch of theology.

    best wishes

  34. Hi Vimothy,

    I post this here, with respect, related to the religious aspect of some economic policy/advocacy. Tom Hickey who often posts meaningful comments here gave me this referral. An economist Robert Nelson (PhD) has studied the religious aspect of the major schools of economics. Has written a book that I am currently reading and here is an article he wrote around the same time as the book (about 10 years ago now) that hits the major themes of his book, excerpt:

    “An economist would say that my evidence of an important theological element in economics is thus far “anecdotal.” It is far from meeting any standard of “proof,” as an economist would expect to see a formal analysis developed. If the arguments thus far have been “soft,” I propose next to make a “hard” argument. My claim is the following: Without certain theological assumptions, some of the most important conclusions of economic theory could not sustained. It is as though a mathematician had developed the proof of a theorem in which some of the key steps in the proof had been left out. On close inspection, moreover, the omitted steps in this case turn out to have a special feature – they are theological in character. Their omission, I further submit, reflects an implicit understanding among economists that these steps can not be defended in “scientific” terms, and an unwillingness to defend them explicitly in theological terms.”

    The book is very interesting in regards to how at least here in the US, the two major schools of economic thought (Cambridge and Chicago), take on actual religious aspects of faith and divergences of approach similar to that of Roman Catholicism vs Protestant reformers.

    Anyway, food for thought. Resp,

  35. Matt, Vimothy, et al.,

    There are numerous scholars, both within and outside the field of economics, who have made the suggestion that neoclassical economics has much in common with theology. One of the more recent is Duncan Foley’s book, “Adam’s Fallacy: A Guide to Economic Theology,” for instance. In short, there’s a good deal of scholarly research that backs up Bill’s views here, including the book Matt/Tom cite. If one isn’t willing to go so far as to use the term “theology” to describe neoclassical economics, there’s in addition probably even more scholarship demonstrating the rather poor “scientific” methodological approach in neoclassical economics.

  36. With respect to model use and economic theology, see Krugman’s recent post Debt, Deleveraging, and the Liquidity Trap

    … models are an enormously important tool for clarifying your thought. You don’t have to literally believe your model – in fact, you’re a fool if you do – to believe that putting together a simplified but complete account of how things work, with all the eyes crossed and teas dotted or something, helps you gain a much more sophisticated understanding of the real situation. People who don’t use models end up relying on slogans that are much more simplistic than the models – debt bad, inflation bad, savings good, all of which are just wrong some of the time.

    These “slogans” are what philosophers call “norms.” They are purported general descriptions that actually function as criteria, hence, they are themselves privileged from error. That’s about as close to dogma as one can get.

    Also check out the summary of his paper at linked in the post.

    From the summary:

    The current preoccupation with debt harks back to a long tradition in economic analysis, from Fisher’s (1933) theory of debt deflation to Minsky’s (1986) back-in-vogue work on financial instability to Koo’s (2008) concept of balance-sheet recessions. Yet despite the prominence of debt in popular discussion of our current economic difficulties and the long tradition of invoking debt as a key factor in major economic contractions, there is a surprising lack of models – especially models of monetary and fiscal policyΒ – of economic policy that correspond at all closely to the concerns about debt that dominate practical discourse. Even now, much analysis (including my own) is done in terms of representative-agent models, which by definition can’t deal with the consequences of the fact that some people are debtors while others are creditors.

    New work that I’ve done with Gauti Eggertsson (Eggertsson and Krugman 2010) seeks to provide a simple framework that remedies this failing. Minimal as the framework is, I believe that it yields important insights into the problems the world economy faces right now – and it suggests that much of the conventional wisdom governing actual policy is wrong-headed under current conditions.

    [emphasis added]

  37. If you ever actually study economics (and not sleep through it!), you will get a much better understanding of the limitations of the models. And the orthodox bogey men that get so little love round here are in general much more cognisant of the limitations of their models for that very reason. If you doubt this, you obviously haven’t given it much thought.

    Vimothy, I am fully aware of the fact that there are a lot of very intelligent and well-meaning economists who operate from within the mainstream framework and who deliver some very ingenious and intricate insights from that vantage point. I am also not deluded to think I could do any better, if that’s what you were implying. On the other hand, I firmly believe (and yes, that is a somewhat theological argument) that models who’s foundations are not at odds with basic empirical evidence will ultimately deliver better results than others, if put in the right hands.

    Furthermore, each profession carries a responsibility towards the public by making sure its basic tenets are as precise as possible. It is not the frills, anecdotes and exceptions but the naked, unrefined models themselves that will be portrayed by the media and politicians and then go on to form public opinion. From that perspective, I would say it is all the more important to make sure that even the smallest and crudest building blocks of theory give an accurate depiction of reality.

    Also, the phenomenon that professionals cling to ideas in face of very strong evidence to the contrary, or that obsolete ideas live on in public despite an opposing professional consensus, is not confined to economics. I can assure you there are very similar tendencies within the architecture profession, although it luckily doesn’t purport to be scientific in any way. But, the lower down the ranks one climbs, the more apparent the negative effects of falsely maintained myths and anachronisms become. So, for us, it is important to step out and sensitize the public to current topics and solutions but also to allow for public feedback to force us to reflect on our own beliefs and arguments more critically. There is a need for more honest exchange and less lofty obfuscation and many economists fail baldy at this to the detriment of us all. I would claim, many even fall for their own tricks.

  38. Matt,

    One of the things that amuses me about books like that is the often naked ideological bent of the writer–even as they complain about the slide of economics into theology, right-wing zealotry, etc, etc. Generally meaning that since orthodox economics types stopped agreeing with me they have become a lot more partisan and close minded. What a surprise eh? People who disagree with me are wrong, bears are Catholic and the Pope he sh…

    Donald MacKenzie, Michel Callon and John Law do serious academic work about the sociology of economics, if you are interested. I highly recommend MacKenzie’s “An Engine not a Camera”, which is focused more on finance than economics per se, but still is very germane and well written (would be quite interested in hearing the MMT take on the performativity thesis, incidentally).

  39. Scott,

    I don’t doubt that there are people who agree with Bill. Do you consider your own position to be quasi-religious in nature?

  40. vimothy:

    “Do you consider your own position to be quasi-religious in nature?”

    What difference would this make? One’s own consideration of what one espouses to be religious or scientific is completely irrelevant.

    I imagine Newton considered some of his insights to be quasi-religious. Good for him, but I don’t care.

    One must test the implications of someone’s position in order to ethically make a determination of scientific or “quasi” religious quality. Otherwise it is not truly argument.

    I don’t know if you intend to actually engage in thinking here, or are just frustrated that others don’t see the world as you wish.

