Tuesday, March 22, 2016

北京麻雀 - Shanghai - 5

A Paper

There is no universally accepted definition of the word "recession" in economics, though a commonly provided rule is "two consecutive quarters of negative GDP." Even less so the word "depression." For a profession famous for its arcana, this is a very strange omission. It would be one thing if these words were treated as vernacular, which had popular, but not professional, meanings. However, recession occurs regularly in journals such as Econometrica, and the past outgoing chairman of the Federal Reserve, is known for being a student of "The Great Depression." Even the common definition is well known for its problems: first, because by the time it occurs, the recession is likely over, second because statistical measurement of GDP is open to question, and often varies between Gross Domestic Product and Gross Domestic Income enough to change the definition of start and end point – even though, in theory GDP and GDI measure the same thing – and third because the centrality of GDP itself has been questioned as being the best measure of economic well being.

It might have been arguable before to say that the confusion of definition was acceptable flexibility in the face of having the lack of unified theoretical basis for the entire business cycle - and its causes. However, in early 2008, when the global downturn was just beginning there were arguments over whether a "recession" had started, and the consensus from many leading economists, was that the business slow down did not constitute a recession, and that mild stimulatory measures were enough. This proved to be one of the most tragic follies of economic policy in modern history, as the attempts to control inflation over stabilizing demand caused the economy and global trade to "roll off the table," and touched off a "freefall" in trade and global output. this was pervaded by members of center left and center right, though not from exact left or right.

This will not do, particularly since it is by no means clear that the structural problems which created this globally synchronized downturn have been addressed, and that there will be a next downturn is a metaphysical certainty. The lack of a clear definition, and of theoretical framework helped push policy makers into poor decisions in the early part of the economic crisis, nd therefore, a sound basis for classifying downturns would be an improvement on present practice.

This paper presents an outline of downturns, and what the meaning of "recession" and "depression" should be in the context of the anatomy of the business cycle. This theory is grounded in Keynesian macro-theory, combined with the use of an iterative model of mesöeconomic behavior using the theory of games. This theory provides a way of understanding that a downturns are based on different mesö-shocks: inflationary, disinflationary, and deflationary. It is shown that the classic post-war recession is a response to a disinflationary shock, while the inflationary shock is a common effect in developing economies. A depression, however, results from the deterioration of the monetary basis, and is the result of a deflationary shock. Each of these are characterized as much by the structure seen at the end of the downturn.

1.0 Micro, Macro and Meso Formalisms

Alfred North Whitehead remarked in Process and Reality that "The safest general characterization of the European philosophical tradition is that it consists of a series of footnotes to Plato." One might make the similar observation that Economics consists of a series of footnotes to Adam Smith, but that the most insightful writer in the margins is John Maynard Keynes. Smith's work laid the foundations for the microeconomic formalism, which relates to the balancing point between supply and demand, and while key points have changed, the most important being the marginal theory of value, his general idea, that under certain conditions, there is a tendency towards equilibrium between supply and demand, can be shown rigorously. The success of the market formalism, is that describes non-market activity better than other competing formalisms, including those based on Marx, or on alternative value theories, such as Austrian economics.

In its modern form the microeconomic formalism carries out its work in an infinitely dimensioned commodity space, between economic actors with certain properties. The actual decision process is modeled as a vector table. Classical micro-economics had one set of vector tables, but these are largely replaced by vector tables drawn from the theory of games. The "least loss" from an optimal barter economy is calculated, and shown under relatively weak assumptions to have a single price market economy as both the local and general minimum of the achievable curve. One can note, in the direction of Arrow, and expanded on by Sen, that this environment must either leave the distribution of welfare disordered, or that single individuals will determine the ordering, or it will not achieve the conditions of the market.

The macroëconomic formalism relies upon the insights of John Maynard Keynes as expressed in the famous General Theory. In the micro-formalism, there is a general clearing of the whole economy which produces the least loss, and for approximately half of the history of political economy and economics, it was argued that this point was also an equilibrium. 

However, the assumptions for a general equilibrium are far more contentious than for a Walrusian clearing of the economy, and Keynes showed that there was no general tendency to equilibrium in the totals of supply and demand. He analyzed the economy in sectors, and showed that absent government, the two valued market for liquidity and commodity is not stable.

