Our second assumption is that the location of noncommodity industries is related to the location of the commodity producing industry. While a determination as to which are basic and which are nonbasic industries for a specific region will depend on specific analysis of that region's economic structure, the simplifying assumption was generally made here that the commodity-producing industries were the basic sector and that the noncommodity industries were the nonbasic sector.

PROJECTIONS OF EMPLOYMENT IN COMMODITY INDUSTRIES.-Our procedure placed primary emphasis on the projection of employment in commodityproducing industries. Manufacturing was projected at the two-digit SIC level of aggregation; agriculture and mining at the one-digit level. We arrived at our employment projections in each of the commodity industries by means of successive approximations embodied in a three-stage, trend-supply-demand approach.

First we extrapolated State industry employment on the basis of our study of past trends in the following manner. The projected addition to (or loss of) national industry employment between 1957 and 1976 was distributed among regions and States in the same proportions as the increment (or loss) of employment felt by that industry was distributed between 1947 and 1957. These amounts were then added to (or subtracted from) our absolute employment figures for 1957.

Such an uncritical extrapolation does not question the casual determinants of employment shifts, but assumes only that the net effect of conflicting locational forces on the differential growth of regions and States will continue as in the first postwar decade. Therefore this first approximation was revised very significantly in our second and third approximations, which provided our final estimates.

In the second approximation, specific attention was devoted to anticipating changes in the relative importance of a number of supply and demand factors deemed to be particularly crucial to determining location decisions in each individual industry. In particular, a judgment was made as to the prospective improvement (or deterioration) in a State or region's competitive position with regard to attracting new plant capacity. These judgments depended on an evaluation of the future of each region's and State's access to skilled labor, adequate raw materials, other intermediate inputs, and low transportation costs.

Attention was also given to the possible impact of forward or backward linkages among industries on the location of new capacity. Such links, as affected by both institutional and technological factors, provide clues to the future location of a given industry once a determination had been made of the expected location of closely associated industries. Finally, an attempt was made to determine which State or region's industries might improve their competitive position by virtue of an improvement in their access to large-scale external economies as that region gained population and approached a well-diversified industrial structure and with it greater ease of supply of semifabricated inputs from related industries.

Our projections were revised further in our third approximation, which took into consideration expected changes in the relative advantages to be offered by the various regions and States in the sizes and types of market that they were expected to offer.

To correct our projections for market characteristics, we found it necessary to have first an approximate idea of total civilian employment and population in each State and region. This first rough approximation of total employment was obtained by using the commodity employment figures reached through our first and second approximations. We tentatively assumed that for each State and region the 1947-57 trend in the commodity-noncommodity relationship could be extrapolated to 1976 after adjusting for the projected decline in the relative position of the commodity sector for the United States as a whole. Rough projections of State population were then derived by extrapolating past trends in the relation of State employment to State population, account being taken of an expected slow decline in the U.S. ratio of employment to population. With these rough estimates of population, our State commodity industry estimates were further revised by comparing State population gains with gains in those industries most strongly oriented to markets within the State. Regional estimates were also revised in the same manner for those industries with important markets within the multi-State region.

After completion of our three approximations, we submitted the State industry projections to two additional tests:

1. For each industry, we compared each State's (region's) 1976 share with a trend extrapolation of its shares over the 1947-57 period; serious divergences forced a reexamination of our estimates.

2. The structure of each State's (region's) commodity industry employment was examined to determine whether the revisions resulted in inconsistent projections of the growth in related industries.

PROJECTIONS OF EMPLOYMENT IN NONCOMMODITY INDUSTRIES.-For each of the noncommodity sectors we obtained a first set of estimates by extrapolating for each region and State past ratios of its employment to total commodity industry employment, taking into account the faster projected growth in the U.S. as a whole of each noncommodity sector than the growth in the commodity sector. These ratios were then modified to take into account specific regional and State information on noncommodity industries, in particular for those noncommodity sectors such as finance which were not closely related to final or intermediate markets within the region. These projected ratios were then translated back into actual employment projections by applying them to our projected U.S. commodity industry employment.

This should not leave the impression that we think that commodity-noncommodity relationships are simple. Of course, each noncommodity sector behaves in a different way under the pull and push of different forces and these naturally influence its relationship to the commodity sector. However, after analysis

of each sector's individual peculiarities, we still found commodity-noncommodity relationships to be useful analytical tools.

From Employment to Population to Employment

The State population projections refer to the total civilian population residing in the State. Armed forces are not distributed by State but shown in the aggregate.

The State population projections depend on the projections of civilian employment. This was done largely but not exclusively by extrapolating past trends in the ratios of State employment to population, taking into account a projected decline in the U.S. employment to population ratio.

