Home > Economics in Two Lessons > Are recessions abnormal (crosspost from Crooked Timber)

Are recessions abnormal (crosspost from Crooked Timber)

November 9th, 2015

I’m on to the macroeconomics section of my book in progress, Economics in Two Lessons. The key point of this section is that, whereas the academic economics profession has wasted most of the last thirty years on the project of founding macroeconomics on (some near approximation of) standard neoclassical microeconomics, the validity of the core results of neoclassical microeconomics depend on the assumption that the economy is operating at full employment[^1]. This observation isn’t original – it was why Keynes saw his theory as saving capitalism from itself. Even the title I used in this post on the macro foundations of microeconomics turns out to be a reinvention of the wheel.

Having noted the importance of the full employment assumption in the abstract, how relevant is it? If the economy is, with notably rare exceptions, at, or close enough to, full employment, then it seems safe enough for economists to continue, as the profession has for 40 years or so, to treat macroeconomics as a special subfield with little relevance to the rest of the discipline.

To put the question simply, are recessions abnormal?

Are recessions abnormal ?

Much economic discussion is based on the implicit assumption that the ‘normal’ state of the economic or business cycle is one of full employment, and that mass unemployment is a rare exception to this state. On this view of the world, recessions are temporary interruptions to a pattern of stable growth. The pattern of economic activity associated with a ‘typical’ recession is ‘V-shaped’, with two or three quarters of sharp contraction followed by an equally rapid expansion which restores the economy to something close to full employment. The widely-used informal definition of a recession as ‘two quarters of negative growth’ reflects this view.

There have, however, been lengthy periods when the economy has behaved quite differently. In deep depressions, however, such as those following the Wall Street Crash of 1929 and the Global Financial Crisis (GFC) of 2008, the contraction is sharper and the recovery, when it comes, is slow and fragile. Even after years of ‘recovery’ employment remains far below normal levels.

During the Great Depression the ratio of employment to population in the US fell from 55 per cent in 1929 to 42 per cent at the depths of the slump in 1933. Despite the expansionary effects of the New Deal, employment remained weak throughout the 1930s, with the ratio only reaching 47 per cent in 1940.

The same is true the ‘Lesser Depression’, which began with the Global Financial Crisis at the end of 2008 and has continued ever since. The ratio of employment to population in the US fell from 63 per cent to 58.5 per cent at the onset of the GFC. Despite years of ‘recovery’, the ratio has remained at or near that level ever since.

There have also been lengthy periods when recessions were consistently mild, so mild that many observers believed the business cycle to have ceased to operate. The longest such period began with the outbreak of World War II in 1939, and came to an end in the 1970s. This ‘long boom’ began when wartime economic planning mobilised all available economic resources. Most economists expected the economy to decline when the war ended, as had happened after World War I. However, under the influence of Keynesian economics, governments in the decades after World War II were committed to maintaining full employment and did so with substantial success.

The Keynesian system of economic policies ran into difficulties during the late 1960s. The 1970s was a chaotic period of high inflation and periodic high unemployment. In the mid-1980s, the economy began to recover, as the Federal Reserve developed new tools for economic management. Recessions continued to occur, as in 1990 and 2000, but they were relatively brief and mild. By the early 2000s, economists discerned a period of relative stability which was quickly christened ‘The Great Moderation’.

However, the Great Moderation turned out to be an illusion. Whereas the Keynesian long boom had lasted for decades, the Great Moderation was already over by the time it was ‘discovered’. The bursting of the Internet bubble in 2000 marked the end of strong employment growth in the US and much of the developed world. The GFC turned slow growth into sharp decline, followed by stagnation.

Taking these disparate periods into account, can we regard full employment as the normal state of the economy, subject to temporary interruptions associated with downturns in the business cycle? The evidence suggests that we can not.

Before looking at the business cycle, it’s important to observe that, even under the conditions normally described as representing full employment, around 5 per cent of the labour force is unemployed and actively looking for work at any given time. In addition, substantial numbers of workers would like to work longer hours while others would enter the labour force and seek work if they thought such a search would be successful.

In treating such a state as one of full employment, the underlying assumption is that, under these conditions, unemployment arises from difficulties in matching workers with jobs, rather than from a shortage of jobs in aggregate. (This will be addressed later on).

Turning to the cyclical data, the United States was the first country where systematic study of the business cycle was undertaken, and therefore yields a long series of data based on consistent criteria. The National Bureau of Economic Research was set up in the 1920s and has long been the source of official estimates of the start and end dates for recessions in the United States. According to NBER estimates, over the 100-year period since 1914, around 25 years have been spent in recession.

However, this classification is, in critical respects an underestimation. The NBER treats recessions as beginning when the economy starts contracting, and ending when economic growth resumes. This treatment works reasonably well for ‘typical’ ‘V-shaped’ recessions where the recovery phase restores full employment within a few quarters.

In deep Depressions, however, such as those following the Wall Street Crash of 1929 and the Global Financial Crisis (GFC) of 2008, economic weakness persists long after the end of the contraction phase. At least from the perspective of labor markets it would make more sense to treat the recession as continuing until the economy returns to its pre-crisis growth path. In particular, as long as the employment-population ratio is far below its pre-crisis level, implying the existence of large numbers of unemployed or discouraged workers, wages do not properly represent opportunity costs.