  41. Vimothy,

    My own position has a sound scientific methodology behind it. And I know this because, unlike 99% (maybe a bit less) of economists, I was actually trained in the philosophy of science in my Ph.D. program. It doesn’t mean I can’t be wrong, but it’s a big difference. The methodology spells out empirical, real-world criteria that, if met, would require that I acknowledge that what I previously thought was correct actually was not. This is fundamentally different from the core of neoclassical economic theory.

  42. Also, Vimothy, it’s not that “people agree with Bill.” It’s that people, quite a few of them, have done serious scholarship with serious methodological approaches that provide evidence to support Bill’s claims here. Again, doesn’t mean they or Bill can’t be wrong or even that they’re necessarily right, but it’s quite a bit more serious foundation for an opinion than your by definition anecdotal “in my experience . . . ” claims to the contrary.

  43. Oliver,

    Those are worthy sentiments, but easier to espouse than practice. And it is not obvious to me that any mainstream economists would disagree with them in principle. To give one example taken from what you write:

    “On the other hand, I firmly believe (and yes, that is a somewhat theological argument) that models who’s foundations are not at odds with basic empirical evidence will ultimately deliver better results than others, if put in the right hands.”

    What do actually mean? If you are talking about empirical statistical relationships, how do you produce reliable out of sample forecasts? Trying to solve this problem is how economics ended up here in the first place (proximately, at least–i.e. the Lucas Critique).

  44. Scott,

    “My own position has a sound scientific methodology behind it.”

    I see. You must realise how this looks to those of us who are unaffiliated. Convenient, no–even if true? Does poor scientific methodology map one-to-one with the set of economists who disagree with you?

    I agree that a better grounding in the philosophy of science would be a good thing for the economics profession. Still, I am often struck by the fact that a better grounding in writing readable prose would be a good thing for the economics profession. I suppose that neither are likely any time soon, but we live in hope.

    As for support for the economic theology thesis, I simply meant that there are those who agree (some of whom are serious scholars and have done serious research) and those who do not (ditto). How am I suppose to evaluate this?



  45. vimothy: How am I suppose to evaluate this?

    You’ll have to get a PHD in economics. But the choice of school and thesis advisor will heavily impact your outlook.

  46. Oliver: Also, the phenomenon that professionals cling to ideas in face of very strong evidence to the contrary, or that obsolete ideas live on in public despite an opposing professional consensus, is not confined to economics. I can assure you there are very similar tendencies within the architecture profession, although it luckily doesn’t purport to be scientific in any way.

    If architects had the success/failure rate of economists, we’d all be buried under rubble. πŸ™‚

  47. Bill,

    I spent (part of) my Friday night reading this post–20 pages of text when printed out in Word. Perhaps it is not obvious while I sit arguing with everyone, but I am glad you write here and find your contribution valuable. I can think of few other econ bloggers who could write a post of this scope and still make it readable and interesting!

    I didn’t see a definition of inflation in there. How do you define inflation? Given this definition, what causes inflation?

    I don’t think I mentioned anything theoretical in the other thread–I was pointing to an empirical regularity: the comovement of money growth and inflation over longer time frames, and the lack of a relationship between money growth and GDP growth over any time frame (this is not a claim about causality–Tom Hickey, I’m looking at you!). I didn’t see you address this in your post. Could you do so here, or point me at what I missed?



  48. Tom,

    I am a postgrad at Manchester, UK. Perhaps a PhD would bring me round, perhaps not. Where did you get yours? Is your position WRT economics religious? I would feel a lot better about this proposition if I could find someone who claimed it for themselves rather than applying it to those who disagree with them.

  49. Vimothy,

    Somebody unaffiliated but with expertise in methodology can decide for himself/herself. Someone unaffiliated without such expertise will hopefully know enough to know he/she doesn’t know enough to make an informed judgment one way or the other. Wishful thinking on my part, I know.

    Regarding 1-to-1, there are many with sound methodologies who aren’t MMT’ers (my own dissertation advisor, for instance), so definitely not 1-to-1. There are also a number of those sympathetic to neoclassical economics in some fashion that are critical of the scientific robustness of its methodology–McCloskey, for instance. Finally, there are some who are sympathetic to MMT but do not have particularly good methodologies in my view.

  50. Scott,

    I think you are right to be cautious. One ought to be careful about blurring the line between methodological and political (or “positive and normative” in corporate econ-speak) critiques for that very reason. It is not obvious that the methodological interests of economics qua economics are the same as the interests of the political coalition you (quite naturally, in sociological terms) form as a consequence of making this argument.

  51. What do actually mean? If you are talking about empirical statistical relationships, how do you produce reliable out of sample forecasts? Trying to solve this problem is how economics ended up here in the first place (proximately, at least-i.e. the Lucas Critique).

    I was thinking of behavioural assumptions (say, profit maximization) that are extrapolated to make generalized predictions. Those parts of the population who don’t follow the ‘rules’ of the model are implicitly marginalized, which can then easily be used against them. So, instead of admitting that the model has a strong normative character, in that it creates new moral categories, the outcome is proclaimed as scientifically sound ‘proof’ for the normative biases of those who invented it. It becomes a dangerous, positive feedback loop, when those who use the models also believe in the absolute truth of the outcomes. It’s a slippery path and I prefer the basic framework of monetary operations and double entry bookkeeping Γ  la MMT as a starting point from which to fill in behavioural and empirical assumptions. If the foundations are solid, the margin of error is greater for all other parts. Mainstream economics seems to prefer building ever more intricate structures on top of unsound foundations.

    If architects had the success/failure rate of economists, we’d all be buried under rubble.

    If architect hadn’t invented civil engineers, we’d all be buried under rubble :-).

  52. Disclaimer: It’s probably clear to you that I’m not a trained economist (beyond a dodgy MBA that I took in one of my weaker moments), but I thought I’d remind you just to be clear. I come here out of personal interest and comment mainly to improve my own understanding, although I’ve been noticing a decline of my learning curve recently. That isn’t to say that I don’t occasionally mix in personal, politically motivated or hyperbolic language and most probably the odd nonsensical assertion. I guess that’s the price professionals like yourself pay for partaking in (or hosting) such an open format – you get to sift through all sorts of different levels of professionalism. But as I said, I’m here to learn, even if I’m biased towards MMT, so feel free to tear my arguments to shreds, if you’ve got the patience (and if you feel they’re worthy of it). All the more to learn.

  53. vimothy, I have a PhD in philosophy from Georgetown U. in Washington, D.C. My dissertation was on the logic of justification in Wittgenstein’s On Certainty, and my advisor was George L. Farre, a physicist turned philosopher who specialized in philosophy of science.

