A diagnostic approach to intra-metropolitan spatial targeting: Evidence from Cape Town, South Africa

Abstract

This article ascribes poor returns from place-based economic policy to prevailing spatial norms and causal assumptions which continue to influence its deployment across South African cities. By elevating the local over the systemic as the cause of and solution to urban problems, spatial targeting in the telescopic mould gives rise to three forms of spatial bias which lock in suboptimal local outcomes and gradually undermine the resilience of the urban system. Place-based policy should instead be guided by a systemic and relational evaluation of local economic potentiality supported by data-driven planning tools. The article introduces one such tool developed by the City of Cape Town, focusing on its theoretical basis, initial findings and implications for intervention. We find that the tool represents a robust platform for policy-makers to make targeting decisions that are more evidence led and hence less arbitrary.

Keywords:

• spatial targeting
• location theory
• place-based intervention
• evidence-based planning

1. Introduction

Amid ever-increasing divergence between urban and rural development trajectories in the Developing South, the centripetal forces that power cities towards innovation, productivity and regional competitiveness (Lucas, 1988; Abdel-Rahman & Anas, 2004; Amin, 2012) are also responsible for fuelling the urbanisation of the rural poor. Here the ‘developmental’ local government has to manage the inherent tension between growth and equity agendas, by establishing spatially-blind institutions and policies, providing infrastructure to drive the spatial connection between leading and lagging regions, and spatially targeting place-based interventions (Turok & Bailey, 2004). In South Africa, an emerging reliance on place-based interventions as a next-best solution to much-needed structural reforms has not been tempered by the inconvenient lesson drawn from the domestic track record; that is, such interventions have generally been ineffective in addressing poverty, stimulating private-sector investment or enhancing the overall efficiency of the urban system at a scale commensurate to their cost (Nel & Goldman, 2006; Peter, 2008). In fact, several authors have found that the opportunity cost of South Africa's legacy of place-based interventions outweighs the most generous estimation of benefits (Wellings & Black, 1986; Todes, 2013).

This article argues that intended policy outcomes are undermined by the prevailing telescopic conceptualisation of the city as an assemblage of discrete and competing neighbourhoods, thus neglecting the role of regional economic topography in shaping local prospects. For this reason, the technical matching of particular areas with appropriate instruments should be supported by data-driven tools that assess local economic potentiality from a systemic and relational perspective – that is to say, in relation to other nodes within the urban regional system. The diagnostic tool introduced in this paper spatially matches place-based instruments to appropriate business districts – and vice versa – based on their performance and location potential relative to other districts within the urban region. We focus on the tool's theoretical underpinnings, mechanics, limitations and the preliminary results when applied to Cape Town.

2. Revisiting place-based policy in South Africa

The prospect of effective place-based intervention in South Africa is hampered by contradictory policy thrusts: whereas the new generation of instruments⁴ concentrate intervention in locations that are both accessible and exhibit demonstrable economic potential, the traditional telescopic approach to spatial targeting⁵ remains entrenched in local and regional policy discourse. In evaluating three forms of spatial bias invoked by the telescopic approach, we underscore the need for a systemic and relational understanding of local economic potentiality within the context of large cities.

2.1 Origins and rationale of telescopic targeting

The telescopic approach is rooted in sociological and political economic theory, emerging from the conception of the urban region as a checkerboard of discrete neighbourhoods vying for allocative preference in a perpetual zero-sum competition.⁶ Local intervention is then premised on the conception of urban location as a socially-constructed development problem giving rise to a politically-constructed solution (Donaldson & du Plessis, 2011; Todes, 2013).

The tendency to define the location problem in social rather than economic terms harkens back to the post-war roots of place-based policy in Europe and the United States, where attempts were made to target residential neighbourhoods which exhibited disproportionate levels of social dysfunction (Smith, 1999; Tunstall & Lupton, 2003). The transmission of social pathologies and anti-social behaviour was observed between individuals and groups clustered in space, giving rise to what Imbroscio (2006:231) calls ‘the downward spiral of poor neighbourhoods’ (Rusk, 1999; Orfield, 2002; Pugalis, 2013). These ‘neighbourhood effects’ significantly diminished the effectiveness of mainstream social policy on local livelihoods (Tunstall & Lupton, 2003).