To see the implications of this, consider the NBER data separately for the periods before and after 1929. Before 1929, contractions and expansions were about equally long, so that the economy was in recession a little under half the time.

Now, in addition to the NBER data, treat the whole of the Great Depression 1929-39 and the years since the GFC as recessions. On that basis, the US economy has been in recession for about a third of the period since 1929, only a modest improvement on the period 1854-1929.

But even this is an underestimate. The post-1929 average is pulled up by World War II when the government actively worked to ensure that everyone capable of working towards the war effort did so, and by the period of Keynesian macroeconomic management from 1945 to 1970. If these periods are excluded, the proportion of time spent in recession is around 40 per cent.

To sum up, except when governments are actively working to maintain full employment, the economy is in recession almost as often as not. The idea of full employment as the natural state of a market economy is an illusion.

[^1]: Full employment doesn’t mean zero unemployment, since some people are always changing jobs, or are in the process of leaving the labor market. Roughly speaking, the employment is at full employment in the sense required here when any additional job creation in one sector of the economy is feasible only by attracting workers away from other sectors.

Categories: Economics in Two Lessons Tags:
  1. Uncle Milton
    November 9th, 2015 at 11:19 | #1

    You might want to build in the new but actually old idea of how secular stagnation is going to keep growth low.

  2. Ikonoclast
    November 9th, 2015 at 11:23 | #2

    I agree basically. The more general question would seem to be this. What is the normal state of a complex system? At the risk of sounding a glib, the normal “state” is change (in fact non-linear change). If it has a repeated regular cycle or irregular cycle, then the cycle considered overall is the “normal state”. Different phases within that must be considered standard phases of the general cycle.

    A lot hinges on the semantics of “normal” so it might be best to avoid that term and talk about “standard recurrent phases” or something like that. Recessions, and the big recessions called depressions, have to date been standard recurrent phases of capitalism; usually more pronounced and more frequent when capitalism is unfettered.

    A lot also depends on where you draw your system boundaries. Are you going to look within national boundaries or at the global economy? When does the global capitalist economy regarded as a whole reject and then re-include say the economies of USSR/Russia and China? Then we get to problems of definition. Where is the dividing line between a socialist economy and a capitalist economy? “Capitalist economy” is rather a broad grab-bag term in any case but pretending different style of economies don’t exist is also a problem for analysis.

  3. Ernestine Gross
    November 9th, 2015 at 11:52 | #3

    I don’t know whether the description of the mental model of ‘economists’ is true in a quantitative sense. But I do recall posts by JQ regarding changes in sectors of the economy which, IMO, made much more sense.

    I don’t like the separation of micro-economics from macro-economics. I prefer theoretical approaches where the outcome (solution) of ‘an economy’ is endogenously determined (ie by all participants in ‘the economy’, including governments). In this area great progress has been made which is helpful for getting a handle on current urgent problems: finite resources’, ‘financial instability’, ‘wealth concentration’, ‘environmental destruction’. Witness the CSIRO report on The Future of Australia, which is joint work of natural scientists and computational general equilibrium economists. GE does not assume ‘full employment’. It at most studies conditions under which this is conceivable. GE does not assume financial stability, but studies conditions under which this is possible. GE does not assume ‘growth’ is a desirable feature. GE does not presume the profit motive is the only relevant motive and so on.

    In other words, the description of the mental model of ‘economists’ is quite strange, or should I say foreign, to me by now. But then I am in no position to comment on the importance of addressing the obviously flawed mental model, which is apparently ‘the norm’.

  4. Ikonoclast
    November 9th, 2015 at 14:13 | #4

    @Ernestine Gross

    Sorry, this might be a stupid set of questions and conjectures. Does a perfect pendulum in motion (no friction, no energy losses) have a GE (general equilibrium) solution? What I am asking, as a general question, is this. Can an ongoing cyclical or oscillating system have a GE solution? Further, can an evolving system (our economy evolves via knowledge and technology changes and for other reasons) be said to have a GE (a) never (b) only at a point or span in historical time or (c) always?

    I wonder more generally, is the GE model best for an economy or should we be considering a homeostatic model? Or a hybrid GE / Homeostatic Model or a hybrid evolutionary / homeostatic model?

    I mention homeostasis because that would seem to encapsulate some of the political economy reactions to events in our economy. We seek, institutionally and socially, to keep certain parameters within bounds; be those parameters interest rates, money supply, unemployment rates, lack of personal income and so on. Our political economy can be said to have both automatic stabilisers (taxes, social spending, charity) and emergency stabilisers (one-off expedients).

    I mention evolution because both our formal systems (ideologies, macroeconomic tools and preferred settings) and our real system (science and technology) evolve over time.

    A final question (which might be too idiosyncratic, poorly asked and even nonsensical). Is GE theory what one might call an “STP theory”? By STP I mean standard, temperature and pressure. I am asking does GE theory have to assume standard institutions and standard settings to derive an equilibrium? What then does it have to say about alternative settings and even alternative, especially non-market, economic systems?