    Wittgenstein viewed his approach to philosophy as a kind of “therapy” to bring logical clarification to areas of confusion. In the logic of justification this involves distinguishing norms (rules) from descriptions (assertions of fact) in various universes of discourse. On Certainty is about the logic of justification in ordinary language, specifically about how norms are rules that delineate the framework of a universe of discourse and provide its boundary conditions. Norms function as rules of inclusion and exclusion in a particular universe.

    “Commonsense” expressed through cultural conventions and institutions furnishes the norms for ordinary language in any time and place. Dogma serves this purpose in institutional religion, while custom does the same in religions that are not formally institutionalized. The purpose of science is to correct errors that arise owing to sloppy thinking in ordinary discourse, but scientists are not above ordinary language and often fall victim to logical error themselves, since it is extremely difficult to think independently of ordinary language and its worldview consistently. As the philosophy of mathematics shows, even brilliant mathematicians have gotten confused about key concepts.

    In economics, the question is whether there are non-empirical norms masquerading as general descriptions (laws) that are privileged because the role they play in the universe of discourse. The existence of “heterodox” economics suggests that there are, in that heterodoxy is defined by orthodoxy as a standard being imposed. I think that Bill has provided ample examples of such statements functioning as norms in a number of his posts, along with a critique of their lack of solid empirical warrant. Their pedigree is logical, that is, based on how they function a priori in model construction. Often, they are not borne out empirically.

    Of course, this is all hotly debated among schools. It is not always a simple matter to verify claims empirically in the social sciences, for example, owing to the difference between nominal (notional) and real (actual), and the difficulty in designing and conducting experiments to test hypotheses empirically. The data that economists use, especially in macro, are estimates, and populations cannot be manipulated for practical and moral reasons. But putting those difficulties aside, there seem to be instances where rules are being imposed arbitrarily.

    As a result, there is no overarching normal paradigm in economics. A normal paradigm is a feature of a developed science. Without a normal paradigm, a field is essentially a lore. “Orthodoxy” is the prevailing universe of discourse, and its norms are culturally (conventionally and institutionally) determinative of what is acceptable. Other views are marginalized or excluded by the prevailing norms. I am reminded here of Mark Thoma’s refusal to debate Bill because Bill rejects the money multiplier. Here the money multiplier is functioning as a norm, similar to a dogma. Why debate with a heretic?

    I am not claiming that economics is “the same as” religion or theology. But I am saying that there are a lot things that resemble theological discourse and religious behavior, and stand in contrast to what is normally accepted as scientific discourse and behavior, even though mainstream economists, especially freshwater types, try to disguise what they do through the use of complicated econometric models. This, of course, recalls priestcraft, where supposed specialists alone have access to privileged knowledge and skills owing to their position, training and initiation. It doen’t take a degree in sociology to see such phenomena manifesting in behavior, especially institutional behavior. Just as there is institutional religion, etc., there is also institutional economics. Mainstream economics is institutional, which is what “orthodox” implies sociologically.

    The problem that I see is that assumptions that are admittedly designed to make the model work are then accepted as somehow proved by the model iaw the coherence theory of truth. If the model appears to be falsified by events iaw the correspondance theory, then an ad hoc explanation is brought in to save the model. Thus, the question becomes, what does it take to falsify an economic model? Why is it that models that were falsified in the past recur? (Krugman and DeLong frequently ask this.)

    In summary, I find a lot of mainstream Neoliberal and New Keynesian economics to be ideological instead of scientific, especially when applied to policy. For example, I grant that a representational agent is a logical construct that is used legitimately to fashion a simple model to reason about complex issues. But then I see a jump from assumption to reality, as it that is the way the economy actually works, when other life sciences contradict this view of motivation and behavior based on empirical research into how human being actually function. Bill has also mentioned this particular example, btw. Bill has also shown how this and other such logical jumps from model to reality have had deleterious effects on policy and are continuing to do so, apparently because of a resistance to acknowledge failure that seems inexplicable other than by ideological commitment.

    See David Sloan Wilson’s series on economics and evolution as different paradigms, for example, for how a life scientist views mainstream economics from the vantage of his field.

  54. With my own caveat emptor in place… You are totally misunderstanding the ontological status of behavioural assumptions in economics. You cannot treat the methodological and political arguments as identical–that’s a category error.

    You must make generalised assumptions about behaviour or your model has no content. This is a property of models and not of economics. You couldn’t model the climate without makings assumptions about the behaviour of its constituent parts, for example. Since that is the case, you cannot simply dismiss mainstream econ on that basis. Specific modifications to models are how the discipline progresses. These modifications are distinct from any political arguments that might or might not be advanced as a consequence of making or not making them. Don’t like an assumption (say, profit maximisation)? What would you replace it with, and it what context? How does this improve the model and according to what criteria? How do validate this modified model? Harder than it looks. It’s not politics either.

    Models are driven by assumptions. This might be profound or it might be trivial, but it is surely no surprise to anyone doing serious research. Do you really think that all these professors at the top schools who have been doing this for decades are suffering some kind of collective hallucination?

    “those who use the models also believe in the absolute truth of the outcomes”

    Apparently so! That’s just absurd–no one believes in the absolute truth of the models. There would be no discipline at all if this were the case. Everyone already doing this stuff is already alive to these problems. How could they not be? Think about the logical consequences of your own assumptions. Perhaps they might benefit from some revisions.

  55. vimothy: this is not a claim about causality-Tom Hickey, I’m looking at you!)

    See Ludwig Wittgenstein, Tractatus Logico-Philosophicus, 6.3-6.375.

    According to TLP 6.32, the “law of causality” is the form of a law. Independent variables and dependent variables stand in a relation given by a function, e.g, y=f(x). These forms are a priori and purely logical. Therein lies their necessity. Whether they correspond to events is an empirical matter. It only takes one counter-instance to disconfirm a universal proposition, so no scientific law is ever confirmed by experiment. This is the relation of theoretical and empirical. Thus, there are no “natural laws” in the sense of what is given. Theories and the models constructed to represent them are means we use for organizing and understanding the given. They are tentative.

    The question is whether there are “laws” of economics and what that means. Calling things laws does not make them so, physics envy notwithstanding. Is Say’s Law a law?

    Is there an economic law (function embedded in a theory) that relates to inflation? Does empirical evidence support the claim?

  56. Tom,

    You are not reading a word I write, I swear. Thanks for the substantive post on Wittgenstein above, though–that was excellent, if misguided.

    I am interested in a sociological perspective on knowledge formation, but it shouldn’t be applied partially. It must be distinct from any methodological arguments.

