GaWC Research Bulletin 176

This Research Bulletin has been published in Urban Studies, 43 (9), (2006), 1625-1630.

doi:10.1080/00420980600831874

Please refer to the published version when quoting the paper.

(Z)

On a Non-Appraisal of the 'Jacobs Hypothesis'

P.J. Taylor

Summary

This is a critique of Polèse recent paper entitled ‘Cities and National Economic Growth: A Reappraisal'. It purports to test the ‘Jacobs hypothesis'. It is argued that both main variables used, ‘national levels of urbanization' and ‘per capita GDP growth', are mis-specified. The critique is related to the state-centric nature of social science and policy.

Although he would ‘[d]eep down …prefer that Jane Jacobs were right', Mario Polèse (2005, 1446) in his ‘Cities and National Economic Growth: A Reappraisal' marshals an impressive amount of evidence and argument – statistical, theoretical and historical – to show that she is probably wrong: it is, after all, nations and not cities that generate economic growth. Providing a balanced review of the arguments in the ‘new economic geography' and ‘new growth theory' literatures for and against the ‘Jacobs hypothesis', Polèse suggests that the growth of cities is the result of them capturing disproportionate amounts of national economic growth, rather than their creation of that growth. However, there is one figure largely missing from this balanced appraisal: Jacobs herself. For students of Jacobs, it is a bizarre notion that her work should even be part of a debate on national economic growth.

THE EXPANSION OF ECONOMIC LIFE

The starting point for Jacobs (1970) is that:

‘A city does not grow by trading only with a rural hinterland. A city seems always to have implied a group of cities, in trade with one another.' (p. 35).

In this situation the most obvious source of city economic growth is to export more to the other cities. This is the expansion of ‘old work' that can increase only insofar as other cities can absorb its products. But for Jacobs (1970, chapter 5; 1984, chapter 2) the source of most economic expansion is through the creation of new work, consequent upon replacement of imports from other cities. For the city, import replacement is like additional imports, previous supplies of what were imported commodities are made available, leaving the city's exports now able to be used to buy a new mix of imports from the other cities. This produces a ‘spurt‘ of growth in the city itself and adds to the overall economic content of the group of cities. For Jacobs (1970, 50), a city is defined by the experience of such economic expansion. Import replacement results from improvisation, adapting local economies to new work, and the growth process is only feasible within a group of economically interacting cities. Jacobs refers to the city sub-process as ‘the little movements at the hubs' and the group sub-process as ‘the great wheels of economic life' (p. 176).

At any one point in time, some cities in the group will be dynamic and experiencing large economic expansion while others will be going through lean times that may result in stagnant or declining city economies. However, as long as some dynamic cities exists the group will continue to prosper as they can provide the source for the stagnant cities to begin anew the process of import replacement. But, with no ‘little movements' operating in any city, the ‘great economic wheels' come to a halt. Thus it is easy to see why this model has been of interest for economic policy making. However, a key point to note is that a ‘group of cities' in this argument is not a national urban hierarchy. For the import replacement process to operate, it does not matter whether the other cities are within or without the political boundaries of the state in which the city is located. Of course, in policy circles the latter is important but more on this below. In her first description of the model, Jacobs (1970) does relate her model to national economic growth rates by treating the latter as the net rate of growth over all cities in a country, some dynamic, some stagnant, some declining. She suggests a corn popper analogy: ‘not all the kernels are popping simultaneously; but all the time the corn is popping' (p. 167). Thus in this argument the national economic growth rate depends on the proportion of a country's cities that are ‘popping' at any given time.

In her second description of the model, Jacobs (1984) is much more scathing in her treatment of states in economic discourse:

‘Nations are political and military entities … But it doesn't follow from this that they are also basic, salient entities of economic life. … Once we … try looking at the real economic world in its own right rather than as a dependent artefact of politics, we can't avoid seeing that most nations are composed of collections of grab bags of very different economies, rich regions and poor ones within the same nation.' (p. 31-2).

RATHER TOO MUCH CONFLATION

But what exactly is the ‘Jacobs hypothesis' (Polèse 2005, 1429) that Polèse is trying to test? There is a conflation at the heart of Polèse's analyses between cities, economies, urbanisation and agglomeration. He begins by identifying ‘the link between cities and national economic growth' (p. 1429, emphasis in the original) as the focus of the argument, discusses ‘Jacobs economies' as first, ‘ dynamic agglomeration economies (emphasis in the original') and then as ‘dynamic urbanisation economies' (p. 1434) which leads finally on to [‘r]igorously testing the relationship between cities (agglomeration) and national economic growth' (p. 1435). The former is operationalised as ‘urbanisation levels': ‘if one wishes to compare nations, one has little choice' (p. 1435). But urbanisation includes declining cities – the UK with its numerous declining cities beyond London are a favourite example of Jacobs and in this analysis they are included in the UK's very high urbanization level as part of the test of ‘Jacobs hypothesis' that more national urbanization equals more national economic growth. It is no wonder the hypothesis falls.

