Explaining Unemployment: Natural Accelerating Rate of Employment Destruction (NARED *tm)

Travis Fast

In the great labour market policy debate of the mid to late 1990s no end of explanations for persistently high levels of unemployment in the face of recovered GDP growth was trotted out. The infamous NAIRU curiously did much of the original heavy lifting. I say curiously because the NAIRU is ill suited to provide an explanation of persistently high unemployment in the face of strong GDP growth. Indeed, the NAIRU, was designed to put a gloss on the problem of stagflation–poor growth and high inflation. At a higher level of abstract policy thinking however, the NAIRU was part of broader assault on demand determined explanations of macroeconomic phenomena including unemployment. Fighting inflation and ultimately conquering it was said to be the sine qua non of renewed GDP growth.

The renewal of GDP growth was in turn said to be the cure to high unemployment (as we shall see towards the end this is the central determinate of unemployment). Yet, by the latter half of the 1990s the term “jobless recovery” had crept its way into the policy lexicon. Here the NAIRU became near useless as a supply side explanation for the persistence of high unemployment. And it was here (then) that I would argue New Keynesianism played a vital role in the continuance and ultimate dominance of supply side explanations of unemployment and what I have called the solidification of the neoliberal consensus.

To that end two explanations were put forward. The hysteresis/skill spoilage thesis maintained that prolonged periods of unemployment had caused a relative degradation in the skills of the unemployed. This degradation of skills in turn was responsible for the inability of of the unemployed to find jobs. In short, high unemployment was the cause of high unemployment (hysteresis). The insider/outsider hypothesis maintained that the persistence of unemployment was caused by the higher than marginal productivity wages of the employed. In a real bastardization of Keynes, involuntary unemployment was caused by the myopic behaviour of the employed (a refusal by existing workers to take a pay cut). If the two explanations did not readily hold together in a general theory sense they certainly held together at a higher level of abstraction: ultimately the cause of unemployment laid with the qualities of workers: employed or unemployed–on the supply side.

And although the twin New Keynesian hypotheses had a hard time dancing on the same floor they nonetheless were seemingly given credibility by inter temporal comparisons of GDP growth and rates of unemployment. In the graph below five year averages for real per capita GDP growth and unemployment rates are plotted.

If one looks at the five year averages from 65-70 through to 95-00, one cannot help but notice that during the 1980s a transition seemingly occurred whereby similar levels of GDP growth failed to produce similar rates of unemployment. The most stark comparisons can in fact be made. During 65-70, a 3.1 percent average growth in GDP produced and average unemployment rate of 4.5 percent whereas in the 95-00 period a 3 percent average growth in GDP produced an 8.5 percent average rate of unemployment–a full 4 percentage point difference. It was from such facts that, I think New Keynesian hypotheses drew their succour.

Why indeed had GDP growth recovered but not employment? But there were a couple of constraints on the types of explanations that REAL economists were allowed to make . You could not blame technological change, nor could you blame free trade and you certainly could not blame employers sweating more effort out of the existing workforce. The key to a righteous explanation laid with the supply side characteristics of workers. Now I am not arguing that there was some kind of conspiracy. Far from it, I am just arguing that the the new classical victory had been so total and that the terms of war had become so solidly placed on a different terrain that New Keynesians were bound to reach for explanations that jived within the increasingly narrow confines of the debate within mainstream liberal economics. Just think of all those more classically inclined liberals who made the reluctant Keynesian Devil Shift during the Golden Age Illusion. After all, once upon a time Friedman was a Keynesian.

But whatever the cause, those observed facts are not sufficient to explain the seeming disparities between the two epochs. In fact, the same disparities can be witnessed within the same epochs! Notice for example that in the period form 60-65 GDP growth was 4.3 percent but unemployment averaged 5.7 percent–GDP growth and the rate of unemployment being 1.2 percent higher than the proceeding period. Were insiders really more active in the 60-65 period than in the 65-70 period? Were workers skills duller in the 60-65 period than in the proceeding period? The sober answer is of course not. And one could beg the same question of the 95-00 and the 00-06 averages. How is it that within such a short period of time we could move from a 3 percent rate of GDP growth and a corresponding 8.5 percent of unemployment to a 2.1 percent average rate of GDP growth and a corresponding 6.8 percent of unemployment? I suppose if one takes the hysteresis hypothesis at face value we could get to a plausible explanation of the present period but that dog won’t hunt for the initial periods under consideration. And it is not much of hunter in the two most recent periods either.

In the graph below I have hyper-trended the data by superimposing a trend on the five year averages. What becomes apparent is the the close relationship between GDP growth and unemployment. But beyond this shocking truth is the more subtle observation that the early 1990s seemingly marked the end of the long secular slowdown in real GDP growth which began in the mid 1960s (for further confirmation see the last graph for long term trends in GDP growth and Unemployment). This slowdown in turn produced a Natural Accelerating Rate of Employment Destruction (NARED).

It would be perverse in this respect to employ hysteresis effects in skill (de)formation to explain increasing unemployment. Notice, however, that hysteresis effects in the degradation of business sentiment might be a plausible explanation of the duration of the long slowdown. Although I have serious doubts about a 40 year long hysteresis effect on sentiment and rather think that some version of the Brenner/Shaikh/Marxian thesis is necessary.

However, leaving the cause(s) of the structural decline in GDP growth to one side, the other observation that can be made from the two graphs presented above is that it seems as though during the early 1990s a reverse process was initiated whereby increasing levels of GDP growth began to have a cascading effect on the level of unemployment so that by the early 2000s the brief drop in GDP growth was not sufficient to change the momentum in employment creation; or, what I have identified as a Natural Accelerating Rate of Employment Creation (NAREC).

On this reading, the present conjuncture is quite significant. Should GDP growth become stuck at low stagnant levels for a period of time the acceleration effects could revert back to a NARED dynamic whereby high unemployment begets even higher unemployment irrespective of the supply qualities of workers.

This analysis and scenario poses serious challenges to the dominant neoliberal policy paradigm. No amount of retraining, mobility or cuts to the reservation wage can remedy a situation in which the creation of employment is weak. Focusing on the skill set or psychological attitudes towards work on the part of workers simply serves to blame the victim and ignores the cause: weak GDP growth overtime creates higher levels of unemployment. Workers do not retreat from the labour market on mass, as any graph of participation rates can demonstrate, because most workers have no alternative but to seek employment. And I suppose in that last regard unemployment is caused by the unemployed.

Note: Sources

Table 282-0002 Labour force survey estimates (LFS), by sex and detailed age group, annual.

Data prior to 1976 is from Statistics Canada 11-516-XIE (FREE!) Historical statistics of Canada.

GDP data prior to 1961 is from Madison The World Economy: A Millennial Perspective