However, this conception remains ideologically contested (Kleinman, 1999; Smith, 1999; Batey & Brown, 2007; Pugalis, 2013) and muddied by ambiguous empirical evidence. The observed ineffectiveness of past place-based intervention is ascribed to, firstly, the inherently arbitrary and political rather than technical nature of defining the location problem (Tunstall & Lupton, 2003; Pugalis, 2013). Secondly, there is a growing recognition that urban poverty is a product of wider social and economic processes that require a broader policy response (Turok, 2004; World Bank, 2009). By viewing urban problems through an exclusively local lens, the telescopic approach ignores:

all the myriad hidden connections and relational things that hold together the contemporary city as an assemblage of many types of spatial information, from economically interdependent neighbourhoods to infrastructures, flows and organisational arrangements that course through the city and beyond. (Amin, 2012:14)

Within the complexities of change driven by the external world, place-based interventions operating in small areas are not going to make a significant difference to local livelihoods, and may, at best, offer ‘sticking plaster solutions’ in pursuit of short-term visible effects (Smith, 1999:37).⁷

2.2 Spatial biases of the telescopic approach

The telescopic approach leaves the policy-maker mired in an analytical limbo, unable to make the critical shift from the normative (i.e. why we intervene) towards the technical matching of particular areas with appropriate instruments (i.e. how and where we intervene). Consequently, three forms of spatial bias are created which reproduce the systemic inefficiencies they seek to address. We show how these biases serve to negate policy opportunities arising from the presence of economic agglomeration, worker mobility and the ever-expanding horizon of integrated spatial data systems.

2.2.1 Dispersive bias

By emphasising the dispersion of economic activity to new nodes instead of reinforcing existing nodes, telescoping targeting fails to take advantage from agglomeration economies. The focus on outward growth draws ostensibly from the political imperative to ‘bring jobs to poor areas’, even though dispersive policies have failed to offset tendencies towards economic concentration (World Bank, 2009). However, there is a lack of theoretical grounding and empirical evidence supporting the assumed efficiency benefits of dispersive policies (Parr, 2004; Turok & Bailey, 2004; Meijers, 2008; Sinclair-Smith & Turok, 2012). We ascribe the dispersive tendency to the inherent organisational bias for bureaucracies to favour duplicative infrastructure (Anas et al., 1998:1458): the prevailing emphasis on new investment is the path of least institutional risk, demanding less information, technical know-how and political capital (Barca et al., 2012) than revitalising existing facilities (Savage, 2008).⁸ The dispersive bias implicit in the telescopic approach neglects the extent to which the fate of large cities is bound up with the fortunes of existing business districts.⁹ Shoring up existing business districts ensures: firstly, better service delivery through economies of scale in public goods provision¹⁰ (Turok & Watson, 2001; Van Huyssteen et al., 2010); secondly, increased productivity through spatially-concentrated knowledge networks (Fujita & Thisse, 1996; Rosenthal & Strange, 2008); thirdly, increasing returns, reinvestment and overall economic growth due to the concentration of economic activity (Lucas, 1988; Shukla & Waddell, 1991; Fujita & Thisse, 1996; Anas et al., 1998; Brakman & Marrewijk, 2013); fourthly, better economic and social integration as a result of compact urban form (Fujita & Thisse, 1996; Turok & Watson, 2001; McMillen & Smith, 2003); and, fifthly, higher overall social welfare compared with dispersive policies (Tabuchi, 1998).