These two formalisms do not directly connect, and attempts to do so have generally required either constructions such as "stickiness," which has no mathematical or theoretical basis, or holographic economic actors, who have the entirety of the information space of the economy inside them. This creates the need for middle formalisms, which include complex economic models, such as the Federal Reserve Board model of the US economy (FRB-US) which relates empirically observed relationships between components and crafts a series of equations to model the observed relationships. These models are neither strictly speaking micro or macro models. Nor are a host of other tools of the financial world, such as the Black-Scholes equation and its offspring.

These formalisms are then middle scale, or meso- structures. The definition of a meso- scale model can be stated simply in that it represents groupings of supply and demand which either violate essential micro- assumptions, or have a differentiation which is not visible at the macro level.

2.0 The Business Cycle as a Real, not Nominal Phenomenon

As noted above there is no generally accepted rigorous definition of the components of the business cycle, even though there are a host of models which predict, or require, the regular rise and fall of business activity. Empirically, such rises and falls are observed in virtually all industrialized economies where there is sufficiently reliable data to observe them. Generally such definitions assert that Real Gross Domestic Product (RGDP) is the essential indicator for a "recession." However, this assertion breaks down on closer observation. First, RGDP can be measured in a variety of ways, and the proxies for intermediaries of the components introduce meso-considerations, such as the means to measure intermediate values of goods among mutually dependent firms. This can be seen from the "statistical discrepancy" between Gross Domestic Product, and Gross Domestic Income. Second, the correlation between declared recession in the United States and falls in GDP is not clear, instead there is an equally good correlation between rapidly rising unemployment rate (U-1) and declared recession. 

Finally, there is the question of provenance of numbers, and the declarations by official bodies themselves. Economic numbers are compiled by official bodies, for the purposes of an official body. Is it a recovery, if no one but a small percentage at the top recover?

Most importantly, the definition is completely in hindsight. There is no way to tell whether a recession is in progress until such time as it has already occurred, and indeed occurred long enough ago that statistics have been compiled and revised. For the NBER, the track record is to declare a recession nearly a year later, with an average of 12 months after a peak or trough since the 1980 recession, and a high of 31 months after the date of its beginning, by which point, the majority of recessions are either over, or nearly so. What is needed is a definition of an economic downturn which allows policy makers and other economic actors foresight, not hindsight, since the safety record of driving out of the rear view mirror is not notably high.

From this we may offer the following principle: definitions should be useful. Applying this principle to the state of economic argument on recessions, we may say that the present cluster of definitions is neither descriptive, nor useful, nor does it follow from theory in any meaningful way, and it is empirically insufficient.

For these reasons, this paper will take rises and falls in business activity as a real, not nominal, phenomenon. That is, real economic activity varies in a rising and falling marginal distribution over long term trends, and this rise and fall manifests itself in measurable ways, including unemployment rates, real gross domestic product, real wages, real returns on investment and so on. However, these manifestations, while linked, lag each other because of the meso-structure of the economy. Such lags are well documented in the data.

What then of the terms "depression" and "recession." Since economics is concerned with recessions and depressions, it is particularly weak to argue that the terms are purely colloquial. Indeed the causes of "The Great Depression" and of Japan's "Bright Depression" of 1992-2007, have been called "The Holy Grail of Macroëconomics." It is difficult to argue for a Holy Grail in one breath, and then to say that it is purely a lay term in the other.
This discussion, unlike some in economics, is not of strictly academic value.
In 2007, Lawrence Summers penned a piece for the Financial Times published November 27th 2007, in it he wrote:

“Three months ago it was reasonable to expect that the subprime credit crisis would be a financially significant event but not one that would threaten the overall pattern of economic growth. This is still a possible outcome but no longer the preponderant probability.”

“Even if necessary changes in policy are implemented, the odds now favour a US recession that slows growth significantly on a global basis. Without stronger policy responses than have been observed to date, moreover, there is the risk that the adverse impacts will be felt for the rest of this decade and beyond.”

