Dynamic Trip Modelling The GeoJournal Library

Volume 84

Managing Editor: Max Barlow, Toronto, Canada

Founding Series Editor: Wolf Tietze, Helmstedt, Germany

Editorial Board: Paul Claval, France Yehuda Gradus, Israel Sam Ock Park, South Korea Herman van der Wusten, The Netherlands

The titles published in this series are listed at the end of this volume. Dynamic Trip Modelling From Shopping Centres to the Internet

by

ROBERT G.V. BAKER School of Human and Environmental Studies, University of New England, Australia A C.I.P.Catalogue record for this book is available from the Library of Congress

ISBN-10 1-4020-4345-7 (HB) ISBN-13 978-1-4020-4345-1 (HB) ISBN-10 1-4020-4346-5 (e-book) ISBN-13 978-1-4020-4346-8 (e-book)

Published by Springer, P.O. Box 17, 3300 AA Dordrecht, The Netherlands. www.springer.com

Printed on acid-free paper

All Rights Reserved © 2006 Springer No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.

Printed in the Netherlands.

To Sue, Kristen and Cameron

Contents

Preface ix

Illustrations xi

Chapter 1: Introduction 1

1.1 Shopping Change 1 1.2 Definitions of Retail Forms Underpinning the Model 6 1.3 The Time-space Convergence 16 1.4 A Way Forward 18

Chapter 2: An Introduction to Retail and Consumer Modelling 21

2.1 Definition 21 2.2 A Justification for Modelling 22 2.3 The Art of Modelling 23 2.4 Model-building and its Weaknesses 29 2.5 Examples of Retail and Consumer Modelling 31 2.6 A Vision for Dynamic Trip Modelling 75

Chapter 3: Dynamic Trip Modelling 77

3.1 Background to the RASTT Model 77 3.2 Space and Time-discounting Shopping Trips 79 3.3 Characteristics of Space-discounting Behaviour 84 3.4 The Time-discounting Model 97 3.5 The Fourier Transform and Aggregate Periodic Trips 116 3.6 Estimating Shopping Centre Hours 126 3.7 Two Dimensional Space-time Modelling 130 3.8 Estimating Market Penetration with an Extension of Shopping Hours 134 3.9 Stochastic Space-time Trips 138 3.10 Space-time Modelling Shopping Trips: A Summary 151 3.11 Dynamic Shopping Trip Modelling 156

viii Contents

Chapter 4: Empirical Testing of the RASTT Model in Time and Space 157

4.1 Introduction 157 4.2 Background to the Research Methodology 158 4.3 The Empirical Method 161 4.4 The Sydney Project: Long Term Time Change of Shopping Trips (1980-1998) 167 4.5 Changes in Time-space Trip Behaviour in the Sydney Project 200 4.6 Application of the RASTT Model to Unplanned Shopping Centres: Armidale in Regional New South Wales, 1995 237 4.7 Application of the RASTT Model to Planned Shopping Centres: Auckland, New Zealand, 2000 247 4.8 Is there a ‘Global’ RASTT Model? 256

Chapter 5: Dynamic Modelling of the Internet 265 5.1 Introduction 265 5.2 The RASTT Model and Internet Transactions 268 5.3 Deriving the RASTT Model for Internet Transactions 273 5.4 Empirical Evidence 277 5.5 Applications to Shopping Transactions 289 5.6 Summary 290

Chapter 6: The Socio-Economic and Planning Consequences of Changes to Shopping Trips 293

6.1 The Problem of Shopping Times and Shopping Places 293 6.2 The Role of Parking and Walking 294 6.3 The Vacant Shop Problem in Australia 298 6.4 The Role of the Large Supermarket or Superstore 304 6.5 The Role of Planned Regional Shopping Centres 317 6.6 Policy Implications for Modelling Shopping Trip Change 321 6.7 Retail Planning as a ‘Wicked Problem’ 326

Chapter 7: Conclusions 327

B ib liography 339

Preface

The thesis of this book is that there are one set of equations that can define any trip between an origin and destination. The idea originally came from work that I did when applying the hydrodynamic analogy to study congested traffic flows in 1981. However, I was disappointed to find out that much of the mathematical work had already been done decades earlier. When I looked for a new application, I realised that shopping centre demand could be like a longitudinal wave, governed by centre opening and closing times. Further, a solution to the differential equation was the gravity model and this suggested that time was somehow part of distance decay. This was published in 1985 and represented a different approach to spatial interaction modelling.

The next step was to translate the abstract theory into something that could be tested empirically. To this end, I am grateful to my Ph.D supervisor, Professor Barry Garner who taught me that it is not sufficient just to have a theoretical model. This book is an outcome of this on-going quest to look at how the evolution of the model performs against real world data. This is a far more difficult process than numerical simulations, but the results have been more valuable to policy formulation, and closer to what I think is spatial science.

The testing and application of the model required the compilation of shopping centre surveys and an Internet data set. I would like to thank the management of the shopping centres for allowing me to survey consumers. I have had to observe, in some cases, a lag period to use some of the data as part of commercial confidentiality agreements. I am indebted to David Marshall who supervised the collection and collation of the shopping centre data. The Internet data was kindly made available by Dr Les Cottrell and Dr Gerod Williams from their network at the Stanford Linear Accelerator Centre. I also appreciate the work on this project by Troy MacKay, Brett Carson and Raj Rajaratnam and their discussions over many cups of coffee. My thanks also extends to Mike Roach who undertook a huge cartographic task and Sue Baker for a difficult proofing job. I am also indebted to the Australian Research Council for grants to complete the shopping centre surveys and Internet analysis.

Finally, thanks to my family, Sue, Kristen and Cameron, my brother Ross, my parents, Ellen and Douglas, and Alpha O without whose support during the good times and bad, this book would not have happened.