2.2.2 Exclusionary bias

Telescopic targeting counterpoises social need and economic potential through a false binary of ‘winners’ and ‘losers’, resulting in a bias towards areas with limited economic potential. The exclusionary principle underlying telescopic targeting is based on the assumption that workers and job seekers are entirely immobile, implying that the benefits of place-based interventions are wholly captured by the immediate community rather than neighbouring communities or the city as a whole (Stiglitz, 1984). However, the presence of spatial mobility within a networked urban system renders this dichotomy reductionist and theoretically false. Integrative policies aimed at lowering aggregate movement cost, such as public transport and densification, strengthen access to location potential where it exists, rather than attempting to re-create it elsewhere at immense cost and with little prospect of success.

2.2.3 Supply-side bias

There is a growing appreciation for the role of economic topography in shaping the effectiveness of place-based interventions (Donaldson & du Plessis, 2011; Barca et al., 2012; Donaldson et al., 2012; Du Plessis, 2013). At the same time, the telescopic approach's implicit neglect of regional market forces weakens the capacity of local government to facilitate and direct private-sector development; instead, it gives way to arbitrary decisions weakly linked to economic data (Batey & Brown, 2007). In the few instances where demand-side factors were adequately assessed (e.g. the Maputo Corridor) or where quantifiable location factors formed part of the eligibility criteria (e.g. the Urban Development Zones¹¹), notable successes were achieved in leveraging private-sector investment (Todes, 2013).

More typically, however, project outcomes are assessed using ‘blunt, outdated and misleading’ indicators (Batey & Brown, 2007; Savage, 2007; Peter, 2008; Donaldson & du Plessis, 2011; Todes, 2013). This creates fertile ground for unrealistic assumptions about the extent to which a proposed intervention will influence local economic fortunes (Savage, 2008; Freund, 2010), a tendency not unique to South Africa (Ansar, 2012). In the long-term, the opportunity to use spatially targeted interventions as a test bed for policy experimentation and learning is compromised, and the scope for replication and mainstreaming is greatly diminished (Smith, 1999; Tunstall & Lupton, 2003; Atkinson, 2007; Donaldson et al., 2013).

Apologists for ‘hit-and-miss’ targeting often cite the lack of credible and spatially disaggregated data as the main culprit behind arbitrary targeting decisions and poor post-implementation assessment. However, the unprecedented growth in integrated administrative data systems allows metropolitan authorities to gauge, by means of ongoing hypothesis-testing against their own evidence base, the applicability of generalised causal assumptions to local conditions, providing policy-makers with insights into what works and under which conditions.

Taken together, the distortionary effects have dire implications for the resilience of the urban system as a whole, leading a number of observers to advise against spatially-targeted interventions (Lupton & Turok, 2004; Barca et al., 2012). However, we propose that one way of elevating spatial targeting methods from the telescopic to the systemic is the application of diagnostic tools which systematically and dispassionately identify and track locational potential in relation to regional market forces.

3. Diagnostic approach to spatial targeting

The diagnostic approach presented here involves the systemic and relational assessment of business districts' current level of functioning. The concept of diagnosis, the identification of the nature and cause of a certain phenomenon, is drawn from the medical model, in that a physician conducts tests, collects vital information on the human system and evaluates this information to recommend a course of action. Similarly, the strategic planner uses specialised analytical tools to collect and synthesise vital information about the overall system to identify the nature of the ‘location problem’ and formulate an appropriate response.

3.1 Theoretical points of departure

The diagnostic tool is based on two main propositions. First, the performance of a business district can be measured in terms of its property market because, as rent is a function of the gross profit of an enterprise, the rents paid in a given district are a reflection of its ability to successfully support business operations. Second, the location attributes of a district play an important role in supporting and growing business functions.

Traditional location theory, crystallised by Isard and Alonso, attempts to identify spatial regularities in urban areas and to develop general laws of spatial structure and urban processes (Scott, 2000). The theory draws heavily on the work of the early classical and neo-classical economists of Smith, Ricardo, Mills and Marshall, the German location theorists of Von Thünen, Weber, Christaller, Hoover and Lösch, the growth pole and regional development theories developed by the French economists Perroux, Boudeville and others, as well as the work undertaken by the urban ecologists of Burgess, Hoyt, Park, and Uhlmann and Harris (Scott, 2000).