However, what was his policy proposal? On January 6th 2008 he wrote again for the Financial Times:

“How large should a programme of fiscal stimulus be? It depends on what else is done to help the economy – a subject to which I will return soon. But a $50bn-$75bn package implemented over two to three quarters would provide about 1 per cent of gross domestic product in stimulus over the period of its implementation. With some multiplier effects the total impact would be in the range of 1 per cent of GDP over a year. This seems large enough to take some burden off monetary policy and yet unlikely, if properly implemented, to risk substantial damage given flexible monetary policy if the economy proves stronger than expected. After many months of behind-the-curve policy, moving to implement such measures seems more prudent than waiting till the necessity of even greater ones has been unambiguously established by further pain.”

“Given that many of the indicators he used to predict the rising increase in the chance of a large 2008 recession in 2007 were on the mark, for example a 25% decline in residential property values, the size of fiscal stimulus he settled on was approximately the size of the one which was passed in the spring of 2008, and by September, he was admitting it was not enough, as evidenced y his testimony before the House Budget Committee, where he again supported a plan which was far too small.”

This is not to single out Prof. Summers in particular as having been singularly out of step. In fact, when he presented these views, there was commentary that he was being to radical and pessimistic. However, since he is now a core part of the economic team making decisions in the United States, the lack of forward guidance, even given his belief that this crisis was quantitatively different, shows that the economic discipline needs to have a more realistic framework for dealing with economic downturns.

Consider the testimony of Ben S. Bernanke, self-described "Great Depression Buff" on February 14th, 2008/9:
“At present, my baseline outlook involves a period of sluggish growth, followed by a somewhat stronger pace of growth starting later this year as the effects of monetary and fiscal stimulus begin to be felt. At the same time, overall consumer price inflation should moderate from its recent rates, and the public's longer-term inflation expectations should remain reasonably well anchored”.

“Although the baseline outlook envisions an improving picture, it is important to recognize that downside risks to growth remain, including the possibilities that the housing market or the labor market may deteriorate to an extent beyond that currently anticipated, or that credit conditions may tighten substantially further. The FOMC will be carefully evaluating incoming information bearing on the economic outlook and will act in a timely manner as needed to support growth and to provide adequate insurance against downside risks.”

Or in more direct terms, the United States would experience a slowing of growth, but not an actual Recession. We now know that the National Bureau of Economic Research (NBER) would declare a recession to have begun in December, of 2007, and that banks did not have "strong capital positions." Some 8 months later, Bernanke would tell the Nancy Pelosi, the speaker of the House of Representatives of the United States, that unless the Troubled Assets Relief Program he and Paulson had written were passed "there might not be a US Economy by Monday."

From this, and numerous other examples, sufficient to fill a book length treatment, it is clear that policy response to the coming economic crisis was insufficient, despite ample warnings from respected, and less than respected, quarters.
We can therefore conclude, from an empirical survey of the core policy makers of the time, such as Treasury Secretary Paulson, Chairman of the Federal Reserve Board Ben S. Bernanke, and former Treasury Secretary, and advisor to future President Obama, Larry Summers, that the mainstream opinion of the time was not equipped to deal with the coming crisis. 

While better tools would certainly not have prevented the crisis, by themselves, because individuals need to use tools that are available, they cannot use tools that are not available, and just before crisis is a poor time to introduce intellectual tools, primarily because the people in charge are going to be the ones that chose the policies that led to the crisis in the first place.

3.0 The General Definition of Downturns and an Empirical Example

In economic literature, there are three distinctly recognized sources of downturns.
The classical post-war recession is rising inflationary pressure being blunted by the monetary authority, generally the central bank, raising interest rates to retail and commercial banks, which, in turn, tighten credit standards, and this, in turn, reduces aggregate demand. Or a disinflationary shock.

A crisis of confidence, where external investors respond to inflation and excessive public indebtedness by rapidly withdrawing from a nation. This can be said to be the fear that induces central bank action in the first case. Or an inflationary shock. If responded to by unlimited expansion of the currency supply, this becomes a hyper-inflation.
A contraction in the money supply sufficient to create a self-sustaining contraction in investment and consumption, that is a liquidity trap. Or a deflationary shock.

The question for the empirical study that follows is whether there is empirical justification for these definitions, and if so, what would an application of them say about how downturns have manifested in the United States.

