Diffusion of Innovation
The paper provides a theoretical model of technology adoption based on the idea that the diffusion of information about a technology depends both on the social structure of the adopters and their degree of assortativity.We propose a framework that - while retaining the core assumptions of epidemic diffusion models -allows for explicit modelling of the social structure via social network and of agents cultural heterogeneity via agent-based simulation. Decision-making takes place in institutional contexts where individual features trigger differentiated imitative responses and societal organization acts as medium on which information flows. Themodel simulates the diffusion of fertilizers in five Ethiopian villages (Peasant Associations), which differ in both political and relational structures and farmers belong to numerous ethnic and religious groups. Starting from survey data we run a compositional understanding simulation with the aim of reproducing observed diffusioncurves on the basis of unobserved individual interactions. By minimizing the divergence from model output and observed diffusion, the exercise of categorical calibration and time series fit identify a set of plausible parametersfor each village. Results highlight the importance of cultural dissimilarities to understand the diffusion processes.
Switching Behavior in the Electricity Market
Switching behavior and the liberalization of the Italian electricity retail market. Logistic and mixed effect Bayesian estimations of consumer choice
Submitted to the Journal of Evolutionary Economics
The paper analyzes the determinants of switching barriers in a sample of Italian households. Inertia, while depressing consumer engagement and the associated benefits, also reduces competitive pressure in the energy retail free market. After the liberalization of the energy retail market in the European Union, the observed switching rates are relatively low: Identifying the causes of consumer inertia is therefore crucial to achieve the goals set by the European Union in the Three Energy Packages. We use a logistic regression model to discover the causes of consumer inertia as listed in the Aspect of Daily Life Survey (ISTAT 2014). Namely, we take into account: (i) loyalty, (ii) lack of information, (iii) difficulties in evaluating the economic benefits deriving from switching, (iv) perception of small savings from switching and (v) concerns for a worsening of the quality of service in case of switching. The main findings relate awareness —knowledge of the possibility to change the retailer—to the level of education and to the frequency of internet use. In addition, we find that households that are different in numerosity, geographical and urban location and in economic resources tend to attribute inertia to different causes. Among these, the most frequently reported are alleged difficulties in comparing offers and concerns about lower quality of service.
Pluralism in Economics
Just another niche in the wall? How specialization is changing the face of mainstream economics, Cambridge Journal of Economics, Volume 42, Issue 2, 23 February 2018, Pages 427–451, doi.org/10.1093/cje/bex003
There is considerable discussion on so-called ‘mainstream pluralism’, that is, on the co-presence of a variety of research programmes in today’s mainstream economics that: 1. significantly deviate from the neoclassical core; 2. are pursued by different, often separate communities of researchers; and 3. have their origins outside economics. The literature tends to regard mainstream pluralism as a transitory state towards a new, post-neoclassical, mainstream. This paper advances a new interpretation: it suggests that the changing and fragmented state of mainstream economics is likely to persist over time under the impact of specialization (as a self-reinforcing mechanism) and the creation of new specialties and approaches, also through collaboration with researchers from other disciplines.
Science of Science
A bridge over troubled water. A reassessment on interdisciplinarity, novelty and impact (work in progress)
Interdisciplinary research (IDR) is increasingly seen and consequently promoted as the appropriate way to cope with the growing complexity of the objects of scientific enquiries. The US National Academies (2005) and the European Union Research Advisory Board (2004) register an increase in IDR and encourage the creation of interdisciplinary research networks, and the funding of interdisciplinary projects. Subsequently, interdisciplinarity appears as a target in the strategic plans of Universities and research centres, and ranks as a reward criterion for several funding bodies. It is therefore important to investigate the IDR to appreciate its relevance in the production of knowledge and to assess the degree to which it translates in impactful, and possibly novel, research. In spite of the abundant literature dealing with these issues, evidence is rather inconclusive. Conflicting evidence can be attributed to the concept of interdisciplinarity itself which is easy to intuit but rather difficult to measure reliably. In addition, the number of proposed measures (e.g. variety, balance, Rao-Stirling diversity score), the different levels of analysis (e.g. paper, journal, institution, and fields), the variety of bibliometric information (e.g., disciplinary memberships of authors, published journals, references) (Wagner et al. 2011) and its elaborations (co-citations and citations, cosine similarity, and combination of knowledge) make results quite difficult to interpret and to compare.The paper focuses on the meaning, properties, and the interpretation of the most acknowledged interdisciplinarity measures and highlights their relationships and limitations. It suggests several ways to overcome the latter by introducing null configuration models and by compounding different aspects of interdisciplinarity. The analysis is conducted at the paper level accounting both for interdisciplinarity in the production of knowledge (i.e. references) and in knowledge diffusion (i.e. citations). The diffusion and impact of IDR in physics are used both as a testbed and as an insightful case study.Results see measures as not entirely orthogonal but revealing different information concerning IDR and identify the presence of important bias in the acknowledged measures. For instance, the use of cosine similarity in computing fields distance provides insights in the structure of the research communities while the computation of the frequency of knowledge combination highlights the direct link among fields. In addition, co-citation reflects the heterogeneity of the citing fields whereas citations measure the distance between the focal and citing papers. As for biases, size and time evolution of the knowledge space are shown to affect IDR indicators.As for Physics, in the observed time span (1985-2005) there is no significant increase in interdisciplinarity for all the considered unbiased measures. However, data reveal a spike in IDR occurring in 1990 following the discovery of High Temperature Superconductivity that has opened a set of new possibility for knowledge recombination. Impact seems to follow the acknowledged inverse U-shape relation with IDR computed both on references and citations although the effect of the latter is stronger. In addition, more impactful papers are those with low balance. This amounts to say that having a prevailing knowledge field decreases the risk of producing knowledge that crosses conventional boundaries but is not recognized as belonging to any of the involved fields. The use of unbiased measures in our data set suggests a wary interpretation of the concept of novelty as separate from interdisciplinarity and highlights the limitation of operationalizing novelty as recombination of (distant) knowledge.
