Structured Programming Research Articles

The impact of pension reform on employment, retirement, and disability insurance claims

We evaluate the 2011 comprehensive reform of Norwegian early retirement institutions using a parsimonious random utility choice model. Conditional on employment at age 60, we estimate a three-state conditional logit model to explain the realized labor market state at age 63 among the alternatives of employment, retirement, and disability program participation. The reform radically changed work incentives and/or pension access age for some (but not all) workers, such that the influence of economic incentives can be identified based on reform-generated variation only. We find that improved work incentives caused employment rates to rise considerably at the expense of early retirement and exit through disability insurance. Improved liquidity through a lower age to access own pension funds on actuarially neutral terms caused a small increase in employment and a large drop in disability program participation. Properly designed pension reforms thus need to take the interplay between old-age pension and disability insurance programs into account.

Does the health of local populations modify occupational differences in employment rates of older workers? Findings from the ONS Longitudinal Study 2001–2011

Poor health is a key reason for early exit from the labour market. Few studies have explored how the health of local populations is related to occupational differences in employment outcomes among older people. Our study used data for England and Wales from the ONS Longitudinal Study linked with 2001 Census measures of the health of the older working age population at local authority level. We included 128,710 people aged 40-64y in 2001 who were in paid work in the previous five years. We investigated the associations of both occupation and area level with two employment outcomes ten years later (in 2011): i) in paid work or not; ii) economic activity (employed (reference), unemployed, retired, sick/disabled, other). People in elementary occupations were more likely to not be in paid work in 2011 compared to those in managerial occupations (RRR 1·55 [95%CI 1·47,1·64]). Compared to the healthiest third of local authority areas, being resident in the unhealthiest third was associated with greater likelihood of not being in paid employment ten years later (RRR 1·25 [95% CI 1·18,1·33]). While area level health was associated with employment outcomes for all major occupation groups, the gap between the healthiest and unhealthiest areas was most marked for skilled trades; process, plant and machine operatives; and elementary occupations. Occupational differences for the economic activity outcome were most marked for the sick/disabled category. Policies to improve the health of local populations may support retention and reduce occupational inequalities in employment rates of older workers.

Navigating identity shifts and well-being in the entrepreneurial exit process: A comparative study of entrepreneurs nearing retirement

Entrepreneurs in their 50s and beyond face significant identity shifts and well-being challenges as they transition out of the businesses they have built. Grounded in Super’s career-development framework and utilizing the Eisenhardt Method for case study analysis, this research explores the entrepreneurial exit process during the maturity and decline stages. With an average business tenure of 11 years, these entrepreneurs encounter emotional and practical hurdles, such as health concerns, financial pressures, and shifting priorities. The study highlights valuing identity diversification and social support in easing the transition, revealing that financial success alone does not guarantee positive post-exit well-being. Gender differences show women prioritize sustainability and family, while men focus on growth and professional identity. These insights underscore the need for early exit planning and holistic support for successful entrepreneurial succession and retirement strategies. JEL CLASSIFICATION: L26: Entrepreneurship; D91: Intertemporal Household Choice; Life Cycle Models and Saving; M14: Corporate Culture; Diversity; Social Responsibility; J14: Economics of the Elderly; Economics of the Handicapped; Non-labor Market Discrimination; I31: General Welfare, Well-being

Predictors of disability retirement - a prediction model for disability pension application risk