The following table is calculated strict from the Bureau of Labor Statistics Establishment Survey of seasonally adjusted payroll positions, or CES0000000001. The methodology is to locate all hiring downturns of more than two months duration in the survey's history, and calculate the employment as a percentage of the peak of employment, and the length of time to return to that peak. This data is not a "return to full employment." Nor does it adjust for population. A more sophisticated version would only increase the size of the effects, since this would adjust for trendline growth. However, for meso-purposes, the raw percentage and return is more correct.

The line across the top represents the period of hiring and recovery, the second number is the number of months from first drop to return to previous peak of employment, though in the case of the present that return has not occurred. The numbers in the column represent the percentage of the previous peak of employment, and the sum of the employment deficit, not trendline adjusted, is the top number in the column.

From the above we can see that there are three distinct downturn types in the survey's existence:

The classical "V" shaped downturns from 1943 through 1981-82, ranging in size from the abortive 1980 downturn to the 1943-46 downturn. Some of these came in close enough proximity to be classified as "double dip" or "W" shaped downturns.
The downturns from 1991 forward, which have a long trough like character.

The first of these categories would seem to fit neatly into the common categories outlined above, as a disinflationary shock. However, the second two groups do not, as groups. The World War II downturn is relatively easy to describe as a drop in government demand and the time required to convert capital to consumer demand and begin normalization of a civilian economy. However, no description has been advanced which explains the 1991-94 downturn, the 2001-2005 downturn, or the present as a group. Nor does the disinflationary shock explain the 1948 downturn, which was accompanied by actual macro-deflation, that is falling general price level.

What are the differences here? For one thing the most obvious which is the slope of rehiring: the first group has a much sharper return of employment. The same is true of the onset, however, not as clearly. The first group of downturns have gone to positive slope between 8 and 12 months, and have positive slopes over .0025. The second does not exceed .002 until well after two year after beginning of falling employment, or more.

This leads to the sheer length of the second type of downturn. The shortest is longer than any second type downturn, and the longest is already destined to be longer than the demobilization downturn.

The entire anatomy of the business cycle is beyond the scope of this inquiry, instead the focus is on three parts: the period of time immediately preceding a downturn, when the prediction of a downturn is most useful, the downturn itself, and the period afterwards. The purpose of the inquiry is to see if there are measurable or observable conditions which separate the different types of downturn. This is not the same as causal relationships, which may be invisible to outside measurement.
In the empirical study provided, there is a clear anatomy of the downturn period, which can be characterized by the slope of change of payroll change. Mathematically de/dt. In the next section the two types of downturn seen are called “U/V” and “L.”

Onset The onset period, which includes months before those shown on the chart include slopes that are very shallow and near 0, with some negative slopes. Somewhere during this period, slopes become consistently negative and the downturn proper begins. U/V recessions generally have a short onset period, and within 3 months of consistently negative slopes, the onset is visibly over. For the L downturns, onsets are much longer, and remain so even after there is a consistently negative slope.
Fall This the period where the bulk of the payroll positions are lost. Slopes become consistently highly negative.
Trough This is the period where job losses end and hiring returns, however it does not have a consistent profile. In many downturn periods, hiring bounces off the bottom sharply, and in these cases there are often set back months with a large number of job losses. In the “L” downturns, hiring slowly reduces its losses, and slowly hiring returns. The exception of the trough in the 2007-Present downturn is entirely attributable to government hiring.

Recovery This is the period which shows the most difference between the two kinds of downturn. In the U/V downturns, hiring slopes are roughly equal to the reduction slopes. In the “L” periods of relatively low hiring, and even job losses can continue for years after the original recession.

From this empirical survey of one manifestation of downturns, the present categorical causations are, at least, incomplete with respect to the American experience beginning from World War II. It is important not to draw specific causation from this data, since it is merely one measurement out of many, does not take into account inflation as a driver of employment, higher wages as an incentive to disemployment, costs of production and so on, even within the labor sector.

This also leads to the conclusion that a more sophisticated anatomy of recessions is needed, one which encompasses a broader range of structures of downturn and return to trendline growth, including those from before modern statistical record keeping in the field of economics.