Illustrations

List of Figures

Figure 1.1 Meadowhall Regional Planned Shopping Centre, Sheffield, 1999

Figure 1.2 The Macellum on the Dupondius coin (AD 65) in the reign of the Emperor Nero

Figure 1.3 Market Share of Pharmaceutical Products in Australia, 1997

Figure 1.4 The Generation of an Internet Tree showing the Aspatial Connectivity from 100,000 Internet Routers and the Hierarchical Structures that develop from a few Highly Connected Nodes

Figure 1.5 The Time-space Convergence Showing the Cone of Time and Space Interaction Relative to Changes in Technology

Figure 2.1 A Flow Diagram Showing the Evolution of the Gravity Model in the Context of Consumer Behaviour

Figure 2.2 A Process of Building Relevant, Testable and Reproducible Models

Figure 2.3 The Regression of Mean Trip Frequency and the β Coefficient of the Gravity Model for Shopping Trips to the Sydney Shopping Centres 1980/82 and 1988/89

Figure 2.4 An Extension of Shopping Hours reduces the Slope of the Gravity Model (β ) where there is an Increased Propensity for Households to Travel to Planned Shopping Centre O rather than Shop Locally. Inset Photographs: (right), Vacant Shops (signed) in the Abbotsford Shopping Centre from Competition from MarketPlace Leichhardt (left), Sydney, New South Wales, May, 1999

Figure 2.5 Loschian Modifications in Christaller’s Hexagonal Trade Areas and the Northwest Retail Hierarchy for Canberra, Australian Capital Territory

Figure 2.6 The Location of 24-hour Coles Supermarkets in Sydney 1996

Figure 2.7 Consumer Equilibrium Analysis for Shopping Time and Visits to a Shopping Centre

xii Illustrations

Figure 2.8 Impact of Price Shifts (top) and Demand Curve Formation (bottom)

Figure 2.9 The Trading Hour Consumption Curve for Shopping Centre Time- Visit Allocations

Figure 2.10 Percentage Composition of Socio-economic Status of Late Night Shoppers (Armidale, NSW, November, 1995)

Figure 2.11 Shopping Preferences for Extended Hours by Socio-economic Groups

Figure 2.12 Analogous Engel Curves Relative to the Size of the Centre (a) Percentage of HDI Respondents in the Sample Plotted with HDI Trip Frequency and (b) a Model showing Normal and Inferior Engel Lines with Centre Scale

Figure 2.13 The Utility of Shopping at a Hierarchy of Malls in Sydney with Trip Distance, 1988/89

Figure 2.14 The Relative Utility Distribution with (k D)/N for Sydney Project 1988/89 and 1996/98

Figure 2.15 The Regression of the Gravity Coefficient β and U2N2/MD2 for the Aggregate Sydney Project in 1988/89 and 1996/98 (excluding the regional Bankstown Square samples)

Figure 2.16 Quadratic Distributions of the Gravity Coefficient (top) and Mean Trip Frequency (bottom) with Centre Size (Sydney Project 1980/82 and 1988/89). The Point of Inflections are at N = 147 and 145 Centre Destinations, respectively, for Small (negative slope) and Large (positive slope) Centre Behaviour

Figure 2.17 The Relationship between Trip Frequency and the Percentage of Multi-purpose Shopping (Sydney Project 1988/89)

Figure 2.18 The Relationship between the Gravity Coefficient and the Percentage of Multi-purpose Shopping Squared Divided by the Transfer Coefficient (Sydney Project 1988/89)

Figure 2.19 The Distribution of MPS changing with Centre Size (Sydney Project 1988/89). The Point of Inflection is at N = 167 Centre Destinations for Small (negative slope) and Large (positive slope) Centre Behaviour

Figure 2.20 The Gravity Model as a Negative Exponential Distribution away from a Shopping Centre for β1 > β2 Illustrations xiii

Figure 2.21 The Trade-off between Constant Logarithmic Supply of Destinations and Constant Negative Exponential Demand for a Shopping Centre C , if Consumers Minimise Trip Distance

Figure 3.1 The Fourier Transform of exp-g2t for Space-discounting Consumers

Figure 3.2 Changing Market Areas for Space-discounting Behaviour for Successive Time Periods

Figure 3.3 Changing Trip Distributions over a Day for Space-discounting Shopping

Figure 3.4 Changing Morning and Afternoon Distributions for Westfield Burwood and Ashfield Mall for pre-Christmas Space- and Time-discounting Trip Behaviour, respectively: Sydney Project 1988/89

Figure 3.5 The Negative Exponential Self-reciprocity between Trip Frequency ( f, 2f, 3f to 6f ) and Post Office Distance for Westfield Burwood (Sample: 15/12/88A), Sydney Data Set, 1988/89

Figure 3.6 The Fourier Frequency Assignment Ψ(f) with Shopping Time (t)

Figure 3.7 Testing the Shopping Time Hypothesis in Figure 3.6 with Distance Zones 1 and 2 in the Westfield Burwood pre-Christmas Rush Sample (15/12/88A) with Box and Whisker Plots

Figure 3.8 The Relationship between Mean Shopping Duration (p) and Destinations Visited m (1988/89) showing the Shift Right towards ‘Large Centre’ Behaviour Figure 3.9 The Regression of Intra-centre Shopping Frequency (f = p/2T) and the Mean Shopping Duration p per Trading Week T ( f = m× k /2T) showing the Shift Right towards ‘Large Centre’ Behaviour Figure 3.10 Comparison for the Theoretical m/2T and m/4T Values with the p/T Empirical Estimates from the Sydney 1988/89 Data Set

Figure 3.11 Regression showing the Relationship between Two Forms of the Intra-centre Shopping Frequency (f = mk/2T)) and (f = Mk ) for the Sydney Project 1988/89

Figure 3.12 Regression showing the Relationship between Two Forms of the Intra-centre Shopping Frequency (f =mk/2T)) and (f = Mk ) for the Sydney Project 1996/98

Figure 3.13 Simulation of the Space-time Shopping Distributions for 49.5, 70 and 100 Trading Hours (per week) for ‘Small Centre’ Behaviour xiv Illustrations

Figure 3.14 Simulation of the Time-space Shopping Distributions for 49.5, 70 and 100 Trading Hours (per week) for ‘Large Centre’ Behaviour

Figure 3.15 The Aggregation of Population Demand Waves (Sf) of Different Frequencies

Figure 3.16 The Demand Wave φ3 showing the Equal Likelihood of Visiting Three Equally-sized Centres (n =3) over the Trading Week

Figure 3.17 (a) The Graph of a sinc Function and the Gaussian Wave Packet (dotted line) and; (b) the Gaussian Wave Packet for T = 0.5, 1.0 and the Probability Density Function for T = 1.0 (dotted line)

Figure 3.18 The Probability Density of the Weekly Grocery Trip as a Gaussian Wave Packet with the Shift Towards More Frequent Trips with Extended Shopping Hours

Figure 3.19 The Higher Frequency Shift with the Extension of Shopping Hours in the Sydney Project from Regulated Hours in 1988/89 to Deregulated Hours in 1996/98

Figure 3.20 The Theoretical and Empirical Frequency Distributions for MarketPlace Leichhardt (8/12/88A) and Bankstown Square (23/3/89M)

Figure 3.21 Bessel Functions of Order Zero and One

Figure 3.22 The Relationship between the Gravity Coefficient and Mean Trip Post Office Distance in the Sydney 1980/82, 1988/89 and 1996/98 Data Sets

