POPULATION PROJECTIONS As described in the section on methodology, changes in area population are projected as a function of changes in area employment, with special adjustments made for "retirement" areas. Because projected employment changes are reconciled with projected changes in earnings of employees, and because earnings form the bulk of personal income, there is a strong correlation between the projected changes in population and those in income. As with the income projections, the projected population movements are generally in line with past trends. The only significant exceptions are in New England and the Rocky Mountains (table 4). New England's population is projected to rise at an above-average rate. The region's population growth was below average during the 1940's and 1950's, when the region was falling behind national growth rates in most economic measures. During the 1960's, when economic growth in the region nearly equaled that in the Nation, population growth was only slightly below average. During the 1970's and 1980's the region's population growth is projected to exceed the national rate by a small margin as people are attracted to the above-average economic opportunities of the region. In the Rocky Mountain States, where population growth was above average from 1929 to 1969, the projections call for growth slightly below average over the next two decades. This stems from the relatively slow growth projected for agricultural income in Montana, Idaho, and Wyoming. Slow income growth means fewer economic opportunities, which mean less immigration and sometimes even net outmigration of population. Population projections for individual States are shown in tables 4 and 6. PER CAPITA INCOME In general, regional population and income growth rates both tend to deviate from the national average in the same direction, although the magnitudes of the deviations may differ a good deal (table 1). However, the observed tendency for regional per capita incomes to converge toward the national average means that some divergence does occur between trends in an area's population and in its personal income. The degree of this divergence appears to be correlated with the level of per capita income. Thus, in regions with above-average per capita income, the population growth rate tends to exceed the national average by a wider margin than does the personal income growth rate. In areas with belowaverage per capita income, population growth relative to the national average is slower than personal income growth relative to the national average. Under both conditions, the per capita income of the region moves toward the national average. The most striking examples of this occur in the Southeast and Far West (table 2). 2. Percent changes calculated from data carried to one more decimal than shown. 3. Alaska and Hawaii are excluded from 1929 data. To achieve comparability, they were excluded from 1950 data in calculating percent change for 1929-50 period. In the Southeast, personal income growth far outpaced the national average from 1929 to 1969, but population growth was below average (table 1). As a result, per capita income in the region rose from 52 percent of the national average in 1929 to 80 percent of it in 1969 (table 2). In the Far West, by contrast, income growth was well above average from 1929 to 1969 but population growth exceeded the national rate by an even greater margin, and per capita income in the region fell from 129 percent of the national average to 112 percent. Continued convergence of per capita incomes is projected for 1969-90. This can be seen in summary in table 5 and in detail in table 6. CONCEPTS AND METHODOLOGY The projections presented here are based on an extension of past relationships believed to have relevance for the future. The choice of relationships to be extended and the methodology for extending them are based on assumptions, some of which are stated explicitly and some of which are implicit in the projection methodology. The assumed conditions are those believed to have the greatest probability of realization. Thus, the projections represent an attempt, imperfect though it may be, to forecast the economic future. In general, long range projections are more likely to prove wrong than are those made for short periods, and projections in detail are more likely to prove wrong than those of broad aggregates. Accordingly, projections for 1990 are probably less reliable than those for 1980; projections for a specific industry in a specific region are probably less reliable than those for the same industry nationally; and earnings projections for a specific industry are probably less reliable than those for total earnings or total income. By the same token, a projection of the labor force at the national level for 1990 is probably quite reliable because that labor force will be drawn almost entirely from a population the size and age distribution of which are known, though projected participation rates may be wrong. However, a projection of the 1990 labor force in a given State is related not only to the current State population but also to future interstate migration, and is therefore much less reliable than the national projection. Assumptions The projections are based on longrun or secular trends and ignore the cyclical fluctuations which characterize the short run path of the economy. The general assumptions that underlie the projections are as follows: (1) Growth of population will be conditioned by a decline of fertility rates from those of the 1962-1965 period. (2) Nationally, reasonable full employment, represented by a 4 percent unemployment rate, will prevail at both of the points for which projections are made; as in the past, unemployment will be disproportionately distributed regionally, but the disproportion will be diminishing. (3) At projection dates, there will be no direct effects on the projections due to foreign conflicts. (4) Continued technological progress and capital accumulation will support a growth in private output per manhour of 3 percent annually. (5) The new products that will appear will be accommodated within the existing industrial classification system, and, therefore, no new industrial classifications are provided. (6) Growth in output can be achieved without ecological disaster or serious deterioration, although diversion of resources for pollution control will cause changes in the industrial mix of output. (7) The composition of personal consumption will continue to change. The regional projections are