Some of the most important problems to which regional economists and planners address themselves involve processes of growth (or more broadly, change) in the economies of regions. Such changes concern, of course, the people dwelling in the region; they concern also business firms and individuals who are choosing a region for their future activities; and they concern administrators and policy makers on the national level.

The objectives and tools of public policy will be examined in Chapter 12; the present chapter inquires why and how regional growth and other major changes occur.

11.1 SOME BASIC TRENDS AND QUESTIONS

Sheer growth of population is sometimes thought to be a measure of progress. More relevant to the idea of developmental advance is, of course, rising income levels. Finally, major changes in regional economic structure seem to accompany development. We shall look briefly at some regional trends in population, per capita income, and activity-mix in the United States in order to identify the most meaningful aspects and issues of regional development.

11.1.1 Relative Regional Growth in Population

Figure 11-1 shows the differences in the population growth rates of the Census divisions’ of the United States over the past century or so. We are concerned here not with the absolute sizes of the divisions¹  populations but only with the question of which areas have shown faster growth than others in each period. Accordingly, the chart is plotted on a logarithmic or ratio scale, so that the slope of a line on it represents the percentage rate of population growth per annum, and the lines for the different Census divisions (see Figure 11-2) are simply stacked in convenient order on the chart for comparison. We are interested not in the vertical position of these lines but only in their relative slopes.

It is immediately apparent that although all divisions have increased in population throughout the period, the rates of growth have been quite different for various divisions at various times. The earliest settled Eastern areas have grown more slowly than the others in the period shown. The West North Central division displayed above-average growth until 1890 but has since lagged, while the West South Central had a rapid growth phase lasting until 1910, and since then has just about kept pace with the rest of the country. The Far Western divisions, especially the Pacific, have grown faster than the national average throughout the period. The 1940s, the decade of World War II, brought sudden surges of population to the Pacific and South Atlantic, and further slackening of growth rates in the Middle Atlantic and West North Central.

It is apparent also that there has been a gradual tendency for the growth rates to become more alike as the pioneer stages of development have passed and the country has become more fully settled and more evenly industrialized. The fastest-growing parts of the country in recent years have been the Pacific Coast, the Southwest, and the Mountain states, while the East North Central and Middle Atlantic regions have tended to lose ground.

11.1.2 Regional Trends in Per Capita Income

We noted earlier (see Table 10-2) a substantial variation among major regions in per capita income, especially in money terms before any adjustment for relative living costs. Is the pattern of differentials historically well established?

Figure 11-3 portrays the changes in the relative levels of regional per capita income that have occurred since 1920. Each region’s per capita income is shown as a percentage of the national per capita income of the same date. The regional breakdown used here is not the same as in Figures 11-1 and 11-2 but is as shown in Figure 9-1.

We observe here again the persistently lower income level of the South. However, after 1929 there is also a general trend toward equalization, or convergence. The regional disparities become narrower.

Table 11-1 gives more detail on income differentials for the period 1929-1980, focusing on Standard Metropolitan Statistical Areas (SMSAs). The trends over this 51-year period are rather striking.

First, we observe that in the United States as a whole, per capita income levels have been consistently higher in metropolitan than in non-metropolitan areas, though the gap has considerably narrowed in recent years. In 1929, people in nonmetropolitan areas had incomes that were only 43 percent of those in metropolitan areas. That ratio has increased steadily, so that in 1980 nonmetropolitan per capita incomes were 74 percent of metropolitan per capita incomes.

Looking at the data for various regions shown in Table 11-1, we see once again the familiar differentials against the South and also marked convergence of the interregional differentials. Metropolitan per capita income in every region, without exception, was closer to the national average in 1980 than in 1929.

Within each region, we see a wide range of per capita incomes for individual SMSAs. In 1929, there was more than a 2-to-1 spread among individual SMSA income levels in six of the eight regions. Here too, convergence is evident. In 1962, 1971, and 1980 only two of the regions showed that much spread.

Finally, it is observable that as a rule the highest-income SMSA in a region was much larger in size than the lowest-income one. There is still a positive association of per capita income with size among metropolitan areas, though the differentials seem to be narrowing. More sophisticated analysis of trends in U.S. per capita income differentials, as reported by Irving Hoch in 1972, also showed incomes to be positively related to city size, and higher in the North and West than in the South, with both the interregional and the urban-size differentials converging between 1929 and 1962.²

A protracted controversy among statisticians and economists, which has produced a voluminous literature dating back at least to the 1930s, concerns the North-South differential in wages and incomes. From time to time someone has proclaimed the differential’s demise, whereupon someone else has reported finding it alive and well.³

Some of the apparent confusion results from the fact that there is more than one differential involved. There is little basis for dispute about continuing (though shrinking) differentials between the South and other regions in terms of per capita and per family incomes. Also, when looking at aggregate wage and earnings rates in most occupations and industries, North-South differentials persist. However, even in these respects it has been claimed that certain high-wage or high-income cities south of the Mason-Dixon Line should be counted out because they are not "really Southern."

But none of the differentials cited really implies that employers’ labor costs are lower in the South or that the Southern worker or resident is worse off than his or her Northern or Western compatriot. These basic questions involve some factors that are difficult to measure quantitatively. Productivity, dependability, trainability, and attitudes of employees are as important to employers as pay scales. On the earners’ side, relative costs of living need to be taken into account, and they are clearly lower in the South; but even the most elaborate consumer price index leaves out many intangibles that affect the desirability of a place to live, such as climate, recreational opportunity, or air quality. Finally, there are differences between large cities and small towns that are even more substantial than interregional differences, and therefore any legitimate comparison among regions has to be made in terms of individual size classes of places or with some other allowance for the interregional differences in degree of urbanization and average size.

One point on which there is universal agreement is that there has been a great deal of convergence of wage and income differentials, both interregionally and among cities of different size classes since the 1930s. Indeed, for the period after about 1962 it has been argued that such differentials as remain can be almost wholly explained away in terms of differences in occupational and population composition, differences in the measured cost of living, and equalizing differences compensating for such factors as air quality and congestion, which are not embraced in the cost-of-living measures. For example, Irving Hoch found that New Yorkers in 1967 had an average income 35 percent above the national average; he explained 9 percentage points of this on the basis of the cost-of-living index, 18 more points as equalizing other preference factors, and the remaining 8 points on the basis of population composition—leaving no differential attributable to disequilibrium of labor supply and demand.⁴

The highly aggregative nature of the data usually employed in analyzing interregional differentials has contributed to the difficulty associated with making valid comparisons. A recent study by Shelby Gerking and William Weirick has made use of data on individual household heads in order to eliminate the confounding influence of aggregation.⁵  For each individual, detailed measures of education, work experience and occupation, as well as information on work place and job characteristics were available. Controlling statistically for these and other factors, Gerking and Weirick find that real-wage or earnings differences for broadly defined geographic areas in the United States are not significant.

Equilibrium in terms of the absence of real differentials, however, does not necessarily imply any net migration; as we saw in the previous chapter, migration from an area depends to a very large extent on population structure rather than on the area’s relative income level.

11.1.3 Regional Structural Changes

Major changes in the activity-mix and other structural features of regions have accompanied increases in population density already indicated. For present purposes, we can focus on the trends in just one major aspect of development: "industrialization," as crudely measured by the relative importance of manufacturing employment for each region.

In Table 11-2 we have a series of location quotients in which each region’s relative industrialization is measured by comparing the percentage of population employed in manufacturing in that region to the corresponding national percentage in the same year.

For example, in 1899 in the United States as a whole, 6.49 percent of the population was employed in manufacturing industries, while in New England the percentage was 15.6, or 241 percent of the national average. This location quotient of 241 percent tells us that New England had 2.41 times as much manufacturing employment as it would have had on a pro rata basis if manufacturing were distributed in the same geographical pattern as was population among the regions. Referring again to Table 11-2, we see that in the same year, 1899, the West South Central region had a location quotient of only 28 percent, indicating a marked underrepresentation of manufacturing in that region.

As we read across to the later years, New England’s coefficient drops. The region’s specialization in manufacturing was diminishing, and New England was becoming more like the rest of the country in its degree of industrialization (or rather, the rest of the country was becoming more like New England). The Middle Atlantic region, another area of relatively early industrial development, showed a similar trend, while the East North Central region became more specialized in manufacturing until around 1950, and then less so. Most of the regions that were far less industrialized than the rest of the country in the earlier period had rising location quotients for manufacturing, so that the overall trend is strongly convergent. Manufacturing has come to be distributed among regions in a pattern more and more similar to the pattern of population distribution. This convergence is brought out by the bottom row of figures in Table 11-2, which shows the range of variation of the location quotients steadily narrowing.

11.1.4 Some Basic Questions on Regional Development

In this thumbnail survey of population growth, income levels, and industrialization, we gather that more recently settled regions have tended to show relatively fast growth for a considerable period, followed by a slowdown—suggesting a pattern of successive phases in a development sequence in which migration plays a prominent role. We see also that interregional differences in income level have been quite persistent but seem to have narrowed a great deal, especially in certain periods such as 1930 to 1970. Indeed, convergence in the sense of a growing similarity among regions is observable in respect to all three of the indicators examined: rate of population growth, level of per capita income, and relative importance of manufacturing employment.

Accordingly, some key questions suggest themselves:

1. Causes of growth. Why do some regions grow faster than others? What are the primary initiating factors responsible, and through what processes do these causes operate? What is the role of interregional trade, migration, and investment in the spread of development from one region to another?

2. Structure. How does regional economic structure relate to growth? What kinds of structure are conducive to growth, or the reverse? What structural changes are associated with growth?

3. Convergence. Why is convergence so much in evidence? Is it universal and inevitable, or is it subject to reversals?

4. Control over regional development. Can regional development be substantially guided by policy? If so, what are defensible objectives and appropriate policies?

The questions on policy will come up in Chapter 12; the other questions are examined later in the present chapter.