  5. John Quiggin
    November 9th, 2015 at 17:56 | #5

    @Uncle Milton

    I really mean to post on secular stagnation, but I think it’s outside the scope of this book

  6. J-D
    November 9th, 2015 at 20:17 | #6

    One thing that strikes me forcibly from your narrative is that the pre-slump ‘strong’ level of employment in the Great Depression example, 55 per cent of population, is lower than the post-slump ‘weak’ level of employment in the GFC example, 58.5 per cent of population. Between the Great Depression and the GFC (on your account) something (or some things) had happened that changed the definition of what level of employment counted as ‘full’ employment.

  7. Ernestine Gross
    November 9th, 2015 at 20:40 | #7


    My best guess is that we are talking at cross purposes. To an extent this is my fault because after making the main point in sentence 1 and 2, I talked about ‘GE’ without making it clear that it is the methodology which matters. By now the more general term is ‘agent models’ and there are simulation approaches.

    You are talking about macroeconomic variables.

  8. Ikonoclast
    November 10th, 2015 at 06:54 | #8

    @Ernestine Gross

    Understood. Not that I know much about “agent models”. The Wikipedia entry notes; “An agent-based model (ABM) is one of a class of computational models for simulating the actions and interactions of autonomous agents (both individual or collective entities such as organizations or groups) with a view to assessing their effects on the system as a whole. It combines elements of game theory, complex systems, emergence, computational sociology, multi-agent systems, and evolutionary programming.”

    I’d better not go off topic. Suffice it to say, it seems to me that working in the direction of complex systems and emergence should be a very useful approach. Just briefly, does it involve only markets and market behaviours or can “moral” agents and “ideological” agents be modeled too?

  9. Ikonoclast
    November 10th, 2015 at 07:14 | #9


    That’s an interesting point but it contains at least two issues in my opinion. The percentages you quote are participation rates. “Unemployment occurs when a person who is actively searching for employment is unable to find work.”

    Participation rates went up as women entered the workforce. More recently participation rates have started to go down again. With unemployment however, in the 1960s anything over 2% was considered significant unemployment. Now it seems anything under 5% is accepted as reasonably good.

    Once, one wage earner per household was considered sufficient. In Australia we used to have the “living wage” or “minimum wage” set by conciliation and arbitration. The rule of thumb was that the minimum wage should be enough to enable a person (a male in more sexist times) to feed, clothe and house a family of four. The entry of women into the workforce (a good thing in itself) was subverted in certain ways. One obvious way was in the increase in house prices and rents relative to wages such that it now requires essentially two wages to house a family and not one (in Australia).

    Real wages in the USA have stagnated since about 1979. So I guess the picture is bigger than even participation rates or unemployment. The picture also revolves around family real income and living costs.

  10. J-D
    November 10th, 2015 at 13:23 | #10


    No, the percentages I quoted are not participation rates. John Quiggin stated that they were figures for the ratio of employment to population. Participation rates are the ratio of workforce (or labour force) to population (or to population in an age range stipulated to be ‘working age’). The size of the workforce (or labour force) is not the same as total employment (if it were, this whole discussion would become meaningless).

  11. Ernestine Gross
    November 10th, 2015 at 13:37 | #11


    Well, ‘moral agents’ and ‘ideological agents’, are, in a sense represented by the specification of their behaviour. No, markets are not the only institutional environment (eg incomplete markets). Hope this will suffice. (The Wiki entry is pretty good, IMHO.)

  12. Ivor
    November 10th, 2015 at 14:37 | #12

    …can we regard full employment as the normal state of the economy, subject to temporary interruptions associated with downturns in the business cycle? The evidence suggests that we can not.

    Fair enough but where is there the data that backs this up?

    You can never have full employment under Keynesian capitalism – you need co-operatives and market socialism. Full employment is then, arguably, automatic (or structurally embedded).

  13. Ernestine Gross
    November 10th, 2015 at 17:01 | #13

    “Full employment” is not a good variable to observe because at zero wage everybody is ’employed’.

    The USA had working poor for a very long time. Over time pockets of working poor have also been observed in so-called high wage countries like Germany and Australia.

    At least one would need a condition where full-employment statistics are conditional on a ‘living wage’, appropriate for the local economy (ie the local relative prices matter).

  14. Greg McKenzie
    November 11th, 2015 at 16:31 | #14

    The existence of full employment equilibrium exist under the pure capitalism theory assumptions of no hoarding and all income is spent! The Income-expenditure analysis equilibrium that came from Francoise Quesnay and the French Physiocrats was always a static theory. With only the major determinants of Consumption, Savings/Investment, budget outcome and net exports; this static theory can never reflect the reality of the Twenty First Century global economy. The transfer mechanisms in place amongst the global money economy alone prohibit full employment equilibrium. With Quantitative easing supplanting monetary policy in many OECD countries, sovereign governments are engaged in currency wars. This is an external shock to domestic real economies the likes of which no free market economist can accomodate with acceptance of Say’s Law. As John Quiggin pointed out, the efficient market hypothesis does not work under globalization.

Comments are closed.