    What you write about models suggests that you are only wrestling with these issues in an abstract sense. Do you really think that models persist despite their obvious flaws because of the stupidity or occlusion of their users? Models persist because they are useful (relatively). What is the best way to make out of sample forecasts? This is a central issue: hit me with your unfairly ignored alternative.

    Re your PhD, I only mentioned it because you seemed to be saying that I needed a PhD in econ in order to be able to form a coherent opinion on the matter, but I don’t understand how you can advance that argument without having one yourself.

  57. Dear vimothy (at 2010/11/28 at 8:22)

    You asked:

    Do you really think that models persist despite their obvious flaws because of the stupidity or occlusion of their users?

    Answer: definitely.

    You asserted:

    Models persist because they are useful (relatively). What is the best way to make out of sample forecasts?

    No they do not! Not one of the mainstream economic models taught to students and which form the basis of the extensive published orthodox literature predicted the recent crisis. Those models were claiming the business cycle was dead – the great moderation. Conversely, Modern Monetary Theory insights predicted that the crisis was coming (we were writing about the impending meltdown from the debt etc in the 1990s).

    Further, not one of the dominant mainstream models gives credence to the only thing that has saved the world from another depression – fiscal policy. The models claim it to be ineffective. Further, they claimed that the type of monetary operations currently being engaged in by the central banks would be highly inflationary. Poor prediction.

    So there is virtually zero predictive value in the mainstream models yet they persist. To understand why you really need to sit through a PhD program in economics to appreciate the subtle and not so subtle indoctrination that goes in those so called “educative” environments. It is pernicious to say the least.

    Models persist because they suit the dominant ideology. The empirical content of mainstream economics (macro) is very low yet its influence is huge.

    best wishes

  58. vimothy, I confess to knowing nothing about economics as a field of study. I took Econ 101 fifty years ago in college. In addition, I never specialized in math. So I don’t pretend to have anything substantive to say about economics, although I can repeat things I have picked up. But I do have what I think are intelligent procedural questions about economics based on philosophy, philosophy of science and philosophy of economics.

    I am interested in economics now for two reasons: First and chiefly, I am a citizen concerned with policy and right now that revolves around economic arguments. I got interested in economics at the time of the GFC, when it seemed clear that to be an informed citizen, one had to understand the economics underlying policy. This pursuit led me to MMT rather serendipitously. Secondly, as a philosopher, I look at economics from two perspectives, as a subset of ethics by way of political philosophy and the life sciences and from the vantage of the philosophy of science as a methodological investigation. I find it curious that there can be so much disagreement in a field that purports to be scientific when if there were a proper method, key questions would be decidable based on methodological criteria. But they don’t seem to be. I am wondering why that is. I am somewhat dismayed to find such divergence of viewpoint and argument even over what constitutes facts in economics. Given this environment, it seems to be to be a stretch to call economics a science in its present state. Again, it seemed to me that the MMT’ers had the best grasp on the methodological questions that need to be answered before a fruitful study can be undertaken.

    You ask why models persist. I don’t know why certain economic models persist. It seems that the case should have been closed on such issues long ago if there were an agreed upon method and criteria for settling such questions in the field. But there does not seem to be. How can it be a science in the absence of a methodology and criteria that contributors to the field agree upon? When there is such a method and criteria, questions can be answered forthrightly and views that do not stand up to scrutiny are dismissed and forgotten. What I see in economics is people picking sides, like in religious controversies. People that went to freshwater schools hold one thing, and saltwater schools another, and all other schools are considered backwater schools. It is a mess for citizens trying to get some notion of policy to find instead of clear answers to crucial questions that can be demonstrated empirically. I get the impressions that people are talking past each other. No wonder the world is in the mess it is in, if these are the grounds on which policy is set.

    vimothy, you are the economics student. You tell me what you think is the best way to make out of sample forecasts. Then tell me what your criterion for “best” is, and why. My job is to critique reasoning and criteria based on logic. You say that I speak abstractly. That is exactly what philosophers do. They examine reasoning critically using logic. That is their method.

    I happen to prefer MMT because it seems to me from the short time I have at this that they have presented a clearer picture than others have about current conditions, since they begin with operations and stock flow consistency rather than theoretical assumptions and models that only vaguely represent reality, most of which failed to predict the debacle. I say this not from an economic perspective because that is beyond my ability. I say it from a logical perspective and with some knowledge of philosophy of science. A good example is what happened when Her Majesty the Queen asked for an explanation from her economists for how they missed the crisis. There was a lot of hemming and hawing, saying that no one could have foreseen it. As Jamie Galbraith retorted, there were plenty who saw it coming. Who Are These
    Economists, Anyway?

    I’ve largely rejected the Austrian approach owing to an aggregation theory of society and a rejection of any substantive role for government as inefficient therefore ineffective. I’ve rejected the Neoliberal approach owing to assumption of “natural laws” like the invisible hand, the notion that macro is scaled up micro, and over-extending simplistic assumptions, as well as an overemphasis on quantification that submerges quality. I’ve rejected New Keynesianism as basically a compromise with Neoliberalism that also lacks appreciation for monetary operations.

    It seems to me that MMT’ers have the logic of operations and stock-flow consistent models figured out. MMT’ers got SFC from Wynne Godley, so it is not original with them. Circuitists like Marc Lavoie (Godley’s co-author of Monetary Economics) also have operations down, as far as I can determine, while it seems to me that many other economists are playing catch up in trying to account for the GFC, for example, since they don’t seem to properly appreciate the role of monetary operations and finance, owing to the assumption of money neutrality, for instance. While I am not in a position to critique the economics, the logic seem to be on the right track. So as a philosopher and a concerned citizen, here’s where I am placing my bets.

    Finally, you say, You cannot treat the methodological and political arguments as identical-that’s a category error. I don’t think that I am doing that. As a matter of fact, this is what I am accusing many economists of doing when they jump from assumption to “law” wrt policy. This is pretty consistently what I am hearing. While this is not all coming directly from economists, it is coming from policy makers and key politicians, and I assume it is what their economic advisors are telling them. What I suspect is that many economists are heavily influenced by their political persuasion and personal investment, and they fashion their economics to fit it. Some are more transparent in this respect than other. This is one source of economic “theology.”

  59. vimothy: One ought to be careful about blurring the line between methodological and political (or “positive and normative” in corporate econ-speak) critiques for that very reason.

    This is exactly what I am accusing many economist of doing. The underlying Austrian notion of society as an aggregation of “free,” that is, essentially unrelated individuals, hence, a rejection of society as a coordinated entity, is normative; the state is “bad.” The Neoliberal notion of “the invisible hand” as embodying the natural laws of markets is normative; liberalism is “good.” Welfare economics is normative; public purpose is “good.”