Figure 3.23 The Relationship between Retail Floorspace and the Number of Retail Destinations: Y = -0.811+ 0.313X , R 2 = 0.92

Figure 3.24 The Aggregate Household Time ‘Doughnut’ for MarketPlace Leichhardt 8/12/88A through an Extension of Trading Hours from 49.5 to 60 hours per week

Figure 3.25 A Venn Diagram showing Set Relationships among Selected Random Processes

Figure 3.26 The Shopping Trip Pattern for Space-discounting Behaviour (α = 2) compared to Non-Space-discounting Behaviour (α = 1.3)

Figure 3.27 State Transition Flow Chart for Trip-chaining from a Residential State to n Shopping Centres

Figure 3.28 The Poisson Distribution for n Shopping States Illustrations xv

Figure 3.29 The Erlangian Distribution for N Stops over a Distance D

Figure 4.1 The Location of the Planned Shopping Centres used in the Sydney Project, Australia with the Period or Year of Sampling

Figure 4.2 Sample Questionnaire used in the Sydney, Armidale and Auckland Projects

Figure 4.3 An Example of the Segmentation of One Kilometre Concentric Aggregation Bands from MarketPlace Leichhardt, Sydney Project 1996/98, where the Respondents Pointed to which Band their Residence is Located

Figure 4.4 Regression between Postcode Centroid Distance and Segment Nominated Distance (Aggregate Sydney Project 1996/98)

Figure 4.5 Box and Whisker Plots for Trip Distance at Bankstown Square for Equivalent Morning (top) and Afternoon Samples (bottom): 1989, 1997 and 1998

Figure 4.6 Box and Whisker Plots for Trip Frequency at the Regional PSC Bankstown Square for Equivalent Morning (top) and Afternoon (bottom) Samples: 1989, 1997 and 1998

Figure 4.7 Box and Whisker Plots for Trip Frequency at the Community PSC Ashfield Mall for Equivalent Morning 1989 and 1998 (top) and Afternoon Samples: 1989, 1997 and 1998 (bottom)

, Figure 4.8 Box and Whisker Plots for Total Populations Perception of the Level of Shopping Satisfaction for 1988/89 and 1996/98

Figure 4.9 Box and Whisker Plots for Equivalent Samples of the Time Spent Shopping (Duration/min/trip) at Pre-Christmas Westfield Burwood Afternoon Samples: 1988, 1996 and 1997

Figure 4.10 Box and Whisker Plots for Total Samples of Shops Visited per trip over the Decade from 1988/89 to 1996/98

Figure 4.11 Box and Whisker Plots for Total Samples of the Changes in the Socio-economic Index over the Decade from 1988/89 to 1996/98

Figure 4.12 Regressions for the Time-space Convergence (β = k 2/M) for the 1980/82, 1988/89 and 1996/98 Data Sets for the Sydney Project

Figure 4.13 The Stages in the Evolution of a New Definition of the Gravity Coefficient (using 1988/89 and 1996/98 data): First Approximation Regresses the Deterministic and Probabilistic Forms of the Transport Coefficient M; Second Approximation Regresses a Revised and Standardised Form of M Eliminating the xvi Illustrations

Autocorrelation; Third Approximation Corrects the Double Counting in the Deterministic Form of M

Figure 4.14 Regression between Raw (top, R^2 = 0.666) and Standardised (bottom, R^2 = 0.84 ) Probabilistic Forms of the Gravity Coefficient. The Dotted Lines are the 95% Confidence Lines from the True Mean of the Regression

Figure 4.15 Aggregate Transfer Mobility (M) with Centre Size (Number of Shopping Destinations) for the 1988/89 Sydney Data Set

Figure 4.16 Aggregate Transfer Mobility (M) with Centre Size (Number of Shopping Destinations) for the 1996/98 Sydney Data Set

Figure 4.17 The Aggregate Curve for Trading Hours Regulated (1980/82 and 1988/89; 15 samples) and Deregulated (1996/98; 17 samples) Data from the Sydney Project

Figure 4.18 (top left) Quadratic Regression of Population Index and Distance Decay from Bankstown Square 3/11/80 (1km Bands); (top right) Log-Linear Regression of Population Index and Distance Decay from Bankstown Square 3/11/80 (1km bands); (bottom left) Quadratic Regression of Population Distance and Distance Decay from Bankstown Square 3/11/80 (1km bands); (bottom right) Quadratic Regression of Population Density and Distance Decay from Bankstown Square 3/11/80 (1.5km bands)

Figure 4.19 The Gravity Trip Distribution for Ashfield Mall 23/3/98 (Afternoon) for (left) 10 Zones with DW-statistic of 0.758 and (right) 9 Zones with an Improved DW-statistic of 1.380

Figure 4.20 The Park Test Regressing the Logarithm of Frequency Variance and Trip Distance and Testing for Significance

Figure 4.21 The Bankstown Square 1997 Afternoon Gravity Regression (left) Including and (right) Excluding the 13.5km Point

Figure 4.22 Quadratic Regression of the Gravity Coefficient-Trading Hour Hypothesis for the Sydney Project 1988/89

Figure 4.23 Quadratic Regression of the Gravity Coefficient-Trading Hour Hypothesis for the Sydney Project 1996/98

Figure 4.24 Three Dimensional Contour Model of Changing Time-space Behaviour at the Community Centre MarketPlace Leichhardt, from the Regulated 49.5 hours in 1988 to a Supply Average of 64.7 hours in 1998 Figure 4.25 Three Dimensional Contour Model of Changing Time-space Behaviour at the Regional Centre Bankstown Square from the Regulated 49.5 hours in 1989 to a Supply Average of 61.6 hours in 1998 Illustrations xvii

Figure 4.26 Linear Regression between (p/2T) and (mw/2T) for (left) the 1996/98 Sydney Data Set and for (right) the Aggregate Sydney Data Set (1988/89 and 1996/98) Figure 4.27 Linear Regression between Inter-location Trip Frequency ( f ) and Intra-centre Frequency k, ( = Mkf ), for (left) 1988/89 and (right) 1996/98 Sydney Data Sets

Figure 4.28 Linear Regression for MPS and Trip Frequency, Sydney Project 1988/89, 1996/98 and Aggregate Regression

Figure 4.29 Quadratic Regression for Percentage of MPS and Centre Scale (Number of Destinations), Sydney Project 1988/89, 1996/98 and Aggregate Regression

Figure 4.30 The Relationship between the Percentage of MPS and HDI Respondents in the Sydney Project (1988/89 and 1996/98)

Figure 4.31 The Relationship between the Percentage of MPS and HDI × Trip Frequency Squared (per week) in the Sydney Project (1988/89 and 1996/98)

