M5 competition uncertainty: Overdispersion,distributional forecasting,GAMLSS, and beyond

The M5 competition uncertainty track aims for probabilistic forecasting of sales of thousands of Walmart retail goods. We show that the M5 competition data face strong overdispersion and sporadic demand, especially zero demand. We discuss modeling issues concerning adequate probabilistic forecasting of such count data processes. Unfortunately, the majority of popular prediction methods used in the M5 competition (e.g. lightgbm and xgboost GBMs) fail to address the data characteristics, due to the considered objective functions. Distributional forecasting provides a suitable modeling approach to overcome those problems. The GAMLSS framework allows for flexible probabilistic forecasting using low-dimensional distributions. We illustrate how the GAMLSS approach can be applied to M5 competition data by modeling the location and scale parameters of various distributions, e.g. the negative binomial distribution. Finally, we discuss software packages for distributional modeling and their drawbacks, like the R package gamlss with its package extensions, and (deep) distributional forecasting libraries such as TensorFlow Probability.     

Risk Preferences and Gender Differences in Union Membership in Late Nineteenth-Century Swedish Manufacturing

Women are generally seen as less inclined to join trade unions. This study matches firm–worker data from the Swedish cigar and printing industries around 1900 and examines information on men and women holding the same jobs; such data are rare but important for understanding gender gaps. The results explain the gender gap in union membership among compositors, but not among cigar workers. Differences in union membership varied considerably across firms, with the largest differences found in low-union-density cigar firms where indirect costs (that is, uncertainty and risk) accrued in particular to women workers. The lack of gender differences in mutual aid membership indicates that women were not hard to organize but avoided organizations associated with greater risk for employer retaliation and uncertain returns according to a cost–benefit analysis.     

Dynamics of Child Malnutrition in Rural and Small‐Town Ethiopia

Improving the status of malnourished children, and preventing children from becoming malnourished, lies at the heart of several of the Sustainable Development Goals. While many cross‐sectional studies examine correlates of stunting, they largely cannot identify drivers of change in stunting. We use two waves of panel data from Ethiopia and incrementally larger sets of fixed effects to control for time‐variant observable characteristics and time‐invariant unobservable characteristics. After controlling for these potential confounders, our analysis reveals that many factors that are associated with stunting in the cross‐section do not impact stunting dynamics. We also estimate individual fixed‐effects regressions, separately, according to baseline stunting status. We find evidence suggesting that while improved societal conditions drive many children out of a stunted state, certain exogenous factors may lead previously healthy children to become stunted. Overall, policymakers and practitioners would be wise to consult research utilising both cross‐sectional and panel data analyses in order to more effectively target already stunted children as well as vulnerable children who may be at risk of becoming stunted.     

Migrants and multinational firms: The role of institutional affinity and connectedness in FDI

We examine how, and to what extent, migrants in a host country attract foreign direct investment (FDI) from firms based in their country of origin (CO). Introducing the notion of institutional affinity, we argue that increased institutional affinity and increased connectedness of institutional environments of migrants’ CO and country of residence, make a location attractive to CO firms. Empirical analysis of FDI and migration panel data shows that in addition to the traditional factors influencing FDI patterns, there is a collective migrant effect on FDI, and this effect is statistically significant and economically meaningful for migrants from developing countries.     

Are USDA reports still news to changing crop markets?

This study investigates whether major USDA reports still provide important news to changing crop markets. The news component of each report, or market “surprise,” is measured as a difference between the USDA estimate and its private expectation in corn, soybeans, and wheat markets. Changes in the relevance of USDA information are assessed by examining changes in the magnitude of market surprises and shifts in the futures price reaction to these surprises, which isolates the impact of each report. The stable size of market surprises over time suggests that competition from alternative data sources has not reduced the news component of USDA crop reports. Increasing price reaction to most reports, including those facing competition from alternative information sources, suggests that value of public information may be enhanced in uncertain markets affected by structural changes.     

