Paul Ormerod

Assessing the Productivity of the UK Retail Sector

Several recent comparative studies have shown a labour productivity gap in respect of UK retailing when compared with other countries, notably France and the USA. This article seeks to identify, through an overview of existing data and related research, the extent to which retail productivity in the UK compares to global competitors and attempts to reach a consensus on the factors that determine retail productivity, while highlighting common performance measures for retailers and Government to use in measuring future productivity trends. Methods employed include a review of published studies; interviews with industry participants in the UK and a small number of leading retailers in the USA; and an analysis of a specially created database of the performance of over 200 US, UK and French retail companies. The authors find that it is unwise to draw definitive conclusions from aggregate international economic analyses of the sector; that a wide variety of efficiency indicators are employed by the sector in practice; and that there are significant differences in the structure, operating and regulatory environment for retailing in the UK which impose costs on retailers that are not necessarily incurred in other countries.

Validation and Verification of Agent-Based Models in the Social Sciences

This paper considers some of the difficulties in establishing verificaction and validation of agent based models. The fact that most ABMs are solved by simulation rather than analytically blurs the distinction between validation and verification. We suggest that a clear description of the phenomena to be explained by the model and testing for the simplest possible realistic agent rules of behaviour are key to the successful validation of ABMs and will provide the strongest base for enabling model comparison and acceptance. In particular, the empirical evidence that in general agents act intuitively rather than rationally is now strong. This implies that models which assign high levels of cognition to their agents require particularly strong justification if they are to be considered valid.

Evolving social influence in large populations

Darwinian studies of collective human behaviour, which deal fluently with change and are grounded in the details of social influence among individuals, have much to offer “social” models from the physical sciences which have elegant statistical regularities. Although Darwinian evolution is often associated with selection and adaptation, “neutral” models of drift are equally relevant. Building on established neutral models, we present a general, yet highly parsimonious, stochastic model, which generates an entire family of real-world, right-skew socio-economic distributions, including exponential, winner-take-all, power law tails of varying exponents, and power laws across the whole data. The widely used Barabási and Albert (1999) Science 286: 509-512 “B-A” model of preferential attachment is a special case of this general model. In addition, the model produces the continuous turnover observed empirically within these distributions. Previous preferential attachment models have generated specific distributions with turnover using arbitrary add-on rules, but turnover is an inherent feature of our model. The model also replicates an intriguing new relationship, observed across a range of empirical studies, between the power law exponent and the proportion of data represented in the distribution.

Power law distribution of the duration and magnitude of recessions in capitalist economies: breakdown of scaling

Power law distributions of macroscopic observables are ubiquitous in both the natural and social sciences. They are indicative of correlated, cooperative phenomena between groups of interacting agents at the microscopic level. In this paper, we argue that when one is considering aggregate macroeconomic data (annual growth rates in real per capita GDP in the seventeen leading capitalist economies from 1870 through to 1994) the magnitude and duration of recessions over the business cycle do indeed follow power law like behaviour for a significant proportion of the data (demonstrating the existence of cooperative phenomena amongst economic agents). Crucially, however, there are systematic deviations from this behaviour when one considers the frequency of occurrence of large recessions. Under these circumstances the power law scaling breaks down. It is argued that it is the adaptive behaviour of the agents (their ability to recognise the changing economic environment) which modifies their cooperative behaviour.

A rapid method for assessing social versus independent interest in health issues: A case study of 'bird flu' and 'swine flu'

Effective communication strategies regarding health issues are affected by the way in which the public obtain their knowledge, particularly whether people become interested independently, or through their social networks. This is often investigated through localized ethnography or surveys. In rapidly-evolving situations, however, there may also be a need for swift, case-specific assessment as a guide to initial strategy development. With this aim, we analyze real-time online data, provided by the new Google Trends tool, concerning Internet search frequency for health-related issues. To these data we apply a simple model to characterise the effective degree of social transmission versus decisions made individually. As case examples, we explore two rapidly-evolved issues, namely the world-wide interest in avian influenza, or bird flu, in 2005, and in H1N1, or swine flu, from late April to early May 2009. The 2005 bird flu scare demonstrated almost pure imitation for two months initially, followed by a spike of independent decision that corresponded with an announcement by US president George Bush. For swine flu in 2009, imitation was the more prevalent throughout. Overall, the results show how interest in health scares can spread primarily by social means, and that engaging more independent decisions at the population scale may require a dramatic announcement to push a populace over the tipping point.

