Research Policy 45 (2016) 81–96 Contents lists available at ScienceDirect Research Policy jo ur nal ho me page: www.elsevier.com/locate/respol Mapping the technological landscape: Measuring technology distance, technological footprints, and technology evolution Barak S. Aharonson a,∗ , Melissa A. Schilling b a Tel Aviv University, Business School, Strategic Management, 6997801 Tel Aviv, Israel b New York University, Stern School of Business, 44 West 4th Street, New York, NY 10012 USA a r t i c l e i n f o Article history: Received 29 April 2014 Received in revised form 23 July 2015 Accepted 3 August 2015 Keywords: Technology landscape Knowledge-spillover Recombinant search a b s t r a c t We develop and apply a set of measures that enable a fine-grained characterization of technological capa- bilities based on the USPTO database. These measures can capture the distance between any two patents, and help to identify outlier patents. They also provide a rich characterization of a firm’s technological footprint, including its depth and breadth. The measures also enable researchers to assess the technolog- ical overlap, similarity, and proximity of the technological footprints of two or more firms. At the level of the macro technology landscape, the measures can be used to explore such dynamics as technology agglomeration, knowledge spillovers, and technology landscape evolution. We show applications of each of the measures and compare the results obtained with those that would be obtained with previously existing measures of firm diversity, similarity and proximity, highlighting the advantages of the measures used here. © 2015 Elsevier B.V. All rights reserved. 1. Introduction A firm’s technological capabilities are central to its identity, its strategies, and its potential for success. Technological capabilities represent what the firm can do in the present, as well as what it has learned in the past. These capabilities also have a significant influence on the trajectories a firm will choose in the future. It is not surprising, then, that there are many streams of research that invoke the notion of technological capabilities, including (but not limited to) those that examine competitive positioning, inno- vation, organizational learning, diversification, and organizational growth. However, the development of measures for constructs such as technological capabilities, technology distance, technological similarity, and technology footprints is a nascent field, with much opportunity for further development. Most of the prior work on such measures can be classified into four categories: (1) measured based on SIC codes, (2) measures based on patent classes, (3) measures based on patent citations, and (4) measures based on textual analysis of patents. Early work on technological relatedness tended to use industry classification (SIC or NAICS) codes, looking at, for example, a firm’s business share across different SIC codes or the degree to which its combinations ∗ Corresponding author. E-mail addresses: aharonson@tau.ac.il (B.S. Aharonson), mschilli@stern.nyu.edu (M.A. Schilling). of SIC codes mirrored patterns in the general economy – what Teece et al. (1994) term “corporate coherence.” The advantage of this kind of approach is that one is not limited to examining only firms that patent. The main disadvantage of this kind of approach is its impre- cision: First, most industry classification systems reflect markets in which a firm competes, not technological capabilities with which they compete. For example, SIC code 2043, “cereal breakfast foods manufacturer,” tells us little about the technology (e.g., cooking, drying, extrusion, rolling, etc.) involved. By contrast, USPC mainline subclass 323.4 “cereal puffing by means of an apparatus adapted to subject cereal to sudden changes in pressure to disrupt the same and produce an expanded or inflated product” gives us consider- ably more technology capability-relevant information and offers greater precision. Second, SIC codes are typically only available at the firm level, and even then usually only at the “primary” SIC code level, which may mask considerable firm activity. One of the most common approaches to handling technology distance or similarity is to count the number of patent classes two firms have in common across their patenting portfolios (e.g., Ahuja and Katila, 2001; Diestre and Rajagopalan, 2012; Dushnitsky and Lenox, 2005). This method is relatively easy to implement and relies on publicly held data that can be regularly updated, however the tradeoff for this efficiency is a loss of granularity and detail. This loss is particularly acute when researchers use only the first class listed on a patent (Benner and Waldfogel, 2008). As we will show, measures based on patent subclasses provide a much richer and more reliable picture of a firm’s technological capabilities. http://dx.doi.org/10.1016/j.respol.2015.08.001 0048-7333/© 2015 Elsevier B.V. All rights reserved.