    Cognitive economics and institutional economics suggest that economics is inherently normative, because of the functioning of brains and behavior. There is no social thinking or action possible independently of norms. Physical science may be amoral. It is non-hierarchical and non-teleological. Life and social science are not; they are hierarchical and teleological. Pretending that economics is physics is silly. Indeed, the claim is normative rather than empirical.

    A central issue in philosophy of economics is whether economics is both normative and descriptive or purely descriptive. Political economy is indisputably both. Whether a “pure theoretical economics” exists is controversial. However, if there were such a paradigm, there would not be the plethora of economic schools disputing key fundamentals of the field, such as whether supply or demand is determinative and whether markets follow natural laws. On the face of it there seems to be no “pure” economics that has emerged as yet. Perhaps there will be a cognitive economics that is developed when more is understood about brain function. But from what we know about brain function now, human do not separate normative and descriptive, subjective and objective discretely. See, for instance, Descartes’ Error: Emotion, Reason, and the Human Brain by Antonio R. Damasio (Harper, 1995).

    What many economists and policy makers are doing is claiming that they are being guided by pure economic principles existing independently of norms; therefore, there is in effect no choice but to follow their prescriptions, or else nature will take its course. Those claims are easily refuted because they either cannot be substantiated, or the justification offered is deficient, as Bill is constantly pointing out here on this blog. These claims are normative instead.

  60. Bill,

    “Definitely”–you are not even prepared to admit the possibility you might be wrong? Everyone who disagrees with you is either stupid or mendacious? You can’t mean that.

    It is true that no mainstream models predicted the crisis. But this is quite different to being of no use and only deployed for religious or sinister reasons. Wise man once said, prediction is hard, especially about the future. When your model predicted the crisis, what weight did you assign to this outcome and why? How did you validate the model? What do your false positives look like? How did you deal with Lucas Critique / “tyranny of regression” type problems? How does your model do when precasting stagflation / associated meltdown for previous macroeconometric models?

    I am already sitting through PhD modules (4 yr PhD here so there is some overlap between programmes), so I have to say that I don’t think you are right when you say that taking a PhD will convince me of the religious nature of economics. I am friends with several current PhD students and we all are studying the same subject, as far as I can tell.

    Can we come back to my questions about inflation and money growth?



  61. Tom,

    Awesome post. Thanks for that. I am sure that I could not do it justice, even if I had the time. But let me have a go at some of the issues you raise…

  62. Tom,

    I forget the exact words, but I’m reminded of the quote (something like), “When I was young, my parents knew nothing; but as I grew older, I was impressed by how much they learned.”

    Your attitude as a concerned citizen is entirely appropriate and to be applauded, IMO. I do not believe that you need to have a lot of economics or maths to have a valid opinion about these matters. Rather, it is all the other stuff that comes along with it that is important. But you are obviously a smart guy and I’m sure you have these qualities in spades. I am not claiming that you should step back-you should step forward; you will better understand where other people are coming from. Get a copy of Wooldridge’s econometrics textbook, say, from Amazon and read it (all the maths you need is in the appendices). The entire text is devoted to dealing with situations that violate the classical model’s assumptions. That’s the subject!

    One of the key contributions of quantitative work in any discipline-be it education or economics-is that it forces you to be rigorous. The people who are actively engaged in this know how difficult it is-because they are actively engaged in it. The problems that you have uncovered are problems indeed, but they are not really news to anyone working in the area. If only it were that simple.

    You do not have a methodological critique. Therefore, there is no general critique beyond the political. This is fine, but different. I realise that you are criticising economists because they mix positive and normative critique, but you are proceeding here on the exact same basis.

    And how do you make good out of sample forecasts? By estimating a structural model and accurately specifying the “deep” parameters, taking due care to not hold anything constant that should not be held constant. Think about the failed CDO models-they extrapolated a trend, but they held something constant that should not have been held constant. Fitting a model to past data is not enough. This is hard. But here you are already through the looking glass, with the religious zealots and the rest of us losers who didn’t predict the crisis.

    We’ve done this before, you know.

  63. Vimothy,

    I don’t want to get into the whole political debate here, but the statement that there is no correlation between money supply and real GDP growth is wrong. Debt (e.g. as measured by either net borrowing from Z.1 or as measured by M2-M1) leads real GDP growth while the monetary base moves contemporaneously with it, in support of my previous argument that as incomes increase, the demand for base money by financial institutions rises and the CB accommodates in order to keep the rate at target. Therefore incomes increase first, as a result of dissaving, and this drives changes in real GDP together with an increase in the monetary base. Therefore nominal effects, or income effects if you will, are very important.

    That income affects are very important is believed by everyone, only asymmetrically. After all, economists argue that lump sum taxation is harmful to output, even though this is a purely nominal effect, right? The private sector does not have more or less to consume as a result of the government raising taxes on the private sector. This is purely a nominal income adjustment, and yet ask an economist who believes in “money neutrality” whether increasing taxes would inhibit economic activity. That will tell you how much they really believe in money neutrality. It will point out a whole host of logical inconsistencies and contradictions in the accepted theory.

  64. “as incomes increase, the demand for base money by financial institutions rises and the CB accommodates in order to keep the rate at target”

    The increase in base money is entirely in the form of currency held by the public. Has nothing to do with bank demand or rate targeting.

  65. unless you’re referring to currency passed through the banks to the public; maybe that’s what you mean

  66. vimothy:

    You said,
    “You couldn’t model the climate without makings assumptions about the behaviour of its constituent parts, for example”

    The differences are pretty huge:

    1. The basis for the majority of climate simulations is actual physics such as thermodynamics and quantum mechanics. They have been tested and retested vast amounts of times in labs, heavy industries, even high school labs. They are not assumptions anymore.

    2. The areas that the physics are too complex, chaotic or unknown to model from first principles are captured by simplified linear or quadratic equations derived from actual data. They are assumptions – but ones extracted from the real world, not made up on the spot.

    3. The models are validated by hindcasting and forecasting. Their predictive skill is quantified, measured.

    Not really a very valid comparison with economics models at all.

  67. RSJ,

    Before we talk about explanations for empirical regularities, we need to agree what they are. Can you link to some evidence please?


    Climate models are as big a joke as macroeconomic models. Everything you say about them is true for modern macro models as well, except the “actual physics” bait. “Validated by hindcasting and forecasting” indeed.

  68. Anon,

    I think RSJ means that total income has increased as a consequence of real output growth, which means that broad money has increased, which means that banks demand more narrow money to manage their expanded agg. portfolio (of course, the public may require more currency as well, but I’m not sure how significant this would be in and of itself).