11.2 WHAT CAUSES REGIONAL GROWTH?

Regional growth and change entail complex interactions among activities within the regional economy, so it is not reasonable to expect that any single cause of such change can be identified. Useful explanations consist mainly of analyses of the ways in which an impetus of change is passed from one region or one regional activity to another, and we have in fact sorted out the various intraregional linkages in some detail already in Chapter 9. Some theories of development, however, emphasize certain kinds of change as especially independent, exogenous, primary, or causal. (All these terms mean much the same.) In particular, we shall see that the external demand for a region’s exports and its supply of labor and other production factors have been stressed as prime movers in some widely accepted theories of regional development.

11.2.1 Self-Reinforcing and Self-Limiting Effects

Our examination of the various kinds of linkages among firms and activities in a region brought to light some effects of a cumulative or chain-reaction character. Both vertical and complementary linkages are generally of this type. Thus external economies of agglomeration (the expression of complementary linkages) attract firms and activities of a similar nature, and this further enhances the agglomeration economies, so that still more firms and activities are attracted, which leads to still more agglomeration economies.

Vertical linkages per se have cumulative effects. If Detroit can increase its automobile sales to other areas, Detroit’s automobile manufacturers will buy larger quantities of inputs locally. Each of the supplying activities will then increase its own local purchases of inputs (for example, automobile workers will spend some of the increased payroll on housing, consumer goods, and services, and Detroit public utilities will need more labor and other inputs). Some of the additional spending in Detroit will take the form of increased purchases of automobiles, which will further contribute to the repercussions of the initial stimulus.

It appears, then, that vertical linkages of activities in a region and also complementary linkages (which are really combinations of vertical linkages) have self-reinforcing effects. An initial change in the level of activity in the region leads to still further change in the same direction and affects a broader range of activities. This applies to decline as well as to growth.

This being the case, how does it happen that regions do not normally expand in an explosive chain-reaction fashion, or wither away to the vanishing point? What are the forces that provide some constraint and stability, by setting up counterreactions to an initial change and thus limiting its total effects?

Part of the answer lies in the horizontal linkages among activities, which as we have seen are characteristically negative, or locationally repulsive, in their effects. In other words, activities in a region are always competing for some scarce local inputs (land, labor, and others); and, particularly in the short run, increased demand raises the cost of these inputs. Other constraints on explosive growth or decline will appear, as we look further into the mechanics of regional adjustment.

11.2.2 Demand and Supply as Determinants of Regional Development

The various kinds of linkages represent ways in which some impetus to regional change is transmitted from one activity to another within the regional economy, leading to overall growth or decline. The next question, then, is where can such impetus originate? What really initiates change?

Here as in almost every economic problem, the dichotomy of supply and demand appears. Regional activity requires both inputs and a market for outputs, and it does not make sense to argue that either supply or demand is the sole determinant of growth.

If we look to demand for the explanation of regional growth, we first inquire where the demand comes from and then trace its impact through the regional economic system. This approach will emphasize backward linkages among regional activities, since such linkages are the way in which a demand for one regional output (say, automobiles) gives rise to demand for other regional activity (say, the making of automobile parts or paint, the generation of electricity, or the employment of labor).

If we look to supply for the explanation of regional growth, we inquire where inputs come from and in what way the supply of, say, mineral resources, capital, or labor in a region leads to regional activity generating a regional supply of, say, coal, electricity, automobile parts, or automobiles. The approach from the supply side will emphasize forward linkages.

Clearly, both approaches are relevant and necessary parts of an adequate theory of regional change and development. Complementary linkages and external economies of agglomeration, as we have seen, involve both backward and forward vertical linkages; and in evaluating the factor of competition for scarce local inputs, both demand and supply have to be considered.

11.3 THE ROLE OF DEMAND

11.3.1 Economic Base Theory and Studies

One approach to an explanation of regional growth is that of the so-called economic base. The essential idea is that some activities in a region are peculiarly basic in the sense that their growth leads and determines the region’s overall development; while other (non basic) activities are simply consequences of the region’s overall development. If such an identification of basic activities can really be made, then an explanation of regional growth consists of two parts: (1) explaining the location of basic activities and (2) tracing the processes by which basic activities in any region give rise to an accompanying development of nonbasic activities. The usual economic base theory identifies basic activities as those that bring in money from the outside world, generally by producing goods or services for export.⁶

The argument advanced for this approach is that a region, like a household or a business firm, must earn its livelihood by producing something that others will pay for. Activities that simply serve the regional market are there as a result of whatever level of income and demand the region may have achieved: They are passive participants in growth but not prime movers. A household, a neighborhood, a firm, or a region cannot get richer by simply "taking in its own washing"; it must sell something to others in order to get more income. Consequently, exports are viewed as providing the economic base of a region’s growth.

A regional economic base study⁷  generally seeks (1) to identify the region’s export activities, (2) to forecast in some way the probable growth in those activities, and (3) to evaluate the impact of that additional export activity on the other, or nonbasic, activities of the region. The result is not only a projection of the region’s prospective growth and structural change but also a model that can be used in evaluating the effects of alternative trends of export growth.

A region’s export activities can be determined with various degrees of precision.⁸  The simplest and crudest procedure is simply to assign whole industries or activity groups to the export or nonexport category without making a specific local investigation. Thus retail trade, utilities, local government, and services may be classed en bloc as nonexport, while manufacturing is considered wholly an export activity.

A more sophisticated approach is to recognize that almost all activities in a region produce partly for export and partly for the regional market, and to try to estimate how much of each activity is for export. The simplest way to make such estimates is by using location quotients. For example, in 1970 North Carolina accounted for 2.45 percent of the national output of men’s and boys’ work-clothing factories, while personal income in North Carolina was estimated at 2.04 percent of the national total. The location quotient is 2.45/2.04=1.20. From this we could surmise that 20/120 or about one-sixth of North Carolina’s output of work clothing was for export to other areas and the remainder for consumption within the state.

This surmise, however, rests on the rather tenuous assumption that a region’s personal income is a good measure of its purchases of work clothing. If we wanted to use the location quotient approach to estimate how much of the Toledo SMSA’s output of metalworking machinery is for export, we would do better to base the location quotient not on personal income or population but on some statistic presumably more indicative of the demand for such machinery: for example, value added by manufacture in metalworking industries.

Location quotients are likely to lead to an underestimate of a region’s exports, since they are necessarily applied to whole industries or even industry groups. Within any industry classification (or for that matter, within any single firm or establishment), there are different specific products, and the region may be importing some and exporting others. Since the quotient estimates only the net surplus of output over regional consumption, it may seriously understate the gross exports of products of that industry.⁹

The location quotient method, however, does have the advantage of taking account of indirect as well as direct exports:

A community with a large number of packing plants is also likely to have a large number of tin can manufacturers. Even though the cans are locally sold, they are indirectly tied to exports. Location quotients will show them as exports.¹⁰

A more painstaking procedure is to get information on actual shipments of goods and services out of the region. In recent years, progress has been made by the Census in collecting and organizing data on manufacturers’ shipments between large regions. For some time, however, there will continue to be a dearth of information on exports from smaller regions such as individual metropolitan areas or counties; and exports of some services pose additional data problems. Many economic base studies have canvassed at least a sample of the firms that are believed to be involved in exporting, in order to get a reasonably accurate measure of the region’s external trade.

Projection of the future trend of exports from a region involves a series of studies of the prospective national growth and interregional location trends of each of the activities concerned, and an evaluation of whether the region’s competitive position is likely to get better or worse. The kinds of location factors to be taken into account in such studies have already been discussed in earlier chapters.

Given some prospective change in the level of export or basic activity in the region, how much overall regional growth in income and employment is implied? This determination requires the tracing of linkage effects. Specifically, it involves the estimation of a regional multiplier, which tells us how much increase in total regional income (or sales, or employment) to expect as a result of each additional dollar of export sales or income, or each additional person employed in producing for export.

At this stage too, there are alternative procedures of varying degrees of sophistication. The simplest method is to derive the multiplier from the "basic ratio." If, for example, one-third of the region’s employment is in basic activities, we simply assume that that proportion will be maintained. Accordingly, every worker added to basic employment will directly lead to the employment of two additional workers in nonbasic activities: the multiplier is 3.

Such a procedure is too easy to be convincing. There is really no reason to assume that the ratio will remain unaffected by export growth, and such ratios vary rather widely. There is a discernible tendency for export multipliers (whether derived by this or by more sophisticated analysis) to be larger with increasing regional size and diversity.¹¹

The view of export demand as the prime mover in regional growth raises some interesting questions that indicate the need for a more adequate explanation. Consider, for example, a large area, such as a whole country, that comprises several economic regions. Let us assume that these regions trade with one another, but the country as a whole is self-sufficient. We might explain the growth of each of these regions on the basis of its exports to the others and the resulting multiplier effects upon activities serving the internal demand of the region. But if all the regions grow, then the whole country or "superregion" must also be growing, despite the fact that it does not export at all. The world economy has been growing for a long time, though our exports to outer space have just begun and we have yet to locate a paying customer for them. It appears, then, that internal trade and demand can generate regional growth: A region really can get richer by taking in its own washing.

Let us next look at the role of imports. In the mechanism of the regional export multiplier, expenditures for imports represent demand leakage from the regional income stream. The greater the proportion of any increase in regional income that is spent outside the region, the smaller is the multiplier.

It follows that if a region can develop local production to meet a demand previously satisfied by imports, this "import substitution" would have precisely the same impact on the regional economy as an equivalent increase in exports. In either case, there is an increase in sales by producers within the region.

It is quite incorrect, then, to identify a region’s export activities exclusively as the basic sector. It would be more appropriate to identify as basic activities those that are interregionally footloose (in the sense of not being tightly oriented to the local market). This definition would admit all activities engaging in any substantial amount of interregional trade, regardless of whether the region we are considering happens to be a net exporter or a net importer. Truly basic industries would be those for which regional location quotients are either much greater than 1 or much less than 1.

This necessary amendment to the export base theory, however, exposes a more fundamental flaw. We are still left with the implication that a region will grow faster if it can manage to import less, and that growth promotion efforts should be directed toward creating a "favorable balance of trade," or excess of exports over imports. Let us examine this notion.