The Medieval inquisition: scale-free networks and the suppression of heresy

Qualitative evidence suggests that heresy within the medieval Church had many of the characteristics of a scale-free network. From the perspective of the Church, heresy can be seen as an infectious disease. The disease persisted for long periods of time, breaking out again even when the Church believed it to have been eradicated. A principal mechanism of heresy was through a small number of individuals with very large numbers of social contacts.

Power law distribution of the frequency of demises of US firms

Both theoretical and applied economics have a great deal to say about many aspects of the firm, but the literature on the extinctions, or demises, of firms is very sparse. We use a publicly available data base covering some 6 million firms in the US and show that the underlying statistical distribution which characterises the frequency of firm demises—the disappearances of firms as autonomous entities—is closely approximated by a power law. The exponent of the power law is, intriguingly, close to that reported in the literature on the extinction of biological species.

The US business cycle: power law scaling for interacting units with complex internal structure

In the social sciences, there is increasing evidence of the existence of power law distributions. The distribution of recessions in capitalist economies has recently been shown to follow such a distribution. The preferred explanation for this is self-organised criticality. Gene Stanley and colleagues propose an alternative, namely that power law scaling can arise from the interplay between random multiplicative growth and the complex structure of the units composing the system. This paper offers a parsimonious model of the US business cycle based on similar principles. The business cycle, along with long-term growth, is one of the two features which distinguishes capitalism from all previously existing societies. Yet, economics lacks a satisfactory theory of the cycle. The source of cycles is posited in economic theory to be a series of random shocks which are external to the system. In this model, the cycle is an internal feature of the system, arising from the level of industrial concentration of the agents and the interactions between them. The model—in contrast to existing economic theories of the cycle—accounts for the key features of output growth in the US business cycle in the 20th century.

Books average previous decade of economic misery.

For the 20th century since the Depression, we find a strong correlation between a ‘literary misery index’ derived from English language books and a moving average of the previous decade of the annual U.S. economic misery index, which is the sum of inflation and unemployment rates. We find a peak in the goodness of fit at 11 years for the moving average. The fit between the two misery indices holds when using different techniques to measure the literary misery index, and this fit is significantly better than other possible correlations with different emotion indices. To check the robustness of the results, we also analysed books written in German language and obtained very similar correlations with the German economic misery index. The results suggest that millions of books published every year average the authors shared economic experiences over the past decade.

Scaling behaviour in the number of criminal acts committed by individuals

We examine the distribution of the extent of criminal activity by individuals in two widely cited data bases. The Cambridge Study in Delinquent Development records criminal convictions amongst a group of working class youths in the UK over a 14 year period. The Pittsburgh Youth Study measures self-reported criminal acts over intervals of six months or a year in three groups of boys in the public school system in Pittsburgh, PA. The range of the data is very substantially different between these two measures of criminal activity, one of which is convictions and the other self-reported acts. However, there are many similarities between the characteristics of the data sets. A power law relationship between the frequency and rank of the number of criminal acts describes the data well in both cases, and fits the data better than an exponential relationship. Power law distributions of macroscopic observables are ubiquitous in both the natural and social sciences. They are indicative of correlated, cooperative phenomena between groups of interacting agents at the microscopic level. However, there is evidence of a bimodal distribution, again in each case. Excluding the frequency with which zero crimes are committed or reported reduces the absolute size of the estimated exponent in the power law relationship. The exponent is virtually identical in both cases. A better fit is obtained for the tail of the distribution. In other words, there appears to be a subtle deviation from straightforward power law behaviour. The description of the data when the number of boys committing or reporting zero crimes are excluded is different from that when they are included. The crucial step in the criminal progress of an individual appears to be committing the first act. Once this happens, the number of criminal acts committed by an individual can take place on all scales.