On Sun, Mar 14, 2004 at 10:08:56PM -0500, Erik Reuter wrote: > I haven't had time yet to write up an interpretation of the data, but > I will post one soon so we can discuss. In the meantime, if anyone > has any interpretation they'd like to post, please do! It will be > interesting to see if we draw the same conclusions from the data.
Here is my promised interpretation of the data. I'd be interesting in hearing alternate ideas. Dan suggested that there were some structural changes in employment going on in America after I suggested that most of the current employment problems were due to overinvestment in the late 90's. After studying the data, I tend to think there is some merit in both statements. First, the structural changes. I looked at two measures of employment, the establishment survey (non-farm payrolls) and the household survey (civilian employment). As suggested by the name, the payrolls measure is more widely accepted; however, the household survey could be a better measure since it actually surveys households rather than drawing conclusions from official payroll data. For example, in the recent booming homebuilding market there are many opportunities for people to work "off the books" and be paid surreptitiously -- such people would likely be counted in the household survey but definitely not the establishment survey. First look at the employment annualized 3-year trailing growth rates: http://erikreuter.net/econ/ce_nfp3.png It seems the peak rate of growth has been decreasing for the past 30 years (the highest peak was about 4% in 1978, followed by 3.5% in 1986 and 2-2.5% in the latter half of the 90's). To see this more clearly, look at the 10-yr trailing graph: http://erikreuter.net/econ/ce_nfp10.png where the trend in payrolls is clearly decelerating since 1980, and in civilian employment the deceleration has been going on since 1973. I think Dan has a point that there are some structural changes. Peak rates of growth in employment are not as high as they used to be. An interesting question is, why? I have a few guesses, but none have much supporting evidence. What do you think?` Now look at overall employment, establishment and household measurements: http://erikreuter.net/econ/cpr_ce_nfp.png The civilian employment is consistently higher than the payroll employment, as expected. Also note how much better the current employment situation looks when measured by household surveys (this is also apparent in the 3-yr trailing growth graph referenced earlier). This would seem to indicate that there are quite a few people working "off the books" (some would argue that the household survey is inaccurate, but I haven't seen any evidence to support such claims). Also, note that by the household survey, that the current employment slowdown doesn't appear significantly worse than the slowdown in the early 80's, nor does it look much different than the slowdown in the early 90's. In short, it looks cyclical. In the payroll data, the current slowdown looks unprecedented, so the question hinges on whether there is indeed a lot more "off the books" employment now than in previous slowdowns. It certainly seems plausible that there is. I have anecdotal evidence of a number of construction and homebuilding jobs happening "off the books". I also suspect there is a lot more unreported consulting work going on. But this is mostly speculation on my part. If anyone has any convincing evidence either way, I'd like to hear it. To make this interesting, I wanted to make a prediction. What will employment do in the future? I needed a leading indicator. Well, I didn't have to rack my brains, since I already thought I had one, which is what started this particular discussion -- investment. Investment increases productivity down the road. The mechanism is that more and better capital allows each worker to produce more per hour. And if productivity increases quickly, then employment does not need to increase as quickly to maintain a given GDP growth rate. So, I predict that high investment growth is a leading indicator for lower employment growth. Here is what I found. http://erikreuter.net/econ/nfp_rgdp_pro.png The above graph plots employment growth and the difference: ( real GDP growth - productivity growth ). Productivity is measured as total output per hour of all workers (business sector non-farm output per hour). The agreement is very good -- this shows that real GDP growth can be well-modeled as the sum of employment growth and productivity growth. So, if I can find leading indicators of real GDP growth and productivity growth, then I should be able to predict employment growth. If you trust the estimates of potential GDP, then the GDP part is easy: http://erikreuter.net/econ/pgdp_gdp.png Although using this estimate to predict employment is a bit circular, since potential GDP estimates are mostly based on estimates of future labor force and productivity, I tend to think it is a reasonable thing for me to do since their estimates tend to be quite conservative (i.e., population growth rates don't tend to change much and productivity growth long run average has been in the low 2 %'s in America) Then the harder part is whether productivity growth can be predicted using investment. I got what I would call "fair" agreement with historical productivity growth by lagging investment growth by 7-years and taking one-third of investment growth : http://erikreuter.net/econ/pro_inv.png The 7 year lag between investment growth and productivity growth is somewhat arbitrary. It seems to fit the productivity rises in 1973, 1992, and 1997 well, the one starting in 1983 is off by a year or two, and there are some problems in the 60's, around 1981, and the dip in the late 90's. Clearly there is a correlation, but it is far from perfect. For some periods, the 7 year lag could be only 3 years. There is no reason why it would be constant over the decades, since different types of investments may pay off sooner or later in productivity. Still, the agreement is good enough that I think it is worth pursuing. So, the moment of truth. Plotting the difference ( real potential GDP growth - investment / 3 lag-7-yr growth) http://erikreuter.net/econ/emp_pgdp_inv.png And the prediction? Employment growth doesn't get going again until 2008. Ouch. But as I said earlier, the 7 year lag is somewhat aribtrary. It could easily be only a 5 year lag this time, and conceivably it could be as little as 3 years. In other words, the prediction is quite vague: employment should start to grow again sometime in the next 4 years. But I'll stand by my original prediction and put it in the later part of those 4 years. Incidentally, the overinvestment during the late 90's was severe. I've seen a lot of evidence of it in the industry I work in (telecom component supplier). We are still quite low on the capacity utilization scale, although it is starting to turn up. http://erikreuter.net/econ/tcu_er.png By this graph, it would seem that employment should start picking up within a year. I think this is the basis for so many economists over-optimistic estimates of employment during the past several months. However, I don't think the current economic growth is sustainable. It is largely fueled by monetary and fiscal stimulus, which are in turn fueled by a heavy inflow of foreign investment and record current account deficits. If brisk GDP growth does not continue, then the capacity utilization will not go up and employment will not grow much. Time will tell. My predictions are on record, so we shall see how I did, in time. Anyone else want to make any competing predictions? This could be interesting... -- Erik Reuter http://www.erikreuter.net/ _______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l