Abstract Background Regarding the objective of preventing early exits from labor market, the late life-course years present an important timeframe. It is essential to identify the employees in the greatest risk of permanently losing their work ability. In practical applications, multifactorial models are needed to evaluate the combined effect of risk factors. In this study, the disability pension risk is modeled based on large unique register data, at the same time assessing the effect of sickness absences (SA) of different lengths. Methods We utilize a logistic model for the risk of filing a first-time disability pension application (DPA) based on register data on 222,321 Finnish municipal sector employees. The data include employment and sickness spells, occupations, and wages in 2016-2018 as covariates. During the 3-year follow-up, 2,969 employees filed a DPA. The data include also short SA spells, a novelty in these kinds of studies. DPA is used as a follow-up to include employees who receive a rejecting decision but still carry a physician’s evaluation of diminished work ability. Results The resulting model can accurately predict the risk of DPA in employee groups. Age, occupation, cumulative SAs preceding the application, and the number of SA spells of all lengths, also short ones, have a significant increasing effect on the DPA risk. Proximity of the retirement age and high earnings decrease the risk of DPA. Surprisingly gender had no effect on the risk. Conclusions The results indicate that register-based sociodemographic data is very useful in modeling the employee DPA risk in public sector employee groups. The resulting model provides a basis for tools for predicting the risk status of employee groups for the use of employers responsible for risk management and early intervention, also allowing for the assessment of economic effects of SAs and disability risk. The presented modeling results are already utilized among public sector employers in Finland. Key messages • Disability pension risk can accurately be predicted in public sector employee groups based on readily available sociodemographic register data. • All sickness absences have a significant increasing effect on the risk of disability pension application, but the effect varies greatly depending on age and occupation.

RAPNet: Resolution-Adaptive and Predictive Early Exit Network for Efficient Image Recognition

RAPNet: Resolution-Adaptive and Predictive Early Exit Network for Efficient Image Recognition

HN1 expression contributes to mitotic fidelity through Aurora A-PLK1-Eg5 axis.

Hematological and neurological expressed 1 (HN1) is homolog of Jupiter protein from Drosophila melanogaster where it functions as a microtubule-associated protein. However, in mammalian cells, HN1 is associated partially with y-tubulin in centrosomes, Stathmin for stabilizing microtubules, and Cdh1 for regulating Cyclin B1 for cell cycle regulation. Moreover, HN1 overexpression leads to early mitotic exit as well. Other molecular functions and interactions of HN1 are not clear yet. Here, based on our previous analysis where HN1 was shown to cluster supernumerary centrosomes and maintain mitotic spindle assembly, we further investigated the role of HN1 in centrosome maintenance and mitotic fidelity in PC-3 prostate and MDA-MB231 mammary cancer cell lines. The maturation-associated roles of HN1 during cell division by examining the AuroraA-PLK1 axis involving a plus end kinesin, Eg5 as well as pericentriolar matrix protein (PCM1) as components of centrosomes were established. We found that HN1 co-localized to centrioles with Eg5 and Aurora A to suppress aberrant spindle formation to ensure the fidelity of centriole/centrosome duplication when overexpressed. Consistently, depleting the HN1 expression using siRNA or shRNA resulted in an increased number of dysregulated mitotic spindle structures, where Aurora A as well as PLK1 co-localizations with Eg5 and PCM1 were disrupted. Further, the PLK1 and Aurora A kinase's phosphorylations also decreased, confirming the hypothesis that the cells struggle in mitotic progression, display nuclear and cytokinetic abnormalities with supernumerary but immature mononucleated centrosomes. In summary, we described the role of HN1 in centrosome nucleation/maturation in PLK1-Eg5 axis and concomitant mitotic spindle formation in human cells.

Antecedents of and experiences with voluntary early exit from working life before the expected retirement age in Sweden.

In most Western countries, senior workers have increased their participation in the workforce. However, at the same time, early retirement also increases. The reasons behind this early exit from the workforce are still unclear. This qualitative study aims to explore the antecedents of and experiences with a voluntary exit from working life before the expected retirement age of 65 in Sweden. Data consist of semi-structured interviews with 18 participants who exited working life between the ages of 61-63. In Sweden, these ages are considered as an early exit from working life since, for many years, the expected retirement age has been 65. Qualitative content analysis with an abductive approach was utilized. The analysis revealed four sub-themes: 1) Health benefits with an early exit from working life (with the categories: own health status, the possibility for recovery time, and avoidance of strain); 2) Having economic conditions that enable an early exit from working life (with the categories: offers from the employer and financial compromises); 3) Social benefits with an early exit from working life (with the categories: enabling more time with my social network and avoidance of unsatisfying social work environment); 4) Self-fulfillment activities during the senior years (with the categories: enabling time for activities beyond work and avoidance of decreased job satisfaction. This variety of antecedents of and experiences with a voluntary early exit from working life before the expected retirement age highlights that the ongoing increased statutory retirement age also increases the risk for extended inequalities among the aging population.