    Interesting if dated non-technical case study of a model’s development here:

  69. Vimothy,

    That credit (however defined, whether M2, M3, or broader asset classes) leads the business cycle whereas M1 does not (and that M2 leads M1, with M3 leading M2, etc.) has been known for some time, and of course the endogenous money folks have been predicting this. For “evidence”, it’s a bit like asking why factor shares are relatively constant. This is a well known stylized fact. The most famous recent paper would be Kydland and Prescott (1990), but there are many other studies that show the same thing, and this has been long before 1990.

    Here’s a St. Louis Fed study:

    Alternately, you can download BEA and FoF data and run your own VAR among output, Net borrowing, M1, M2-M1, etc.

    Growth in credit leads the monetary base as well as the business cycle, to the complete befuddlement of the mainstream. OK — there is Bernanke and Gertler (1989), but that doesn’t explain why the broader aggregates lead the base even in the absence of a monetary policy shock, nor does it explain why modern economies cannot sustain de-leveraging, or even a slowing in the growth rate of debt, without this being invariably followed by recessions and a decline in real income, together with a decline in the monetary base.

    But from an accounting view, this is a trivial result.

    Spending = Income, and a variable component of spending is deficit spending, which does not subtract from other expenditures but adds to them.

    Therefore you would expect a decline in deficit spending to lead to a real income decline unless consumption spending suddenly increased to pick up the slack.

    But households smooth consumption, so this isn’t going to happen. But if you are using the funny accounting in which production = income, rather than spending, then it’s not obvious why M2 should lead M1, or why credit growth in general is required in order to have real output growth.

  70. RSJ,

    We’re talking about different things. I posted the data I was referring to in the other thread. Here it is again:

    “Few empirical regularities in economics are so well documented as the co-movement of money and inflation. Chart 2 shows that this relationship is true for broad money as well as the monetary base. The other side of the coin to this close relationship between money and prices is the absence of a long-run relationship between money and output growth, shown in Chart 3. Over the 30-year horizon 1968-98, the correlation coefficient between the growth rates of both narrow and broad money, on the one hand, and inflation, on the other, was 0.99. Correspondingly, the correlation between the growth of narrow money and real output growth was -0.09 and between broad money growth and output was -0.08.”

    See associated charts.

  71. Anon,

    “monetary base” is both currency held outside bank vaults as well as bank vault cash and reserves on deposit at the Fed. The point is that as prices rise, then currency demand will rise, as will bank vault cash holdings in order to meet the increased currency demand (cet. par.) as will checkable deposits (because households will keep a proportion of their wealth in the form of checkable deposits in order to make purchases, and therefore the checkable deposit amounts will rise when the prices of goods purchased goes up) and therefore so will reserves.

    Everything will rise because the nominal amounts go up. Here, it doesn’t matter whether you use the adjusted monetary base or the source base, as the same argument passes through. Obviously that is a cet. paribus argument, and it wont hold in the current situation, in which the AMB has been pushed upward due to CB intervention, rather than allowing the AMB to float in response to private sector demand.

  72. right

    required reserves and vault cash are so immaterial in the scheme of things, I forgot about them

  73. Vimothy, I understand what you are saying, but my point is that your correlation doesn’t tell you a whole lot. I agree with the long run correlation. You don’t need to cite charts — it’s pretty obvious that there must be a long run correlation. If prices are rising, then you cannot support that on a constant monetary base, the base has to go up, too, as do all the other aggregates. But the point of contention is whether:

    1) an increase in M1 causes inflation to rise.
    2) money is neutral in the long run

    Neither 1) nor 2) follows from the long run correlation between inflation and the base, or between inflation and M2 — Friedman’s favorite aggregate.

  74. RSJ,

    I totally agree!!!!!

    Phew. That felt good.


    “The other side of the coin to this close relationship between money and prices is the absence of a long-run relationship between money and output growth, shown in Chart 3…. The other side of the coin to this close relationship between money and prices is the absence of a long-run relationship between money and output growth, shown in Chart 3.”

  75. Sorry, should read:

    “The other side of the coin to this close relationship between money and prices is the absence of a long-run relationship between money and output growth, shown in Chart 3… the correlation between the growth of narrow money and real output growth was -0.09 and between broad money growth and output was -0.08.””

  76. Right. And that’s what I was responding to. His claim is observationally equivalent to the statement that credit drives both real output growth (over the short run) and growth in the base (over the long run).

    And frankly, the whole article that he wrote is completely irresponsible. He knows well that current theory does not assume that growth in the base causes inflation. The actual theories out there are much more complicated than that — even Friedman did not believe (and openly mocked) the “crude quantity view”, as Friedman argued that velocity was a function of the real interest rate, and NK folks would argue that it is a combination of the real interest rate and inflation expectations that drive inflation, and not the size of the base at all. But King is following a well-worn tradition of simplifying theory to the point of being misleading in presentations to non-experts.

  77. RSJ,

    Not only does he know that, he states it explicitly. Really, I sometimes think you guys just love to tilt at monetarists, and who cares what anyone else actually says or thinks.

  78. OK, but in terms of “long run” neutrality, if you believe that the economy follows a unit root — i.e. that it mean reverts to constant exponential growth, then everything is neutral in the long run.

    And historically, economies have done this.

    I don’t know whether they will continue to follow this path, but if you assume that they will, then all of economics is “neutral” in the long run, by assumption. Taxes, monetary policy, fiscal policy, current account deficits, none of it matters in the “long run” by the exact same methodology.

    But if you believe that only under the “correct” policy response can maximum output be achieved, then you would not say that money is neutral in the long run. What you would say is that the correct fiscal and monetary stance is necessary in order to get that 0 long-run correlation.

  79. “Not only does he know that, he states it explicitly.”

    He states it, but then says this:

    Despite appearances, however, these new models give no less weight to money than the older versions. Irrespective of whether the central bank uses base money or interest rates as the policy instrument, the quantity theory of money still applies. In the new models, monetary quantities play no independent role in the transmission mechanism over and above that summarised in interest rates. But, equally, in the old models too, monetary policy impacted on the economy through its effects on interest rates. The key question is not whether the central bank uses the monetary base or interest rates as its policy instrument. It is whether the equations which are embedded in both the old and new models of monetary policy exclude important channels through which monetary policy works.

    And this just isn’t true. You can change rates significantly with zero meaningful adjustment to the quantity of money. The two metrics (quantity of money vs. interest rates) do not measure the same thing at all. And the fact that there is no short run correlation between the quantity of money and inflation, whereas policy is short run, should tell you that the two measures are different.

    The difference is in the assumption that the CB is changing interest rates by changing the monetary aggregates, but it’s really a series of reversible transactions that have no material impact on the monetary aggregates. Quantity versus rates is not different views of the same thing, and in the models, only rates appear, but quantities don’t.