Figure 4.32 The Regression of the Gravity Coefficient β and U2N2/MD2 for the Sydney Project Excluding Bankstown Square Data (left) 1988/89 and (right) 1996/98

Figure 4.33 The Location of Armidale, New South Wales

Figure 4.34 The Scatter Plot of the Five Armidale 1995 Samples Compared to Sydney 1980/82, 1988/89 and 1996/98 Regressions of β = k2 /M

Figure 4.35 The Scatter Plot of the Five Armidale 1995 Samples Compared to Sydney 1988/89 and 1996/98 Regressions of f =± Mk

Figure 4.36 The Scatter Plot of the Five Armidale 1995 Samples Compared to the Aggregated 1988/89 and 1996/98 Regression of p/2T = m × k/2T for Sydney

Figure 4.37 The Scatter Plot of the Five Armidale 1995 Samples Compared to Sydney 1988/89 and 1996/98 Aggregate Regression of MPS = h k

Figure 4.38 Quadratic Regression of MPS Percentage and Centre Destinations showing the Armidale Samples as Positive Type 2 MPS

Figure 4.39 The Scatter Plot of the Five Armidale 1995 Samples Compared to the Sydney 1988/89 and 1996/98 Regression of MPS = HDI × k2 xviii Illustrations

Figure 4.40 The Scatter Plot of the Five Armidale 1995 Samples Compared to Sydney 1988/89 and 1996/98 Regressions of U= D× k / N

Figure 4.41 Location Map of the Three Planned Shopping Centres Sampled in the Auckland Survey, Thursday April 6, 2000

Figure 4.42 The Scatter Plot of the Six Auckland 2000 Samples Compared to Sydney 1980/82, 1988/89 and 1996/98 Regressions of β = k2 /M

Figure 4.43 The Scatter Plot of the Six Auckland 2000 Samples Compared to Sydney 1988/89 and 1996/98 Regressions of f = ± Mk

Figure 4.44 The Scatter Plot of the Six Auckland 2000 Samples Compared to Sydney 1988/89 and 1996/98 Regressions of p/2T = m × k/2T

Figure 4.45 The Scatter Plot of the Six Auckland 2000 Samples Compared to the Aggregate Sydney 1988/89 and 1996/98 Regressions of MPS = h k

Figure 4.46 The Scatter Plot of the Six Auckland 2000 Samples Compared to Sydney 1988/89 and 1996/98 Regressions of MPS = HDI × k2

Figure 4.47 The Scatter Plot of the Six Auckland 2000 Samples Compared to Sydney 1988/89 and 1996/98 Regressions of U= D × k/N

Figure 4.48 The Regression of Sydney 1980/82, 1988/89 and 1996/98, Armidale 1995 and Auckland 2000 for β = k2/M

Figure 4.49 (top) The Aggregate Regression of Sydney 1988/89, Armidale 1995, Sydney 1996/98 and Auckland 2000 of f = ± Mk (bottom) Post-1993 Extended Hours Data (Excluding Sydney 1988/89 Points) (28 Samples) showing an Improved R-squared Value of 0.60

Figure 4.50 The Regression of Sydney 1988/89, Armidale 1995, Sydney 1996/98 and Auckland 2000 of p/2T = m × k/2T

Figure 4.51 The Regression of Sydney 1988/89, Armidale 1995, Sydney 1996/98 and Auckland 2000 for MPS = h k

Figure 4.52 The Regression of Sydney 1988/89, Armidale 1995, Sydney 1996/98 and Auckland 2000 for MPS = HDI × k2

Figure 4.53 The Regression of Sydney 1988/89, Armidale 1995, Sydney 1996/98 and Auckland 2000 for U= D × k/N

Figure 5.1 The Location of the hepnrc.hep.net.gif Monitoring Site (top) and the Remote Hosts (bottom)

Illustrations xix

Figure 5.2 (top) A Range of Possible Time-space Distributions that could apply to Internet Demand are Simulated for β = 0.0001, T = 24 hours, x0 = 0 to x0 = 10,000 km and a scaled φo max = 10 for a Sequence of k Values where k =0.1, 0.2,... 1.0 (bottom). A Three-dimensional Plot Visualising a likely form of the Demand Wave for k= 0.1 (Baker, 2001)

Figure 5.3 The Equal Likelihood of J umping Forwards in Time to Sites in Auckland or Backwards to Perth from the ith Sydney Site defines the Underpinnings of the Type of Differential Equations in Equations (5.16) to (5.18) (Baker, 2001)

Figure 5.4 Contour Density Plot for a Simulation of the RASTT Model with k = 0.385 (left) and a 3-Dimensional Plot (right)

Figure 5.5 The Ping Time Distribution for Traffic in the hepnrc.hep.net.gif Network for 2000

Figure 5.6 The Internet Traffic Wave for hepnrc.hep.net.gif using Packet Loss Averages for 2000. The Range 0 to 168 hours represents Monday to Sunday. The US Origin-destination Pairs (-1300 W to -60 0 W Longitudes) show the Capacity to handle Peak Traffic Times with Small Amplitudes. This is not the Case with Connections to Europe and Asia

Figure 5.7 The Cumulative Frequency of Traffic Contributions from Successive 75km Zones from the hepnrc.hep.net.gif Monitoring Site Relative to Latency Periods 5-15ms to 75-85ms.

Figure 5.8 (top) The Gravity Model for Traffic Densities for the hepnrc.hep.net.gif Monitoring Site in 2000 for Average Ping Times for Latencies less than 25ms; (bottom left) the Three-dimensional of φ =A exp-0.005D sin 0.12t showing Time Gaussian Behaviour over 24-hours and Two-Dimensional Density Plot (bottom right)

Figure 5.9 (top left) The Gravity Model for Traffic Densities for the hepnrc.hep.net.gif Monitoring Site in 2000 for Average Ping Times 5-15ms; (top right) the Three-dimensional Plot for 24 hours for φ =58 exp-0.015 D sin 0.208 t; (bottom left) the Three-dimensional Plot for 168 hours for φ =58 exp-0.015 D sin 0.208 t; and (bottom right) a Density Plot

Figure 5.10 (top left) The Gravity Model for Traffic Densities for the hepnrc.hep.net.gif Monitoring Site in 2000 for Average Ping Times 15-25ms; (top right) the Three-dimensional Plot for 24 hours for φ =6.9 exp-0.004 D sin 0.108 t; (bottom left) the Three-dimensional plot for 168 hours for φ =6.9 exp- 0.004 D sin 0.108 t; and (bottom right) a Contour Density Plot xx Illustrations