Demographic considerations were not neglected. For example, Florida, because of its rich endowment of amenities, is expected to attract an increasing share of U.S. employment, which tends to reduce its ratio of population to employment. However, the attraction of these same amenities to retired persons caused us to slow down the pace of reduction in Florida's population to employment ratio. Similarly, the expected availability of a sizable pool of female labor in New England caused us to revise upward our projections of apparel employment in that region, which contributed to maintaining New England's projected share of U.S. population, although its ratio of population to employment does not increase.

Unemployment Estimates

We did not make a detailed study of the projected State structure of unemployment, partly because of incomplete statistical coverage of State unemployment in past years. Historical series of insured unemployment are available by States, while total unemployment figures are available only for 30 States in 1959 and 1960. For the remaining States, total unemployment for past years was estimated on the basis of an estimated relationship between insured unemployment and total employment.

Unemployment by States for the year 1976 was projected on the assumption that each State's unemployment rate would bear the same relationship to the U.S. rate-projected at 4 percent for 1976-as it did in 1957. Thus the current depressed areas are expected to remain relatively depressed. State labor force projections can be estimated as the sum of State employment and unemployment.

Projections of Gross Product Originating, Income, and Consumer Expenditures

Projections of gross product originating, personal income, disposable income, and consumer expenditures were derived by relating them to the employment projections. The detailed one- and two-digit projections of employment were the basis of the regional GPO projections. The income and expenditure projections were based upon both the employment and the GPO projections.

The gross product originating in all of the regional industries is taken as our measure of regional

outputs. A complete discussion of the GPO in national industries is provided in "Industry Output, Employment and Productivity in the Years 1976 and 2000."

GPO BY REGION AND STATE.-The regional GPO is the sum of the gross product originating in the region from each of the one-digit industries. Each one-digit regional GPO was obtained by multiplying the 1976 regional industry employment by the projected GPO per employee in that industry and region. These measures of 1976 regional industry "productivity" were derived by extrapolating the trend in the deviation of each regional industry's GPO per employee from the national industry.

Using our regional GPO projections as control totals, we projected State GPO through trend extrapolations of each State's percent deviation from the regional ratio of GPO to personal income. Having projected the State ratio of GPO to personal income, personal income by State, and regional GPO, we could obtain 1976 GPO estimates for each State.

PERSONAL INCOME BY REGION AND BY STATE.State personal income was derived by two separate methods which yielded similar results.

In one method the trend in the percent deviation of each State's per capita income from U.S. per capita income was extrapolated. These were then related to the projected national per capita income to derive the projected State per capita incomes, which were then multiplied by the State population projections to yield the first set of State and regional personal income estimates.

The second method made use of the regional GPO projections by extrapolating the trends in the ratio of State GPO to personal income, as adjusted for differing expectations in the components of both personal income and GPO. The two sets of income estimates were then reconciled by successive approximations.

DISPOSABLE INCOME BY STATE AND REGION.Disposable personal income was derived by extrapolating the State and region trends in the ratio of disposable to personal income, bearing in mind that the area projections must add to the 1976 control totals for the United States.

CONSUMER EXPENDITURES BY REGION.-Consumer expenditures were derived from the projections of disposable personal income and projections of regional personal savings rates. The extent to which the 1976 regional personal saving rates might deviate from the projected U.S. saving rate was based on data gathered from a "Study of Consumer Expenditures" for the year 1950, published by the University of Pennsylvania.

Employment and Income Projections

for the Year 2000

The regional projections of employment, gross product originating, and income for the year 2000 followed procedures which were generally simplified. versions of the methodology employed for the 1976 projections. However, greater emphasis was usually

put on trend extrapolations, with less attention to specific casual determinants of locational changes.

As is the case of the 1976 projections, our procedure started with employment. We projected regional and 'State employment for the year 2000 principally on the basis of extrapolating the same trends in manufacturing, mining, and agriculture which we found most valid for the projections to 1976; the employment projections for the noncommodity sectors were based on trend extrapolation of 1957-76 relationships between each of them and the commodity sector, always keeping in mind that the projected U.S. totals for each sector for the year 2000 were controlling.

The regional projections for all other indicators were derived from our employment projections in

the same manner as described earlier for our 1976 projections.

Standard Metropolitan Areas

Some data were projected for selected standard metropolitan areas. The SMA population projections were obtained by adjusting the ORRRC SMA population projections on the basis of our national population projections. The SMA employment was obtained by assuming that the projected ratio of SMA population to State (in which the SMA is found) population would be the same as the SMA to State employment, The SMA income data were projected on the basis of the current relationship between family incomes in and outside SMA as estimated from data found in "The Conference Board Business Record."