The central assumption of the theory is that firms seek to maximise profits by locating in places where the costs associated with production and accessing markets are minimised. In turn, those firms that can minimise costs and maximise revenue and profits in a particular location will be able to pay higher rents and effectively outbid other users for a particular site (Goodall, 1972; Cohen, 2000). A location is seen to include both the broader elements of an area and the specific characteristics of a particular site within the area (Cohen, 2000). The theory therefore predicts that a firm's location will be determined by the internal functional requirements of the firms, the need to exchange and interact externally, and the cost of locating in a particular location. Where these costs and returns lie will depend on the nature of the product or service being produced and therefore, depending on the nature of the business, a firm will make a trade-off between locations such that the net profit maximising location is achieved. In making these trade-offs, a firm will also attempt to capitalise on the benefits of internal (economies of scale) and external (economies of urbanisation and localisation) economies that occur when similar and related firms co-locate (Goodall, 1972; McDonald & McMillen, 2010).

The above theory, however, has been criticised for assuming that decision-makers are rational, well-informed, profit maximisers, that space is a ‘neutral’ plane upon which location decisions are made, and that decisions regarding production processes are taken before the actual location is considered (Krumme, 1969; Scott, 2000). Furthermore, the cost of relocation in terms of direct costs, loss of production, and loss of labour and customers means that often firms will be ‘trapped’ by inertia and therefore they will often locate where their needs are satisfied but not necessarily maximised (Simon, 1956; Townroe, 1972). Structurally, the theory has been criticised for ignoring the inherent contradictions and forces associated with the capitalist mode of production (Scott, 2000). These criticisms have led to the development of ‘alternative’ structuralist (Molotch, 1976; Massey, 1984; Scott, 2000; Davies, 2002; Harvey, 2011), behavioural (Gregory, 1981; Thrift, 1983; Giddens, 1984; Massey, 1984; Scott, 2000; Corpataux, 2007), territorial (Thrift, 1983; Massey, 1984; Aydalot, 1986; Porter, 1990; Camagni, 1991; Keeble & Wilkinson, 2000) and institutional (Nelson & Winter, 1982; Amin & Thrift, 1992; Ball, 1998; Cooke & Morgan, 1999; Keeble & Wilkinson, 2000; Corpataux, 2007) theories to explain the location of economic activities.

More recently, partially in response to the lack of rigour and the descriptive nature of the ‘alternative’ theories, and partially as a result of the development of modelling techniques that allow general equilibrium theories to be modelled, a new approach known as New Economic Geography has emerged (Krugman, 1991). The central endeavour of this approach is to explain how economic agglomerations occur in space as a result of competing centripetal and centrifugal forces (Krugman, 1991; Martin, 1999). The approach has, however, been criticised for making unrealistic modelling assumptions, and for ignoring the role of agency, territory and institutions in the creation of spatial patterns (Martin, 1999).

As the diagnostic model is strongly based on traditional location and New Economic Geography theories, the critiques of such theories provide useful insights into the limitations of the model and give direction as to what the model should and should not be used for. In particular, there is a danger of assuming that property market indicators give a total indication of the desirability (or lack of) of a location to a firm. In many cases, firms will make location choices, which are often sub-optimal, based on what options are available to them, and these options are often the outcome of particular private and public actors using their disproportionate power (financial, political, networks, etc.) to crowd investment into locations that enhance their positions and wealth (Molotch, 1976). Therefore, the use of the diagnostic model should be the first step in any analysis and such analysis seen as one of many inputs that should be taken into account when intervening in the city.

3.2 Model specification

The objectives of the diagnostic tool are to: identify the business districts of the city; identify a set of indicators that will measure the current performance and future potential of an area; generate a typology of areas based on the above indicator set; understand the relationship between the performance of an area and its locational attributes; and develop a set of interventions based on the above typology.

