Today on the podcast a brief note on the forecast and then a preliminary report on some startling results from our research on productivity and unemployment.
First, more than a few big financial houses have economists trotting out rosy scenarios of robust growth beginning in the third quarter of 2009 and extending for several quarters out. Not too long ago we featured an economist on Idiot of the Week who suggested the recession ended in June. And if you've peeked at our current forecast, you see that our curves point skyward as well, after a point.
But the Demand Side premise is that the economy is driven from the demand side, and we do not see incomes, employment, or household balance sheets recovering in the near future, so we do not see any sort of robust recovery taking place from the consumer side. Past recoveries have had engines of growth. There is no easily visible engine to help the economy out of the current downturn.
Take the two most recent recessions, 1991 and 2001. Incomes were helped in the first case by falling mortgage rates and a wave of refinancings, then by business investment in high tech and the so-called dot.com boom, and throughout by collapsing oil prices and resulting cheap energy. The recovery of the early 2000s was pulled along by the biggest housing bubble in history, a function of Alan Greenspan's one percent interest rates and attendant financing and debt bubble. Housing assets are now in the process of breaking in half.
We may have the cheap money, but nobody can get it.
Demand Side has been a great proponent of the Recovery and Redevelopment Act, the $700 billion Obama plan, and we expect it will have a significant effect going forward. But business investment has collapsed, the financial sector has become zombie land as far as the real economy goes, and the stimulus is too little to carry the day.
Recovery has a technical meaning, which is upward movement. If the patient can raise his hand, many will consider him to be in recovery. We envision an owl recovery, a series of small W's that means the economy is bouncing along the bottom. Only when we accept that we need a public sector engine of growth will we get going again.
Top line GDP may increase, but if it is because of the trillions in federal deficits, it ought not be considered a true recovery. Take a look at our net real GDP metric. When this begins to go up, then we have recovery.
Multifactor Productivity Solved
Multifactor quote unquote productivity is the official statistical designation for "We don't know what the hell" -- as in productivity with no obvious driver. We titled our podcast "multifactor Productivity Solved" in a self-consciously grandiose way of announcing some success in our search for the missing half of productivity improvements not accounted for by economic science.
I must say that when I first saw the trendlines on the Excel chart, I was amazed. So amazed that fifteen minutes later I played a computer game, which I haven't done in a while, and after that went and scrubbed some pots in the kitchen. Having allowed reality to settle in this way, I returned to the chart, but it had not changed. It still seems a bit improbable.
This is from among the first cuts, using the most commonly cited data for the unemployment rate and productivity -- output per hour -- directly out of the Economic Report of the President.
Several weeks ago, you may remember, I challenged our listeners to help with an investigation into productivity and its causes. One courageous young man took me up on it, and we began to explore whether productivity changes could be explained by changes in unemployment and changes in oil prices. Alex and I began what will likely be an extended trip. But this past weekend, we produced what I think is a jaw-dropping chart. I put it up on the blog yesterday and will get a larger version on the web site soon.
The correlation between a smoothed line of change in productivity and a smoothed line of the unemployment rate is nearly complete. They essentially mirror each other around a central trend of 4.0.
So without further ado, Demand Side presents:
The Rule of Eight
Eight minus the unemployment rate equals the change in productivity. Eight minus the change in productivity equals the unemployment rate.
There is a very close correlation between a smoothed line of unemployment and a smoothed line of productivity. As productivity rises, unemployment falls. As unemployment rises, productivity falls. Whether or not one causes the other, the correlation is nearly complete. Emphasizing that this is for a trendline described by a polynomial equation.
Our postulate here, and the hypothesis we set out to test, is that the rate of unemployment causes a change in productivity because as unemployment falls, workers are shifted to more productive tasks, retrained as necessary, equipped better, or simply used more efficiently. This idea would not be surprising to a worker or a manager, since they would be familiar with the dynamics of the workplace. Not being able to hire appropriate skill levels, currently employed skills are improved. But so far as I am aware it does not appear in the economics literature.
Here we would like simply to publish the fact of our being first to the finding and to sketch the outlines of its implications. The chart displays the irrefutable correlation. In subsequent pieces we will identify the appropriate statistical metrics.
A change in the rate of unemployment influences productivity. In tight employment climates, managers shift current assets -- including labor -- to more productive tasks.
