> "In spite of the substantial liberalizing measures of the past decade, > developing countries still attract less than a third of world FDI > flows, and these flows remain highly concentrated. In 2001, the five > largest host countries in the developing world received 62 percent of > total inflows and the 10 largest received three-quarters. The level of > concentration of FDI in developing countries has in fact risen in > recent years."
Yes, that is the trend, which is one reason why you have to be a bit wary about "globalisation" theories. For the statistical definition of FDI, see for example http://www.oecd.org/dataoecd/56/1/2487495.pdf If you carefully consider the meaning of this definition, you will have to conclude that a large lump of foreign investment is not being captured by these criteria, and that therefore focusing exclusively on FDI data, will not tell you what is really going on, it is one indicator only. It has to be supplemented with other investment data and indirect evidence on capital movements. There is, besides, a heck of a lot of capital that chases around the world looking for the highest rate of interest, a lucrative currency fluctuation, a sudden rise in share values, a once-only business deal etc. but this capital might even not be invested for a full year or more, but for a month or some months, a week, a matter of days in some cases, making accurate statistical measurement extremely difficult for that reason alone, never mind the sophisticated legal constructions involved in sluicing capital from one country to another (income, transfers, investment and consumer expenditure might masquerade for each other, i.e. the precise origin and purpose of funds may be very difficult to establish). This mobile capital runs for all I know into many trillions of dollars, but I don't think the IMF, the World Bank or the International Bank of Settlements can reliably estimate it, you can give only an order of magnitude, a rough estimate. There is a lot we simply do not know about it, because no institutional framework exists that would allow a truly accurate measurement device to be operated. I can give you an anecdote about all this. In post-WW2 New Zealand, all significant foreign transactions had to be centrally registered with government agencies, and the Reserve Bank kept close tabs on this, the trade & industry secretary's (William "Bill" Sutch) idea being that local industries would be developed behind a wall of protectionist measures (tarriff barriers and so on). >From 1985/86 onwards however all the walls were broken down, and the legislation changed drastically, a Chile-type experience without Pinochet. Instead of regulating foreign transactions, the Reserve Bank was from now on just under a legal requirement to keep price inflation below a few percent, I cannot remember exactly, 2 percent or something, they revised the mandatory maximum slightly once I think. Anyway, the implications for statistical staff like me were that, in order to calculate BOP and SNA statistics, resident enterprises (and some NZ entities overseas) had to be individually surveyed with survey forms, digitally or on paper, you didn't have any central administrative data base anymore, just enterprise directories etc. So far so good. But, in order to do that, you had to be able to identify distinct enterprises and institutions engaging in foreign financial transactions of any sort. Well you could capture a lot of that, but what if some entity like a "shelf company" consists only of a postbox address, and changes in some respect during the year, in terms of ownership, location etc. ? How do you actually know the consolidation methods of fast-changing, intricate and complex company or corporate structures, when you are dealing with companies which via-via own (part of ) other companies which own (part of) many other companies, and so on ? That is one side of the problem you have to cope with. Another side of the problem is: how do you develop a survey instrument which collects reliable data ? In one survey questionnaire design I tested with BOP staff, talking with some accounting people in a huge corporation, we discovered that, because we did not understand in advance the precise form the transaction took, the terminology involved, and the precise way it was accounted for, that as a result the survey question was inadequately worded, in turn resulting in a misunderstanding by the respondent, and a data discrepancy of several 100s of millions of dollars debit/credit from what it theoretically ought to be (I do not remember the exact figure). Now, we could do a limited amount of testing, to arrive at an acceptable survey instrument, but even so, we could not say with absolute certainty what precisely the margin of response error was, we did not have the resources to do that. Well, you might say, as long as you get the bulk of it correct with a wellconsidered methodology, you still have a valid indicator, you compare your result with other data sources and so on, adjust it within the realm of plausibility and credibility. But these days there can be sudden very large inflows and outflows of funds which do not conform to any past trend, that is the point. And the assumption is always, that companies present their accounting information in an honest way. There are many other aspects of data accuracy that could be discussed, but I won't go into that here. Well anyway, I had among other things a Charlie Brown cartoon on my office noticeboard where Charlie visits Lucy at her toy psychiatrist clinic desk, and explains his problem, and Lucy says something like "for 38 percent of people this is no problem" or something like that, and Charlie asks, "where did you get those figures ?" and Lucy says "I just made them up". Reference: Oskar Morgenstern, On the Accuracy of Economic Observations, 2nd edition, Princeton University Press, 1965. Jurriaan