If a region’s earnings from exports exceed its outlays for imports, on net there is an exodus of productive resources from the region (as embodied in goods and services traded). In this sense the region is loaning its resources to other areas,¹²  and its people and businesses are building up equities and credits in those areas. Thus the region is a net investor, or exporter of capital. By the same token, if imports exceed exports, the region is receiving a net inflow of capital from outside.

It is patently absurd to argue that the way to make a region grow is to invest the region’s savings somewhere else, and that an influx of investment from outside is inimical to growth. If anything, it would seem more plausible to infer that a region’s growth is enhanced if its capital stock is augmented by investment from outside—which means that the region’s imports exceed its exports.

In any event, regional development is normally associated in practice with increases in both exports and imports. There was, in fact, a tendency among United States regions between 1929 and 1959 for increases in both per capita and total income to be greater in capital-importing (import-excess) regions,¹³  though there is no reason why this need always be the case.

We shall come back to this relationship later. The important point here is that explanations of regional growth based exclusively on demand lead to absurd implications, so that a broader approach is called for, along lines to be indicated later in this chapter.

11.3.2 Regional Input-Output Analysis

The economic base approach has been described in its simplest terms. Actually, various types of models of regional economic interaction have been developed to trace the impact of demand on a region’s income and employment. They all involve some framework of "regional accounts" describing transactions between the region and the outside world and among activities within the region; and nearly all include some type of multiplier ratio that sums up the relation between an initial increase in demand and the ultimate effect on regional income or employment. Some of these procedures are primarily relevant to short-term variations, while others are more relevant to long-term regional growth trends. We shall confine attention here to models using an input-output or interindustry framework.

The essence of the input-output schema is a set of accounts representing transactions among the following major economic sectors:¹⁴

• Intermediate—private business activities, within the region. The sector is broken down into individual industries or activities (such as mining, food processing, construction, and chemical products). It is sometimes referred to as the interindustry sector because much of the detail refers to transactions among the separate industries within the sector.

• Households—individuals and families residing or employed in the region, considered both as buyers of consumer goods and services and as sellers (primarily of their own labor).

• Government—state, local, and national public authorities, both within and outside the region.

• Outside World—activities (other than government) and individuals located outside the region.

• Capital—the region’s stock of private capital, including both fixed capital and inventories.¹⁵

These are, of course, transactions both among sectors and among the activities within each sector (for example, among households, or among different processing activities and regions in the "outside world"). But not all categories of transactions are of equal interest to us in analyzing a given region. The form of account illustrated in Table 11-3 represents a usual abridgment, where the lower right-hand portion is not filled in. What we have, then, is simply an itemization of the inputs and the outputs of each of the designated activities in the intermediate sector.

In order to express all these transaction flows in a common unit, they are stated in terms of money payments for the goods or services transferred. Thus the purchase of labor services from the household sector is shown as wages and other payroll outlays; inputs from the government sector are represented by taxes and fees paid to public authorities; and inputs from the capital sector are represented by depreciation accruals plus inventory reductions.

The accompanying schematic chart, Figure 11-4, may help in understanding the mechanics of the input-output model. The flows shown there are goods and services passing from one sector to another; money payments for those goods and services go in the opposite direction. The gray line represents the regional boundary; as noted earlier, the government and capital sectors are partly inside and partly outside the region.

Activities within the intermediate sector engage in interindustry transactions with one another (and also each with itself, since each activity includes a variety of firms with somewhat different kinds of output). Sales by the intermediate sector to other sectors are called sales to "final demand." At this point, the outputs are considered to be in their final form, not destined for further processing, and ready for their final stage of use as far as the region is concerned—namely, export, delivery to household consumers or the public sector, or incorporation into the stock of capital. They are leaving the region’s stream of current processing activity. The input-side counterpart to final demand is "primary supply": Imports and the services of labor, capital, and public authorities are entering the region’s processing system for the first time.

The abridged set of accounts in Table 11-3 shows total receipts and payments for only the activities in the intermediate sector, since transactions among all the other sectors are ignored. Thus we cannot read total regional personal income from a table such as this, since it omits the incomes that individuals receive from government jobs, pensions, property ownership, or sources outside the region. Nor does this table show total regional exports or imports of goods and services, since interregional transactions by the household, government, and capital sectors are omitted.

This kind of input-output table is particularly useful, however, in tracing and evaluating certain cumulative effects of vertical linkages in the region. It is easy to construct a set of "input coefficients" (see Table 11-4) showing that for each dollar’s worth of output of industry A in the region, that industry buys 1.2 cents worth of industry B’s output, 23.3 cents worth of industry C’s output, and so on.

Now let us suppose that industry A increases its sales outside the region by $1000. To furnish this added output, industry A will (according to Table 11-4) need to spend $12 more on inputs from industry B, $233 more on inputs from industry C, $442 more on labor payrolls, and so on. But industry C’s sales have now increased by $233, so it will have to spend $233 x .032 for additional inputs from A, $233 x .323 for additional inputs from B, $233 x .097 more for imported inputs, and so on. As each of the activities in the intermediate sector feels the impact of the increase in demand for its outputs, its own purchases in the region will increase. The chain of repercussions, or "indirect effects," is in principle endless; but this does not mean that the initial $1000 increase in A’s sales will snowball into an infinitely large growth in the region’s activities. The total effect, in fact, will be at most only a few times the size of the initial final demand increase. The ratio in this case is called the regional "export multiplier."

The reason that the multiplier is not infinitely large is that there are so-called demand leakages from the regional economy. Each time one of the intermediate activities experiences an increase in sales, it has to allocate part of the extra revenue to purchasing inputs not from other intermediate activities but from primary supply sectors. Money paid for additional imports leaves the region, and its stimulus to regional demand is ended. Similarly (in the simplified model portrayed by our input-output accounts), disbursements for payroll, taxes, and depreciation simply drop out of the stream of "new money" that is being circulated among the processing activities. The stream gets smaller at each round and finally peters out altogether.

We can, in fact, gauge exactly what the total stimulus will be, on the basis of our hypothetical input coefficients. Table 11-5 shows the amount by which each processing activity’s sales are increased as the ultimate result of a dollar’s increase in the final demand sales of any intermediate activity, including the whole sequence of multiplier effects described earlier.¹⁶  These effects are naturally largest for the activity experiencing the initial final demand increase, since that increase is part of the total increment. This explains why the figures on the diagonal of the table are especially large. In the case we assumed (an initial $1000 increase in export sales by industry A), we see that as a result A gets a total direct and indirect increase of sales amounting to $1118, while B, C, and D come out with smaller increments: $126, $297, and $68 respectively.

The total increase in sales for the whole intermediate sector is $1609. Since all this resulted from an assumed initial $1000 increase in A’s sales to final demand, we could identify here a multiplier of 1.609. This is a specific multiplier ratio, evaluating the effects of an initial increase of final demand sales by industry A.¹⁷

This estimate of the multiplier, however, is almost certainly too small. Our evaluation of indirect effects took into account only the vertical linkages implied by transaction relationships among activities within the intermediate sector. A more sophisticated estimate would have to allow for vertical linkages involving other sectors, as well as for the positive effects of complementary linkages and the negative effects of horizontal linkages.

Perhaps the most obvious omission involves the household sector. With all this increase of intermediate sector output, payrolls must also increase, and it would be unrealistic to assume that all the added pay will be saved, taxed away, or spent outside the region. Instead, we should expect a roughly proportional increase in consumer demand for the outputs of the region’s intermediate sector, and this in turn would be magnified in its ultimate effect by the workings of the multiplier.

It is somewhat less certain that increased purchases from government and increased use of the region’s fixed capital and inventories would automatically induce either increased purchases in the region by government or a step-up in investment activity. And it seems rather unlikely that increased imports would lead (through raising incomes in other regions) to any significant increase in the demand for the region’s exports.

The upshot of these considerations is that final demand (except perhaps for the export component) is not really independent of primary supply, as our abridged set of input-output accounts assumed. The modifications or adjustments that might be called for would depend on the particular regional situation. But we might well decide that it would be more realistic to assume an automatic feedback from household supply to household demand than to assume no feedback at all. To incorporate this new assumption, we could simply take households out of final demand and primary supply and put them into the intermediate sector as an additional, fully interacting activity. Referring to Table 11-3, this would mean supplying numbers to fill out the presently incomplete "households" row and column.¹⁸  The successive steps and results are set forth in Appendix 11-2.

The possibility of shifting households out of the final demand category makes it clear that the decision about what activities to include to final demand (and primary supply) is not preordained or arbitrary but reflects our judgment about what relationships are important and relevant to the question at hand. Final demand in the input-output accounts framework really has the same implications as basic in the simple economic base model, and an input-output model with export demand as the only final demand category can be thought of as a more detailed description of an export-determined regional economy.

The inclusion of government in final demand does not represent any major departure from economic base principles. Government is a basic source of income if public expenditures in the region vary independently from total regional income. This is true of most federal and state government expenditures; perhaps a case could be made for putting local government in the intermediate sector.

The role of investment in regional economic change is not really spelled out in the simple form of input-output model that we have been considering; since by convention, sales to the capital sector of final demand include all sales of capital goods, whether within the region or outside or to governments. There are other, more complex, varieties of input-output tables, as well as more general systems of regional income and product accounts, that do lend themselves to analysis of the mechanisms of saving, investment, and interregional capital flow. These will not be discussed here,¹⁹ but it is appropriate to ask whether investment in a region should more logically be considered (1) an exogenous factor initiating growth of regional income and output or (2) a response to other changes in the regional economy.

The answer depends on whether we are concerned with the short run or the long run. In the short run, rates of investment can vary widely and suddenly relative to levels of output, and decisions by major firms in the region to make extensive additions to their facilities can almost immediately convert a depressed region into a prosperous one. The question in the short run is the degree to which existing regional labor and productive facilities are fully employed, and changes in investment outlays can be a major determining factor. Thus a short-run regional model should certainly treat investment as primarily an exogenous or basic element.

For the long-run development of a region, however, it is reasonable to regard investment at least partly as a reflection of regional size and growth, rather than as a sufficient explanation in itself.

