Machine learning for financial transaction classification across companies using character-level word embeddings of text fields

An important initial step in accounting is mapping financial transfers to the corresponding accounts. We devised machine-learning-based systems that automate this process. They use word embeddings with character-level features to process transaction texts. When considering 473 companies independently, our approach achieved an average top-1 accuracy of 80.50%, outperforming baselines that exclude the transaction texts or rely on a lexical bag-of-words text representation. We extended the approach to generalizes across companies and even across different corporate sectors. After standardization of the account structures and careful feature engineering, a single classifier trained on 44 companies from 28 sectors achieved a test accuracy of more than 80%. When trained on 43 companies and tested on the remaining one, the system achieved an average performance of 64.62%. This rate increased to nearly 70% when considering only the largest sector.     

Model-free portfolio theory: A rough path approach

Based on a rough path foundation, we develop a model-free approach to stochastic portfolio theory (SPT). Our approach allows to handle significantly more general portfolios compared to previous model-free approaches based on Föllmer integration. Without the assumption of any underlying probabilistic model, we prove a pathwise formula for the relative wealth process, which reduces in the special case of functionally generated portfolios to a pathwise version of the so-called master formula of classical SPT. We show that the appropriately scaled asymptotic growth rate of a far reaching generalization of Cover's universal portfolio based on controlled paths coincides with that of the best retrospectively chosen portfolio within this class. We provide several novel results concerning rough integration, and highlight the advantages of the rough path approach by showing that (nonfunctionally generated) log-optimal portfolios in an ergodic Itô diffusion setting have the same asymptotic growth rate as Cover's universal portfolio and the best retrospectively chosen one.     

Inertia Risk: Improving Economic Models of Catastrophes*

We model endogenous catastrophic risk in a new way. We call it “inertia risk”, which accounts for delays between physical variables and the hazard rate – a characteristic often observed in reality. The added realism significantly affects optimal policies relative to the standard model of catastrophic risk. The probability of a catastrophe occurring at some point in time can span the entire interval [0,1], and is not 0 or 1 as is typical in standard models. Inertia risk can also generate path dependences. We illustrate the implications for policy in a simple model of climate change.     

Effects of consumer cohorts and age on meat expenditures in the United States

Meat demand is likely influenced by the birth cohort and age of the individual. In this study, we examine the demand for beef, pork, poultry, and other meat in the United States using the 1984–2012 Consumer Expenditure Survey and the almost ideal demand system with the incorporation of age, period, and cohort (APC) effects. We find that the model with APC effects performs better than the models without APC effects. The results indicate that cohorts born in earlier time periods are expected to purchase significantly less poultry compared to cohorts born in later time periods, when they are measured at the same age. Over the life cycle, purchase of poultry is expected to increase with age while the opposite is true for red meat. We also find that the own-price elasticity for beef is highest among the products examined, while the own-price elasticity for other meat is lowest and the inclusion of APC effects increases the absolute value of the own-price elasticities for beef, pork, and poultry, but reduces the own-price elasticity for other meat. Our forecasts indicate that the aggregate poultry purchase will continue to increase until 2022, while the aggregate purchase of red meat will slightly increase until 2017, but will either decrease or stay at same level from year 2017 to 2022.     

Multi-directional program efficiency: the case of Lithuanian family farms

The present paper analyses both managerial and program efficiencies of Lithuanian family farms, in the tradition of Charnes et al. (Manag Sci 27(6):668–697, 1981) but with the important difference that multi-directional efficiency analysis rather than the traditional data envelopment analysis approach is used to estimate efficiency. This enables a consideration of input-specific efficiencies. The study shows clear differences between the efficiency scores on the different inputs as well as between the farm types of crop, livestock and mixed farms respectively. We furthermore find that crop farms have the highest program efficiency, but the lowest managerial efficiency and that the mixed farms have the lowest program efficiency (yet not the highest managerial efficiency).     

Predictors of glycemic control and diabetes-related costs among type 2 diabetes patients initiating therapy with liraglutide in the United States

Objective:

Liraglutide has been shown to significantly improve glycemic control and reduce body weight while minimizing the risk of hypoglycemia in adult patients with type 2 diabetes (T2D). This study aimed to identify characteristics that predict clinical and economic outcomes associated with liraglutide therapy in clinical practice in the US.

Methods:

Using the Truven Health MarketScan Laboratory Database, glycemic control (A1C <7%) and diabetes-related costs were evaluated in T2D patients initiating liraglutide between January 1, 2010 and June 30, 2012. Patients were required to have ≥1 post-index claim for liraglutide and A1C values at baseline and 6 months follow-up. All valid values of baseline A1C were included. Patients previously treated with GLP-1 receptor agonist(s) or insulin, or with evidence of type 1 diabetes, pregnancy, or gestational diabetes during the study period were excluded. Multivariable regression models were used to identify predictors of glycemic control and diabetes-related costs.

Results:

Of 417 patients newly treated with liraglutide, 54.0% achieved glycemic control (A1C <7%) during follow-up. Factors associated with increased odds of glycemic control during follow-up were: being female, POS/EPO health plan type, baseline A1C, early liraglutide initiation (0–1 prior oral anti diabetics [OADs] vs ≥2), adherence to liraglutide (defined as the proportion of days covered [PDC]), and diabetic retinopathy. Being female, earlier liraglutide initiation (0–1 prior OADs), and higher patient share of liraglutide costs were associated with significantly lower diabetes-related costs during follow-up. Factors associated with significantly higher post-index diabetes-related costs were: higher baseline A1C, baseline use of sulfonylureas, and diabetic retinopathy.

Conclusions:

Within this commercially-insured population of T2D patients treated with liraglutide, gender, baseline A1C, early liraglutide initiation (0–1 prior OADs), diabetic retinopathy, better adherence, and patient share of liraglutide costs were associated with increased odds of achieving glycemic control and the odds of having higher or lower diabetes-related costs.     

Does Peacetime Military Service Affect Crime?

Draft lottery data combined with Danish longitudinal administrative records show that military service can reduce criminal activity for youth offenders. For this group, property crime is reduced, and our results indicate that the effect is unlikely to be the result of incapacitation only. We find no effect of military service on violent crime, on educational attainment, or on employment and earnings, either in the short run or in the long run. These results suggest that military service does not upgrade productive human capital directly, but rather affects criminal activity through other channels (e.g., by changing attitudes to criminal activity).     

Seasonal employees' intention to return and do more than expected

Seasonal employment is important in the service industries. Having motivated and satisfied employees is fundamental when front-line employees play a key role for the customers' perceived service quality. Seasonal work differs from permanent employment in many aspects and if managed properly, this could be a competitive advantage and contribute to the success of an organization. This study focuses on seasonal employees' intention to return and to do more than expected at work as well as the relationships between certain motivational factors and job satisfaction. A survey was carried out at four ski resorts with 477 respondents. Correlation and regression analyses indicate that job satisfaction influences the intention to return but does not have a strong influence on organizational citizenship behaviour (i.e. doing more than expected at work). The results improve our understanding of seasonal workers' motivation and behaviour. The concluding part discusses managerial implications.