Polèse's key figure plots the two variables for 95 countries to see how the urbanization level in 1960 relates to real growth in GNP per capita for 1960-2002 (Polèse 2005, Figure 3, p. 1436). A broad scatter of points results with no discernible relationship or pattern. But Polèse does identify 14 ‘outliers', countries with high levels of growth that range across the whole gamut of urbanization levels. Polèse reasonably suggests that ‘it is difficult to find any common denominator: Botswana, South Korea, Malta, Singapore, China, Oman …' (p. 1436). This leads directly on to the second section of the paper where he provides a lengthy argument that it is ‘the national institutional environment which largely determines the rate of economic growth' (p. 1437-8). This ‘institutional environment' concept is ‘a catch all' including history, culture and other effects on policy making (p. 1438); subsequent discussion includes urban infrastructure, sound financial system, criminal violence, and corruption (pp.1438-9). The message is straightforward: ‘without national institutions that work, the productive potential of urban places will remain limited' (p. 1438). Since Polèse's Figure 3 features countries rather than cities, it seems to me that it is rather better at ‘testing' as Polèse's ‘national hypothesis' rather than Jacobs city hypothesis. And yet this part of the discussion does not refer back to the figure. In Table 1 I have abstracted Polèse's outlier countries: presumably these particular ‘nations' should have ‘national values, ideals and collective perceptions' as the ‘essential elements in orientating social action and individual behaviour towards capitalist” pursuits' (p. 1439). Of course, the countries listed in Table 1 look just as poor as evidence for this national hypothesis than for the original urbanisation one. In fact this is exactly what a true Jacobs hypothesis would expect; ‘national economies' as a ‘grab bag of very different economies':

‘It … affronts common sense, if nothing else, to think of units as disparate as, say, Singapore and the United States, or Equador and the Soviet Union, or the Netherlands and Canada, as economic common denominators. All that really have in common is the political fact of sovereignty.' (Jacobs 1984, 32).

And we could add, Botswana and South Korea or China and Malta or Lesotho and Japan, and so on.

FURTHER CRITICISMS

The previous section provides my core critique of Polèse's conclusion but there are three further criticisms that need to be aired: one other mis-specification, the use of historical examples, and the question of methodology.

Jacobs' concern is for the expansion of economic life and this is not the same as increase in per-capita gross domestic product. For Jacobs (1970, 49), development is the addition of new kinds of work, not the accumulation of more and more old work. Thus she distinguishes between ‘development work' and ‘production work' (p. 90). For instance, the location of new branch plant factories in Lesotho to take advantage of cheap labour is not development, it is somebody else's production work, old work for cities geographically far away. (This is another way in which the economic power of cities is projected to produce remote ‘grotesque economies', in this case not simple agricultural or resource supply regions but simple industrial supply regions instead (Jacobs 1984, chapter 7).) Of course, such redistribution of old work does increase per capita GDP where it happens to land; hence Lesotho's appearance in Table 1. It follows therefore, that not only is the cities/urbanization/agglomeration variable mis-specified as shown above, the other variable in Polèse's (2005) Table 3 is equally problematic for testing ‘the Jacobs hypothesis'.

The basic reason for such mis-specification is that there are no available statistics that measure ‘new work'; it is hard to see how such continual revisions of divisions of labour could be incorporated into employment statistics (Jacobs 1970, 94-5). In such research contexts it is, of course, wholly legitimate to conjure up surrogate measures and what Polèse has done can be interpreted in this way. However, such surrogates must be in the spirit of the original model and I hope I have shown above that this is not the case here. But the issue is more than a problem of measurement. Jacobs is postulating a specific mechanism for expansion of economic life and therefore it is difficult to see how extensive research - Polèse's (2005, 1430) calls it ‘'to concentrate on the big picture' – can be useful here (Sayer 1992). Polèse is explicitly looking for causation (p. 1430, 1445) and this is precisely what an extensive analysis of outcomes (statistics) cannot describe. As much is admitted by Polèse when first he refers to impossibilities and difficulties in putting together the data for Figure 3 (p. 1435), and second he is consummately modest and careful in drawing conclusions (p. 1445), but neither can ever overcome mis-specification of variables and method.