Figure 5.11 The Linear Regression between Latency ∆t and Distance Range ∆x showing Time Gaussian Behaviour for the hepnrc.hep.net.gif Site

Figure 5.12 The Plot of Periodic Pairs for the hepnrc.hep.net.gif Site Relative to the Phase Line Standardised to the Earth’s Rotation

Figure 6.1 (top left) Frequency Distributions for Number of Shops Visited; (top right) Total Walking Distance (compared to a normal distribution); (bottom left) Maximum Walking Distance (compared to a normal distribution); and (bottom right) Time Spent Shopping (compared to a normal distribution)

Figure 6.2 Log-linear Regression of Total Walking Distance from Carparks in the Tracking Armidale 1995 Data Set

Figure 6.3 The Location of Sample Centres in the New South Wales Retail Hierarchy

Figure 6.4 The Regression of UCL 1996 Population and Population Change (1996-2001) for the Selected Case Studies

Figure 6.5 Main Street Mayfield showing a Vacant Shop, Financial Planner and Pawnbroker forming a Sequence of Shops in what was Prime Retail Space a Decade earlier. The former Cake Shop, now Vacant, offers the First Three Months Rent Free (notice on the door)

Figure 6.6 The Redeveloped Supermarket Site in the Centre of Main Street Mayfield showing the Current Tenants as the Salvation Army Second Hand Shop and a $2 Shop

Figure 6.7 Extra Employees per One Hundred Thousand Dollars of Turnover, NSW

Figure 6.8 The Collapse of Rental Income from Prime Retail Properties surrounding Regional Shopping Centres (1990 to 1995)

Figure 6.9 Vacant Shops in Sheffield in 1999, nine years after the Opening of the Meadowhall Regional Shopping Centre

Figure 6.10 Vacant Shops in Morley, Leeds in 1999, 12 months after the Opening of the White Rose Centre, 3km away

List of Tables

Table 1.1 Classification of Planned Shopping Centres

Table 1.2 Proposed Floorspace Assignments for the Proposed Woolworths Supermarket, Inverell, New South Wales 2000

Table 1.3 Change in Independent Supermarkets and Food Specialty Stores in Australia 1992-1999

Table 2.1 Survey of Tenancy Changes in Local Centres, Canberra, 1998

Table 2.2 Shopping Centres used in the Canberra Household Shopping Preference Survey 1996, 1997

Table 2.3 A Review of the Important Attributes Considered in Studies of Consumer Behaviour

Table 2.4 An Individual’s Assessment of their Level of Shopping Satisfaction (Utility)

Table 2.5 Characteristics of ‘Small’ and ‘Large’ Centre Behaviour

Table 2.6 Source Matrix of Literature Associations for Multi-purpose Shopping

Table 3.1 A Classification of Relevant Equations using Space-Time Operators

Table 3.2 Time-space Characteristics of Three Malls in the Sydney 1988/89 Data Set

Table 3.3 Various Estimates of the Intra-centre Shopping Frequency (per week) in the Sydney 1988/89 Data Set

Table 3.4 SE and FA Estimates of Mean Trading Hours for the 1988/89 Sydney Data Set

Table 4.1 Occupation Weighting for the Index of Disposable Income (IDI)

Table 4.2 Number of Retail Outlets for Sampled PSCs Sydney Project 1980/82, 1988/89 and 1996/98

Table 4.3 Sample Sizes, Sydney Project, Equivalent Time Samples (shaded) for 1988/89 and 1996/98

Table 4.4 Trip Distance Comparison, Sydney Project, Equivalent Samples (shaded) for 1988/89 and 1996/98

xxii Illustrations

Table 4.5 Mean Trip Frequency Comparison, Sydney Project, Equivalent Samples (shaded) and Pre-Christmas Samples (bold) for 1988/89 and 1996/98

Table 4.6 Comparison in the Mean Level of Shopping Satisfaction, Sydney Project, Equivalent Samples (shaded) and Pre-Christmas Samples (bold) for 1988/89 and 1996/98

Table 4.7 The Percentage of Samples in the 1988/89 and 1996/98 Data Sets (The Westfield Chatswood Sample in 1989 of 17.61 is included in the Mean Shopping Satisfaction, but not in 1996/98 because of refurbishment.)

Table 4.8 Comparison in the Mean Shopping Time, Sydney Project for Equivalent Samples (shaded) and Pre-Christmas Samples (bold) for 1988/89 and 1996/98. The Total Time Spent Shopping (Frequency × Time Spent Shopping per trip) per week is in brackets (the asterisk describes the Mann-Whitney Significance at the 0.05 Level)

Table 4.9 Comparison in the Mean Shops Visited for 1988/89 and 1996/98 in the Sydney Project, Equivalent Samples (shaded) and Pre-Christmas Samples (bold)

Table 4.10 Comparison in the Socio-economic Index for 1988/89 and 1996/98 in the Sydney Project, Equivalent Samples (shaded) and Pre-Christmas Samples (bold)

Table 4.11 A Summary of Occupational Types from Respondents from Sydney Project (Ordinary Font, 1988/89; Bold Font 1996 or 1997; Bold Italic Font 1997 and 1998 Samples)

Table 4.12 Summary of Nature of Trip Purpose from Respondents from Sydney Project (Ordinary Font, 1988/89; Bold Font 1996 or 1997; Bold Italic Font 1997 and 1998 Samples)

Table 4.13 Changing Behaviour from Particular Trip Purpose and Socio- economic Groups from 1988/89 to 1996/98 (1988/89 Samples, Normal Font; 1996/98 Samples, Bold Font) (HI High Income; LI Low Income, according to the Index of Disposable Income)

Table 4.14 Population Index and Density Assignments for 1km Bands: Bankstown Square Morning Sample, 3/11/1980

Table 4.15 The Estimation of the Gravity Coefficient for Bankstown Square using Two Assignment Procedures (DW = Durban-Watson Statistic); BSM- Bankstown Square Morning Sample, BSA- Bankstown Square Afternoon Sample

Illustrations xxiii

Table 4.16 Mean Trip Frequency and Variance per Concentric Band: Sydney 1996/98 Data Set

Table 4.17 Comparison in Variance in Trip Frequency for 1988/89 and 1996/98 Samples in the Sydney Project for Equivalent Samples (shaded) and Pre- Christmas Samples (bold)

Table 4.18 Comparison in the Spatial and Time-based (in brackets) Gravity Coefficients for 1988/89 and 1996/98 in the Sydney Project, Equivalent Samples (shaded) and Pre-Christmas Samples (bold)

Table 4.19 The Sample Sizes in the Armidale Survey, November 1995

Table 4.20 Comparison between Armidale LGA, Primary Trade Area and NSW LGA Averages for Selected Socio-economic Groups, 1991