3.2.1 Identification and delineation of business districts

The delineation of business districts is fraught with problems (Kowalski & Paraskevopoulos, 1990; Anas et al., 1998; Anderson & Bogart, 2001; Adair et al., 2005; Amin, 2012; Davoudi, 2012), but should – at its core – reflect the ‘functional economic geographies’ of formal economic activity (Pugalis & Townsend, 2013:703), rather than administrative and political boundaries. We identify business districts on the basis of contiguous concentrations of commercial and industrial land use filtered through a set of technical thresholds derived from standards developed by McDonald (1987). The technical thresholds include areas with more than 1000 morning commuter arrivals and an aggregate value of business properties greater than ZAR50 million. The spatial delineation of boundaries was based on available land-use data and adapted where necessary to functional boundaries as perceived by local stakeholders and property brokers. When the above criteria were applied to the City of Cape Town, over 60 districts were identified (see Figure 1).

Figure 1: Map of Cape Town business districts

3.2.2 Indicator selection

The functioning of each business district was measured in terms of the performance of its property market between 2005 and 2013, on the one hand, and its long-term location potential on the other. Whereas indicators associated with market performance provide us with our dependent variable, location potential indicators were employed as explanatory variables.

A pragmatic approach was adopted in the final selection of indicators, involving three sequential stages of assessment. Firstly, with respect to the property market, price indicators and those influencing price levels (either directly or as a proxy) reflect the supply and demand of space in the market over time and hence indicate how well it is performing. A panel of expert property economists and brokers then assessed these ‘dependent’ indicators to determine whether they were reliable and relevant indicators of property market performance. Secondly, drawing on business location surveys in the literature (Wellings & Black, 1986; Karakaya & Canel, 1998; Ansar, 2012), indicators were chosen that generally reflect the degree to which an area facilitates access to the factors of production and markets. The identification and weighting of these ‘explanatory’ variables were then verified by the degree of positive correlation with the property market ‘dependent’ variables. Lastly, all of the indicators were filtered according to whether datasets were systematically and regularly collected on a citywide scale and available at the appropriate scale. The potential data bias resulting from this approach warrants consideration when interpreting model results.

Property market data were collated and organised to estimate the short-term to medium-term performance of the property market (Table 1). Similarly, data indicating economic agglomeration, labour supply, room for growth, accessibility, infrastructure, crime and the surrounding residential catchment were collated for each business district to determine location potential (i.e. the degree to which an area's attributes corresponds to the location requirements of the notional firm) (Table 2).

Table 1: Market performance indicators

Sub-indicators	Detail	Source
Industrial rentals (2013) Industrial rental growth (2005–13)	Prime industrial rentals, R/m^2, for 1000 m^2	Rode & Assoc.
Office rentals (2013) Office rental growth (2005–13)	Average prime office for Grade A office, R/m^2
Street-front rental (2013)	Average R/m2 rental for 100 m^2
Shopping centre rental (2013)	Average gross rental (R/m^2)	Corporate reports
Industrial vacancy (2013)	Industrial stand vacancies	Rode & Assoc.
Office vacancy (2013)	As % of Grade A office stock	South African Property Owners Association
Building development completed (2005–13)	Internal floor space added, by m^2; includes new build industrial, office, shopping; and non-residential additions and alterations	City of Cape Town
Building plans received (2012–13)	Internal floor space to be added, by m^2; includes new build industrial, office, shopping; and non-residential additions and alteration	City of Cape Town
Number of property sales/overall number of properties (2005–13)	Single-dwelling residential properties excluded	Deeds Office
Industrial vacant land sales (2005–13)	By value and number of sales