Early exit deep neural networks for distorted images on edge environments

Deep Neural Networks (DNNs) are widely used for image classification but can struggle with distorted images, leading to reduced accuracy. Moreover, these applications often demand meeting a strict deadline. To this end, an alternative involves employing an adaptive offloading based on early exit NNs (EE-DNNs). EE-DNNs have branches inserted into their middle layers at the edge device. These branches provide confidence estimates. If the classification is sufficiently confident, the inference terminates at the edge device. Otherwise, the edge offloads the inference task to the cloud which runs the remaining layer. Moreover, multiple branches can compose an ensemble that collectively produces a more accurate inference. This work analyzes whether EE-DNN-based solutions can make DNN inference more robust against image distortion, while also satisfying deadline requirements in an adaptive offloading. Our results show that, in terms of accuracy, EE-DNNs and the ensemble approach are as sensitive as conventional DNNs for blurred images. However, given the edge usage, these EE-DNNs approaches can reduce the inference time, enabling them to better fulfill deadline requirements compared to conventional DNNs (i.e., with no branches).

Menopause and the workplace: consensus recommendations from The Menopause Society.

Menopause is a natural life transition experienced by half the world's population. Women aged 50 years and older are the fastest growing demographic group in many countries, making essential contributions to the workforce. Although menopause is a universal and natural life transition, the symptom experience is highly variable among women. Some women may experience few or no symptoms, whereas others may be bothered by moderate to severe symptoms for a decade or longer, which can adversely affect quality of life, relationships, job satisfaction, and career advancement. Indeed, menopause symptoms, including vasomotor and genitourinary symptoms, as well as sleep and mood disturbances are associated with multiple adverse work outcomes. Studies to date have demonstrated that these adverse work outcomes related to menopause symptoms include a compromised ability to work, reduced work productivity, absenteeism, and even loss of employment or an early exit from the workforce. Further, the relationship between menopause symptoms and work may be bidirectional, with certain aspects of the work environment being linked with a greater menopause symptom burden, such as insufficient restroom facilities, unpredictable or long work hours, the inability to take breaks, and confined or crowded workspaces. Thus, workplace solutions may need to be tailored based on women's individual needs, the work environment, and the type of work.The Menopause Society, in conjunction with an expert panel of medical and legal experts and women's health advocates, has developed a set of consensus recommendations that challenges employers to create a menopause-supportive workplace for their employees. These recommendations include, among other things, suggestions for employers to review policies and healthcare plans and benefits and to consider flexibility and accommodations that may be needed for some women with menopause symptoms. Guidance for women with menopause symptoms that affect them at work in terms of understanding their resources and empowering them to be self-advocates are also provided, as well as recommendations for what occupational health professionals should know and do for women with bothersome menopause symptoms in the workplace.

ToEx: Accelerating Generation Stage of Transformer-Based Language Models via Token-Adaptive Early Exit

ToEx: Accelerating Generation Stage of Transformer-Based Language Models via Token-Adaptive Early Exit

Building a Defence Manpower Planning Tool Using Survival Analysis Models

ABSTRACT A vibrant economy results in strong competition in the marketplace for highly skilled and versatile workforce. Therefore, the retention of such workers is an important issue for most organisations, including the Royal Navy, which is sensitive to economic pressures from the civilian economy. This paper considers early exits from the Royal Navy as the termination of an individual’s career in the naval force at any time before the termination of their contractual agreement with the organisation. The paper uses a survival analysis model to forecast/predict early exits from the Royal Navy and help build a defence manpower planning tool to manage retentions. At the micro-level, it provides two sets of predictions, the first on who is more likely to leave early, and the second on when that event is most likely to occur. At the aggregate level, the predictive model provides a management tool for performing stress testing in relation to potential changes in economic and labour market conditions and how the heterogeneous skilled workforce within the organisation is likely to respond to different changes.

