Neoclassical modelling as a strategy to ‘unsee’ involuntary unemployment
Neoclassical modelling as a strategy to ‘unsee’ involuntary unemployment
– the example of the neoclassical analysis of the Great Depression in the USA. By Merijn Knibb
What does the post WW II decline in hours worked in the USA teach us about the economy – and about economics? Why do economists gloss over this momentous event and why are they often not even aware of it? It’s not because of its size, as this decline was of the same magnitude as decline during the Great Depression [graph source: Higgs (2009)]
This lack of interest in the comparison between the 1929-1933 and the 1944-1949 period becomes even more remarkable when we look at unemployment. Between 1929 and 1933 unemployment went up from 3% to a devastating 25%. But after August 1944 unemployment only increased from 1% in august 1944 to 6.6% in 1949, an increase of 5.6% of which 2%-point took place in one single month, September 1944. Why aren’t economists more interested in this at first sight totally anomalous behavior of the economy? In some way or another, economic policies and events prevented the rise of mass unemployment – what does this period teach us. Why aren’t economists more interested in this question!
Below, I will, using the work of Harold Cole and Lee Ohanian, two economists who published extensively on the Great Depression in the USA, argue that the lack of interest in the 1944-1949 period might be due to the a-historical nature of much economic theory. Economic models and economic theory have difficulties incorporating changing circumstances – which leads and enables economists to gloss over idiosyncratic events like this one. One aspect of this is the absence in neoclassical macro-economics (Ohanian and Cole label themselves as such and even use the phrase ‘neoclassical’ in the title of their papers) of the concept of ‘involuntary unemployment’. ‘Involuntary unemployment’ is (looking at present day questionnaires) what we actually measure as ‘unemployment’, measurements which clearly show that, Greece and Spain being two recent examples, involuntary unemployment can rapidly rise to double digit and even 25%+ levels – just as it did during the Great Depression in the USA. There clearly seems to be a mayor discrepancy between our measurements and neoclassical macro theory. The facts about unemployment during the thirties in the USA are thus clear in that the absence of ‘involuntary unemployment’ in the neoclassical models can’t be explained by any kind of ‘accident’ – it’s not in the models on purpose. What kind of modelling strategies are used by neoclassical economists like Cole and Ohanian to ‘unsee’ events like the 22%-point increase of unemployment after 1929 and, as a consequence, to accept a highly idiosyncratic period like 1944-1949 as ‘normal’? As all data on the Great Depression show an unprecedented drop in the number of hours worked as well as an unprecedented increase in unemployment, Cole and Ohanian did have quite a problem: how did they manage to explain 25% involuntary unemployment away? And how does this relate to a period like 1944-1949, when a very large decline in hours did not lead to 20%+ unemployment? How did they do it? And why? While they do mention high rates of unemployment (first link), they try to disconnect this from demand and to redefine this as a supply side reaction. What enables them to do this?
Cole and Ohanian use two somewhat conflicting but deeply neoclassical strategies. One is the ‘representative household’ concept. This concept states that all households act like one and with perfect foresight, which means that ‘involuntary unemployment’ is impossible. The household might, choose to work a little less – but that’s not unemployment, that’s leisure (for the innocenti: I’m not kidding. That’s what the models assume). The good thing about Cole and Ohanian is that they operationalize this concept. They look at the average number of paid hours worked per capita, which as the population of the USA increased with roughly 1% a year declined even faster and more than total hours! And, consistent with neoclassical economics, they see this sudden, unprecedented and momentous decline as a conscious, rational choice for leisure… As Robert Higgs bluntly states (p. 152) : “By using hours worked as our measure of employment, we avoid the necessity of distinguishing who is employed and who is unemployed”. Higgs says that unemployment is a poor measure because people are not just employed or unemployed, there can be differences in hours worked. He then says that even when unemployment fell, this did not equate to a return to earlier levels of hours worked. Higgs is totally right to point this out. Of course, the number of hours worked is an important variable which has to be used to investigate the severity of the Great Depression and to understand what caused the decline of unemployment. It is, however, an average and surely during very severe depressions it is also important – not to say a necessity – to look at the unemployed, as it individual unemployment which causes hardship. Remarkably, I find myself taking a more individualistic stance than either the neoclassical economists or even libertarians like Higgs! Which underscores the question: why do they try to explain unemployment away? It’s like the well-known example of the man who tries to cross the river which, on average, is only 0,5 metre deep – and drowns in a deep hole.
If we accept the idea that a 25% decline in hours worked per capita is voluntary this of course implicates a totally different lifestyle – but did people in the thirties really prefer to be dirt poor or to be a hobo’s to having a regular job and a prospering family? If so, why did all these hobo’s suddenly disappear during World War II, when labour was scarce and jobs were plentiful? And why did preferences of the USA population representative household suddenly change again after 1933, when the number of jobs number of hours started to grow at a 5 to 10% level a year? Cole and Ohanian are mumbling a bit about people wanting to be unemployed because they did not want jobs in ‘competitive’ sectors and wanted jobs in protected sectors but are we sure that 20%+ of the population really preferred unemployment to even poorly paid jobs? When we look at the supply side, the growth of hours worked can be neatly explained by the concept of involuntary unemployment in combination with a recovering economy. It would however take till World War II before the USA surpassed the 1929 level of hours worked – Cole and Ohanian do have a point when they point out that productivity absolutely surged, during the thirties, which meant that even 8 to 10% growth of demand and production led to a somewhat disappointing increase of employment… But this only serves to underscore the insufficiency of their modelling strategy which basically boils down assuming that people didn’t want to work when unemployment was high and wanted to work when unemployment was lower. However, taking unemployment seriously would, in combination with the rapid increase in productivity, have taught them demand and not supply was the problem.1 Is that why they are bending over backwards to unsee 25%+ unemployment?