Table 4.21 Population Change in Armidale and Region 1991–1996

Table 4.22 Numbers of Students Enrolled at Armidale Campus of UNE

Table 4.23 Full-time Equivalent Staff Numbers at UNE, Armidale

Table 4.24 Summary of Statistics of Armidale Surveys (bold), November 1995, Compared to the Sydney Project 1996/98

Table 4.25 Sample Sizes of Centre Surveys, Auckland, April 6, 2000

Table 4.26 Summary Statistics of Auckland Surveys, November 1995, Compared to the Sydney Project 1996/98

Table 4.27 Comparison between Auckland Samples with Selected Sydney Samples in 1997/98 taken on the Thursday before Easter

Table 6.1 Walking Distance Statistics from Armidale Carparks

Table 6.2 Selected Case Studies of Retail Vacant Shops in Non- metropolitan New South Wales. All Elements not stated as CBD Values are Main Streets only

Table 6.3 Multi-purpose Retail Functions within Oberon Retail Establishments

Table 6.4 Changes in Landuse of Retail Establishments in Oberon

Table 6.5 List of Shop Closures and New Businesses within 18 Months of the Major Supermarket Re-locating to an Edge-of-centre Site, Mayfield NSW 1995-1997 xxiv Illustrations

Table 6.6 Comparison in Retail Employment Statistics between Australian States 1995-2000

Table 6.7 Ratio of Full-time to Part-time Employment for NSW and WA, 1980 and 1992

Table 6.8 NSW Employment and Retail Structural Ratios

Table 6.9 Changes in Employment Structure at Coles Supermarkets, ACT 1992-95

Table 6.10 Changes in Employment Structures at Coles, Armidale

Table 6.11 Floorspace Equivalent per Person Employed for NSW and Queensland

Table 6.12 Westfield Marion, Adelaide Floorspace and Turnover Statistics compared to Westfield Carindale, Brisbane (both regulated hours and no Sunday trading) and Westfield Miranda and Parramatta, Sydney (deregulated hours) for 1999 and 2001

CHAPTER 1

Introduction

1.1 Shopping Change

Shopping is an essential part of our day-to-day existence and the organization of retailing has implications for every household. The Concise Oxford Dictionary defines ‘shopping’ as ‘goods purchased in shops’ and ‘retail’ as ‘the sale of goods in small quantities direct to consumers’ (Brown, 1992a). These definitions are suggestive of the interdependence between ‘shopping’ and ‘retailing’, where the former is ‘consumer-based’ and the latter ‘producer-based’. Shopping is underpinned by connecting consumer demand to producer supply and the mechanism for this exchange is the shopping trip. This implies the importance of flows of consumers through a time-space fabric of shopping opportunities. Conversely, the nature of retailing is producer-focused, encompassing the ways in which firms interact with each other and sell goods to consumers. Whether the dynamics of this demand and supply interaction is driven by consumer or producer sovereignty is a hotly contested issue. Nevertheless, it is the thesis here that consumer trips to and within shopping centres are what maintain their viability, much like the analogy of blood-flows to different organs of the human body. If flows are stopped or reduced, precincts die. It is for this reason that it is important to study the dynamics of these consumer flows at different places, times and scales so that retail planning can be pro-active rather than reactive to sustain healthy shopping precincts.

The nature of shopping is not only an integral part of our society, but also an indicator of the rapid societal changes in the socio-economics of households and technological innovations. Consumers can access the Internet at home to shop globally, penetrating national borders or different time zones. The idea of the traditional neighbourhood shopping centre has been replaced, in part, by the one- stop shop at supermarkets, hypermarkets or planned shopping centres (Figure 1.1). The result has been substantial structural change in landuse patterns within retail hierarchies over a short time period.

Town centres that were once vibrant have been replaced by chains of vacant shops and urban obsolescence. The decline of town centres and neighbourhood stores have therefore been one of the major ramifications of this new global retail order where consumers have fundamentally changed ‘when’ and ‘where’ they undertake their shopping trips. The reasons for this retail ‘shock wave’ enveloping many centres are complex, but could involve an interplay of socio-economic change with household mobility and expenditure, technological change in the nature of flows and regulatory change in the ideology of public policy. Any one or combinations of these factors can fundamentally affect the nature of consumer flows and the health of precincts in retail hierarchies.

1 2 Chapter 1

The main function of retailing is to act as an intermediary between the consumer on one hand and the producer or wholesaler on the other, in the physical distribution of particular goods or services. Such a perspective requires some qualifications. Retailers are now not only fundamental to the buying, selling, storage and delivery of goods, but are increasingly becoming multi-range and multi-purpose in what they offer, and when and where they offer them. Consequently, there has been a fundamental change, from retailers originally selling specialised merchandise, to a multiplicity of goods and services. The supermarket is very much part of this transformation. It now offers a diversity of food and non-food merchandise within its floorspace. In other words, the supermarket has become a shopping centre within a shopping centre. Consumers can now conduct their banking, buy a book or flowers on the floorspace of a supermarket, twenty-four hours-a-day, seven-days-a-week. This structural change should have implications for the nature of the multi-purpose trip.

A particular dynamic model will be used extensively in this study (the so-called retail aggregate space-time trip or RASTT model) because it can be applied to a range of trip configurations, from local trips to a neighbourhood centre to global flows of Internet traffic. The model will provide a unifying structure to the analysis because it is process-based rather than subject-based and therefore it has the potential to look at retail change over different scales of space and time. The RASTT model was developed back in the mid-1980s and was a key component of the ‘Storewars’ debates on retail trading hours and location of floorspace in the 1990s within Australia. ‘Storewars’ is a term coined to describe the process of powerful retail corporations taking market share of smaller traders using various market strategies and public policy positions (Wrigley, 1994). Particular attention in this study will focus on changes in retail locations and shopping times. The RASTT model has therefore been intimately linked in public policy, by showing the predicted affects and accrued advantages to large retailers of changing the length and structure of the shopping week (Baker, 1994b) and by locating retail developments away from town centres (Baker, 1995). This process will be examined in this study, showing the link between a model and its policy implications.

Many of the issues that became globally significant in the 1990s (such as out-of- town retail developments) are not new. Indeed, such issues have been recorded within antiquity. For example, in ancient Rome, Claudius had asked consuls to license him to hold a market on his estates (Sherwin-White, 1966). Other cases sometimes led to protests. In Book V, Letters of the Younger Pliny (Radice, 1963, 137-138), in a letter written in 105AD to Julius Valarianus, Pliny records:

A praetorian senator named Sollers asked the Senate for permission to hold a weekly market on his property. This was opposed by representatives of the town of Vicetia, with Tuscilius Nominatus acting on their behalf, and the case was adjourned until a subsequent meeting of the Senate.

