From: glevy@PRATT.EDU
Date: Tue Nov 08 2005 - 13:49:00 EST
---------------------------- Original Message ------------------------- Subject: [OPE-L] [Jurriaan] Notes on gold From: "Jurriaan Bendien" <adsl675281@tiscali.nl> Date: Tue, November 8, 2005 10:52 am ----------------------------------------------------------------------- Hi Paul, I can see your point. In my experience, few statisticians believe their estimates to be totally accurate. Rather, they concentrate on quantitative margins of error, and the plausibility of estimates, by collating many different data sets. In addition, the argument is, that if the survey methodology is constant over time, then even if some items are overlooked by the survey, it is nevertheless possible to show the real trend in a variable. But BOP data are notoriously prone to error and revision, because of the number and complexity of transactions (as I know from professional experience). In the case of the United States, I think I could create a good peaceful job for myself just improving the questionnaire documents used to collect financial data from business people. I have looked at quite a few of them, and from a professional point of view they're often just not user-friendly documents, and that increases the likelihood of response error, the magnitude of which remains unknown, unless specifically tested for with split-ballot, chi-square type experiments (unlike e.g. sampling error where you can often estimate a magnitude of error). The additional problem in the USA is, that a number of different statistical agencies collect their own macro data, without there being any central agency providing an integrated approach to data collection and processing. And thus BEA will compile its estimates from a lot of different sources, with different data collection policies. The number of national standard classifications are few. Here in Holland, a lot of CBS economic surveys are now integrated, digitalised and simplified, but in many countries it's still the old paper thing. Years ago, I did an exercise on cumulative revisions of New Zealand's annual GDP data, comparing the first official release with the last revised estimate, and I discovered that the discrepancy between the two was in some cases close to the annual increase in GDP. Yet, "the markets" attach great significance to current reports of growth rates in real GDP, even although that is not really warranted. All this suggests that the use of statistics is an "art" of sorts, in that you have to know what exactly is being measured, and how a statistic can be validly used. Mostly in my own writing I just stick to percentages and totals providing an indication of magnitude or proportion, knowing that often the degree of data accuracy does not warrant more. My point of view on value relations between products in Marx's sense is, that they reflect social relations between people (not infrequently, relations of bargaining power), but value relations cannot usually be measured directly, only indirectly (as quantities of labour time, as real or ideal prices, or as trading ratios). Because value relations can only be inferred in this sense, not observed directly as sense-data, the naive empiricist will say that they do not exist. But that is bad science; in science, there are many phenomena which we cannot observe or experience directly, although we know, as realists, that they exist, and can measure them through the observable effects they have. Even so, it could be argued that value relations can be observed directly, simply by observing how economic actors empirically interact, but the point is, that this remains an interpretation of that interaction, a theory or theory-laden observation. If that wasn't the case, we'd all be Marxists now. I aim to write all this up one day more exactly in an article about the transformation problem, but I have not yet had time to do it beyond writing a few notes on it, and looking at some of the voluminous literature. My initial argument is that the Marxian transformation of value-added into profit is no less problematic than the official transformation of profit into value-added, and that you cannot group, relate, compare and calculate prices without at least implicit reference to a value theory that provides rules for price aggregation. That's just an insight that is basic to accounting of any sort. I think that Piero Sraffa came close to this insight, in his contribution to the "capital controversy" (the so-called "aggregation problem") but he did not elaborate the wider theoretical implications. Using certain assumptions, you can build complex mathematical models, but if we do not provide good arguments about what entitles our assumptions in the first place, then no amount of mathematical wizardry proves anything much. In other words, in official economics, prices are regarded as an entirely unproblematic notion, but on closer inspection the notion of prices is much more complex and problematic. This did not bother Marx at all, because he viewed prices simply as symbols expressing value relations more or less accurately - but I think a more precise discussion of the nature of prices can resolve some of the theoretical controversy. And if I get the time, I aim to trace out the link between value relations and moralities as well. This however is not so much a value-form theory, but a price-form theory. Jurriaan
This archive was generated by hypermail 2.1.5 : Wed Nov 09 2005 - 00:00:01 EST