Home   Editorial   About   
The Tarshis paradox

Hector McNeill1

This is a re-issue of an article written in 2012 (past updates are recorded below) and I have brought it up to date to reflect its relevance to the development of the Real Incomes Approach.

I take advantage of this opportunity to reflect more on what Lorie Tarshis was communicating; it was important.

Lorie Tarshis
During the period 1967 through 1969 I attended lectures provided by Dr. Lorie Tarshis2, Head of the Department of Economics at the University of Stanford. His lecture procedure would be to present on the black board (in fact green) some aspect of Keynesian analysis. I should add that Tarshis did not limit himself to Keynesian analysis but would pop up cameos of the Monetarist or Classical schools' take on whatever he was talking about. Then for the last half hour or so he would sit on the desk at the front of the lecture room, swinging his legs under himself and fielding questions from students. These question and answer sessions were chatty, very lively and informative since the ground covered was immense and students were at liberty to ask any question on economic policy.

It all depends on what happens to ....

The impression left with me of all of those sessions was that in answer to just about any question, concerning what would happen to some aspect of the economy by applying a policy variable, Tarshis would begin with the words, "It all depends on what happens to...." prices, employment, interest rates, levels of debt... the list of dependencies seemed to be never-ending. All of this was intriguing in that the sense transmitted was that nothing is entirely what it seems, indeed, this lack of a black and white demarcation or predictability of outcomes added to the intellectual challenge we faced as students trying to grapple with macroeconomics; all of this was very stimulating and almost every question did not present a simple answer but rather, threw up a paradox. The main lesson I took from this was that as things stood economic policy is a massive economic and social experiment where things can, and often do, go wrong. The question arises as to the underlying professional responsibility of economists giving advice to governments which in practical terms has a significant element of risk. This risk, of course, is borne by the social and economic constituencies and can even involve job losses, losing their homes, declines in income and businesses failing. The most that can happen is the governing party will lose the next election only to appear at some later date and the economists advising them sink into the background. No one is accountable for the prejudice suffered by the social constituency.

Slumpflation as a counter-intuitive shock

A first major demonstration of the significant weakness of macroeconomic policies was the slumpflation crisis intensified by three major international petroleum price hikes which occurred in 1973, 1979 and 1980. These were all exogenous or externally-generated impacts. Slumpflation, also known as stagflation, was characterized by a combination of simultaneous rises in unemployment and inflation. This for Keynesians was completely counter-intuitive since conventional aggregate demand management policy for the economy was unable to address this issue, at first, even on theoretical grounds. To address unemployment by higher government expenditure would have made the inflation worse and to reduce demand to lower inflation would have increased unemployment. Before the petroleum price shocks of the 1970s, the United Kingdom economy showed a more or less stable relationship between unemployment levels and inflation rates known as the Phillips curve (after William Phillips) and first presented in 1958 as shown on the right as the curve p-c. This meant that raising interest rates or increasing tax to reduce aggregate demand might reduce inflation but at the cost of increasing unemployment. However by 1970 this simple relationship was breaking down and by 1979 had shifted to a specific point which combined high unemployment and inflation, slumpflation (as shown in diagram on the right).

The theory doesn't work!

I tried to apply "received theory" in economics, including what I learned at the School of Political Economy at Cambridge, to identify what I would consider to be an appropriate policy to control slumpflation; I failed abysmally. Having attended Lorie Tarshis's course I should not have been so surprised. But at first I did have doubts as to whether I had understood what I had been taught. However, I came to my senses when I observed that the "leading economists" were also not coming up with convincing solutions. Many stated that outcomes, that Tarshis had described as well within the bounds of possibility, were "unexpected" or they pointed to something that had not been predicted. The notable characteristic of the analyses they put forward was the uni-variable myopic analysis, common to all conventional analyses, trying to explain why variables were moving in the opposite to expected directions.

Initiation of work on the Real Incomes Approach

I decided that, indeed, I had understood economic theory but that it did not really represent a working model of the economy and therefore provided no predictable basis for tackling major market and financial failure. I became extremely frustrated, I would say incensed, at the unacceptable suffering of the social and economic constituencies, losing jobs, homes, the closing up of businesses, all caused by ineffective economic policies. The question for me became, "What are economists for?" No longer a student able to "play" with theory and enjoy the intrigue of uncertainty of Tarshis paradoxes, I set out to develop a model of the economy from the ground up starting with an analysis of supply and demand. As a result of 12 months research many problems became evident, not only with Keynesian theory but also with Monetarist theory being advanced by Milton Friedman of the University of Chicago. I later reviewed supply side economics only to find that it was a variant of fiscal policy with little supply side content. For a critique of supply side failings it is worth referring to Bill Clinton's recent University of Georgetown lecture: Some evidence on the failure of supply side economics. I proposed the name of KMS policies for Keynesianism and Monetarism and Supply Side because they have very similar views of the basic economic paradigm with each of these schools viewing the same aggregate demand model (ADM) through their own specific kaleidoscopic perspective.
The KMS Maze

The walls of the KMS Maze are Nominal Monetary Aggregates which impede progress because the basic economic paradigm does not provide an appropriate freedom nor incentive for economic units to take beneficial actions. Arbitrary values are cast in the form of broad nominal monetary values leaving the constituency to respond as best they can. But because such policies are not designed to reflect the extreme diversity of the conditions of economic units, any response, as far as the nation is concerned, will be suboptimal and to the detriment of significant parts of the economic and social constituencies.