The issue was whether markets held weekly on private land could compete with the public daily markets in forums (Figure 1.2) and thus change the nature of consumer flows. Ancient Rome gave the Senate the ability to adjudicate on this issue because Introduction 3 of questions of public order (Sherwin-White, 1966, 319). Such a basis of order still underpins Planning Acts of Parliament today. Further, this example shows time- space retail developments have been a source of debate not just peculiar to 21st century policy. The negative idea of retail change and loss of market share was the same for the representatives of Vicetia as for small business in places such as the UK, Australia and the US. The location of shopping destinations and their time of trading was crucial then as it is now.

Figure 1.1 Meadowhall Regional Planned Shopping Centre, Sheffield, 1999

Figure 1.2 The Macellum (Rome’s shopping mall) on the Dupondius coin (AD 65) in the reign of the Emperor Nero 4 Chapter 1

For space, classical economics often avoids the ‘where’ question for the location of economic activity, yet this has a major bearing on the cost and efficiency of distribution. The space variable is also central to any analysis of a retail market and to avoid it is to present an irrelevant view of retailing. This is why a geographic approach is just as important as an econometric approach to retail analysis. Space has many meanings, ranging from the perception of space to the mathematical conception of space. We will view space largely in a narrow physical context and our variables will be measurable through various survey methods. This is not to say that the perception of space is not important to retailing. Stilwell (1994) argued that a key aspect for political economists is the interaction of space with society. Space is seen as a social product and does not occur randomly. Since the 1970s, human geographers have dwelt on what constructs determine ‘place’ and what the relationship is between ‘space’ and ‘place’. Is place just personalised space and is it diametrically opposed to the freedom of space? Such a sense of place is important, particularly to the design of shopping centres. People have a simultaneous need for security of place and freedom of space and this seems to be part of the contradictory dualism in the human condition. People’s shopping activity is not solely price-driven (as neo-classical economics tells us) so an understanding of these issues is important to retail analysis but not necessarily to this study.

Time paths very much underpin the evolution of the movement theory of this study and follows on from the conception of time geography developed by Hagerstrand (1970) and the Lund School in the 1970s. Time geography assigns an individual’s activities in a 24-hour day and ‘travel’ as a continuous temporal sequence of activities within geographical space (Kwan, 1999). The location and trading hours of a shop are time-space constraints on these activities and their regulation limits the person’s freedom of choice. There is a time supply and demand for shopping opportunities and the trade-off, at an aggregate level, depends on societal priorities. This allocation of shopping time opportunities affects accessibility and the viability of location. Consequently, there is a strong link between shopping times and retail landuse. This connection has often been missed in retail policy, but it is an important corollary of dynamic trip modelling.

A further important question to arise in large scale spatial interaction is whether technological advances in production and transportation have lead to a situation whereby the constraints imposed by distance on economic life tend to evaporate over time. Karl Marx made the prediction of ‘the annihilation of space by time’ (Stilwell, 1994). Is global access of the Internet the ultimate mechanism for the annihilation of the ‘tyranny of distance’? This is a common view of the Internet and its growth highlights the rapid changes in technology and communication, restructuring the nature of shopping on a world scale. This is leading to the acceleration, in a ‘time- space compression’, of retail form and process (Janelle, 1968; Baker, 2002). Shopping and consumption are just a click away on a computer screen. Yet there is still stability in traditional time-space shopping relationships. For example, despite technological change and deregulation, a majority of people still shop once a week at

Introduction 5 a supermarket and a substantial number at their nearest shopping centre. What provides the ‘glue’ for this stability?

The reasons for this are difficult to disentangle from the tapestry of human behaviour. One ingredient that we could measure is the frequency of spatial interaction. The very concept of place and the human identification with specific spaces, depends on a certain repetitiveness. The frequency of shopping trips can be an indicator of store or centre loyalty. A person who shops at a particular centre regularly can be counted in a quantitative assessment. Further, the definition of a place depends on certain predictability in behaviour. Frequency, therefore, not only underpins place but also can be assigned a number and can be used for prediction. It is argued here to be an essential ingredient in the realisation and reproduction of space and time (and place). This connectivity is a fundamental part of the time-space model and we use frequency to solve the assignment problem of ‘where’ and ‘when’, on average, people shop. Changes in space and time in the differential equation (the engine driving the RASTT model) are equal to the trip frequency. ‘How often’ is therefore a proxy for understanding the nature of trip behaviour.

Another ingredient is the mobility underpinnings of socio-economic groups. Elderly consumers will access shopping opportunities by adapting distance minimisation trips to reduce the total effort in shopping. Likewise, women with young children may prefer smaller shopping centres rather the regional planned shopping centres with parking and access problems. Their aim would also be to minimise walking distance. Such groups are more likely to shop in the mornings than the afternoons. Conversely, ‘time poor’ consumers, such as ‘professionals’ in two income households, prefer to shop in the afternoons and evenings, particularly on the way home from work. These households provide instability in the time-space relationship because of the fluidity in their time budgets. They are more likely to shop on the Internet and are the targeted market for retailers in the evolution of the new retail order of shopping. Therefore, if such groups are substantially present in samples, then they would have a strong influence on the time-space characteristics of the shopping distributions. Likewise, the timing of the sample and its location would also affect these distributions, but this could also be autocorrelated with socio- economics. Therefore, the socio-economic mix of the sample could be a major factor in determining the ‘when’, where’ and ‘how often’ shopping occurs in aggregate distributions.

The blood supply exchanging oxygen, carbon dioxide, energy and waste with the organs of the body provides an analogy to the flow of people from an origin to a destination. When shopping, people exchange money for goods and services and help maintain viable retailers and healthy shopping precincts. This exchange occurs (like the different organs of the body) at a range of retail forms, such as the supermarket or discount department store. Therefore, to understand the flows of consumption we need to understand the structure of these retail units at the points of exchange and how these are changing over time.

6 Chapter 1

1.2 Definitions of Retail Forms Underpinning the Model

1.2.1 PLANNED SHOPPING CENTRES

Planned shopping centres (PSCs) are a consequence of the rise in household mobility through increasing car ownership and the decreasing attractiveness and accessibility of the Central Business District (CBD) of large cities (due to such factors as, increasing congestion, changing population characteristics and suburban sprawl). They are synonymous with carparks and one-stop shopping (Figure 1.1). These centres cater not only for daily needs, but also for higher order specialised goods and services for a large segment of the urban population. They represent a quantum jump in retail capitalism, replacing traditional shopping centres that evolved from central places in previous centuries. Shopping functions have devolved from the central business district (CBD) of large cities to the suburbs, and this suburbanisation of retailing has placed significant stress on Local Government in their location planning for new retail capacity. PSCs are designed to operate as a unit at a single time to meet the trading requirements of a specific trade area under single management. They are a major generator of trips to satisfy consumer demand and are the basis for much of the dynamic trip modelling in this study.