Exploring work ability, psychosocial job demands and resources of employees in low-skilled jobs: a German cross-sectional study

BackgroundExtending working lives due to labour market and pension regulations makes maintaining and promoting work ability necessary. The coronavirus pandemic has shown that employees in low-skilled jobs (no qualification required) contribute significantly to society and the economy. Research on these employees has been neglected in Germany for many decades despite demanding working conditions. Therefore, we investigate the relationship between low-skilled jobs and work ability. Moreover, we explore this relationship’s variation by psychosocial work demands and resources.MethodsWe use two waves of the German Study on Mental Health at Work (S-MGA). We calculate Ordinary-Least-Squares (OLS) regression models with pooled data (n = 6,050) to analyse the relationship between job requirement level and work ability. We also explore the contribution of job demands and resources on this relationship with interaction models. We use the Copenhagen Psychosocial Questionnaire (COPSOQ), to assess psychosocial work demands and resources.ResultsEmployees performing low-skilled jobs report significantly less work ability than those in medium- or high-skilled jobs. Interaction models show significantly greater work ability for employees in medium- and high-skilled jobs with high influence on their work (amount or tasks). Unexpectedly, employees in low-skilled jobs have lower work ability with more influence on their work. Furthermore, high role clarity, describing responsibility, authority and work goals, is associated with lower levels of work ability among employees in low-skilled jobs.ConclusionsThe moderating effect of role clarity on the work ability of employees in low-skilled jobs can possibly be attributed to skills mismatch and limited responsibility, as well as a lack of self-perceived collective purpose of the job. The moderation of the influence on work dimension supports results of previous studies. Too much job autonomy can have negative effects under certain circumstances and is therefore perceived as a job demand in some studies. Consequently, mechanisms concerning psychosocial work demands and resources must be investigated in further studies with different theoretical approaches. The imbalance of job demands and resources shows that employers should invest in preserving the work ability to prevent early exit from the labour market in an aging society.

Comparing the microbiological characteristics of early vs late enteral feeding tube exit site infection

Comparing the microbiological characteristics of early vs late enteral feeding tube exit site infection

Hierarchical Training of Deep Neural Networks Using Early Exiting.

Deep neural networks (DNNs) provide state-of-the-art accuracy for vision tasks, but they require significant resources for training. Thus, they are trained on cloud servers far from the edge devices that acquire the data. This issue increases communication cost, runtime, and privacy concerns. In this study, a novel hierarchical training method for DNNs is proposed that uses early exits in a divided architecture between edge and cloud workers to reduce the communication cost, training runtime, and privacy concerns. The method proposes a brand-new use case for early exits to separate the backward pass of neural networks between the edge and the cloud during the training phase. We address the issues of most available methods that, due to the sequential nature of the training phase, cannot train the levels of hierarchy simultaneously or they do it with the cost of compromising privacy. In contrast, our method can use both edge and cloud workers simultaneously, does not share the raw input data with the cloud, and does not require communication during the backward pass. Several simulations and on-device experiments for different neural network architectures demonstrate the effectiveness of this method. It is shown that the proposed method reduces the training runtime for VGG-16 and ResNet-18 architectures by 29% and 61% in CIFAR-10 classification and by 25% and 81% in Tiny ImageNet classification, respectively, when the communication with the cloud is done over a low bit rate channel. This gain in the runtime is achieved, while the accuracy drop is negligible. This method is advantageous for online learning of high-accuracy DNNs on sensor-holding low-resource devices such as mobile phones or robots as a part of an edge-cloud system, making them more flexible in facing new tasks and classes of data.