    If he was really serious about presenting those models accurately, he wouldn’t make any charts of the quantity of money against inflation, he would be charting real interest rates against inflation, and then you wouldn’t be linking to him. So I think it’s pretty fair to say that the point of this article is to oversimplify and make people think that it doesn’t matter whether you talk of quantity of money or of interest rates.

    For more honest presentation, take a look at the New York Federal Reserve document:


    “During the heyday of the monetary aggregates, in the early 1980s, analysts paid a great deal of attention to the Fed’s weekly money supply reports, and especially to the reports on M1. If, for example, the Fed released a higher-than-expected M1 figure, the markets surmised that the Fed would soon try to curb money supply growth to bring it back to its target, possibly increasing short-term interest rates in the process.

    Following the introduction of NOW accounts nationally in 1981, however, the relationship between M1 growth and measures of economic activity, such as Gross Domestic Product, broke down …

    By the early 1990s, the relationship between M2 growth and the performance of the economy also had weakened…

    …A variety of factors continue to complicate the relationship between money supply growth and U.S. macroeconomic performance…

    In 2000, when the Humphrey-Hawkins legislation requiring the Fed to set target ranges for money supply growth expired, the Fed announced that it was no longer setting such targets, because money supply growth does not provide a useful benchmark for the conduct of monetary policy. ”

    So the NY Fed says “money supply growth does not provide a useful benchmark for the conduct of monetary policy”, whereas King says “Despite appearances, however, these new models give no less weight to money than the older versions.” and then he goes on to give charts citing long run correlations between money and inflation, not interest rates and inflation.

    “Really, I sometimes think you guys just love to tilt at monetarists, and who cares what anyone else actually says or thinks.”

    I am criticizing KIng for misleading the public into thinking that an increase in the quantity of money is going to cause inflation. That’s what this is all about, right? I can’t help if its fun to tilt at monetarists — there are so many internal contradictions and they have the power. These are the guys running the CBs. They are the ones that should be held to account.

    You seem to be doing your own share of jousting here, but I’m not sure for what reason. I don’t remember criticizing you for anything, frankly, and I’m not sure why you are being so antagonistic. It doesn’t help make your argument.

  80. Vimothy,

    I am interested in the actual calculations. Quite suspicious of the numbers. The numbers you quote are high.

    Is there an averaging ? Lets go back to basic stuff, and build everything from scratch instead of quoting.

    The paper Mervyn King quotes such as some from the Fed seem to suggest that the relation holds for the US as well. Last time I checked some data, I didn’t find anything interesting.

    How is “long run” defined ?

    Also are you using logarithms instead of changes ? Important because the former suffers auto-correlation effects and the latter doesn’t.

    Looks like … a shady interpretation/usage of statistics.

    For example, take the CPI changes for the US and take the M2 changes, say monthly changes. Find the correlation of these time series. Don’t think you will get 0.99. Not even close!

  81. What I am saying @3:12 is

    Take series A and B. A is growing around 3% and B around 6% with some fluctuations.

    If you take logarithms and do correlation, you find a correlation of 0.99. If you do correlation for *changes* you will may find 0.7 or something.

    In spite of the fact that A is growing at 3% and B is growing at 6% !

    I can then make a plot look like its clustered around 1.

    Am serious about this btw – try it on excel.

  82. RSJ,

    Sorry if I’ve been short with you–at times it feels like we’re jut going round in circles. I think I should probably sit this one out for a while.

  83. Vimothy,

    Further to my comment @3:29 …

    Take a fresh excel file

    In cell A1 type =20*rand()
    In cell B1 type =A1*(1-0.8*rand()) – you can take other values instead of 0.8 as well.
    In cells C1 and D1 type 1
    In cell C2 type =C1*(1+A1/100)
    In cell D2 type =D1*(1+B1/100)

    So the series in C and D are the absolute values constructed out of the changes series A and B

    In cell E1 type =ln(C1)
    In cell F1 type =ln(D1)

    Drag the formulas for some 100 cells or whatever you wish

    Now calculate the correlation between the series in E and F…. you get 0.99!!

    Even though … The series in C rises to some astronomic value compared to the one in D.

    Now plot A vs B … you will see “clustering around the 45 degree line”.

  84. LOL, no problem Vimothy. We may have been going round in circles, in which case the fault would be mutual, or concentric.

    And I admit to having a particular dislike for King, and the paper you cited seemed (to me, at least) as a type of apologia for the old quantity of money view, or an attempt to argue, without explaining why, that the quantity view is still valid even though the current models accept only interest rates as parameters. But in fairness, he doesn’t actually cross the line and say this. He says “no money, no inflation”, not “money causes inflation”. He is careful, which kind of made me angrier. The abstract says “Nevertheless, there are real dangers in relegating money to this behind-the-scenes role.”, but the paper doesn’t explicitly mention of what these dangers could be; He just gives charts relating the quantity of money to inflation. The whole paper struck me as an exercise in disinformation, and I went off on him, and by proxy the profession. This is all in the context of our QE debates, obviously, with the open letters to the CB and crazy gold bugs.

    But I could have misread the whole thing, and in any case that was a side issue to the original debate about money neutrality and inflation. If you would have cited some other random paper with the same data, the discussion would have been more interesting.

    Ramanan — There was once a commentator here at Bilbo’s that gave a reference to a book in which Blinder argues that there is a long run correlation and points out that this is observationally equivalent to the money supply being endogenous, or demand driven. If, over the long run, the dominant type of inflation is demand-led inflation, which I believe, then you would expect the correlation. That the commentator was you. Why are you now disputing this correlation, and for what purpose? In all cases it would be more interesting to come up with a model explaining inflation, and then I think you would see that the model would predict a correlation.

    The only such model that wouldn’t do that is Mosler’s crazy idea that inflation is due to oil prices, so that if we all drive really slowly, there won’t be any inflation. In that case, there would be a long run correlation between, say, auto-accidents and inflation. But barring that model, pretty much every model assumes a correlation, and the meat of the discussion is in debating the model, not the correlation.

  85. RSJ,

    Don’t exactly remember the reference but anyway …

    What I am arguing is that the correlation is spurious. You use logarithms instead of changes, you get 0.99!

    As we move forward in time, everything increases prices rise and the money stock increases. How one interprets is important.

    The point about quoting numbers such as 0.99 or something gives a misleading picture. The money supply can be rising at a much higher rate than prices even though wrong extrapolation from correlation analysis seems to suggest something else.

    Also in my previous comment, I was trying to suggest that a graph can be made to look like money is neutral or such a thing but if you look at the bare numbers, you get a different picture. You just get biased by some points which lie along the 45 degree line.

    For example, the CPI increased by a factor of 3 or something from 1980 for the US but M2 has risen much much more.