Input-output clearly represents a big advance over the simple economic base approach to regional growth; not only because it traces repercussions in a more sophisticated and detailed fashion, but also because it recognizes possible initiation of growth from various elements of final demand other than export sales.

For simplicity’s sake, we looked at the elementary single-region set of input-output accounts. More comprehensive and impressive models can be made if the "outside world" is broken down by areas and activities; and progress has been made in various countries toward complete multiregional accounts systems tracing flows among economic sectors and activities within each region and among regions as well.²⁰

Such accounts lend themselves to a wide array of useful impact analyses. Starting almost anywhere in the system, we can make a change "on paper" and see what happens. We can hypothesize, say, that the sales by some activity in some region increase; or the regional incidence of government expenditures and taxes is shifted; or some major investment project is executed; or consumer expenditures are changed in one or more regions by virtue of demographic change or shifts in spending habits; or new technology alters some of the input coefficients of individual activities. Starting from any such change, we can with an interregional impact model trace the initial and subsequent economic repercussions through the various economic sectors and regions affected.

11.4 THE ROLE OF SUPPLY

In the accounts shown in Table 11-3, the intermediate sector is shown delivering outputs to the various final demand sectors and receiving inputs from those same sectors in their capacity as primary suppliers. Money payments for these goods and services flow in the reverse direction, from final demand sectors to the intermediate sector and then to primary supply.

In tracing changes, we can follow the flow of money payments "backward" from purchaser to seller, or we can follow the flow of goods and services "forward" from producer to user. The scheme is symmetrical with respect to supply and demand, or input and output. It does not indicate whether we should look for the initiating causes of regional growth and change in final demand, in primary supply, or within the intermediate sector; and we might reasonably infer that change can originate in any of these three areas.

In view of this basic symmetry, it is striking that the techniques of input-output and multiplier analysis have nearly always been applied in just the backward direction, tracing the effects of changes from final demand to the intermediate and primary supply sectors.²¹  The implication in locational terms is that market orientation and backward linkage are all-important, with no attention being paid to input orientation or to forward and complementary linkage effects.

Because an input-output table is a reasonably comprehensive and neutral image of a regional economy, we can use it as a point of departure for the consideration of supply factors as well as demand factors. The demand-driven model discussed above emphasizes final demand, backward linkage, and output orientation of activities. Now let us reverse the emphasis to focus on the roles of primary supply, forward linkage, and input orientation.

When considering the effects of demand on regional activity, we implicitly assumed that supplies of inputs would automatically be forthcoming, at no increase in per-unit cost, to support any additional activity responding to increased demand. In other words, supplies of inputs, such as labor, capital, imports, and public services, were taken to be perfectly elastic and consequently imposing no constraint on regional growth. If export demand for a region’s steel output increased, the region could freely import as much additional fuel or iron ore as might be needed; if the demand for labor exceeded the region’s labor force, more workers would join the labor force or move in from other areas.

Conversely, a supply-driven model of regional growth takes demand for granted (that is, it assumes that there is a perfectly elastic demand for the region’s products) and thus makes regional activity depend on the availability of resources to put into production. Accordingly, the starting point in the process of change now becomes primary supply rather than final demand. Availability of labor, capital, imported inputs, and government services (infrastructure) makes possible, through forward linkage, certain intermediate activities oriented to such primary inputs. Increase in output by an activity that sells in the region can encourage, through further forward linkages, increases by other activities, giving rise to what may be called a "supply multiplier" effect. This effect is limited by the existence of supply leakages. At each stage, some of the increase in regional outputs is drained off into exports, investment, deliveries to governments, and household consumption—in other words, to the final demand sectors.

This supply-driven process sounds very much like the converse of the demand-driven process discussed earlier, whereby an initial increase in final demand gives rise to indirect growth of income and employment in the region and increased drafts upon primary supply. Conceptually, the symmetry is complete.

There is, however, an important operational difference. In practice it would not be feasible, save perhaps under quite special circumstances, to calibrate a supply-driven regional model simply on technical coefficients derived from the basic input-output table. The reason seems to lie in technology itself. Goods normally become more specialized in character as they pass through successive stages of processing and handling. We can legitimately use such input coefficients as, say, the amount of steel needed to make a pound of nails, because there is not much flexibility in the nature and amount of input required for a given output. By contrast, if we have an extra pound of steel, we cannot say whether it will be used to produce more nails or more steel sheets or automobile parts or whatever. Output coefficients are a weaker reed than input coefficients. Consequently, the forward-linkage and multiplier impacts of supply increase, though quite genuine, cannot normally be spelled out in terms of specific products and activities by input-output analysis, and with presently available techniques they can be estimated only in relatively impressionistic terms.

The demand-driven and supply-driven models should be viewed as complementary rather than as conflicting or rival hypotheses about regional economic change.²²  Each of the two model types in itself is one-sided and can be seriously misleading; for full insight into real processes, both need to be combined. As yet, however, there is no analytical model that adequately incorporates this union of the two complementary approaches.²³

11.5 INTERREGIONAL TRADE AND FACTOR MOVEMENTS

Systems of accounts do not in themselves tell us anything about where growth starts; they merely help us to trace impacts. But we can see already that a region’s growth involves at least three kinds of external relationships of the region: (1) trade, or the import and export of goods and services; (2) migration of people, both in their capacity as consumers and in their capacity as workers; and (3) interregional "migration" of other production factors, notably investment capital. A fourth external influence, to which some attention will be paid in the next chapter, is the national government’s revenue collection and expenditure in the region.

Trade among regions has, as David Ricardo noted a long time ago with respect to nations, the beneficent effect of allowing each region to specialize in those activities for which it is best fitted by its endowments of resources and other fixed local input factors, with all regions sharing to some extent in the economies of such specialization. Recognition of this effect helps to place the value and limitations of the export base theory in better perspective. When the local market is so small as to limit seriously the productivity gains that can be realized by specialization, exports may be necessary for growth. Thus the weakness of the export base theory lies not in recognizing exports as being important for growth, but rather in focusing on exports exclusively and failing to recognize that it is trade (imports as well as exports) that permits the realization of economies due to specialization.

This specialization of regions is limited, of course, by interregional transfer costs as well as by ignorance, inertia, and the like, and the simplified model implied here fails to take into account the economies of scale and regional agglomeration. But so far as it goes, the effect of freer interregional trade is likely to be in the direction of equalizing not only commodity prices among regions but also wages, incomes, and the rates of return to capital. The reason for this is that a region in which capital is scarce relative to labor can, with interregional trade, specialize in "labor-intensive" lines of production requiring much labor and little capital while importing the products of "capital-intensive" activities from regions better endowed with capital or less well endowed with manpower.

This substitution of trade for production-factor mobility is of course only partially effective. Considerable differentials persist in the rewards of labor and the returns on capital among the regions, leaving an incentive to further equalization by migration of those factors of production.²⁴

11.5.1 Mobility of Labor and Capital Among Regions

Determinants of labor mobility have already been explored at considerable length in Chapter 10. The rate of return to labor (real wages) is indeed a major determinant; but migration and regional manpower supplies depend also on the handicaps to movement imposed by uncertainty, ignorance, cost of moving, and social distance. Moreover, a person’s mobility varies widely according to his age, marital and dependency status, education, skills, and recent migration experience; and migration flows between places depend on such additional factors as previous flows (the beaten-path effect), the size and diversity of labor markets, and the effectiveness of interregional job-information and placement systems.

The mobility of capital is affected by a quite similar array of considerations. The prospective rate of return is, again, a major determinant; inertia, ignorance of opportunities, and social distance act as limiting factors in much the same way as they do for manpower mobility. The effectiveness of organization of the national financial system (including clearing arrangements, facilities for transferring funds from one region to another, securities exchanges, and interregional markets for still other types of investments and obligations) tends to set a limit on how much interest rates and other rates of return on capital can vary geographically within a county. Increased effectiveness of the national financial system in most countries has been evidenced by a trend of interregional convergence in money rates, though such rates still tend to be somewhat higher in places more remote from the chief national financial centers²⁵  and in smaller urban places. Something analogous to the beaten-path effect on labor mobility appears to affect capital mobility as well. Funds flow more readily and in response to a smaller rate-of-return differential from one point to another if there has been a great deal of previous investment following the same path.

Still another similarity appears in the effect of regional or community characteristics on the outward mobility of both labor and capital. A young area with a previous experience of inward migration and rapid growth shows more outward mobility of both factors than does a more settled and ingrown community.

Perhaps the most important difference between the processes of capital and labor migration lies in the fact that most capital has to be "sunk" or invested in durable forms such as site improvements, buildings, and production equipment before becoming useful. This major portion of the capital stock has virtually no spatial mobility. Movements of capital are thus confined to (1) newly created capital awaiting selection of a fixed-investment opportunity, and (2) working capital and other floating funds that remain in the form of paper assets or fairly easily movable types of commodities and thus retain interregional mobility.

The phenomenon of sunk capital would be roughly matched in terms of labor mobility if a high proportion of workers signed up for life on their first job. In Japan this is characteristic of employment practices, at least as they pertain to male workers in major corporations. There, the normal course is for the tenure of employment to extend over one’s entire working life as a matter of informal contract between employer and employee. More generally, however, people do lose mobility rather suddenly once they become established in an occupation, a community, and a family; and mobility thereafter declines further with increased age. In part this is due to the fact that information and skills relevant to a particular line of work, company, or community may be very specific, in the sense that they are not of comparable value elsewhere. So the contrast in this regard between the mobility of people and the mobility of capital is not as absolute as it might appear.

Scale and agglomeration economies affect the migration of both labor and capital, and in not too dissimilar fashion. A location that might be highly advantageous if enough manpower and/or enough capital could be concentrated there may never get over the threshold imposed by the higher costs of an initial small-scale operation or an insufficiently developed production cluster.

Finally, it can be observed that the migration of people from one place to another facilitates the movement of capital along the same route, and conversely. Each factor helps beat a path for the other. To some extent this reflects the fact that migrants normally bring some personal capital with them and often some business capital as well. A further explanation is that the increased familiarity with the other area, which comes from the movement of either labor or capital, enhances the mobility of the other factor along the same path, eroding the barriers of uncertainty and social distance.