Labour market trajectories and unemployment of older workers in Europe after the Great Recession

AbstractMany European countries want to overcome the early exit from labour market which was widespread since the 1970s through pension reform and labour policies. The extension of working life is hindered by factors that the literature overlooks. This article focuses on the discontinuity of the late‐career caused by unemployment. The aim is to investigate whether older workers are at risk of unemployment and what are the transitions after spells of unemployment. The labour market trajectories of workers over 50s in 11 European countries from 2006 to 2019 were analysed, specifically after the Great Recession and before the Covid‐19 epidemic. It is based on longitudinal survey data from the generated Survey of Health, Ageing and Retirement in Europe Job Episodes Panel. Event history analysis and sequence analysis were performed. Older employees are unlikely to face unemployment. The transition rate highlights the difficulty for older workers into re‐employment. A low education level increases the risk of unemployment and remaining unemployed. Data by country confirm the trend towards overcoming the transition from unemployment to early exit. Divergences concern the higher risk of being trapped in unemployment in Southern European countries such as Greece and Italy.

Dynamic hierarchical intrusion detection task offloading in IoT edge networks

AbstractThe Internet of Things (IoT) has gained widespread importance in recent time. However, the related issues of security and privacy persist in such IoT networks. Owing to device limitations in terms of computational power and storage, standard protection approaches cannot be deployed. In this article, we propose a lightweight distributed intrusion detection system (IDS) framework, called FCAFE‐BNET (Fog based Context Aware Feature Extraction using BranchyNET). The proposed FCAFE‐BNET approach considers versatile network conditions, such as varying bandwidths and data loads, while allocating inference tasks to cloud/edge resources. FCAFE‐BNET is able to adjust to dynamic network conditions. This can be advantageous for applications with particular quality of service requirements, such as video streaming or real‐time communication, ensuring a steady and reliable performance. Early exit deep neural networks (DNNs) have been employed for faster inference generation at the edge. Often, the weights that the model learns in the initial layer may be sufficiently qualified to perform the required classification tasks. Instead of using subsequent layers of DNNs for generating the inference, we have employed the early‐exit mechanism in the DNNs. Such DNNs help to predict a wide range of testing samples through these early‐exit branches, upon crossing a threshold. This method maintains the confidence values corresponding to the inference. Employing this approach, we achieved a faster inference, with significantly high accuracy. Comparative studies exploit manual feature extraction techniques, that can potentially overlook certain valuable patterns, thus degrading classification performance. The proposed framework converts textual/tabular data into 2‐D images, allowing the DNN model to autonomously learns its own features. This conversion scheme facilitated the identification of various intrusion types, ranging from 5 to 14 different categories. FCAFE‐BNET works for both network‐based and host‐based IDS: NIDS and HIDS. Our experiments demonstrate that, in comparison with recent approaches, FCAFE‐BNET achieves a 39.12%–50.23% reduction in the total inference time on benchmark real‐world datasets, such as: NSL‐KDD, UNSW‐NB 15, ToN_IoT, and ADFA_LD.

A flexible BERT model enabling width- and depth-dynamic inference

Fine-tuning and inference on Large Language Models like BERT have become increasingly expensive regarding memory cost and computation resources. The recently proposed computation-flexible BERT models facilitate their deployment in varied computational environments. Training such flexible BERT models involves jointly optimizing multiple BERT subnets, which will unavoidably interfere with one another. Besides, the performance of large subnets is limited by the performance gap between the smallest subnet and the supernet, despite efforts to enhance the smaller subnets. In this regard, we propose layer-wise Neural grafting to boost BERT subnets, especially the larger ones. The proposed method improves the average performance of the subnets on six GLUE tasks and boosts the supernets on all GLUE tasks and the SQuAD data set. Based on the boosted subnets, we further build an inference framework enabling practical width- and depth-dynamic inference regarding different inputs by combining width-dynamic gating modules and early exit off-ramps in the depth dimension. Experimental results show that the proposed framework achieves a better dynamic inference range than other methods in terms of trading off performance and computational complexity on four GLUE tasks and SQuAD. In particular, our best-tradeoff inference result outperforms other fixed-size models with similar amount of computations. Compared to BERT-Base, the proposed inference framework yields a 1.3-point improvement in the average GLUE score and a 2.2-point increase in the F1 score on SQuAD, while reducing computations by around 45%.