A corresponding absence of focus occurs when employment markets are looser, as when the unemployment rate rises. From the Demand Side perspective, a rise in unemployment relates to a drop in demand. Lower demand means current assets are not used to previous capacity, which means a fall in productivity.
A close look at the data will show that officially measured productivity seems to anticipate the change in unemployment. We believe this is due to output being a function of labor inputs over a longer period than the currently measured productivity. That is, the organization of plant and facilities and systems over time contributes to output measured in a specific quarter. Following from this, when demand falls and workers are let go, the surviving smaller workforce may be credited with production that actually involves more people over a longer time frame.
A good illustration of this is in the productivity data for Q2 2009, which showed an historically high reading for productivity, but obviously driven by a complete collapse of hours worked. This simultaneous fall in labor input and rise in productivity as measured by current methodology is frequently observed. This exclusively short-term phenomenon has led to confusion around the relationship between unemployment and productivity, and sometimes to the fatuous attempt to describe higher unemployment as good for productivity. The best workers are more productive and the rest are being carried. In fact, as we see here, the relationship is systemic and opposite.
Simply reflecting on the fact that capacity utilization is lower during times of falling demand and rising unemployment should abort this line of reasoning. In any event, as noted, the long-term correlations are in exactly the opposite direction.
One might suggest other causal connections for why the short-term and long-term relationships act in the opposite direction. For example, as one industry prospers, the wages in that industry rise and people train to obtain the skill sets necessary to that industry. Should that industry subsequently decline, the cohort of trained workers is left behind to accept lower level positions. That may have been the situation with the high-tech boom and we may see it soon following the financial sector boom and collapse.
The decline in the measured level of unemployment may bring this unemployment-productivity relationship to the "Rule of Seven" from its inception in 1948 as being the Rule of Eight. We believe the methodology for calculating unemployment has deteriorated over time, and so the long-term decay is a result of measurement, not of dynamics.
Alternatively, and still to be investigated, manufacturing may be more amenable to automation and other productivity improvements than services, so a shift to a service-based economy may influence the general trend of productivity downward.
Another explanation could be the shift to overseas suppliers of low-productivity manufacturing, leaving the higher productivity work to be counted domestically.
Increasing productivity with declining unemployment implies a contradiction to the Phillips Curve, which suggests a lower unemployment rate leads to higher inflation. Since productivity increases ought to subtract from inflation pressure, the rule of eight implies lower inflation with lower unemployment. This should be the finding in all cases where supply is not constrained.
Polynomials are complex equations. The equation describing the trend in unemployment is
y = 4E-07x5 - 5E-05x4 + 0.0024x3 - 0.0448x2 + 0.3476x + 3.7556
with an R2 of 0.4357
But P = 8 - f(U) where P is productivity and U is the unemployment rate
is not complex.
It turns out that the mirror of a fifth order polynomial curve for the unemployment rate explains changes in productivity. Not a parallel line, a mirror image. When unemployment, or the polynomial trend for unemployment goes up, productivity goes down. And vice versa.
What does this mean?
We have long suspected that tighter labor markets generate more efficient use of labor assets and thus higher productivity. The data for the unemployment rate and productivity in raw form do not exhibit contemporaneous correlations in the same direction, even if massaged.
But the inverse relationship between unemployment and productivity is incontrovertible. Whether this is a causal relationship or a situation in which both are functions of another dynamic is not provable. A frequent error in economics arises from the assumption of cause and effect from simple correlations.
Our hypothesis is that low unemployment causes the natural shift of workers to more productive occupations and to improvement in the industry-specific skill sets of workers.
As I told Alex, I am putting this up on the blog and web site to essentially claim ownership ASAP. I very much appreciate his signing on and getting this line of inquiry off the ground.
Has anybody else ever made this comparison? I don't know. I never heard of it if they did. At first blush it appears to be a very substantial finding, and we are going to pretend it is until somebody sets us straight. There is more to do, of course. Particularly along the lines of understanding the mathematics.
But for the moment we are suspending our investigation of the energy cost line of attack and focusing on this trade-off between unemployment and productivity.
We are looking for feedback on this issue. I hate to use the blog or website mail. Spam is so very aggressive. A good contact is kleinbattle at gmail dot com.