4.0 Mesotheory

In Book II Chapter 5 of The Wealth of Nations Adam Smith argues that capital is of one of four kinds: extractive, manufacturing, transporting, or distributing. And so he begins an inquiry into the features and requirements of each, and in so doing examines a scale of the economy between the great movements of money, and the individual action of prices and demand. Or, mesoëconomic thinking. He also makes meso-arguments about which ventures need to be “joint stock companies.” He concludes that only a few enterprises require the access to capital of what we would think of as a corporation.

A mesotheory is one which relies on the different features of particular classes of activity, or of particulars of demand. Dopfer and Holland have both argued that the economy controlled by firms, between households and government, constitutes the “meso-” economy. An example of meso-theory in Smith is his discussion of who pays ground rents, showing them to be a particular kind of natural monopoly. One example comes from “A new approach to business fluctuations: heterogeneous interacting agents, scaling laws and financial fragility.” Which posits that the hetrogeny of firm size is the driver of different sized downturns. The model is flawed, in that during downturns, it is not that larger firms fail, in fact, these fail at lower rates in larger downturns than in medium sized ones compared to all business failures. However, this is a positive feature of meso-models that the provide rather direct measurable tests such as this, rather than arguments over the statistical likelihood of data fits.

The reason for turning to meso-models is that the nascent idea of deflationary versus inflationary downturns is not sufficiently backed by evidence. The explanation of inflation fighting by the central bank, that is a disinflationary shock, causing a downturn explains tolerably well the 1955-1991 period. However, it does not explain the 1948 downturn well, which showed, before and during, periods of deflation as measured by both the GDP deflator, and the Consumer Price Index for Urban consumers, the CPI-U, as well as the CPI-W, which for technical reasons is a more accurate gauge of inflation starting in the 1980's. This will be discussed later in the book.

Instead the macro-economic theoretical reason for the concern over the present downturn in the United States is the problem of the liquidity trap: the Federal Reserve cannot lower interest rates any more than they already have. As Dr. Paul Krugman is fond of quipping, there is a magic number for interest rates, and that number is zero. Thus while deflationary, the 1948 downturn was not caused by monetary and fiscal policy being ineffective, but, instead, that there was a positive willingness to reduce indebtedness, then at a high level as a ratio of GDP because of the just concluded World War, at the cost of employment. what has happened to employment levels, is not the same as in the direct pre-war period, again when to a pattern which is different from it.

But this indicates that inflation reduction and employment are coincident of the same underlying real economic process. To emphasize this one can take the CPI-W, and the percentage change in payrolls, and see that during the entire study period, they move in remarkable lockstep. Thus it is not inflation or deflation, per se that shapes hiring slumps or recessions. Nor is it interest rates alone, because several periods of prosperity have had higher interest rates than poor recoveries.
However, one cannot argue that low Federal Reserve Interest rates are a sign of an “L” shaped recovery either, because they were relatively high at the onset of the 2001-2005 hiring downturn. In short, the standard explanations require significant epicycling to explain the plain empirical data selected. It is possible to make arguments, but these then become qualitative descriptions, of the kind that orthodox microëconomics and macröeconomics pretend to eschew.

This kind of qualitative description includes the observations of Krugman in Depression Economics and the descriptions of Larry Summers quoted above: that this downturn is more similar to the 19th century panic than the recessions of the post war era. From the gross point of view of the data, there are certainly similarities, particularly the long period after the end of contraction where hiring is slow, even as GDP may or may not recover. While there is a resistance in the economic community to going before modern statistical record keeping, this produces a paradox that normal experience is conditioned by a particular policy regime, and assumptions of self-righting tendencies based on previous data are largely invalid, because the data used to support these safe conclusions of innate equilibrium are drawn from a time when government worked very hard to maintain equilibrium. After all, until the present downturn the median price of US residential residences had not fallen in the post-World War II era.

The organization of this book may seem to go the long way around to answer the cluster of questions asked here: what is a recession? Is there a difference worth using the word “depression” as opposed to recession? What are the ways of measuring recessions and depressions? What are the policy implications? Are downturns recessions or depressions from the beginning, or does the policy response shape the downturn and its aftermath?

Instead we ask a different question first: what is money?