based on the following additional assumptions: (1) The factors that have influenced historical shifts in "export" industry location will continue into the future but the sharpness of the shifts will diminish. (2) Trends toward area self-sufficiency in local-service industries will continue. (3) Workers will migrate to areas of economic opportunities and away from slow growth or declining areas. (4) Regional earnings per worker and income per capita will continue to converge toward the national average. (5) Regional employment/population ratios will tend to move toward the national ratio. Projection procedure The State income and employment projections were made in six major steps. First, total national population, employment, GNP, personal income, and earnings were projected. Second, the projected national output, employment, and earnings were broken down into industry detail on the basis of projected trends in industry shares of the national totals. The third step was to allocate the projected industry totals of employment and earnings to the 173 economic areas into which BEA has divided the country. The methodology of this step for each area's "export" industries-those that mainly produce goods and services for export from the area to other areas-was different from the methodology for the area's other industries, i.e., its "residentiary" industries. Basically, however, earnings and employment in both types of industry were projected by extrapolating past trends. The nonearnings components of personal income were projected for each area by a method similar to that used for residentiary industry earnings. The fourth step was to derive area population totals from projected area employment. The fifth step was necessitated by the fact that many of the 173 BEA economic areas cross State boundaries. In those cases, it was necessary to divide the projected area figures into State segments. Sixth, the area projections were aggregated to State totals. The projections procedure is by no means entirely mechanical: At various points in the process, it is essential that judgment be brought to bear, both in estimating the future rate of change in the industrial composition and location of output, and in checking the consistency of the projections. In particular, with employment and earnings projected separately, it is necessary to review for reasonableness the implied industrial and regional patterns of earnings per worker. The decision to derive regional projections through the disaggregation of national totals instead of through the independent projection of each component in each region is based on the assumption that the larger the economic area, the more adequate and reliable are the available statistical measures and the more reliable are the projections that can be made. This assumption applies also in the decision to derive projections of industrial detail at the national level by first projecting national employment, output, and earnings and then disaggregating into national totals for individual industries. Of course, it should be obvious that the disaggregation approach is also subject to substantial error-as is any procedure for forecasting the economic future. Gross national product The initial step in preparing the national projections was the projection of the gross national product. This was done by multiplying projected manhours worked by projected output per man-hour. The variables which entered the determination of man-hours worked include the working age population, labor force participation rates, general government employment (civilian and military), and hours worked per year per man. The Bureau of the Census has made several different population projections, with the birth rate assumption the varying element. In light of all the factors that could be ascertained in mid-1969, when the decision was made regarding the projected population to be used, the "C" series was selected. Of the five Census Bureau projected population series, this one has the second fastest growth rate. It assumes a total fertility rate of 2,787 births per 1,000 women in 1990, which is below the rate of 3,300 in 1962-65, but above the rate of 2,111 per 1,000 which would maintain a constant population, and above the preliminary 1970 rate of 2,472. The "C" series shows national population increasing from 203 million in 1970 to 270 million in 1990, or about 33 percent. The "E" series, which projects lower fertility than the "C" series and for that reason would be the choice of many persons today, is only 4.4 percent lower than the "C" series in 1990. The working age population, labor force participation, unemployment rate, and hours worked per man per year were each projected separately and the best available expertise was utilized in each case. The population of working age is, of course, a subset of the total population; the labor force was de rived by applying age- and sex-specific participation rates developed by the Bureau of Labor Stastistics to the working age population. A 4-percent unemployment rate was adopted as representing full employment nationally. House worked per man per year in the private economy were projected to decrease by 0.25 percent per year, compared to the post-World War II average decrease of 0.4 percent per year. The slower rate assumes that the "easy" reductions in hours of work have already been made. Output (real gross product) per man-hour in the private economy increased at a compound annual rate of 3.2 percent from 1950 to 1968. The projection puts productivity growth at a compound annual rate of 3 percent from 1968 to 1990. The projected rate is somewhat lower than the 1950-68 rate to allow for the fact that some part of the productivity growth in 1950-68 was attributable to a massive movement from farm to nonfarm work which cannot be repeated on a similar scale in the future. There is a variety of opinion on the merits of using the 3-percent rate; some forecasters would lower the projected rate still further because of the projected shift in work force distribution away from higher productivity manufacturing to the lower productivity service industries. Private gross product was projected by multiplying private man-hours by output per man-hour in the private economy. Constant dollar government gross product was projected in accordance with conventional national income and product accounting proctice as the number of general government employees times average compensation in the base year. The sum of projected private and projected government gross product is