Creation and Diffusion of Knowledge in Top Economic Journals
The creation and diffusion of scientific knowledge have a great impact on economic prosperity of countries and regions (Romer, 1990; Phelps, 1996; Grossman and Helpman 1991) and the geographic location of top scientific research and its rate of spatial diffusion has important implications for the evolution of science and for science policy. In the policy arena, public support of scientific research emphasizes the role of excellence in science. The economic benefits of this public support depend upon the fruits of this research, the ability to stay ahead in research and to learn from excellence. Thus for both modeling science evolution and research policy purposes it is important to understand the geographic and temporal dimensions of the spread of newly created scientific knowledge and understand the specific evolution of the different fields. We tackle this issue studying scientific progress in economics. Exploiting the increased availability of large bibliometric databases, a set of recent papers has provided some quantitative evidence on the relative growth of different fields in economics and the degree of geographic concentration of publications in top journals (Angrist et al. 2017; Card and Della Vigna, 2013; Hamermesh, 2013; Kim et al. 2006; Claveau and Gingras 2016; Kosnick, 2015). The general results are the growing importance of empirical vis à vis theoretical work concerning most of the different fields within economics. In addition even if scientific knowledge is typically treated as codified knowledge that diffuses quickly in the global network of scientists, excellence in economics remains highly concentrated and there is scant evidence on the rate of geographical diffusion of different fields in economics. This paper aims at filling this gap and tackles three research questions. First, we ask which topics economists are researching, and which ones are represented in a set of top journals, using topic modeling: a methodology that provides some advantages with respect to the more commonly used JEL codes. Secondly we ask how these topics are distributed across geographical areas, studying whether there is specialization in producing knowledge related to a given topic. Finally we aim at measuring the speed of diffusion and decay of knowledge in economics within and among all combinations of these geographical areas and we explore which topics diffuse more rapidly and are more influential. In particular we estimate the citation lag distributions and describe the citation patterns among all combinations of three large geographical areas US, Europe and Rest of the World. This paper exploits 13,233 papers from 7 top journals in economics (Conroy et al., 1995) and 780,180 citations from 1985 to 2015. We study the papers by topic and geographical area, eliciting the thematic structure of the articles through topic modeling analysis on full-texts (Latent Dirichlet Allocation; Blei et al. 2003). Papers and topics are then assigned to countries and geographical areas via the authors’ affiliations. We analyze the process of diffusion and obsolescence of knowledge contained in the papers estimating the citation lag distribution for 18 different topics and three large geographical areas. To perform this task we adopt a quasi-structural model as proposed by Caballero and Jaffe (1993) and discussed in Jaffe and Trajtenberg (1996) and Hall et al. (2001) for patent data. It combines two exponentials to model the likelihood of citations taking into account different attributes of the cited and citing publications. Our results can be summarized as follows. There is a prevalence of papers from researchers affiliated in the US. This prevalence declines between 1985 and 2012 from almost 80% to less than 70% with a corresponding increase of the European share, which approaches ¼ of the papers at the end of the observation period. The estimated shape of the citation lag distribution in economics suggests that the modal lag on average is about 6.7 years. Citations to articles in top journals in economics have a slow rate of decay. On average after 30 years the estimated probability to be cited is still 46% of its maximum value. Finally we find that there are important localization effects. For example a publication originated in Europe is 39% more likely to get a citation from an average European publication than is a random US publication. This figure is 35% for US publications.In addition we show that citations in the US come faster and the modal citation lag is systematically higher when the citing papers come from Europe or the RoW. So our paper claims that publications in the US become obsolete earlier and that scientific progress proceeds at higher speed. These results give a precise quantitative expression to commonly held perceptions about the dynamism of the economic discipline in the US vis à vis other countries (Frey and Eichenberger 1993, Cardoso et al. 2010). In parallel papers from Europe and the RoW cite relatively more US papers but with a longer lag. We find that the probability that a publication in Europe or RoW would cite - one year after the publication date - a publication originated in the US is respectively 40% and 33% lower than citations originated in the United States, but 30 years later the figures turn out to be 21% and 16% higher. These results measure the extent of the initial localization and the speed of fading in the US and the lasting impact in Europe and RoW. Finally, we find that the diffusion and decay path of the different topics are very different: some topics (like Growth and Technology) are highly cited during the first years but have a quick obsolescence, and other topics like Business Finance and Banks and Education display relatively lower obsolescence rates.
This paper is essentially a rebuttal of the view that neoclassical economics can handle complexity. I have coined the locution ‘oil spot dynamic’ to denote the neoclassical ability to subsume each and every new perspective. The main part of the paper is devoted to showing why the oil spot dynamic cannot work with the complexity approach, which is seen as a coherent stand-alone research program that stems from the SFI Economics Program and manifests itself with different nuances. The fallacy of the oil spot dynamic is relevant in this period, in which economists are beginning to realize that the Neoclassical Samuelsonian Paradigm no longer represents the common language of their profession. The spread of the complexity approach and the dissolving notion of ‘mainstream’ are here interpreted as indicative of a changing economics. A short foray into the features of the process of change completes my arguments by showing that the shift from one paradigm to another has many interrelated dimensions, and that there may be rigidities internalizing changes.