Implementation of Lightweight Convolutional Neural Networks with an Early Exit Mechanism Utilizing 40 nm CMOS Process for Fire Detection in Unmanned Aerial Vehicles.

The advancement of unmanned aerial vehicles (UAVs) enables early detection of numerous disasters. Efforts have been made to automate the monitoring of data from UAVs, with machine learning methods recently attracting significant interest. These solutions often face challenges with high computational costs and energy usage. Conventionally, data from UAVs are processed using cloud computing, where they are sent to the cloud for analysis. However, this method might not meet the real-time needs of disaster relief scenarios. In contrast, edge computing provides real-time processing at the site but still struggles with computational and energy efficiency issues. To overcome these obstacles and enhance resource utilization, this paper presents a convolutional neural network (CNN) model with an early exit mechanism designed for fire detection in UAVs. This model is implemented using TSMC 40 nm CMOS technology, which aids in hardware acceleration. Notably, the neural network has a modest parameter count of 11.2 k. In the hardware computation part, the CNN circuit completes fire detection in approximately 230,000 cycles. Power-gating techniques are also used to turn off inactive memory, contributing to reduced power consumption. The experimental results show that this neural network reaches a maximum accuracy of 81.49% in the hardware implementation stage. After automatic layout and routing, the CNN hardware accelerator can operate at 300 MHz, consuming 117 mW of power.

Early Exit Decisions in Finland: Investigating the Limbo between Unemployment and Retirement

Despite efforts to extend careers, workers continue to withdraw from working life before reaching the old-age pension age. The complexities behind early exit decisions have been extensively studied to understand the individual factors influencing the decision-making processes. However, fewer studies have attempted to understand how underlying cultural notions of age shape these decisions. Using qualitative longitudinal interview data from Finland, this article examined how individuals in the Finnish unemployment pathway to retirement position themselves between unemployment and retirement by drawing on age-specific discourses. Findings show that early exit was framed as the least problematic career option, although colored with ambiguity and anxiety. Individuals did not view themselves as retirees, and some expressed interest in continuing work. Nevertheless, age norms together with experienced, internalized, and expected ageism pushed individuals toward exit, highlighting the need to better understand and support individuals experiencing late-life unemployment.

ConsistentEE: A Consistent and Hardness-Guided Early Exiting Method for Accelerating Language Models Inference

Early Exiting is one of the most popular methods to achieve efficient inference. Current early exiting methods adopt the (weighted) sum of the cross entropy loss of all internal classifiers as the objective function during training, imposing all these classifiers to predict all instances correctly. However, during inference, as long as one internal classifier predicts an instance correctly, it can accelerate without losing accuracy. Thus, there is a notable gap between training and inference. We propose ConsistentEE, an early exiting method that is consistent in training and inference. ConsistentEE formulates the early exiting process as a reinforcement learning problem. A policy network is added to decide whether an instance should exit or continue. The training objective of ConsistentEE only requires each instance to be predicted correctly by one internal classifier. Additionally, we introduce the concept "Memorized Layer" to measure the hardness of an instance. We incorporate the memorized layer into reward function design, which allows "easy'' instances to focus more on acceleration while ``hard'' instances to focus more on accuracy. Experimental results show that our method outperforms other baselines on various natural language understanding and generation tasks using PLMs and LLMs as backbones respectively.