Table 2: Location potential indicators

Sub-indicators	Detail	Source
Existing internal floor space of non-residential building stock	Gross internal floor space for office, street-front and shopping centre retail, industrial and warehousing	City of Cape Town Municipal Valuations Roll
Aggregate property value	Aggregate value of improved industrial and commercial properties
Property value intensity	Same as above, normalised by geographic extent (m^2)
Count of improved industrial and commercial property parcels	Includes sectional title properties
Estimated number of work spaces	Based on employment density function
Projected additional bulk supply	Available future bulk for commercial and industrial activity between 2012 and 2032 as identified by district planners	City of Cape Town Urban Growth Model
Extent and affordability of vacant industrial land	Extent of vacant, serviced land zoned for industrial activity	Municipal valuations roll (Rode & Assoc.)
Travel distance to regional markets and freight corridor gateways	Average of peak and off-peak travel distance and travel time	Tomtom Live Traffic data
Road congestion	Average difference between peak and off-peak travel time
Public transport connectivity and level of service	Level of public transport connectivity, and operations (frequency and capacity)	City of Cape Town Interchange Classification 2012
Infrastructure capacity	Level of constraint on the following infrastructure networks: electricity, water reticulation, waste water treatment capacity, sewer and storm water	City of Cape Town
Security	Incidence of burglaries and robberies reported by businesses, normalised by number of business parcels	Derived from statistics published by South African Police Service
Skills catchment	Number of workers with higher education qualifications, and postgraduate workers, within 5 km radius	Census 2011
Morning commuter arrivals	Number of morning arrivals (by all modes)	City of Cape Town Transport Model
Income catchment	Aggregate and per-capita disposable income within 5 km radius	Census 2011

These indicators were populated with data extracted and aggregated to the business districts and were standardised. Business districts were categorised according to their predominant use (i.e. commercial, mixed use and industrial) and scored according to sub-indicators relevant to that categorisation. Predominantly commercial areas (where >85% of internal floor space is dedicated to commercial activity) were scored in terms of retail and office use, industrial areas according to industrial uses, and mixed use where neither commercial nor industrial activity occupies >85% of internal floor space.

3.3 The four-quadrant diagnostic model

The composite scores are then plotted onto a scatterplot graph, with the X-axis representing market performance and the Y-axis representing location potential. As area scores are standardised, each axis represents the mean score for the sample.

3.3.1 Interpreting the results

The plotting of each district's composite scores on a XY scatterplot enables each to be classified into four quadrants according to the recent performance of a business district's property market and its medium-term to long-term potential (Figure 2): low-performance, high-potential districts; high-performance, high-potential districts; high-performance, low-potential districts; and low-performance, low-potential districts. The classified districts are then mapped as shown in Figure 3.

Figure 2: XY scatterplot showing the results for Cape Town

Figure 3: Classification of business districts

The model assumes a state of equilibrium corresponding with the correlation coefficient line in Figure 4, suggesting that over time the relationship between a district's property performance and its attributes should tend towards a ‘state of equilibrium’ where those districts with location potential have well-performing property markets, and vice versa. This state is reached over time through the (re)development of commercial and industrial space.

Figure 4: Propulsive drivers of business district transformation

3.3.2 The urban development process

Urban development is the result of the demand and supply for space mediated by financial, institutional and political forces. The demand for space is driven by the willingness and ability of firms to pay for space, which in turn is driven by the degree to which a district's location attributes meet the location requirements of a firm and by the nature and performance of the firm occupying space in the district.

The supply of space is driven by the willingness and ability of property owners and developers to invest in the provision of such space, which in turn is a function of whether the value created by the demand from firms is higher than the cost to provide the space (DiPasquale & Wheaton, 1996). The outcome of this demand and supply of space is a continual process of depreciation and appreciation of the built form over time (Visser, 2002; The Appraisal Institute, 2008).

3.3.2.1 Depreciation and disinvestment

It is possible to trace this process over time as follows: assume as a starting point that a business district has good location attributes and consists of profitable companies paying relatively high rentals. Over time, the profitability of the firms may decline for two reasons. Firstly, the clustering of firms may cause diseconomies of agglomeration (congestion, skills shortages, etc.), resulting in a mismatch between the firms' requirements and the attributes of the district – a functional depreciation. Secondly, profits may fall due to a decline in the economic sector of the firms due to exogenous factors (e.g. tariff policies, global competition, etc.). This decline leads to a decreased willingness and ability of firms to demand space in the district, which results in lower rental growth and property values – an economic depreciation. The consequence is a decline in the supply of space (management of existing and construction of new space) because the lower values are a disincentive to invest in existing and new property. The lack of investment by both private (maintenance, upgrades and new construction) and public (infrastructure and urban management) sectors leads to a further physical depreciation, which exacerbates the above-mentioned functional depreciation and creates a spiral of decline.