Reynolds (1993) points out that there is no uniform definition of a planned shopping centre. For example, PSCs in the UK, typically, have over 4 650 sq m of gross leasable floor space (GLFS) with carparking provision and three or more retail units. However, in Germany, this area is set with a GLFS of 15,000 sq m and is defined by several factors, namely, a spatial concentration of specialist non-food, food or service outlets of various sizes; a number of smaller specialist outlets in combination, as a general rule, with one or more dominant operators; a large shared parking area; a central management; and a set of common functions (such as marketing and publicity). Within Australia, planned centres have a lower floorspace boundary, similar to the British definition, and they can be classified broadly into a retail hierarchy based on floorspace and the number of anchors (Table 1.1). Traditionally, within such larger PSCs in the US, there is, typically, a large department store, which is the major attractor of trade, and smaller specialty and convenience outlets that share the trade attracted by the larger stores. In Australia (and some centres in the UK and Canada), supermarkets are also major anchors and this difference in the mix of anchors underpins what type of model is selected to study trips to the PSC.

Planned shopping centres (PSCs) were a retail innovation arriving from the US to Australia in the 1960s (with Roselands opening in Sydney in 1966). Stimson (1985) states that, in general, the retail industry in Australia has been adaptive of innovations that have occurred overseas, especially from North America. Furthermore, he argues that it is likely that the pressures within the industry to adopt increasingly frequent innovations in the industry will become greater as market share competition becomes more aggressively pursued by ownership entities in the industry. However, PSCs were more slow to spread to the UK, because of what Reynolds (1993) described as,

Introduction 7 the most rigorous and restrictive land-use planning systems. This changed dramatically after 1985 when 6.6 million sq m of floorspace was added to UK stock during 1986- 92 (Reynolds, 1993). Likewise, shopping hour liberalisation occurred at around the same time allowing out-of-town locations to be time accessible to consumers and therefore profitable. Deregulation therefore accelerated structural change and allowed for the proliferation of PSCs as a result of a parallel shift from regulatory to deregulatory policy for both space and time.

The RASTT model fundamentally looks at trips to and from a planned shopping centre (or mall). The PSC is considered to be, theoretically, a point density of demand where these trips are focused in space and time. They converge at the mall because of the benefits of shopping there in terms of time, effort and choice in agglomerating shopping opportunities under the one roof. PSCs offer one-stop shopping and this is increasingly attractive to mobile and ‘time-poor’ households. Why visit ten shops in a chain of visitations when they can be grouped together in one trip and accessible from a centralised carpark? The assumptions of the RASTT model (such as, nearest neighbour shopping and continuity of demand) are quite defensible at planned shopping centres because of the agglomeration of supply points and the convergence of demand. Further, each centre can be defined by common trading hours, although in Australia it has been found that the major super- market chains have traded substantially outside the hours set for the other retailers. Planned shopping centres are therefore very amenable for modelling because they allow for partitioning demand in a defined space and time with realistic assumptions.

1.2.2 DEPARTMENT STORES

The department store, first developed in Europe in the mid-nineteenth century (see Adburgham, 1981), was popularised in Australia by consumers of the 1920s requiring the choice of an extended range of quality stock with the aura of European and North American fashion and culture. It was the market for overseas and upmarket fashion, where the latest trends from the continent could be found on the clothing racks of the department store. By combining a whole range of retail functions under the one roof, it was the forerunner of the planned shopping centre. Convenience, quality and choice were the trademarks of these stores. Each had their own cafeteria for shoppers to sit and discuss what they had observed shopping and were self-contained in terms of credit and service.

Yet one of the major casualties of attitudinal and socio-economic change in household shopping behaviour has been the decline of this form of retailing. Consumers had a preference in the 1990s for shopping in specialty shops for quality, discount department stores for price and ‘category killers’ for choice. Category killers are an interesting creation from the US where the product choice, relative to an increase in floorspace is maximised. For example, a department store may allocate 1,000 sq m and offer 3,000 lines, but a category killer would have 3,000 sq m with 15,000 items saturating the consumer with choice. This decline of the department store was a consequence of the time constraint of households, and their budgets were

8 Chapter 1 more inclined to rank competitive prices ahead of the latest fashions. There were more women in the workforce and more single parent families and less time for shopping. Consumers wanted quality but at lower prices. They were more cynical, discerning and less loyal in their shopping choices. Department stores have had to restructure quickly with such rapid change in household socio-economics and taste. Low demand categories such as dress fabrics, carpets, hardware and food have been discontinued and stores have targeted fashion apparel, homewares and leisure.

Discount department stores and supermarkets expanding their product range have made considerable in-roads into department store turnover. Their strategy of quality merchandise at lower prices is what the market is currently demanding. Brand names, once only found at a department store, are now an integral part of merchandise at a discount department store. Their strategy is lower margins for higher turnover. There was once a social stigma of buying at a discount department store, but today this has evaporated. The department store has had to reposition itself.

Centre Minimum Example Floorspace Category Requirement (sq m) Super-regional Department Store, David Jones, Grace 100,000 + Two Discount Bros, Target, Big W, Department Stores, Franklins, Woolworths, Two Supermarkets, 16 Cinema Complex, Mass Entertainment 300 Specialty Shops Regional Department Store David Jones, K-mart, 45,000- and Two Woolworths, Franklins, 100,000 Supermarkets 200 Specialty Shops Sub-regional Two Discount Target, Big W, Coles, 20,000- Department Stores Chandlers, 100 45,000 and at least one Specialty Shops Supermarket Community Supermarket, K-mart, Coles, 50 12,000- Discount Specialty Shops 20,000 Department Stores or Two Supermarkets Neighbourhood Supermarket Jewels and 20 5000-12,000 Specialty Shops

Table 1.1 Classification of Planned Shopping Centres in Australia

The department store is still a major anchor of PSCs and this is one way of distinguishing US malls, in particular, because of the lack of supermarkets as co- anchors. This means that trips to malls in the US should be less periodic than in Australia (where both co-exist in PSCs). Indeed, the RASTT model allows for the definition of a distinctive type of behaviour (termed ‘space-discounting’ behaviour) where shoppers travel there to save time in satisfying demand. This is because choice is maximised by the agglomeration of shopping opportunities at the department store and the competing specialty shops. Department stores were the anchor of choice-