projected constant dollar GNP, which grows at a compound annual rate of 4 percent between 1968 and 1990. Since measures of gross regional product have not been constructed, it was necessary to translate GNP into measures which could be prepared regionally. Personal income and earnings The measures chosen for this purpose are personal income and its earnings-of-persons component (the sum of wages and salaries, other labor income, and proprietors' income). The choice rested on three considerations. First, personal income has a comparatively constant relationship to gross national product; second, its regional location is clear and can be measured with current data sources; and, third, the methodology for preparing regional estimates of personal income had already been developed. Projected personal income was derived from the relationship between constant dollar personal income and constant dollar GNP. A function was fitted mathematically to past values of the income/GNP ratio and extended to 1990. The projected 1990 ratio was applied to projected GNP to derive projected personal income. In a similar manner, the ratio of earnings of persons to total personal income was projected and applied to projected personal income in constant dollars to projected earnings in constant dollars. Industry detail The projected values of three national aggregate measures were disaggregated industrially. The three measures are gross product (which at the industry level is gross product originating, or GPO), earnings of persons, and employment. The disaggregation was into the 37 industry groups for which local area data on earnings and employment are available. (When the final projections were assembled at the State level, the 37 industries were combined into 28.) The disaggregation was done by extrapolating 1948-68 trends in the industrial compo sition of gross product, earnings of persons, and employment; that is, each industry's share in total GNP, total earnings, and total employment was extrapolated and applied to the projected all-industry totals of GNP, earnings and employment. The resulting projections of GNP, earnings, and employment for each industry were then reconciled. The reconciliation focused on two major considerations. First, projected industry GPO, earnings, and employment were examined in the light of historical trends in the relationships among earnings per worker, GPO per worker, and earnings as a percent of gross product. Second, those ratio relationships for each industry were reviewed in the light of the corresponding allindustry ratios, to judge how well the projected data adhered to the empirical observation that interindustry differences in earning per worker and GPO per worker diminish over time. The national projections of industry employment used the data on "persons engaged in production" that are calculated by BEA as an adjunct to the national income and product accounts (table 6.6 in July issues of the SURVEY). These data are conceptually consistent with the series on gross product originating and earnings. However, the "persons engaged in production" series is not available on a regional basis. The only employment data with adequate industrial detail now available for local areas are from the decennial censuses of population. It was necessary, therefore, to convert the projected national industry employment to the Census employment concepts. This was done by first eliminating government workers from the various industries. The resulting 1960 Census employment figures for each industry were then extrapolated forward by the projected change in the "persons engaged" series (1970 Census employment data are not yet available for all States). Independently projected estimates of government employment were then added to projected private employment in the appropriate industries to yield national totals of employment for distribution to economic areas. Economic areas Once projections had been made at the national level for output, earnings, and employment by industry, the next step was to allocate the national industry totals to subnational areas. This was done using the 173 economic areas into which BEA has divided the country as part of its program of regional measurement, analysis, and projection. Each area has an urban center and surrounding counties where economic activity is focused, directly or indirectly, on the activity of the center. Each area combines place of residence and place of work as nearly as possible so that there is a minimum of commuting acros area boundaries. Each economic area has two types of industries. The "expert" industries produce goods and services that are for the most part exported to other areas, providing the earnings with which the area purchases the specialized goods and services of other areas. "Residentiary" industries produce most of the services and some of the goods required by local business as intermediate products and by the household sector. Each economic area approaches self-sufficiency with respect to its residentiary industries. There is general similarity among economic areas in the interindustry relationships among "export" and residentiary industries-within each area. Moreover, these interindustry relationships within areas exhibit substantial stability over time, although they do change as a result of secular trends and developmental thresholds (points at which local markets for intermediate or consumer products become large enough for local production to supplant all or a portion of imports). These characteristics of similarity and stability make the BEA economic areas superior for projection purposes to other geographic areas delineated in accordance with noneconomic criteria. For example, the relationships among industries located within a single county may appear to be meaningless and random. Such relationships would acquire meaning, of course, if data were available on the county's imports and exports so that total input requirements of each local industry could be calculated. Use of the BEA economic areas for projections and analyses makes it unnecessary to have such data for residentiary industries, though export-import information is still needed for the "export" industries. Local area economic measures The local area economic measures used in the projections are population, total personal income, earnings by industry of origin, and employment by industry. Estimates of total personal income, earnings by industry, and the non-earnings component of personal income in each SMSA and non-SMSA county have been |