Characteristics and implications of Chinese macroeconomic data revisions

The research examining macroeconomic data for developed economies suggests that an understanding of the nature of data revisions is important both for the production of accurate macroeconomic forecasts and for forecast evaluation. This paper focuses on Chinese data, for which there has been substantial debate about data quality for some time. The key finding in this paper is that, while it is true that the Chinese macroeconomic data revisions are not well-behaved, they are not very different from similarly-timed U.S. macroeconomic data revisions. The positive bias in Chinese real GDP revisions is a result of the fast-growing service sector, which is notably hard to measure in real time. A better understanding of the revisions process is particularly helpful for studies of the forecast errors from surveys of forecasters, where the choice of the vintage for outcomes may have an impact on the estimated forecast errors.     

Reviewing the DATAS of aviation research data: Diversity,availability, tractability,applicability, and sources

The field of aviation research is entering the era of big data. While data-driven advancements in aviation have clearly brought about applicable models and results with immediate implications, we argue that the influx of aviation data should be better characterized and documented to enable more efficient and standardized usage. To this end, we examine 200 well-cited research articles from sub-disciplines ranging from revenue management to air traffic control published on or after 2010 in order to analyze the diversity, availability, tractability, applicability, and sources (DATAS) of data utilized in aviation research. We find high levels of data diversity within aviation research, with 16 data categories ranging from air traffic flow management-type data to data from distributed sensors in line with the Internet-of-Things (IoT) paradigm. We identified a dominance of proprietary, non-public data in aviation research, with 68% of the 200 research articles utilizing solely proprietary data in deriving their results, and a further 8% utilizing a mixture of proprietary and publicly available data. The pervasiveness of proprietary data has implications on reproducibility and extending research results. We also highlight the increasing tractability of the data by surveying the computational power required to process the data sets, and present vignettes of applications and results that stem from these data-driven studies. Finally, we propose several recommendations regarding standardizing data source nomenclature as well as increasing the availability of and usage of publicly available data.     

An innovation to improve health outcomes in Amajuba district,KwaZulu-Natal,South Africa

Systems’ thinking places high value on understanding the context. This study focused on the collection of disaggregated data in order to understand the context, to facilitate improvement of health outcomes. The aim of this article was to assess the implementation of municipal ward-based health data collection (disaggregated data) and health care workers’ perceptions of this data collection process. This cross-sectional study used mixed methods in Amajuba district. The participants were professional nurses at the Primary Health Care level. Of the 131 respondents, 123 (93.9%) collected municipal ward-based health data, and found it useful. Opportunities for improving data collection were identified. Disaggregation of the data at ward level contributes to a better understanding of the target population’s health, assists planning for health needs and enables provision of targeted interventions in order to improve health outcomes, to prevent financial regression and waste of health resources.     

Effects of scale economies and production seasonality on optimal hub locations: the case of regional fresh produce aggregation

Interest in supporting local and regional food systems is rising and food hubs have attracted considerable attention among Federal, State, and local policymakers. This study explores the problem of endogenous hub location in fresh produce value chains in the Northeastern United States. To overcome limitations in the literature, we incorporate the effects of economies of scale and production seasonality into our models. Three experimental models are designed to examine the effects of alternatively applying yearly, quarterly, and monthly data on model solutions. We explicitly assess how interactions of scale economies and seasonality influence the structure and spatial attributes of an optimal regional produce aggregation hub system. The three models generate marketed different solutions and in many respects they lead to different conclusions about developing local/regional supply chains. The monthly model allows for production seasonality and actual hub operation cycle frequency and thus leads to more efficient hub solution with rich policy implications.     

Evaluating discounts as a dimension of customer behavior analysis

Today, increased competition between organizations has led them to seek a better understanding of customer behavior through identifying valuable customers. Customers’ expectations about the price and quality of products and services play an important role in their selection process. In online businesses, competition and price differences between suppliers is high, so discounts will attract different customers. As a result, discounts and the frequency and amount of purchases can lead to better understanding of customer behavior. Customer segmentation and analysis is essential for identifying groups of customers. Hence, this study uses a model based on RFM called RdFdMd, in which d is the level of discount used to analyze customer purchase behavior and the importance of discounts on customers’ purchasing behavior and organizational profitability. The CRISP-DM and k-mean algorithm were used for clustering. The results indicate that using the R_dF_dM_d model achieves better customer clustering and valuation, and discounts were identified as an important criterion for customer purchases.