3.3.2.2 Appreciation and investment

However, the reverse may occur when a district grows on the back of the rising profitability of the firms located in it. If the economic sector of the firm is inherently healthy, then profits may be improved if functional and physical appreciation occurs as a result of public and/or private investment in the district's attributes. However, where a district's firms represent an economic sector that is in decline, then improved profitability will usually only occur when these firms are replaced by a new set of firms from a different and growing economic sector that can successfully take advantage of the attributes of the district. In this case, a functional appreciation occurs due to a repositioning of the types of firms in the district. This clustering of a sector in an area is often due to the benefits of agglomeration: production costs can be reduced and output maximised when processes, firms and sectors are co-located as a result of the internal economies (economies of scale) and external economies (economies of urbanisation and localisation) that are achieved as a result (Porter, 1990; Krugman, 1991). The resultant increased profitability of the firms in the district leads to an increased ability and willingness to pay higher rentals, which increases the value of property in the district – an economic appreciation.

This increased value then leads to profit-seeking by developers and investment in that area. Profit-seeking occurs when profits are generated by closing the ‘rent gap’: profits can be made by acquiring properties with low land rentals (values) and (re) developing them to their highest and best use (production) (Visser, 2002). In other words, closing the gap between ‘capitalised ground rent' – the economic return from the rights to use land (given its present use) – and ‘potential ground rent' – the return that could be earned if the land was put to its optimal, highest and best use (Lees et al., 2013). This process drives both the development of new, green-field sites, where a change of land-use rights and infrastructure provision can increase the potential ground rents, and the redevelopment of existing districts where decayed property can be bought at relatively low prices (capitalised ground rents) and redeveloped to satisfy the demand of new inhabitants and economic sectors allowing for higher potential ground rents and profits. The consequent investment results in physical and functional appreciation, leading to further economic appreciation as more firms demand space and precipitating a spiral of agglomeration and growth.

3.4 Accounting for market disequilibria

The question is therefore whether the above urban development process always results in a position of equilibrium, since a range of factors such as sunk costs, path dependence, the symbiotic co-evolution of firms and surrounding communities, property ownership structures, the inflexibility of the building stock and the impact of development controls may inhibit the extent to which local firms are able to respond to changing conditions (Fujita, 1989; Blair, 1991; Ansar, 2012). Districts, in other words, exhibit varying degrees of spatial fixity and in some cases the performance of a business district's property market indefinitely refuses to respond to its location assets. We tender two discrete but mutually inclusive explanations for this – model misspecification and market failure.

3.4.1 Model misspecification

The model does not take into account differentiation of business location preferences between firms, but rather treats firms as sharing the same set of general preference. In order to select and weight location factors according to the relative merits of alternative locations for a specific kind of firm, one would need an understanding of the specific business activity involved (Hoover & Giarratani, 1971). However, the heterogeneous nature of location requirements across and within sectors, combining generic, sector-specific and idiosyncratic aspects, is a significant encumbrance to modellers. While it is theoretically possible to model such preferences, the data requirements make it extremely difficult to do so.

The second factor is that as the model only tracks measurable economic phenomena, certain types of economic activities such as home, informal and visitor-based activities are excluded. However, as these are relatively small in scale and geographically correlated to formal activity, it is argued that they have a limited impact on the model (Valodia & Devey, 2012).

3.4.2. Market failure

By inferring a hypothesised state of equilibrium, the model betrays an assumption that markets are generally efficient, ceteris paribus. However, as Stiglitz (1984) argues, market failure is often the rule rather than the exception and therefore urban systems are typified by various forms of urban inefficiencies, and therefore the deviation from the trend-line is not principally a product of misspecification but rather evidence of market failure that may require government intervention (Figure 5).