In spite of policy

Robin Matthews
1927 - 2010

In 1968 Robin Matthews published a paper in the Economic Journal on why Britain had full employment since the war. In this he argued that far from injecting demand into the system, governments in the so-called "Golden Age of Keynesianism" had persistently run large current account surpluses, instead of the budget deficits that would have been the expected manifestation of a Keynesian stimulus. From this he inferred that fiscal policy had been not only deflationary, but strongly so in the post-war period, therefore something other than Keynesian fiscal policy must have been responsible. The answer, he felt, lay in demand arising out of wartime destruction and an unusually prolonged private sector investment boom. So success was more to do with private initiative, in spite of macroeconomic policies.

Significant policy gaps

Much of the gain in productivity and economic growth was related to investment in technologies but Keynesian theory represented this element by blank pages. What emerged in the development of the Real Incomes Approach is that there were and are two gaping holes in conventional economic theory and policy models:
  • failure to base economic growth on the incorporation of the specific contributions of technology, learning, technique, investment and innovation
  • failure to use compound indicators for policy to maintain control over critical variables

It came as a surprise to me to find that the main contributing factors for economic growth in the economy, that is, the role of technology, technique, learning, tacit knowledge, explicit knowledge and innovation are represented by blank pages in macroeconomic theory and practice. These factors are not just a major source of growth, they are also a major source of adaptability, flexibility and an ability to reduce risk. They are key factors that need to be taken into account in the decision analysis concerning the different determinants of the outcomes of transactions involving short term buying and selling, medium term investment and long term performance enhancement. I wrote and published a short paper in 1981 as a contribution to the series,"Charter House Essays on Political Economy," entitled: "On the Problem of Technological Ignorance amongst KM Economists"

This of course had been worked on before by Kenneth Arrow, Robert Solow and Nicholas Kaldor in the 1950s, all of whom developed interesting "general growth models" that provide insights into the mechanisms of growth. However, these models were not part of macroeconomic theory or policy practice.

What was needed was a policy model able to accommodate a highly diverse economic and social constituency so as to enable them to respond freely to market conditions by making the most effective use of their abilities and resources. This means that each economic unit needs to be free to select actions that are rational in terms of the feasibility, risk and using available resources and capabilities. What is not required is a one-size fits all macroeconomic policy which sets arbitrary monetary nominal aggregates and impacts incomes (taxation) and costs (interest rates) that have no specific relationship to the unique conditions of each economic unit.

Compound indicators

It is apparent that the outcome of the broad brush macroeconomic approach applied to a diverse social and economic constituency is the generation of winners, losers and those who end up in neutral policy impact states. This significant differential in policy impacts explains why traction of KM policies is so poor. It is therefore necessary to develop policy targets or indicators whose values are established by economic units as a function of their capabilities and objectives. This can help move all constituents towards a state of positive systemic consistency (see Positive systemic consistency). This requires compound variable or indicators. For example if one dwells on the two variables of unemployment and inflation it is easy to lose sight of the fact although they move in variable directions, that in all cases just one thing is happening to a single compound variable, real incomes:
  • If inflation rises at falling unemployment then real incomes decline
  • If inflation declines with rising unemployment then real incomes decline
  • If unemployment and inflation rise together (slumpflation) then real incomes decline more
The main finding of the Real Incomes Approach has in fact been the identification of a single compound variable or indicator in the form of real incomes. In all cases of historic and contemporary economic crises the issue has been a failure to focus on real incomes as a policy target. Real incomes can simultaneously represent an appropriate macroeconomic policy objective as well as an appropriate microeconomic objective at the level of each firm. In this way a coherence is achieved between macroeconomic policy and the free operation of microeconomic operations enabling the achievement of a positive systemic consistency. This is discussed in more detail in "Why real incomes?

The important point to realize is that rather than look to isolated but cross-related variables the important issue is to have a single indicator or compound variable whose state provides guidance on which determinants or contributory variables need to be managed more effectively. This is the only way to eliminate the generation of winners, losers and those unaffected by policy.

1 SEEL - Systems Engineering Economics Lab.

2 Lorie Tarshis (1911-1993) was a Canadian economist born in Toronto (BA Toronto University, MA and PhD in economics Cambridge, University). At Cambridge, he studied under John Maynard Keynes. He served with the USA War Production Board and then as an operations analyst at USAF bomber commands in Libya, Tunisia and Italy during WWII. He started teaching at Stanford University in 1946 ending up as head of the Department of Economics at Stanford University between 1950 to 1970. He wrote the first American textbook on Keynesianism "The Elements of Economics" (1947). Some in the McCarthyist anti-communist drive considered the work to sympathize with communism causing the university circulation to fall and resulting in Paul Samuelson's book "Economics" (1948) becoming the dominant Keynesian text in the USA.

Updated: 10 August, 2014: Shortened some sentences to improve clarity.
Updated: 18 October, 2014: added the phrase - ";in policy sweeping generalities serve little, it is the more complex detail that is important."