Cole and Ohanian however also have a second strategy, which at least accept the idea of different households. This is how they describe it:
A household member either works in the competitive sector, (nft),works in the cartel sector (nmt) (if the household member already has a cartel job), searches for a job in the cartel sector (nut), or takes leisure. Since the cartel wage will be higher than the competitive wage, household members compete for these rents by searching for cartel jobs. Searching consists of waiting for a vacant cartel job, and search incurs the same utility cost as working full time. If a cartel job vacancy arises, the job is awarded randomly at the start of the period to an individual who searched the previous period. We denote the probability of obtaining a cartel job through search in period t. To build in job turnover arising from life-cycle events such as retirement or disability, we assume that cartel workers face an exogenous probability of losing their jobs at each date.
Translated: people are never fired. They do quit jobs, but only to indulge in leisure or to search for a job with a higher wage. This by necessity means that all unemployment is caused by wage differences between competitive and protected economic sectors and is, by definition, voluntary. Never mind that during the thirties many, many unemployed were fired and accepted wages which were much lower than their previous wages – this modelling strategy is entirely consistent with neoclassical economics which, therewith, enables you to unsee involuntary unemployment.
This strategy of course backfires when unemployment rapidly declines. Or, like after the war, rose much less than was to be expected when we look at hours data only. The War economy was of course very overheated and the decline of hours worked after 1944 was to an extent of course a return to normalcy: i.e. quite a number of women who left the labour force altogether to care for their families (which was normal, remembering that families were larger in those days while washing machines were in their infancy), less work on Sundays, etc.. Overheating seems to have stopped quite fast – in September 1945 the unemployment rate jumped by 2%-points to about 3%, a level not too different from low pre- 1930 unemployment rates. After September 1944, however, the average number of hours worked per week per employee also declined by about 10%, compared with the thirties and even more when compared with WW II. Between 1929 and 1937, nothing of the kind happened (based upon Kendrick, Productivity trends in the United States (Princeton 1961, p. 316), who gives data for 1929, 1937 and 19482). This alone makes up for half the decline in hours worked. The irony is that this indeed seems to have been a voluntary decline of labour supply. Did ‘New Deal’ rules, higher wages and more union power and enhanced job security enable this development? Was mass unemployment prevented by the weekend? These are tantalising questions – but to answer them we do have to take the idea of involuntary unemployment seriously.
And is that also the very reason why neoclassical economists do not take this idea seriously? Can’t they accept that in a market economy, involuntary employment exists and persists? Can’t they accept the fact that solving it sometimes requires government action, if only to enable households to solve coordination problems connected to, in the 1944 case, a preference to work less (paid work that is). This requires, of course, more thorough investigation than is possible here. But such a proper investigation of household behavior in this time has to be based upon the total number of hours worked – but also on taking the concept and measurement of ‘involuntary unemployment’ seriously. Neoclassical models do not allow this – as they still seem to be engineered to unsee unemployment, measured in individual persons. Look here for the New Area Wide Model of the European Union and the Eagle model used by the ECB. Look here for an infamous ‘soup kitchens caused the Great Depression’ article by Prescott, look here for a takedown of that idea. The stark differences between the 1929-1937 and the 1944-1949 periods are another reminder of the large difference between (the causes and consequences of) voluntary declines in the amount of hours and involuntary declines – i.e. unemployment and the importance to take historical circumstances and institutional developments into account when analyzing the economy. And let’s stop talking about ‘involuntary’ unemployment. Unemployment as we measure it is, by definition, involuntary. We do have to take that seriously.
Doris Day as an angry but confident worker in ‘The Pajama game’. Did this musical capture the ‘Zeitgeist’ of this age and was this confidence caused by Union Power and the New Deal?
1 At the time, it took a total maniac to take advantage of this situation. In any other epoch, the NAZI plans would have led to runaway inflation. In the thirties (that is, after 1933) exceptional increases of productivity prevented this and enabled an extremely rapid rearmament. This foreshadowed the extreme increase of production of the USA economy after 1941.
2 Note Higgs p.152, citing Kendrick: “With regard to hours worked, the profile looks somewhat different, however. Total hours worked fell substantially from 1929 to 1932. Then, unlike the standard depiction of the economy’s course, they hit bottom and stayed put in a virtually flat-bottomed trough for three years, 1932, 1933, and 1934. They then rose substantially until 1937, dropped by 7 percent in 1938, then rose again thereafter. However, even as late as 1940, total hours remained below the 1929 level by 6 percent, and only in 1941, with the population vigorously engaged in mobilization for war, did total hours exceed the 1929 value, by 3 percent.”
From: pp.5-7 of World Economics Association Newsletter 5(5), October 2015