On 9 Oct 02, at 16:14, Davies, Daniel wrote: > > > -----Original Message----- > From: Charles Jannuzi [mailto:[EMAIL PROTECTED]] > Sent: 09 October 2002 15:13 > To: [EMAIL PROTECTED] > Subject: [PEN-L:31148] Re: employment > > > >best you could say it was an argument from > >previously established authority > > Absolutely, because I have no real specialist knowledge of the subject of > unemployment statistics and no real prospect of having the time to get any. > But look at it this way: > > What's the knock-down argument to say that people who only skim want ads > *should* be counted as "actively looking for work"? I haven't heard it. > Same with the long term sick. I don't even understand the argument about > the unemployment rate which seems to be arguing that people who have jobs in > the armed services ought to be counted as unemployed. What I do know is > that there are a lot of people doing good, honest work on this subject, > trying to measure what effect various kinds of non-worker populations have > on the operations of the labour market, and that the most and the best of > them work for the BLS. So in the absence of anyone making a contrary > argument to me, I'm going to assume that the inclusion of these groups makes > the BLS number worse, rather than better, as a measurement of what it's > meant to measure. My understanding of what the BLS unemployment rate is > meant to measure is this: it's meant to measure the number of people who > would be employed if the labour market were to clear at the current > prevailing wage rate, but who are not employed. It's a measure of labour > market disequilibrium. > > There are, I think, two further important questions which arise from this, > and part of the reason why we're all talking past each other is that we're > taking these two questions out of order. The questions are: > > 1) Assuming that "unemployment" is used by the official statisticians to > measure the extent to which the labour market has failed to clear, should it > be measuring something else? > > and -- and this can most likely only be answered conditionally on a specific > answer to 1) above -- > > 2) Can adjustments be made to the official statistics in order to transform > the BLS number into something which works well as a measure of whatever it > is that the unemployment rate ought to be naming? > I am a little curious of how dd comes to the conclusion that BLS statistics are the best in the world and on what basis. They are perhaps the quickest to be published and perhaps the most voluminous but, if I remember correctly, there have been quite huge changes in recent months to growth and productivity measures due to statistical revisions and the reason given was the rush to get the data out means that the input figures to the statistics are themselves preliminary and subject to adjustment. One result of such a revision was a drastic fall in the rate of productivity growth in the latter 1990s which contradicted the assertion of a 'new economy.' In any case, a little nationalism here, I believe the UN system of national accounting was adopted from that developed by Statistics Canada at the end of the 2nd WW, because of the quality of the Canadian statistical services. Indeed, I use Stats Canada statistics a lot and though I often curse them because of changes in definitions etc. their quality is excellent and they always give full details of how each are collected and the margins of error etc. I am not sure that the US stats are any better and, given their speed of release, unrevised US data may be less reliable than that available from stodgy Statscan. They also publish quarterly a journal "Perspectives on Labour and Income" which does in depth studies of such things as unemployment exploring all the variables that we have discussed on this thread and incorporating a lot of statistics that are otherwise not reported -- including stuff on the quality of jobs and what people do with their 'leisure;' also on the grey economy and so on.
This raises a second point. Some economists measure macroeconomic unemployment not by the unemployment rate, but by the employment rate and its divergence from the potential employment rate. Just recently I saw graphs (I don't remember where) showing the divergence of US employment rates over the last ten years from the long term trend. What they showed was not a large rise in unemployment, but rather a sharp drop in the employment rate coinciding with the recession. Thirdly, and this has only been hinted at on this thread as I recall, (I could be wrong), one problem is that the unemployment rate (strict definition or otherwise) does not take into account institutional changes. But to be a real measure of the welfare cost one has to consider the institutional context. In 1911, the definition of potential labour force included those 10 years of age and over. This was revised to 14+ and then 15+ somewhere around 1951 I think. Perhaps it should again be revised to 16+ to account for school leaving laws and behaviour. This would considerably affect the *measured* unemployment rate. Another example, the quality of unemployment was much worse in the 1930s when we had no unemployment insurance and mostly single earning families than it is today with unemployment insurance (no matter how inadequate) and where the majority of families are two earner families and the family sizes are so much smaller. And so on. It seems to me that these are the issues that those of us who are labour economists should be aware of and use in order to "qualitize" the statistics. But there is no single figure that can encapsulate all these institutional factors and we, as economists, should automatically dismiss any figure that purports to do so. Paul Phillips