Figure 5: Hypothesised sources of deviation from equilibrium

Furthermore, Abdel-Rahman & Anas (2004) and Boventer (1978) argue that the assumption that planners and developers dictate urban structure is unrealistic and idealised, and that urban development is heavily influenced by atomistic and bandwagon decision-making by firms and consumers. Therefore, whereas theory based on economic rationality looks for optimal solutions through systematic methodologies, locational decision-making under conditions of bounded rationality suggest that information is only a small component of the decision process and that the process itself is invariably ‘multifaceted, emotive, conservative and only partially cognitive’ (Keen, 1981:25).

For the reasons described above, the explanatory power of the model remains qualified in isolation; instead, its value resides in its ability to collate and integrate data and test spatial hypotheses in conjunction with qualitative research about the role of social, political and institutional dynamics in shaping urban development outcomes.

4. Implications for intervention

While the classification of an area is based on the outcome (symptom measured by an indicator) of various urban drivers, the interventions should be based on the drivers themselves and must respond to a market failure associated with these drivers. Furthermore, the interventions need to be clear as to: firstly, whether they are intervening in the economic activities occurring in the district or the development of the built environment in the district; secondly, whether they are focusing on the consumption or production processes of either of the above; and, finally, how the intervention responds to a specific problem identified in any of these processes.

Figure 6 shows a prioritisation framework which matches various place-based strategies and associated instruments to particular business districts depending on their diagnostic assessment. In this instance, an audit of existing and proposed place-based policy instruments available to the Cape Town metropolitan government was conducted and organised into four place-based strategies: regeneration, growth management, stabilisation and repositioning.

Figure 6: Prioritisation framework for spatial targeting

5. Conclusion and future research

This article identifies the need for a flexible, data-driven planning tool which matches particular place-based instruments with appropriate business districts, and vice versa. Recognising the role of regional economic topography in shaping the effectiveness of place-based policy, we introduce a theoretically-sound conceptual architecture which evaluates local economic potentiality from a relational and systemic perspective.

Further investigation is needed to determine the extent to which this model may be enhanced beyond the realm of descriptive and exploratory statistical modelling. Given that urban systems are complex, dynamic and experience feedback, it is incumbent upon data scientists to cast the net well beyond generalised linear modelling to harness the full predictive power of city data.

This is not merely an academic pursuit: the extent to which large city authorities are able to beneficiate data to inform strategic spatial decision-making will become decisive in determining the level of resilience of increasingly complex urban economies amidst rapidly changing macro-conditions in the twenty-first century.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work is supported by the City of Cape Town and Mistra Urban Futures, a global research and knowledge centre in sustainable urban development, funded by the Swedish International Development Agency and the Mistra Foundation for Strategic Development.

ORCID

Robert McGaffin http://orcid.org/0000-0003-0323-6898

Notes

⁴For example, the National Spatial Development Perspective, Urban Development Zones, the National Development Plan and the attendant Urban Network Strategy.

⁵Typified by Local Economic Development policy, the Urban Renewal Programme and Industrial Development Zones.

⁶See Rodríguez-Pose & Arbix (2001) for discussion on ‘fiscal wars’.

⁷ See Dreier for a discussion on ‘the limits of localism' (2001:171).

⁸See Molotch (1976) and Imbroscio (2006) for a discussion on the role of local property developers in perpetuating the ‘local growth machine’.

⁹Ten of the 71 business districts in Cape Town account for over one-half (51%) of estimated workplaces, of which only two were established in the last 30 years.

¹⁰Dispersed economic development tends to be associated with ‘greater capital costs related to school construction, the extension of road, water and sewer lines, and storm water drainage systems' (OECD, 2008).

¹¹Todes (2013) refers to a 2013 Demacon study which assesses the impact of the Urban Development Zone incentive. Private calculation shows that in Cape Town the Urban Development Zone has generated a leverage ratio of 1:18, which contrasts sharply with the dismal 1:0.13 ratio for the problem-oriented Khayelitsha Urban Regeneration Programme.