Polèse (2005) does not rely entirely on the statistical data and analyse reviewed thus far; he claims an historical ‘bias' (p. 1430) that is evidenced in his Argentina/Canada comparison discussed previously and which continues into a more general historical discussion (p. 1442). Of particular interest here is the greater emphasis on specific cities. However, this is spoilt by the use made of the examples – leading cities failing to sustain economic growth is taken as evidence against the importance of cities for economic expansion. There are, of course, many great cities that declined and this is totally in keeping with Jacobs' ideas – the cities cease to continue import replacement. Jacobs model is as much about city decline as city growth: in fact she devotes a full chapter to mechanisms of decline entitled “Transactions of Decline (Jacobs 1984, chapter 12) as well as peppering her text with case studies such as the demise of Dinant (Jacobs 1970, 130). Thus the fact that the Dutch cities that created the ‘golden age' in the seventeenth century, but did not generate the industrial revolution in the eighteenth century, is not proof of the failure of agglomeration ‘to ensure sustained growth'. as Polèse (2005, 1442) argues by contrasting it against the industrial success in a less urbanised England. In any case, it was not London, by far the largest urban agglomeration in England, that produced the industrial revolution. The point is that, as noted earlier, in any group of cities location of the dynamism moves over time: in the early modern European city network Dutch cities lived off their wealth in the eighteenth century with no need for new innovations (Israel 1995). Arrighi (1993) has a very interesting historical discussion of these processes of decline but the point here is that the fact of multiple city declines is completely consistent with Jacobs' model.

WHY SUCH MISAPPLICATIONS OF JACOBS' IDEAS?

My answer to this question relates to the nature of social science. I have argued elsewhere (Taylor 1996) that social science is endemically state-centric: the ‘national' is taken-for-granted, sometimes even treated as ‘natural'. Therefore, Polèse's (2005) article is like a breathe of fresh research air for its explicit argument in favour of study at the state scale. Debate is opened up and this can only be a good thing. But his work also indicates to me how deeply embedded the state is even in such clearly thought out arguments as Polèse's. I am not here thinking about references to national urban hierarchies being ‘natural' (p. 1445), but rather the link Polèse makes between social science and policy (p. 1446-7).

For Polèse (2005, 1446), the value of social science research ‘rests on its ability usefully to inform public policy'. And, of course, the social sciences arose about a century ago to take up the challenges of informing policies for reform in economic, political and social arenas. I have argued that this is the primary source of state-centrism in the ensuing intellectual discourses (Taylor 1996). ‘Public' usually elides into ‘national' as in national ‘location of industry' policy because surely public policies should be for the ‘national good'? This is implicit in Polèse's treatment of public policy and his earlier emphasis on the ‘national institutional environment' (1437-8). How could it be otherwise when he argues that ‘[c]ities … are a reflection of the societies into which they were born' (p. 1447 - although actually most major cities are older than the states in which they are located). Nevertheless, he concludes by affirming the importance of good planning and governance in cities before ending with the need for ‘appropriate national macroeconomic and monetary policies' (p. 1447). But introducing public policy has a deeper meaning than up-scaling to where most of the ‘public money' is. It directly influences the form that social science research commonly takes.

What public policymakers want from social scientists are relatively simple causal relationships that can be planned into policy as ‘levers', a cause that will produce a desired effect. Instead of lever, Polèse (2005, 1441) uses the equivalent metaphor of ‘trigger'. According to Polèse, the ‘primary shortcoming' of Jacobs model and associated research ‘is that they are difficult to translate into effective public policies' (p. 1446). But this is precisely because cities are treated as complex entities that cannot be so easily planned – this is where Jacobs (1962) entered the debate on the nature of cities – see the final chapter of The Death and Life of Great American Cities entitled ‘The Kind of Problem a City is'. In total contrast, Polèse's treatment of cities as ‘places' rather than ‘primary economic organs' (p. 1436-7), culminates in a most inappropriate instrumental comparison of cities to roads:

‘It is as difficult to imaging sustained economic growth without cities (agglomeration) as it is without roads. But the presence per se of cities (or roads) is not a sufficient condition to generate long-term economic growth.' (p. 1446)

Thus public policy should make cities, presumably like roads, ‘as efficient … as possible' (p. 1447). This is exactly the opposite of what Jacobs (1984) argues – this point is so important she devotes a chapter to it entitled ‘The Valuable Inefficiencies and Impracticalities of Cities' (chapter 3). But for Jacobs' ideas to be taken on board into social science they need to be trimmed and packaged as hypotheses about national urbanization thus making ‘[l]ife … much easier, certainly for policy-makers' (Polèse 2005, 1446). What a pity.

REFERENCES

ARRIGHI, G. (1993) The Long Twentieth Century . London: Verso.

ISRAEL. J. (1995) The Dutch Republic: Its Rise, Greatness and Fall, 1477-1806 . Oxford: Oxford University Press

JACOBS, J. (1962) The Death and Life of Great American Cities . London: Cape

JACOBS, J. (1970) The Economy of Cities . New York: Vintage

JACOBS, J. (1984) Cities and the Wealth of Nations . New York: Vintage

POLÈSE, M. (2005) Cities and National Economic Growth: a Reappraisal, Urban Studies , 42 (8), pp. 1429-1451

SAYER A. (1992) Method in Social Science: a Realist Approach . London: Hutchinson

TAYLOR, P. J. (1996) Embedded statism and the social sciences: opening up to new spaces, Environment and Planning A , 28, pp. 1917-28

Table 1: Polèse's ‘outliers: Grab bags of very different economies

Source: Derived from Polèse (2005, Figure 3)

Edited and posted on the web on 20^th September 2005