Comparing automated text classification methods

Online social media drive the growth of unstructured text data. Many marketing applications require structuring this data at scales non-accessible to human coding, e.g., to detect communication shifts in sentiment or other researcher-defined content categories. Several methods have been proposed to automatically classify unstructured text. This paper compares the performance of ten such approaches (five lexicon-based, five machine learning algorithms) across 41 social media datasets covering major social media platforms, various sample sizes, and languages. So far, marketing research relies predominantly on support vector machines (SVM) and Linguistic Inquiry and Word Count (LIWC). Across all tasks we study, either random forest (RF) or naive Bayes (NB) performs best in terms of correctly uncovering human intuition. In particular, RF exhibits consistently high performance for three-class sentiment, NB for small samples sizes. SVM never outperform the remaining methods. All lexicon-based approaches, LIWC in particular, perform poorly compared with machine learning. In some applications, accuracies only slightly exceed chance. Since additional considerations of text classification choice are also in favor of NB and RF, our results suggest that marketing research can benefit from considering these alternatives.     

Telling and selling the value of diversity and inclusion—External consultants' discursive strategies and practices

On the basis of a qualitative study of a subgroup of diversity professionals, external diversity and inclusion (D&I) consultants, we explore D&I consultants' discursive strategies and practices situated within organisational structures, relations, and interactions of power and knowledge. Theoretically, the research reveals how D&I consultants' own discursive strategies interact with existing organisational and societal discourses of diversity, incrementally shaping their continual evolution. A classification is developed, which sets out four approaches taken by consultants with regard to their discursive strategies in relation to clients. The findings suggest that HR practitioners need to work in tandem with external consultants to develop strategies to improve the status and legitimacy of diversity work if the field is to progress the organisational D&I agenda.     

Persistence of Price–Cost Margin and Technical Efficiency in the Indonesian Food and Beverage Industry

Abstract

This research investigates the persistence of price–cost margin (PCM) and technical efficiency (TE) of firms and the relation between these two factors in 44 subsectors of the Indonesian food and beverage industry in the period 1980–2014. Data envelopment analysis with a bootstrapping approach is applied to estimate TE. An autoregressive model, accounting for endogeneity, is applied to estimate the persistence coefficients of PCM and TE. A cross-sectional regression model is also applied to estimate the relation between the persistence of the PCM and the persistence of TE. The results show that for these food and beverage firms, high PCM and low TE persist. Furthermore, the persistence of PCM negatively affects the persistence of TE in the industry.     

Reverse Payments and Risk of Bankruptcy Under Private Information

Abstract

In the pharmaceutical industry, a reverse payment (pay-for-delay) is a payment from an originator to a generic producer to delay her entry. In some recent cases, the US and EU antitrust authorities have banned these agreements per se, while in others they have used a rule of reason. This paper analyzes their dynamic effects and shows that a ban per se may reduce consumer surplus when the generic producer may go bankrupt and her financial situation is private information. Reverse payments are more beneficial when competition among few players is soft, the economy is in a downturn, and the period of drug usage after patent expiry is long. Results suggest that a rule of reason is more suited than a ban per se.     

Effects of state-sponsored human capital investment on the selection of training type

Since the 1990s, firms in Japan have reduced their human capital investment in the workplace to minimize costs. Moreover, in response to the increase in the number of non-regular employees and turnover rates, workers need to have greater incentive to make the self-motivated investment in themselves for their self-protection. In this study, we first estimate the effects of workers’ self-motivated investment in themselves on wage rates. Next, we explore who is likely to participate in which training type and accordingly estimate the effects of the self-motivated investment on wage rates by training type. Our estimates controlling for individual-level fixed-effects indicate that the return is significantly positive and particularly high for practical training related to workers’ current jobs, and regular workers tend to self-select these higher-returns programs, while non-regular workers are more likely to enroll in lower-returns programs, such as schooling. This trend in investment in oneself could potentially increase the wage inequality between regular and non-regular workers through the self-selection of training types. Our estimates reveal that receiving the training and education benefit raises the likelihood for workers to participate in a high-return training program regardless of whether they are non-regular or regular workers. This suggests that government benefits on self-investment change workers’ self-selection of training type and serve to promote practical trainings that lead to high returns.     