  86. Further,

    I took the CPI and M2 data from Fred (St. Louis Fed) and here is what I get.

    CPI as on 1981-05-01 was 89.80 and that on 2010-10-01 is 218.71 so a 140% increase.

    M2 as on 1981-05-01 was $1648.4 billion and that on 2010-10-01 was $8664.3 billion so 430% increase.

    So money beats price by 3 times!

    Now, who said prices and money move 1:1 ?????????

    I take the correlation of ln(CPI) and ln(M2) and what do I get 0.98!!!!

    I take the correlation of changes and I get minus 0.1!!!

  87. vimothy:

    “Climate models are as big a joke as macroeconomic models. Everything you say about them is true for modern macro models as well, except the “actual physics” bait. “Validated by hindcasting and forecasting” indeed.”

    No, these models are actual physics simulations. The source code is public. ( for example). For an easy to use model with a GUI, have a look at

    They have done a remarkable job predicting the climate, including one from ’88:

    Where is an economic model that has done as well? It’s all well and good asking for empirical proof, but if even actual science backed by reams of physical evidence isn’t good enough for you, I suspect nothing me or Bill have to say will ever change your mind.

  88. Ramanan,

    If you have two time series that are both increasing exponentially (and that’s what we have here), then you want to take the log before finding the correlation. In general, you want to only take the correlation of time series that can be co-integrated somehow; in the case of exponentially growing functions, they will be in AR(infinity), but after you take the log, you’ll get AR(1) (hopefully :)). Then you try to find coefficients so that the difference of the logs is in AR(0) — is stationary. Then you can argue that what you have is a meaningful correlation.

    The way I think about it is that if you have two time series A(t), B(t) that grow at different order, or grow at an order whose difference is not stationary, then the set of points (A(t), B(t)) is not going to carve out a line, but a curve. And that means that your error minimizing slope is not defined — your correlation is an artifact of your sample. But exponential functions have infinite order, so first take the log, reduce the result to finite order, and then see if you can get a stationary residual. If you can’t then you haven’t even found a valid correlation to begin with.

  89. OK, the above suggested that the residual had to be stationary, but that’s not true, just lower order is enough, but it would be unusual to find an exponential time series that grows as e^(t^n) for n > 1.

  90. RSJ,

    No disagreement.

    I am just trying to tell Vimothy that bare data tell another story than pasting some statistical numbers such as 0.99

  91. Dear TomH, V et al,

    I think I can offer a simpler explanation: from the age that a human being first recognises itself as existing it begins to work at relations with others to get what it wants. The mind is simply a tool used to facilitate this. Unfortunately most people think that they are their mind; they also think that what they want is what they need – blithely ignoring the fact that it is already possessed by many – (and these have moved on to wanting something more of something else). Multiply this by 6.8B and you have planet ‘fiasco,’ the hotel we all call home.

    In many, the want is so powerful it eclipses such niceties as logic, integrity, honesty, selflessness etc. Ergo, you can erect any edifice of conceptual thought that you like, slice and dice it however you want – but it’s human nature that needs to change if you want things to work. People play with, create and follow all sorts of formulae if they believe that will get them what they want. Watch any ten-year old manipulate norms, subjects, objects, descriptions and concepts – anything to get what they want. They are happily, not locked down by convention.

    Every now and again you will find some truly courageous soul, who tires of battling with themselves, the world, and others, and dares to desire peace. They desire ‘freedom’ (a word loved and cherished in every language and culture around the earth).

    They also are aware of the absolute necessity for people to learn how to live with themselves, and extend this to each other through cooperation. Selfishness is seen as the old child-like paradigm, leading only to chaos (everybody jostling to get what they want); cooperation – because of new circumstances humanity find themselves in – the essential new, if:

    – the species is to survive with dignity intact as human beings;
    – the ‘skin of the planet” is to be sustained beautiful and vibrant; and
    – humans are to press on freed from conflict, pain and suffering, visited upon their own kind.

    Depends on how ‘kind’ and ‘intelligent’ this crown-of-creation turns out to be, n’est-ce pas?

    Because of constant proximity and utilisation of the mind, its very hard for academics to understand: concepts are secondary (should be written across the portal of every education institution)! There are endless discussions and books on this earth, and many more are being written. It’s human beings and ‘what they want’ that is the engine. The desire nature is the current motivating power or force in humanity; it captures the human will. And those who think that concepts can direct and temper this stupendous force are kidding. Today’s world is a testimony to this! Whatever conceptual rules the intellect may levy to direct this flow, the banks will be broken. All such building work shall be swept away. I have never ever seen academics consider a remedy to this, busy as they are in their constructions!

    It is the direction of this torrent that needs to be changed; the polarity has to shift from selfishness to cooperation, guided by ‘intelligence’ – a deliberate act of will. The step taken because the sky is blue and the earth is green: this is our reality, understood – and there is no other way forward, as far as everybody can see. Competition is the blind-fold of empowered selfishness which it does not wear for itself, even on level playing fields.

    Note I am associating the word ‘intelligence’ more with the power to ‘see’ something (‘in’-sight) than the faculty of intellection (which more often than not goes on in the dark).

    So, using the mind to clothe ideals in ideas is fine, or playing around with the mechanics or constructs of logic is fine, but if you want to see the (icon) material result on earth, then it’s across the river of the desire nature that the conceptual bridge must be built: people have to understand that a river in flood is a problem. Human beings are actually THE PROBLEM: not our belief systems, not the environment, not our theories about just about everything; and not the politics or the religions, the sciences or the economy – these are just our imperfect expressions of our imperfect selves. The first step towards sobriety is taken when the inebriate recognises within themselves both the problem and the solution. This is Intelligence! Taking responsibility is the first sign of its emergence. Then, and only then can the mind become useful as a tool of discrimination and creativity.

    There are people on this blog with incredible minds and faculties – but it’s what you create with them is of significance. Simple people know this instinctively: with wry remarks about ‘ivory towers’ and ‘clouds’. The mind can be used to generate more and more layers of theory (generally known by the great unwashed as procrastination). Or used as a mirror to reflect the reality of what actually is. To reach people’s understanding, and take them with you (as Bill does so determinedly on this blog)! Things are usually a lot simpler than the mind makes them out to be – human nature hasn’t changed that much over the millennia. It’s the ‘obviousness’ of things that matter.

    The burden of sharing that the Irish, or poor Latvians should be taking upon their shoulders, is the yoke of reasoned cooperation to lift their country, consciously – not blindly absolving the debts of predators and fraudsters, under the corrupt auspices of those who lie to lead and deceive, divorced from their humanity.

    Bill is right to call them out, and their bloody models!

    Cheers ,

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top