Given we seem to be stuck in fairly heady economic times it seems worthwhile to me to put out another post on the subject of employment, labour force growth and unemployment. In this post I am going to revisit the question of labour supply and demand and then take a closer look at the related issues of structural unemployment and the rotting skills (hysteresis) and dependency thesis that has gained so much popularity in policy circles since mid 90s.
In a former post I mentioned the surge in labour supply during the 70s and 80s yet the graph I produced was a little underwhelming because it took a look at the underlying demographic growth of potential labour supply and not actual labour supply. So this time I have have subtracted total labour force growth from total employment growth. Again, a positive reading indicates demand growth is outstripping supply growth and thus a decrease in the unemployment rate. The inverse is that a negative reading says labour supply growth is greater than labour demand growth and thus leads to an increase in the unemployment rate. Here is the graph:
This is fairly sobering stuff. The jobs boom of the 60s gave way to the deluge in the labour supply of 70s from which the employment market not to mention the welfare state has never fully recovered. All that talk in the 90s about the need to reform UI/EI was not really about workers gaming the system it was about a welfare state that could not and would not provide the kind of insurance necessary to cover an over 70 % participation rate in which workers were more likely to be unemployed. The graph below shows just how ugly it can get.
Clearly the deluge in the labour supply 70s had profound impact on long term unemployment which amounted not to an institutionally determined behavioural switch in workers propensity to work but rather a structural shift in the labour supply curve sans an equally profound shift in the demand curve for labour. We know for example that for two decades, from the late 70s to late 90s, that real wages were stagnant indicating a loss in the bargaining power of Canadian workers. Yet despite the loss in wage bargaining traction, demand for labour was not forthcoming until the dot com and later housing booms of the late 90s and 2000s. At the end of the day it was the bubble economies which finally, and fictitiously managed to work through the supply boom of the 70s.
If there is any doubt that demand for labour is the most important determinant of structural unemployment rates we need only look south to our American cousins. In the graph below I have plotted US long term unemployment rates along side the Canadian rates.
Despite or because of higher US productivity during the 2000s the US had already begun to move towards higher levels of structural unemployment than Canada during the early 2000s something not seen since the mid 70s. Clearly the severity of the last recession in US accounts for most of the difference in long term unemployment rates. I have not seen any research which argues that the US has a significantly more generous unemployment insurance system than Canada.
The popular myth promulgated throughout 90s was that structural rates were high because they were high. In the literature this is called hysteresis. The basic idea is that the longer a worker is unemployed the less employable they become because their skills degrade like a piece of fresh fruit on the kitchen table. The hysteresis argument was used in turn to argue for more restricted access to EI and lower benefits to encourage workers to get back to work before their skills rotted. This all had the air populist plausibility particularly when combined with the trend toward increasingly individualistic explanations for collective social problems in the social sciences especially economics. But whatever the strange brew of populist folk wisdom–that workers prefer the dole to working–and academic fads the problem is, as the graph above demonstrates, high structural unemployment and hysteresis seem to magically disappear when there is strong demand for labour. And that demand, as the graph above also suggests, is determined by the general health of the macro economy and has nothing particularly to do with the supply characteristics of labour.
To put a fine point on the argument what needs to be demanded of the purveyors of the hysteresis hypothesis is just what reversed the rotting of workers’ skills that magically made them employable after 1991-92. Was there a steady increase in the funding of retraining programs for example? And what about the US what was going on after 2001 did the Skills of US labour all of sudden just start rotting?
The alternative narrative to the rotting skills / lazy labour thesis is that the 70s was a period of structural realignment in which secular period of decreased GDP per capita growth set in matched by a deluge of labour supply. The 80s and 90s were thus periods of adjustment to this new reality. The UK the US and Canada were early supply side reformers which consciously sought to re-enforce the punitive logic of capitalist labour markets in order to assure price stability (tame inflation) and break workers bargaining power. In other words, neoliberal macro-policy, despite protestations to the contrary, had nothing much to with solving the problem of long term and high structural rates of unemployment. It was a feature not a glitch that the solution to that problem would have to wait until the supply side reforms delivered up their magic via a tsunami of consumer credit and control fraud serving up a ponzi economy that could approach something close to what could be called full employment.
As this report from the BLS (p.23 A-12) demonstrates long term structural unemployment has gotten worse in the US not better since 2009. Most sober economists will concede that this has everything to do with the health of the US economy. But it will not be long before the rotting skills thesis is floated into the stream of policy discourse as a cover for the fact that like with the financial sector the solution is to kick the problem down the road and blame the victims along the way.
*Note that Stats Can defines long term unemployment as 1 year + whereas the BLS defines it as 27 months weeks +. By using 6 months + which conforms to the format of the data reported by the OECD I have more less deployed the BLS definition.