More basically, the relation between labor and capital flows is affected by the way in which these factors combine to produce goods and services. Labor and capital can substitute for one another in production if it is possible to choose between a labor-intensive and a capital-intensive technique, depending on which factor is relatively cheap. The substitution relationship in itself would imply that a larger supply of capital in a region would lessen the demand for labor, since there could be a shift to more labor-saving production methods and more capital-intensive activities. Similarly a larger labor supply would lessen the region’s capital requirements.

But this picture is clearly unrealistic in many cases, because the factors of production are at the same time complementary. Using more of one may lead to using more of the other as well. The complementary relationship in itself would imply that a larger supply of capital in a region would augment the demand for labor, since production would expand in response to the enhanced competitive position of the region’s activities. Similarly, a greater supply of labor in a region would create a demand for additional investment in production capacity to take advantage of this more ample and perhaps cheaper labor.

In terms of regional growth, the effect of the substitution relationship may be viewed as equilibratory, whereas the effect of the complementary relationship may be viewed as equilibratory. To the extent that the complementary relationship between labor and capital predominates, interregional capital and labor flows can lend themselves to long spells of continued self-sustaining regional growth (or, on the other side of the coin, cumulative decline).

The effects of trade and of factor movements on regional structural differences can often be opposite. Trade in itself permits more intensive regional specialization and thus a widening of regional structural differences.²⁶  Interregional movements of labor and capital, on the other hand, would seem in general to weaken one of the main bases for specialization (that is, relative regional supplies of labor and capital) and thus promote convergence of regional structural differences.

This last surmise is subject to qualification, however, since it ignores the effect of regional differences in endowment of really immobile factors (land or natural resources). To the extent that such resources are complementary to labor and capital in production processes, regional specialization and structural differentiation based on fixed-resource endowments will be enhanced by greater interregional mobility of labor, capital, or both. This is likely to be true of mineral resources when they occur in regions with few other natural advantages—movement of capital and labor into such regions helps them to develop exploitation of their mineral resources as a regional specialty.

11.6 INTERREGIONAL CONVERGENCE

It is not difficult to find plausible explanations for the convergence of regional income differentials. Such convergence would seem to be a natural result of the gradual development and maturation of areas once on the frontiers of settlement, the greatly reduced relative importance of farming as a means of livelihood, the improvement of transport and communications and the enhanced mobility of both capital and labor, and the rise of more activities not closely oriented to natural resources and consequently enjoying a wider choice of possible locations. Increased interregional trade resulting from improved transport can also promote convergence by permitting regions to share to a greater extent the benefits of the production economies of other regions.

The story is not quite this simple, however, and we cannot infer that convergence will always be the order of the day. There seem, in fact, to have been two periods in United States history in which interregional income differentials either widened or remained about the same: from 1840 to 1880, and from 1920 till sometime in the 1930s. In the earlier period, the development of railroads brought rapid concentration of industrial activity in larger plants and larger industrial centers, and an increase in regional specialization. This raised incomes in the Northeast, where the bulk of the industrial activity was concentrated, compared to the basically agricultural and undeveloped parts of the country. In the period 1860-1880, the disruption of the Southern economy by the Civil War dramatically widened the gap between Southern and Northern incomes. Between World Wars I and II, there were at least two special reasons for divergent regional income levels. One was the low level of farm product prices and the consequently depressed condition of agriculture.²⁷ A major part of the regional income differentials, especially in that period, simply reflected differences in the relative importance of agriculture in the various regions. In addition, the virtual cutting off of the influx of cheap immigrant labor after World War I removed a constraint on rising wage levels in the industrial areas that had previously been employing the bulk of that labor.

The inevitability of convergence can be questioned on more general grounds. To be sure, migration flows predominantly in the direction of higher income levels, and at least in the United States it seems to be, on balance, an equalizing factor. It is not always true, however, that in-migration from a region tends to lower income levels, and it is even less sure that out-migration tends to raise them.

Interregional capital flows may also be destabilizing. Actually, shifts of employing activities are an equalizing factor only to the extent that the activities are primarily oriented to labor supply, so that capital is drawn to low-wage regions. Consequently, changes in production and transfer technology, availability and use of resources, economies of agglomeration, and other location factors can either narrow or widen income differentials according to circumstances. For example, if agglomeration economies assert a powerful influence on capital movements, regions realizing these economies would grow most rapidly, thus creating more agglomeration economies and promoting continued growth (see the related discussion concerning vertical linkages).

The potential for movements of capital and labor to be stabilizing or destabilizing as circumstances dictate is brought out clearly in research related to a model developed by Moheb A. Ghali and his colleagues.²⁸  They describe economic growth for U.S. regions as being explained by the growth of capital and labor inputs. In this model, interregional movements of factors of production are determined by regional earnings differentials and differentials in the rate of growth of output, which they use as a proxy for employment opportunities. These researchers are able to determine empirically that factor movements tend to be equalizing for U.S. regions. Applying the same model for an analysis of regional growth in Indonesia, Soeroso finds that interregional factor movements encourage income divergence.²⁹ In both cases the response to larger earnings differentials is positive. Although this response is much weaker in Indonesia than in the United States (perhaps reflecting different stages in the development of a well-integrated market economy), it has a stabilizing influence in each. In Indonesia, however, differentials in the rate of growth of output are a much more powerful influence on factor movements and contribute to widening earnings differentials; the more prosperous areas grow fastest and attract productive resources, in a cumulative and destabilizing process.

Changes in the make-up of demand for goods and services may also affect income differentials in either direction. For example, the practically universal tendency of demand for agricultural products to grow more slowly than demand for manufactured products and services in a progressive economy seems more likely than not to widen the differential between incomes in farming areas and those in industrialized and urban areas.

Whether it is the lure of employment opportunities and job-search behavior,³⁰  agglomeration economies,³¹  or some other cause at the root of this process, this much is apparent: We cannot take the experience of the United States in the last 50 years as being representative of a general tendency toward the convergence of regional incomes.

We might summarize by quoting Richard Easterlin:

It is by no means certain that convergence of regional income levels is an inevitable outcome of the process of development. For, while migration and trade do appear to exert significant pressure towards convergence, they operate within such a rapidly changing environment that dynamic factors may possibly offset their influence. One may argue, of course, that migration and trade may become progressively more important during growth, as a result, for example, of improvements in transportation, and hence that the pressures towards convergence will tend increasingly to predominate. But whether this is generally the case cannot be settled on a priori grounds.³²

Consideration of all the factors influencing regional income inequalities leads to an interesting hypothesis relating convergence and divergence systematically to the stages of the development process. Specifically, the early stages of national economic development are associated with increasing regional income disparities, while regional income levels tend to converge in a more maturely developed national economy.

In the present age, the crux of what we call development and attainment of self-sustaining progress is the transition from an agrarian economic basis to a basis of secondary and tertiary activities, with accompanying urbanization. A wide gap exists between the new and the old in terms of income levels, ways of life, and location factors.

When industrialization is in its early stages, most of the rise in overall productivity and per capita income comes from the change of mix—that is, the increasing importance of the nonagricultural sector relative to the agricultural. The new activities cannot take root everywhere at once but are highly concentrated at first in a few key cities—generally the places with the most active contact with more advanced countries and the largest and most diverse populations. At this stage of development, most regions still lack the necessary local market potential and the necessary local inputs to engage in the new and unfamiliar types of activity. Migration is likely to be heavy from the backward areas to the industrializing cities. This migration is highly selective (see the discussion of migration selectivity in Chapter 10), and on the whole this selectivity is prejudicial to the areas of out-migration. The result is the next stage: progressive agglomeration of modern industry in the principal urban areas and an accentuation of regional differences in economic structure, productivity, and income. Such conditions appear to have prevailed in the United States during the divergence period 1840-1880, which was lengthened by the destructive effects of the Civil War upon the economy of the South; and they prevail today in many Third World countries of Asia, Africa, and South America.

As development proceeds, more and more regions acquire the market potential, attitudes, and access to capital and know-how required to surmount the threshold of industrialization. A stage of interregional convergence in economic structure, productivity, and income sets in. This convergence may be made cumulative because migration is likely to become less selective, and national government policies will be less preoccupied with the objective of getting industrialization started in the country as a whole and more sensitive to the political pressures arising from regional inequality.³³

11.7 THE ROLE OF CITIES IN REGIONAL DEVELOPMENT

In Chapter 8 we were able to account for the rise of cities on the basis of the internal and external economies of agglomeration, discussed in Chapter 5, and the transfer advantages of location at major multimodal transfer nodes, discussed in Chapter 3. But this does not tell us why cities are so important in a dynamic way. The existence of sizable urban centers seems to be a necessary (though not always sufficient) condition for the transition from a basically agrarian economy to an advanced economy with high productivity and a wide range of productive activities. The question now is not why cities exist, but why they lead the way in regional and national development.

First of all, there is the relatively cosmopolitan aspect of large cities. They are a region’s eyes and ears perceiving the outside world. "Foreign" ideas, goods, and procedures have much to contribute to the development of even the most advanced region; and cities, as entrepôts for interregional transfer, are the main points where these vitalizing inputs gain admittance. This aspect of cities is most abundantly evident in the less developed countries, where the principal cities impressively resemble their counterparts in advanced countries, even though a few miles away we can step back centuries in time.

Quite apart from their interregional contact function, cities serve an important role in the development process simply by being places in which people from other parts of the same region or country are brought together in densities and living conditions sharply contrasting with those of the rural areas. Conservative traditions and outlooks that persist in the hinterland tend to dissolve rather quickly in the urban melting pot; the results are always conducive to more rapid social and economic change, though they are often painful and destructive in terms of personal satisfaction and orderly social and political adjustment. Social effects of urbanization that can be of major importance in overpopulated countries are the mutually reinforcing tendencies toward smaller families and toward greater labor force participation of women. For the working population as a whole, urbanization represents exposure to a way of life in which work is more scheduled and organized, monetized transactions and impersonal relationships play a larger part, literacy and adaptability to change are more valuable personal assets, and the choice of occupations and lines of individual development is widened.

