ACTA SLAVICA IAPONICA

Volume 16 (1998)

The Regional Problem and the Break-Up of the State:
The Case of Yugoslavia
*
╚aslav OciŠ

The Nature and the Scope of the Regional Problem
Regional Policies and Changes in the Institutional Framework
Regional Development Levels: Grouping of Regions
Structural Change: Shift-Share Analysis
Efficiency: Shift-Share Analysis
Interregional Relations: Autarky
Some Other Results of the Regions' Development
Regional Development Costs: Ratios of Investment
Interregional Income (Re)distribution
(D)evolution
Regional Convergence or Divergence?
Equality: The Failure of the Positive Discrimination Model
"The National Question" and Nationalism
Separatism: Economic and Political
Federalism
A Long Journey from Utopia to Dystopia
Selected Bibliography
Data Sources & Documents
Notes
Appendix  (1)  (2)

Regional Convergence or Divergence?

A good illustration of this is the results of a quantitative analysis of the extent to which the objective of interregional equality was achieved. Unlike Williamson,15 whose international comparison of regional disparities was made in terms of per capita GNP only, here we have included employment per 1000 working-age inhabitants and fixed assets per working-age inhabitant -with V1 and V2 being measures of relative regional differences, and M a measure of absolute regional differences. Moreover, V1 is a weighted measure of regional differences, since squares of the deviations in regional indicator values and indicator values for Yugoslavia are weighted by the share of working-age population or total population in the corresponding aggregate at the Yugoslav level. The measure of absolute differences is also a weighted quantity, with the weights being the same as in the calculation of V1. In order to determine the pattern of regional differences over the observed period (1952-1990) each series of obtained values was regressed with relation to time, i.e. the trend functions were estimated. For each series of values of regional differences we specified and estimated three basic functional (co)relations, with time as an independent variable: linear, log-linear and semilogarithmic. The criterion by which a trend function for each series of values of the dependent variable was chosen was the statistical significance of the estimated parameter ▀ and the statistical significance of the estimated function measured by the coefficient of determination.
The trend of regional differences in employment per 1000 working-age inhabitants in terms of the V1 measure shows several subperiods. From 1952 to 1961 regional differences in employment fluctuated following a downward trend. From 1961 to 1964 they were on the increase, then from 1964 to 1972 they fluctuated again but followed no marked downward or upward trend. From 1972 to 1979 a constant growth in regional differences was observed, and from 1979 to the end of the observed period (1990) they decreased each year. It is this continuous decline in regional differences over the last nine years that mostly determined the downward trend for the whole period. From the type of trend function (a semilogarithmic one) it can be inferred that relative regional differences in employment per 1000 working-age inhabitants measured by V1 were rapidly decreased over the whole observed period (1952-1990). A similar trend of relative regional differences was obtained for the V2 indicator. There is also a significant decrease in relative regional differences. Here again, the type of trend function shows an accelerated decline in relative regional differences. However, the estimated value of coefficient ▀ in this function is smaller than in the case of the V1 indicator, as a logical result of the fact that the V1 indicator was calculated by weighting the squares of deviation.
The trend of relative regional differences in the value of fixed assets per working-age inhabitant measured by both indicators (V1 and V2) clearly shows two subperiods. In terms of V1, relative regional differences decreased over the first subperiod (1952-1971), but then increased over the second subperiod (1971-1988). The trend of relative regional differences over the first subperiod is best described by the semilogarithmic trend function, which means that these decreased at a diminishing rate. The trend of regional differences over the second period is best described by a linear trend function, which means that differences increased at a constant rate ▀. In terms of V2, however, as early as 1967 the trend of relative regional differences in the value of fixed assets reversed. They had decreased up to this year, and then started to increase. The trend of relative regional differences over the first subperiod (1952-1967) is best represented by a linear trend function, which suggests that differences decreased at a constant rate ▀. A linear trend is also characteristic for the regional differences over the second subperiod (1967-1990), but the value of the estimated parameter ▀ is positive, which means that differences widened by a constant coefficient. But when the whole period is considered in terms of both indicators (V1 and V2) the downward trend of relative regional differences per working-age inhabitant prevails. In both cases trends are best depicted by the semilogarithmic trend function, which indicates that over time regional differences decreased at a diminishing rate.
Relative regional differences in GNP per capita clearly follow an upward trend, either measured by V1 or V2. In both cases this trend is best described by the semilogarithmic trend function with the logarithmically computed dependent variable. This means that relative regional differences in per capita GNP widened at an increasing rate.
In regard to absolute regional differences in employment per 1000 working-age inhabitants there are four subperiods with different tendencies. During the 1952-1964 subperiod absolute differences increased, during the 1964-1971 period they decreased, then increased again in the 1971-1979 period. Finally, from 1979 to 1990 they diminished year by year. When the whole (1952-1990) period is considered, absolute differences in terms of this indicator clearly demonstrate a downward tendency. This is confirmed by the estimated function of the semilogarithmic trend, according to which absolute regional differences in employment per 1000 working-age inhabitants diminished at an increasing rate.
However, absolute regional differences in the value of fixed assets per working-age inhabitant display no common tendency for the observed period as a whole (1952-1990). This is confirmed by an insignificant value of a parameter estimated against time in all trend functions which were estimated for the entire period. There are four subperiods. First, from 1952 to 1954, when differences grew at a constant coefficient; second, from 1954 do 1962, when differences declined at a constant coefficient; third, from 1962 to 1974, when the absolute differences between regions increased; and fourth, from 1974 to 1990, when absolute differences in the value of fixed assets per working-age inhabitant increased again, but faster than in the preceding subperiod.
In terms of per capita GNP as an absolute indicator, regional differences have the same trend as in the case of the V1 and V2 indicators. Namely, absolute regional differences also display an upward tendency over the whole period (1952-1990). Judging by the form of the trend function that best describes the tendencies in these absolute differences, the latter rapidly increased.
Results of the analysis show that both relative and absolute regional differences in employment and fixed assets declined during the observed period. Moreover, the decline of differences in employment was steep, while in fixed assets it was gradual. In the last decade, however, both absolute and relative differences between regions in terms of fixed assets increased. In terms of GNP, both relative and absolute differences rapidly widened during the entire period observed.