The impact of German-speaking online media on tourist arrivals in popular tourist destinations for Europeans

This empirical study analyzes the relationship between the sentiments in online media with regard to travel destinations and corresponding tourist arrivals. We expect the media reports on political and economic instability and turmoil to enhance tourist arrival nowcasts and forecasts, as they can probably complement them with information on disruptions and shocks. Therefore, we believe this research will help to build better models for tourism demand nowcasting and forecasting. We use the sentiment in the German-speaking online media because the German-speaking region is the most populated in Europe and has the largest group of travelers visiting destinations in and around Europe.

An artificial neural network is used to analyze the mood of the media. The software classifies news items regarding potential tourist destinations with either positive or negative labels. The number of positive and negative news items is used to build sentiment indices for popular tourist destinations for Europeans.

Our results show strong correlations between the mood concerning tourist destinations and tourist arrivals in these countries. Indeed, disruptions and shocks prevalent in the news are reflected in similar ratios in both tourist arrivals and sentiment indices. These results can be used as a new explanatory variable for tourism demand modelling.     

水生态系统服务供需视角下的水网乡村 适应性规划策略——以吴江张鸭荡片区为例

苏南乡村地区一直是中国乡村建设的先行区域，特殊的纵横交错的水网结构构成了一个错综复杂的系统，呈现出其他地区不具有的复杂性和生态特殊性。随着城镇的扩张，乡村的发展建设使其水网空间的平衡发展面临极大的挑战。传统单一静态的规划方法逐渐显示出无法适应经济、社会等发展要求的缺陷。苏南乡村地区以水为核心，从水生态系统服务供需关系的视角下探究水网乡村的适应性规划策略更加适应当前的乡村现状和需求。以传统水网乡村空间形态转译为基础，建立水生态系统供需服务评估体系，在评估水生态系统服务供需能力的基础上，分析供需分异模式及供需矛盾。从构建乡村水域空间生态格局、乡村水域空间功能分区规划，以及乡村水域空间多情景预判3个方面，提出苏南水网乡村的适应性规划策略，并为水网乡村的生态实践提出新思路。     

Consumer Quality Perception in Traditional Food: Parmigiano Reggiano Cheese

Abstract

The paper aims to classify the quality cues and attributes of grated Parmigiano Reggiano cheese, an Italian traditional food cheese, exploring if they affect consumers’ quality perceptions and expectations. Analysis is based on a questionnaire administered in hypermarkets where grated Parmigiano Reggiano is sold. A factor analysis, using varimax rotation, and a cluster analysis, were performed, using the Stata 12 software package. The clusters used were: geographical origin and packaging (cluster 1); price, brand and quality certification (cluster 2); all sensory attributes (cluster 3); and a combination of sensory attributes with price, brand, and quality certification (cluster 4). This is the first study to examine the quality cues and attributes affecting quality perception and expectations of cheese for grated Parmigiano Reggiano from a consumer perspective. It is also the first to classify cheese attributes following a quality approach, encompassing previously studied sensory and other attributes.     

Discouraging Rainforest Transformation: An Ex-ante Policy Impact Analysis

Despite extensive efforts made by national and international certification agencies, Indonesian smallholder farmers’ participation in palm oil certification schemes adoption remains low. A fundamental obstacle is the smallholder practice of rainforest transformation into oil palm plantation which is forbidden by the agencies. In this context, we investigate three policies that could lead to a reduction in rainforest deforestation by smallholders: price premium on certified palm oil; the provision of environmental information; contributor recognition. In order to evaluate the influence of the policies ex-ante, we conduct a social dilemma experiment involving rubber and oil palm smallholders in Jambi Province, Sumatra, Indonesia. The findings indicate that the price premium and provision of context-specific environmental information could reduce rainforest transformation. However, a statistically significant effect of contributor recognition was not found.     

Economic policy uncertainty and mergers and acquisitions: Evidence from China

This study examines the relationship between economic policy uncertainty and mergers and acquisitions (M&As) in China. Using all listed Chinese companies on the Shanghai and Shenzhen Stock Exchanges as well as 4188 ​M&A deals from the period of 2001–2018, we show that Chinese firms are more likely to make acquisitions during periods of high economic policy uncertainty, which contradicts the behavior of US firms. We further show that state-owned enterprises (SOEs) are less likely than non-SOEs to make acquisitions during periods of high economic policy uncertainty. SOEs are less likely to use only cash for their acquisitions during periods of high economic policy uncertainty. These results indicate the prudence of SOEs regarding acquisitions relative to non-SOEs during periods of high economic policy uncertainty. Moreover, acquisitions during periods of high economic policy uncertainty are associated with an increase in shareholder wealth for acquirers, and this wealth effect is more pronounced for SOEs.