Such considerations as these go far to explain why cities (especially large ones with far-reaching external contacts) have been the main seedbeds of innovation; in economic terms, this involves the genesis of new techniques, new products, and new firms. Such places provide the exposure to a wide range of ideas and problems from which solutions emerge. They represent large concentrations of customers and suppliers most receptive to new products and requirements. They provide the diversified supply of skills and supporting services that enable a producer to start small and concentrate on a narrowly specialized function (see the discussion of external economies in Chapter 5). They provide a social and business climate in which impediments of tradition and personal inertia are minimized, and initiative and innovation carry prestige; and in which the innovator can learn much from day-to-day contacts with competitors and can most easily tap the stock of accumulated know-how, exploiting inventions arising not only in his or her own city but elsewhere.³⁴

Major cities are the locations at which the newest types of activities can most easily get a foothold within any region or country, and the advent of industrialization in an undeveloped country is generally accompanied by explosive growth of the largest centers and a heightening of economic and social contrasts between those centers and the rural backwaters.³⁵ But as development proceeds, two things happen. Some of the sensitive infant industries of yesterday attain maturity: Their techniques become less experimental and their products more familiar to a wider market. As a result, these activities are no longer so dependent on the special advantages that large cities provide. The fledglings are ready to leave the nest. At the same time, the positive incentives to decentralize out of the initial large city concentration tend to increase. With a larger and wider market for the product, a location pattern involving a number of regional production centers offers economies in distribution costs without undue sacrifice of the economies of scale. The external economies of cluster become less important with the increase in financial and technical resources of firms and the greater standardization of process and product. Costs of labor and other local inputs in the initial large-city location now appear unnecessarily high in relation to what these inputs cost in smaller places. And in the original centers where the industry developed, maturity may well have meant some development of rigidities and loss of initiative because of the aging of both business leaders and the labor force and the growth of defensive practices to protect seniority rights, painfully acquired but obsolescent skills, positions of power, and other accumulated perquisites. Thus the city that hatched the industry and saw it through its infancy may lose it altogether when it grows up.

The evolutionary process sketched here in terms of locational change accompanying the birth, infancy, and maturing of an activity is a strikingly familiar one. Wilbur Thompson has referred to it in terms of "a filter-down theory of industrial location" and "the urban-regional growth loop."³⁶  As he points out, "New York has lost nearly every industry it has ever had—flour mills, foundries, meat-packing plants, textile mills and tanneries." Pittsburgh pioneered and then lost preeminence in a long series of major industries including oil refining, aluminum, electrical machinery, and steel.

Historically, large cities have been characterized by a disproportionate component of new and small "growth industries" in their mix of activities, but they generally fail to maintain their share of activities that are past the early stages. Smaller cities and towns, and less-advanced regions, have been more likely to show competitive gains in the sense of increasing their share of the national total of employment in the activities represented there; but growth in these areas has historically been held down by the fact that their mix of activities is often weighted with slow-growth and low-wage activities.³⁷

However, there are some signs of change in this historical pattern. In Chapter 8, we found that there had been substantial growth in smaller cities and towns relative to the growth in metropolitan areas, at least since 1970. Previously, the nation’s population growth had been characterized by rapid growth of metropolitan-area populations (see Table 8-3).

This turnaround has an additional spatial dimension. Table 11-6 shows that during the 1970s, population growth in the United States was dominated by growth in the South and West. The figures indicate that the decline of areas in the nation’s industrial "core" (represented in Table 11-6 by the Northeast and North Central regions) and the gains of areas once characterized as the nation’s "periphery" are closely linked to changes in the pattern of urban growth. Smaller cities, towns, and unincorporated places have had strong growth in the South and West. Additionally, in the Northeast and North Central regions, unincorporated places and smaller incorporated places have shown stronger growth (or less decline) than metropolitan areas.

R. D. Norton and J. Bees have undertaken an extensive analysis of changes in the spatial patterns of growth in the manufacturing sector.³⁸  Their research indicates that the core region’s long-standing relative decline gave place to an absolute decline in manufacturing jobs after 1969.³⁹  They attribute this trend to two factors: (1) an acceleration in the rate at which production processes that have become rather standardized have moved from the core to the periphery and (2) decentralization of the nation’s innovative capacity, so that some new and rapidly growing industries have become less highly concentrated in the core.⁴⁰

Thus the evolutionary process we have described, whereby growth is transmitted down the urban hierarchy, has itself evolved. The relative rates of growth of major cities versus smaller places depends to an important extent on how rapidly the filter-down or dispersion of maturing urban activities proceeds compared to the initiation of new urban activities. In Thompson’s words, faster development of the smaller, less developed urban area "would seem to require that it receive each successive industry a little earlier in its life cycle, to acquire the industry at a point in time when it still has both substantial job-forming potential and high-skill work."⁴¹ Additionally, it now seems that the filter-down process is less important overall, as innovation and high concentrations of fast-growing activities are less exclusively characteristic of growth in the nation’s largest metropolitan areas.

Further implications will be explored in Chapter 12, in connection with growth-promotion policies and the focusing of development promotion efforts on urban growth centers.

11.8 EXTERNAL AND INTERNAL FACTORS IN REGIONAL DEVELOPMENT

We have seen that the development of a region—in terms of its size, income level, and structure—is affected by external conditions of two types: (1) demand for the region’s outputs, or more broadly, external sources of income for the region, and (2) supply of inputs to the region’s productive activity. We have also seen that the impact of these external factors is conditioned by the size and maturity of the region and by the internal relationships of its various activities in the form of vertical, horizontal, and complementary linkages.

Since all regions contain a variety of activities, it is to be expected that some of these activities will be determined mainly by external conditions based on demand (such as export markets), while others will be particularly sensitive to supply conditions. The regional economy as a whole, then, is always subject to a variety of growth determinants. Although there may be one principal factor affecting its overall level of activity (as, for example, the nationwide demand for automobiles is the principal factor affecting the prosperity and growth of Flint, Michigan), there is never just a single determinant.

How much influence on a region’s development can be exercised from within the region itself? This question is basic to the discussion of regional objectives and policies in the next chapter. As far as growth determinants in the form of final demand are concerned, the latitude for regional initiative is ordinarily limited. But perhaps export demand in some lines can be stimulated by sales promotion campaigns, or the region can better its access to external markets by lobbying or other pressure to get more favorable freight rates or transport services for its exports, improved waterways, or high-speed highways. Improvement of the region’s own terminal and port facilities may also have some effect on export demand and thus on regional growth. Houston, with its ship channel to the sea, is a dramatic illustration of a successful effort of this type.

A region has some leverage also on primary supply inputs. By persuasion, pressure, and subsidy, it may secure better and cheaper inbound transport for its imported materials and may be able to attract activities with strong forward linkages that will have a supply multiplier effect. Governmental and private research centers and universities are increasingly valued as local suppliers of services, people, and ideas providing the basis for new growth industries. Regions where demand for labor tends to exceed supply can stimulate immigration by campaigns of advertising and recruitment, publicizing both job opportunities and whatever amenities the region has to offer.

Finally, regional growth can be significantly affected through changes in intraregional input supplies and interactivity relationships, which are more immediately subject to local choice and action. The quality of the labor supply can be enhanced by a variety of education and training programs and by removal of barriers to occupational mobility and technical change (including racial and sex discrimination, restrictive work rules, and job entry requirements). The region’s limited land and other natural resources can be managed so as to increase their contribution to productivity. Local public services, an important input to almost all activities, can be made more efficient and conducive to productivity and amenity. The region’s economies of agglomeration can be enhanced by appropriate action involving both public and private sectors (for example, in the planned development of new and improved office centers, regional shopping centers, produce markets, health centers, research parks, and the like).

In the next chapter, we turn to a consideration of how these and other ways of influencing regional development are used in the pursuit of objectives involving regional structure and growth.

11.9 SUMMARY

This chapter addresses itself to such basic questions about regional growth and change as the causes of growth, the roles of trade and of the movement of labor and capital, the relation of regional economic structure to growth, and the convergence of regional differentials in incomes and structure.

Processes of regional economic change work through the various types of linkage examined in Chapter 9. In general, vertical linkages are involved in self-reinforcing growth or decline tendencies, whereas horizontal linkages have a stabilizing influence. Various theories about the generation of regional growth have emphasized either demand for the region’s outputs and backward linkage, or supply of inputs and forward linkage.

The simple economic base approach identifies exports as the generator of growth in a region; nonbasic, or local market-serving, activities are assumed to grow only in response to the local demand generated by the export sector and to maintain a more or less fixed ratio to the latter.

With input-output analysis, it is possible to trace the impact of an increase in business receipts from exports or other components of "final demand" on payments and incomes in the region through local spending for payrolls and purchases from other businesses in the region. The total increase in regional income generated per dollar of initial increase in final demand receipts is the regional income multiplier.

The input-output model treats final demand as the initiator of growth and change. Exports are always part of final demand, but the household, government, and investment sectors are sometimes taken at least partially out of the final demand category and treated as responding to regional demands.

A regional economic model in which all growth and change must come from demand and be transmitted through backward linkage is one-sided. A more adequate model would assign major roles to supply factors and forward linkage as well, but input-output analysis is less well adapted to deal with changes originating on the supply side.

A still broader multiregional view of the development process focuses on the roles of interregional trade and factor movements. The migration of capital is subject to determinants closely analogous to those affecting labor migration, though the patterns of interest and wage differentials are quite dissimilar, as are the patterns of capital and labor flow.

Interregional trade can serve as a partial substitute for labor and capital flows in equalizing returns to those factors. Flows of labor and capital can either substitute for or complement one another; the substitutive relation exerts an influence toward stability in relative regional growth, while the complementary relationship can be the basis of self-reinforcing and cumulative tendencies.

