METHODS OF REGIONAL ANALYSIScannot

METHODS OF REGIONAL ANALYSIS

cannot be assumed to be always valid for counties within a state, but also because difficult problems of multicollinearity and serial correlation arise in connection with the use of a set of “independent” variables.45

3. INCOME ESTIMATES FOR RIGIONS OF OTHER TYPES

The excellence of the Department of Commerce state income estimates has led to their use for most regional income estimates in the United States. This has been the consequence whether we consider regions composed of integral states, of several counties within one or several states, or of integral states and some counties of other states. As one example, various government agencies very frequently define a region as a set of integral states and obtain regional income by summing the incomes of the relevant states. As a second example, the Tennessee Valley Authority constructs income estimates for any of its subregions by summing the incomes of the counties contained by that subregion.46 As a third example, the Federal Reserve Bank of St. Louis (which is an observer of county boundaries. but not of state) constructs an income estimate for its region by adding to the income of Arkansas the incomes of counties in each of six other states.47 Even when a region cuts across county boundaries, the tendency has been to allocate the county income derived previously from state income to the parts of the county.48
In short, there has not been any larger-scale effort in the United States to construct regional income estimates, entirely independent of national, state, and county income estimates.49 In view of the expensiveness of making independent income estimates, in view of the failure to develop a more fruitful conceptual framework, and in view of the available data collected by traditional political administrative units, this is understandable.

45 For further discussion of these latter problems, see standard advanced texts on statistics, such as those cited on p. 20.

46Some consider this income estimation procedure suitable for metropolitan regions as well.

47 See W. Hochwald [33. 34]; also see [l7].

48 It should be added that for such an apportionment reliable allocators are difficult to obtain: and in urban areas the situs problem is formidable when dealing with parts of counties. (L. R. Salkever [61].)

49This has been generally true of work in other developed nations. For example, see P. Deane [l4], F. R. E. Mauldon and A. M. Kerr [51], W. E. G. Salter [62], W. E. G. Salter and R. V. Peters [63]. and R. W. Peters [60]. The papers by Mauldon and Kerr. Salter and Peters, and Peters throw considerable light from a dill'erent angle on the conceptual and procedural problems in regional income estimation. Their studies also illustrate how excellent local data on the output of ram commodities and their costs or production can be combined with allocation procedures with respect to nonagricultural sectors to derive local income.