First Look

Abstract—The purpose of this chapter is to relate the theory of task networks and technology set forth in previous chapters to theories of firm boundaries from economics and management. Complementary goods have more value when used together than separately. Complementarity may be strong or weak. Strong complements are specific and unique goods that have no value (or greatly diminished value) unless all are present in use. In the task network, dense technical interdependencies create strong complementarity, but it can arise for other reasons as well. Transaction cost economics and property rights theory advise that strong complements should be placed under unified governance, for example, through common ownership. Agency theory suggests that weak complementarity can be handled via arms-length transactions and contracts. Furthermore, strong or weak complementarity are not innate properties of tasks and assets but can be the result of choices regarding task networks, incentives, and job design. Supermodular complementarity exists when more of one input makes more of another input more valuable. Distributed supermodular complementarity (DSMC) exists when two or more independent actors can create complementary value by pursuing their own interests and will not find it advantageous to combine in order to coordinate their actions. I derive formal conditions under which DSMC holds as a consistent pattern in a dynamic equilibrium. Given DSMC, clusters of firms making different complementary goods, including open platforms with surrounding ecosystems, can survive and compete effectively against integrated firms that control all complementary inputs.

Download working paper: https://www.hbs.edu/faculty/Pages/item.aspx?num=55132

Abstract—Organizations are formed in a free economy because a person or group perceives value in carrying out a technical recipe that is beyond the capacity of a single person. Technology specifies what must be done, what resources must be assembled, what actions taken, and what transfers made in order to convert stocks of material, energy, and information into products of value to someone. The purpose of this chapter is to build a robust and versatile language that is capable of representing large technical systems. The language is based on elements I have labeled functional components. The language is more abstract than the language of technical recipes and task structures, thus it is capable of hiding details that may be distracting. However, the language also makes it possible to “track back” from each named functional component to a technical recipe (or the lack of one).

Download working paper: https://www.hbs.edu/faculty/Pages/item.aspx?num=55133

Abstract—The purpose of this chapter is to lay the groundwork for a comprehensive theoretical investigation of open platform systems. To do this, we must first recognize that, although there is a strong family resemblance among all platform systems, there are different types of open platforms, each with its own set of technological requirements and challenges. I first develop a taxonomy of open platform types and then provide a brief history of digitally enabled open platforms. I go on to argue that the competitive success of open platforms against closed platforms gave rise to the "vertical-to-horizontal" transition in the computer industry in the 1980s and 1990s. This transition was one of the organizational "surprises" highlighted in Chapter 1. In this case, newly competitive technology of open platforms not only shaped individual organizations but changed the structure of the entire industry.

Download working paper: https://www.hbs.edu/faculty/Pages/item.aspx?num=55092

Abstract—We describe how to employ machine learning (ML) methods in theory development. Compared to traditional causal inference methods, ML methods make far fewer a priori assumptions about the functional form of the underlying model that best represents the data. Thus researchers could use such methods to explore novel and robust patterns in data, which could in turn lead to inductive theory building. ML’s strengths include replicable identification of novel patterns in data. ML methods also address several issues raised by scholars (such as “p-hacking” and confounding local effects with global effects) pertinent to the norms of empirical research in the fields of strategy and management. We provide a step-by-step roadmap that illustrates how to use four ML methods (decision trees, random forests, K-nearest neighbors, and neural networks) to reveal patterns in data that could be used for theory building. We also illustrate how ML methods can illuminate interactions and non-linear effects better than traditional methods. In summary, ML methods could serve as a complement to both existing inductive theory-creating methods, such as multiple-case inductive studies and traditional methods of causal inference.

Download working paper: http://www.hbs.edu/faculty/Pages/item.aspx?num=55043

Abstract—We introduce an exponential smoothing model that a manager can use to forecast the demand of a new product or service. The model has five features that make it suitable for accurately forecasting product life cycles at scale. First, the trend in our model follows the density of a new distribution called the tilted-Gompertz distribution. This model can capture the wide range of skewed diffusions commonly found in practice—diffusions of innovations described as having “extra-Bass” skew. Second, its parameters can be updated via exponential smoothing; therefore, the model can react to local changes in the environment. This model is the first exponential smoothing model to incorporate a life-cycle trend. Third, the model relies on multiplicative errors, instead of the additive errors primarily used in existing models. Multiplicative errors ensure that all quantile forecasts are strictly positive. Fourth, the model includes prior distributions on its parameters. These prior distributions become regularization terms in the model and allow the manager to make accurate forecasts from the beginning of a life cycle, which is notoriously difficult. The model's skewed shape, time-varying, regularized parameters, and multiplicative errors can make its quantile forecasts more accurate than leading diffusion models, such as the Bass, gamma/shifted-Gompertz, and trapezoid models. Fifth, the model's estimation procedure is based on an efficient optimization routine, which can be used to forecast product life cycles at scale. In two empirical studies, one of search interest in social networks and the other of new computer sales, we demonstrate that our model outperforms leading diffusion models in out-of-sample forecasting. Our model's point and other quantile forecasts are more accurate. Accurate quantile forecasts at different horizons are critical to many operational decisions, such as capacity and inventory management.

Download working paper: https://www.hbs.edu/faculty/Pages/item.aspx?num=55096

Abstract—Firms rarely cut compensation, so little is known about the after-effects when compensation reductions do occur. We use commission reductions at a sales firm to estimate how work effort and turnover change. In response to an 18% decline in sales commissions, corresponding to a 7% decline in median take-home pay, we find turnover increases for the most productive workers. We detect limited effort responses. Turnover and effort responses do not differ based on workers' survey replies regarding expectations of firm fairness or future promotion. The findings indicate that adverse selection concerns on the extensive margin of retaining workers drive the empirical regularity that firms rarely reduce compensation.

Download working paper: https://www.hbs.edu/faculty/Pages/item.aspx?num=54422

By late 2017, Brazilian retailer Magazine Luiza's CEO was convinced that the company could significantly grow sales and accomplish its aspirations of digital transformation. What was unclear in his mind was whether he should act as a tech company and grow as fast as possible (e.g., high double digits) or be more conservative and grow sales at a financially healthy rate, like traditional retailers did (e.g., single digits). The primary way e-retailing companies achieved these abnormally high rates of growth was through lowering prices and foregoing profitability. Historically, mass retailing had razor-thin margins. It was thus unlikely that he could have it both ways: grow fast and be profitable. Should Trajano opt for more aggressive growth or proceed more conservatively?

Purchase this case:
https://hbsp.harvard.edu/product/519009-PDF-ENG

By June 2012, Stephan Aarstol felt that he had successfully passed the first critical stage of his ecommerce business. As the founder and CEO of a standup paddleboard (SUP) business, he had built a strong relationship with Asian manufacturers, built a small warehouse and fulfillment center, designed an innovative line of inflatable SUPs, and built an ecommerce website that sold boards and accessories to consumers. After the rising trend in interest for the sport provided a strong wave of growth in sales, Aarstol contemplated the next stage at Tower Paddle Boards. Should he partner with Amazon to sell his full line of boards—manufactured under his brand—and accessories—manufactured by other brands? Should he sell to Amazon? Should he sell on Amazon marketplace? Or should he avoid the powerful online retail giant altogether?

Purchase this case:
https://hbsp.harvard.edu/product/517047-PDF-ENG