The observed convergence in regional income levels and structures in recent decades is not a universal trend. Interregional trade as well as labor and capital movements, though commonly promoting convergence, can in some situations have the opposite effect; and technological dynamics can just as well promote divergence as convergence. According to one plausible hypothesis of development stages, divergence is likely to characterize the youthful stages of a country’s industrial development, and convergence the more mature stages.

Large cities have played a crucial role in regional and national economic development, in their capacity as transmitters of ideas and practices from the outside world and also as places where people from diverse parts of the home region or country are brought into close contact and exposed to new institutions and challenges and a wider variety of opportunities. Innovation has flourished in such a germinating ground. New industries and other activities that started in large cities have historically tended to decentralize at a later stage to play a role in the development of other regions or parts of the region. Evidence suggests that decentralization is occurring more rapidly in recent years and that the capacity of smaller places to support innovative industries has increased.

SELECTED READINGS

George H. Borts and J. L. Stein, Economic Growth in a Free Market (New York: Columbia University Press, 1964).

John Friedmann and William Alonso, Regional Policy. Readings in Theory and Applications (Cambridge, Mass.: MIT Press, 1975).

R. D. Norton, City Lift-Cycles and American Urban Policy (New York: Academic Press, 1979).

Harvey S. Perloff et al., Regions, Resources, and Economic Growth (Baltimore: Johns Hopkins University Press, 1960).

Allen R. Pred, The Spatial Dynamics of U.S. Urban-Industrial Growth, 1800-1914 (Cambridge, Mass.: MIT Press, 1966).

Harry W. Richardson, In put-Output and Regional Economics (New York: Wiley, 1972).

Harry W. Richardson, Regional Economics (Urbana: University of Illinois Press, 1978), Chapters 4-7.

Horst Siebert, Regional Economic Growth: Theory and Policy (Scranton, Pa.: International Textbook Co., 1969).

APPENDIX 11-1

Further Explanation of Basic Steps in Input-Output Analysis

(See Section 11.3.2)

The input coefficients (Table 11-4) are derived from the information on interactivity purchases and sales in Table 11-3 as follows: The total output of activity A is 4300 (dollars per month, or other appropriate money/time unit). A purchased 50 units of output from B. Therefore, for each unit of A’s output, 50/4300 or .012 units of B output is called for. In similar fashion, we find that each unit of A’s output involves the following further purchases of A:

• From A itself (that is, sales from one A firm to another): 300/4300=.070.

• From C: 1000/4300=.233

• From D: none

• From households: 1900/4300=.442

• From government: 200/4300=.047

• From outside: 200/4300=.047

• From capital: 650/4300=.151

Similarly, every unit of output by B involves purchases by B from A amounting to 400/2850=.140 units, and so on. In this fashion, we can derive the rest of the coefficients in Table 11-4.

The figures in Table 11-5 (total direct and indirect effects) are derived as follows from the input coefficients in Table 11-4. Let us denote the outputs of activities A, B, C, and D simply by those letters. Then we can write the entire distribution of A’s output as follows:

A =.070A + .140B + .032C + .192D + F_A

where F_Arepresents A’s sales to final demand sectors. The foregoing equation can be restated more simply as:

.930A — .140B — .032C — .192D — F_A =0                               (1)

and, applying the same procedure to the other three intermediate activities, we get

.930B — .012A — .323C — .115D — F_B =0                                (2)

.968C — .233A — .070B — .269D — F_C =0                                 (3)

.808D — .281B — .065C — E_D =0                                               (4)

We now have four simultaneous equations, (1)—(4), which can be solved for the output levels A, B, C, and D in terms of the final demand sales levels F_A, F_B, F_C, and F_D. This solution ("matrix inversion") of the simultaneous equations is laborious (for even as few as four equations) if done by hand, but it can be done quickly and cheaply on a computer. The solution is as follows:

A =1.118F_A + .289F_B + .157F_C + .359F_D                                      (5)

B =.126F_A + 1.234F_B + .439F_C + .352F_D                                      ((6)

C =.297F_A+ .284F_B+ 1.171F_C ± .501F_D                                       (7)

D =.O68F_A + .452F_B+ .247F_C + 1.400F_D                                      (8)

These coefficients are entered in the upper part of Table 11-5.

The figures in the lower part of Table 11-5 are obtained as follows, taking the first figure (.6 14) as an illustration. From the table of input coefficients (Table 11-4), we see that households sell .442 units to A for every unit of A’s output. From equation (5) (or from the first figure in Table 11-5), we see that A must produce 1.118 units for each unit that A sells to final demand. Consequently, each unit that A sells to final demand will require purchases by A from households amounting to .442 X 1.118 units. But as we see from (6), (7), and (8) (or from Table 11-5), each unit of A’s sales to final demand calls for the following further purchases by A:

From B: .126

From C: .297

From D: .068

For each unit that B produces, B buys .105 units (Table 11-4) from households; there is thus an additional indirect demand via B for .126 X .105 units. Similarly for C and D. The total additional sales by households resulting from a one-unit increase of final demand sales by A is therefore

1.1118 X .442) + (.126 X .105) + (.297 X .323) + (.068 X .154)=.614 units,

which is entered as the first figure in the lower part of Table 11-5.

APPENDIX 11-2

Example of an Input-Output Table with Households Included as an Endogenous Activity

As indicated in the text (households may be included as another activity in the intermediate sector if we care to assume that household expenditures are linearly related to household receipts. This first requires filling in some additional cells in Table 11-3, and we shall use the following figures:

Table 11-3 will now appear as follows, using H to denote households, and with all new figures in italics:

From the figures in the foregoing table, the input coefficients can be calculated in the same fashion as was done for Table 11-4. The revised version of Table 11-4 (with headings abbreviated and with all new figures in italics) will look like this:

Finally, the total direct and indirect effects of an increase in final demand are derived in the same way as for Table 11-5. The revised table follows. In it all the figures are new. All the ratios are much larger than in the original version of Table 11-5, since the new calculation includes a large additional multiplier effect involving feedback through the household sector (additional employment induces additional household expenditure for local products, imports, capital, and taxes). No such effect was allowed for in the original version of Table 11-5, in which households were a final demand sector.

* Figures in this column show the impact of an added dollar of aggregate final demand sales by all intermediate activities, apportioned in the same proportions as those activities shared in the final demand sales shown in the second table above. Specifically, this means added final demand sales of 26¢ by A, 22¢ by B, 19¢ by C, 7¢ by D, and 26¢ by H, totaling $1.00.

It may be noticed that in each column of this last table, the sum of the figures in the primary-sector rows (government, outside, and capital) comes to 1 (ignoring some trivial rounding-off discrepancies). The same holds true in Table 11-5. The reader will find it a useful exercise to explain this fact.

ENDNOTES

1. The areas involved are mapped in Figure 11-2. We shall sometimes in this chapter refer to the individual Census divisions as "regions," despite the fact that the Census Bureau (as shown in Figure 11-2) groups them into still larger areas, which it calls "Census regions."

2. Irving Hoch, "Income and City Size," Urban Studies, 9, 3 (October 1972), 314.

3. Contributions to the discussion include Phillip R. P. Coelho and Moheb A. Ghali, "The End of the North-South Wage Differential," American Economic Review, 61, .9 (December 1971), 932-937; and a critical comment by Mark L. Ladenson and a rebuttal by Coelho and Ghali in American Economic Review, 63, 4 (September 1973), 754-762.

4. Hoch, "Income and City Size," p. 315.

5. Shelby Gerking and William Weirick, "Compensating Differences and Interregional Wage Differentials," Review of Economics and Statistics, 65, 3 (August 1983), 483-487.

6. Exporting in this sense does not necessarily imply that the goods or services are sent out of the region by their producers. They may instead be consumed in the region by outsiders who occasionally come for that purpose. Selling of recreational and other services to tourists from outside is a major "export" activity in some regions. What is relevant for the region’s development is the income, rather than the movement of the output.

7. For a careful and readable description of the purposes and techniques of such studies, see Charles M. Tiebout, The Community Economic Base Study, Supplementary Paper No. 16 (New York: Committee for Economic Development, December 1962).

8. For a comprehensive discussion of the methods used to estimate the exports of a region, see Andrew M. lsserman, "Estimating Export Activity in a Regional Economy: A Theoretical and Empirical Analysis of’ Alternative Methods," International Regional Science Review, 5, 2 (Winter 1980), 155-184.

9. See Tiebout, Community Economic Base Study, Table 10, P. 49, for a series of examples of such understatement, involving eight different industry groups and six different community economic base studies. In each case, a questionnaire survey of business firms provided the more accurate data against which the location quotient estimate was checked.

More generally, there is an aggregation effect involving the offsetting of "surpluses" and "deficits" that restricts the comparability of location quotients. As a rule, the use of a finer activity classification or a smaller region will make quotients larger, whereas a coarser classification or larger region permits more offsetting and reduces the quotients.

10. Ibid, pp. 48-49.

11. For a series of estimated export multipliers for a dozen cities of assorted sizes, see Charles L. Leven, "Regional Income and Product Accounts: Construction and Applications," in Werner Hochwald (ed), Design of Regional Accounts (Baltimore: Johns Hopkins University Press, 1961), Table 1, p. 179.

12. If the region’s net exports are always positive, the loan is, in effect, permanent!

13. See J. Thomas Romans, Capital Exports and Growth Among U.S. Regions (Middletown, Conn.: Wesleyan University Press, 1965), p. 118.

14. The example to be given here is the simplest input-output table applicable to a region. For many years input-output tables constructed for the country as a whole were of identical form. With the completion of the 1972 U.S. input-Output tables, a more general accounting framework was adopted, which includes the schema presented here as a special case. See Bureau of Economic Analysis, U.S. Department of Commerce, Survey of Current Business, 59, 2 (February 1979), 34-72.

15. In practice, the entries involving this sector comprise sales to and purchases from firms (both inside and outside the region) on capital account. The "outside world" entries refer to export or import transactions with nongovernmental parties on current account.

16. For an explanation of the calculations, see Appendix 11-1 or any general reference on input-output analysis, such as William H. Miernyk, The Elements of Input-Output Analysis (New York: Random House, 1965). For any table of substantial size, such calculation is best done on a computer.

17. So far, we have a whole set of specific regional multipliers, since the assumed initial impact that gets multiplied can be taken as an increase in final demand sales of any of the several intermediate sector activities. Sometimes it is desirable to settle on a single overall regional multiplier figure. Such a figure can be derived by starting from an "across-the-board" unit increase in final demand; that is, each intermediate activity’s final demand sales rise by the same proportion. The last column in Table 11-5 illustrates this calculation, giving an overall multiplier of 1.977.

These and other types of regional multipliers have been estimated at different times for many regions. Although, as we might expect, there is considerable variation, there is a rather consistent tendency for multipliers to be greater for larger and more fully diversified regions. This is logical: Such a region "takes in more of its own washing," and the sequence of indirect and induced effects is subject to less demand leakage than would be the case in a smaller or more narrowly specialized region. Presumably, the minimum multiplier (1) would be most closely approached in a community, such as a mining camp, devoted to a single exporting activity.

In a comparative study of export employment multipliers in American cities, it was found that the following characteristics were associated with higher multipliers: city size, growth rate, female labor force participation, income per capita, ratio of nonlabor to labor income, and diversity of activities. Andrew S. Harvey, "Spatial Variation of Export Employment Multiplier: A Cross-Section Analysis," Land Economics, 49, 4 (November 1973), 469-474.

18. There are alternative ways of allowing for the multiplier effect via household income and expenditure. In any case, it is customary to refer to this effect as "induced" to distinguish it from the indirect effects resulting from interindustry transactions in the narrower sense.

19. There is a large literature on systems of regional accounts and models for analysis and policy guidance; for a well-rounded treatment, see Harry W. Richardson, Regional Economics (Urbana: University of Illinois Press, 1978).

20. See, for example, Karen R. Polenske, The US. Multiregional Input-Output Accounts and Model (Lexington, Mass.: Lexington Books, D. C. Heath, 1980); and Jan Oosterhaven, interregional input-Output Analysis and Dutch Regional Policy Problems (Hampshire, England: Gower, 1981).

21. Probably one reason for this overemphasis on the role of demand in regional growth is that modern regional growth theory, input-output analysis, and multiplier analysis were influenced by the contemporary development of Keynesian theories of what determines the degree of utilization of given resources in the short run. A more balanced approach, taking into account long-term growth factors on both the supply and the demand sides, has appeared in some theoretical work; notably in Horst Siebert, Regional Economic Growth: Theory and Policy (Scranton, Pa.: International Textbook Co., 1969); Romans, Capital Exports; and G. W. Borts and J. L. Stein, Economic Growth in a Free Market (New York: Columbia University Press, 1964).

One of the first regional economists to question the primacy of exports and call for a balanced theory was Charles M. Tiebout, "Exports and Regional Economic Growth," Journal of Political Economy, 64, 1 (February 1956), 160-164. His article was prompted by a forceful statement of the export doctrine by Douglass C. North, "Location Theory and Regional Economic Growth," Journal of Political Economy, 63, 3 (June 1955), 243-258. The whole North-Tiebout controversy, including both these articles, North’s subsequent reply, and Tiebout’s final rejoinder, is reprinted in John Friedmann and William Alonso (eds.), Regional Policy: Readings in Theory and Application (Cambridge, Mass.: MIT Press, 1975), pp. 332-357. Another good statement emphasizing supply factors is Richard T. Pratt, "Regional Production Inputs and Regional Income Generation," Journal of Regional Science, 7, 2 (Winter 1967), 141-149.

22. See, for example, Richard F. Muth, "Migration: Chicken or Egg?" Southern Economic Journal, 37, 3 (January 1971), 295-306.

23. Readers interested in a more detailed presentation and critique of supply-driven input-output analysis are referred to Frank Giarratani, "The Scientific Basis for Explanation in Regional Analysis," Papers and Proceedings of the Regional Science Association, 45 (1980), 185-196; and Oosterhaven, Interregional Input-Output Analysis, Chapter 8.

24. Ricardo’s theory and much subsequent theorizing on international trade assumed immobility of production factors—an assumption clearly inappropriate in reference to relations among regions within a single country.

25. Such a differential is suggested, though without empirical evidence, in Richardson’s discussion of interregional capital mobility; see Harry W. Richardson, Regional Economics (New York: Praeger, 1969), p. 305. Lösch found a marked regional differential pattern in money rates in the United States in the 1920s and 1930s, long after the Federal Reserve System had been put into operation. See August Lösch, The Economics of Location (New Haven, Conn.: Yale University Press), pp. 461-476. Additional material on changes in the geographic structure of the U.S. financial system may be found in Beverly Duncan and Stanley Lieberson, Metropolis and Region in Transition (Beverly Hills, Calif.: Sage Publications, 1970), Chapters 11-12.

26. For an excellent discussion of the effects of transport changes on industry location and regional concentration in the United States, see Benjamin Chinitz, "The Effect of Transportation Forms on Regional Economic Growth," Traffic Quarterly, 14,2 (April 1960), 129-142; and Chinitz, Freight and the Metropolis (Cambridge, Mass.: Harvard University Press, 1960).

27. The farm price parity ratio (index of prices received by farmers to prices paid by farmers, on the base 1910=100) averaged 104 in 1911-1920; 88 in 1921-1930; 76 in 1936-1940; and 107 in 1941-1950. U.S. Bureau of the Census, Historical Statistics of the United States, Colonial Times to 1957 (Washington, D.C.: Government Printing Office, 1960), p. 283.

28. Moheb A. Ghali, Masayuki Akiyama, and Junichi Fujiwara, "Factor Mobility and Regional Growth," Review of Economics and Statistics, 90, 1 (February 1978), 78-84.

29. Soeroso, The Distribution of Economic Activity over Space and Economic Growth in Indonesia, (Ph.D. dissertation, University of Pittsburgh, 1982).

30. See Michael P. Todaro, "A Model of Labor Migration and Urban Development in Less Developed Countries," American Economic Review, 59, 1 (March 1969), 138-148.

31. See Harry W. Richardson, Regional Growth Theory (London: Macmillan, 1973), Chapter 7; and Richardson, Regional Economics, (Urbana: University of Illinois Press, 1978), Chapters 4-7.

32. Richard A. Easterlin, "Long Term Regional Income Changes: Some Suggested Factors," Papers and Proceedings of the Regional Science Association, 4 (1958), 325. See also Easterlin, "Interregional Differences in Per Capita Income, Population and Total Income, United States, 1840-1950," Trends in the American Economy in the Nineteenth Century, vol. 24, Conference on Research in Income and Wealth, Studies in Income and Wealth (New York: National Bureau of Economic Research, 1960); and Easterlin, "Regional Income Trends, 1840-1950," in Seymour Harris (ed.), American Economic history (New York: McGraw-Hill, 1961), pp. 525-547.

33. See Jeffrey G. Williamson, "Regional Inequality and the Process of National Development: A Description of the Patterns," Economic Development and Cultural Change, 13, 4(2) (July 1965), 3-45; reprinted in L. Needleman (ed.), Regional Analysis (Baltimore: Penguin, 1968). Williamson presented and substantiated the hypothesis described here, using nineteenth-and twentieth-century data for a number of countries both cross-sectionally and in terms of changes over time. He also investigated the degree of income inequality among counties within states in the United States and established that its changes have been closely correlated with changes in interstate income inequality.

34. For a penetrating and well-documented analysis of the interaction between industrial innovation and urban concentration in the United States, see Allen R. Pred, The Spatial Dynamics of US. Urban-Industrial Growth, 1800-1914 (Cambridge, Mass.: MIT Press, 1966), Chapter 3.

Pred's analysis covers the period up to 1914, and as he suggests, the nature and strength of some of the cumulative forces of urban-industrial agglomeration have subsequently changed. His study is noteworthy in its stress on the interaction between concentration and innovation: That is, technological advance and innovation flourish in the large urban-industrial center, and give rise to new industries that are established in those same centers, and stimulate their further growth. There are many examples of this process in the nineteenth century: the inception of the electrical equipment manufacturing industry in New York, Boston, and Pittsburgh, the making of scientific instruments and optical equipment in Rochester, N.Y., and so on. But (as Pred points out) under more recent conditions, innovation at one location can as easily give rise to new industry in other locations. This is true because technical knowledge is much more diffused and transferable now, and also because large multiplant and multi-industry corporations now play a commanding role in the research and development that give rise to new processes and industries. Such a corporation is perfectly free to choose locations for the new industry quite remote from the headquarters or research-center city.

This and other changes help to explain why specific manufacturing industries are no longer as strongly or persistently concentrated in their "parent cities" as they used to be; and perhaps also, why the fastest-growing metropolitan areas today are not the very largest but those in the intermediate and smaller size classes. (See Section 8.5.)

35. As a sobering touch of historical realism, we must note here that leading cities in which wealth, power, and foreign influence are concentrated have not always been an unmixed blessing to their regions and countries. In some situations of old-style colonialism in particular, the dominant externally oriented metropolis has been parasitic on its hinterland. Preexisting native industries, economic and social institutions, and cultures have been damaged to an extent that, for a considerable period at least, is not compensated by the growth-generative effects. See Bert Hoselitz, ‘Generative and Parasitic Cities," Economic Development and Cultural Change, 3 (1955), 278-294.

36. Wilbur R. Thompson, "The Economic Base of Urban Problems," in Neil W. Chamberlain (ed.), Contemporary Economic Issues (Homewood, Ill.: Irwin, 1969), pp. 6-9.

37. See Appendix 12-1 for some discussion of measures of regional economic growth in terms of components reflecting activity-mix and competitive gain or loss.

38. See R. D. Norton and J. Rees, "The Product Cycle and the Spatial Decentralization of American Manufacturing," Regional Studies, 13, 2 (1979), 141-151; and R. D. Norton, City Life-Cycles and American Urban Policy (New York: Academic Press, 1979).

39. Norton and Rees. "The Product Cycle," p. 142.

40. Ibid., p. 147.

41. Thompson, "Economic Base of Urban Problems," p. 9.