Climate change affects agricultural land productivity. This risk is more pronounced in agrarian economies, where land constitutes farmers' primary asset. Existing research has largely focused on the effects of climate change on crop yields or income, and with limited attention to how climate change is capitalized into parcel-level land prices in emerging markets. This study addresses the gap by estimating price discounts associated with RCP4.5/8.5 climate scenarios for individual land parcels, controlling for parcel-level micro-location attributes. A sample of land price data (2019–2024) from the coastal state of India, Andhra Pradesh, reveals that markets incorporate climate risks into land prices. Specifically, each 1% increase in rainfall is associated with 3% increase in land price, but a 1°C increase in projected temperature corresponds to about 47% decline in price. The net effect diverges across segments: rural parcels benefit under extreme climate change scenarios, whereas median and urbanizing segments experience 20–35% price declines. The study also finds that micro-location factors explain up to 60% of price variation, exceeding the explanatory power of district fixed effects. Parcel size shows diminishing returns: a 1% increase raises price by only 0.66–0.72%. Proximity to highways generates 12–15% price premiums, while proximity to expressways is associated with premiums exceeding 130%.
Celebrity capital refers to the celebrity's level and frequency of media visibility and consists of four stages: acquisition, consolidation, decline and resurgence. Two between-subject experiments with non-student respondents find that consolidation-stage endorsers are the most effective in positively influencing consumer evaluations, followed by acquisition-stage endorsers. Advertisements featuring decline-stage celebrities are no different from endorser-less ads. Furthermore, while consolidation-stage celebrities are more effective for unfamiliar brands, acquisition-stage celebrities are as effective as consolidation celebrities for familiar brands. Our findings suggest that brands should only retain consolidation or acquisition-stage endorsers.
This study aims to examine research trends in Library and Information Science (LIS) in India through a systematic review of relevant literature. Thirty-three research publications focusing on LIS research trends in the Indian context were evaluated. Each article was reviewed to identify the key topic and the research methodologies employed. All the main LIS themes were then categorised into their respective decade for further analysis. The results indicate a significant focus on bibliometric and scientometric studies, as well as research on informationseeking behavior in recent decades. In the earlier years (1950-1960), LIS research primarily centered on themes such as classification, cataloguing, information retrieval and the LIS profession. However, the prominence of these topics has declined over time. In addition to bibliometric and information-seeking studies, researchers have also explored areas including various types of libraries (academic, university, public and special libraries), LIS education, digital libraries, library automation, electronic resources and webometrics. The study reveals a notable deficiency in methodological transparency, as 58 % of the examined articles do not describe the research methods used to derive their findings. This suggests a need for greater methodological rigor and reporting in LIS research conducted in India.
We study the nature and effects of cultural biases in choice under risk and uncertainty by comparing peer-to-peer loans the same individuals (lenders) make alone and after observing robo-advised suggestions. When unassisted, lenders are more likely to choose co-ethnic borrowers, facing 8% higher defaults and 7.3pp lower returns. Robo-advising does not affect diversification but reduces lending to high-risk co-ethnic borrowers. Lenders in locations with high inter-ethnic animus drive the results, even when borrowers reside elsewhere. Biased beliefs explain these results better than a conscious taste for discrimination: lenders rarely override robo-advised matches to ethnicities they discriminated against when unassisted.
This paper proposes a Linear Programming (LP)-based local search framework for fine-tuning pretrained transformer models with explicit control against overfitting. The approach formulates transformer fine-tuning as a bilevel optimization-based regularization problem, in which model parameters and regularization hyperparameters are jointly updated. Information collected during initial warm-up iterations, including validation gradients and training Hessian information, is used to construct a local descent direction by solving an LP that minimizes a scaled directional derivative while preserving training optimality. This validation-aware descent direction enables focused local updates of both parameters and regularization hyperparameters, reducing overfitting without requiring repeated full retraining cycles. The resulting method, termed Linear Programming-based Fine-Tuning (LiFT) for transformers, differs from conventional fine-tuning by systematically identifying task-specific updates rather than relying on heuristic or grid-based hyperparameter selection. Experiments on GPT-2 Small fine-tuned on WikiText-2 demonstrate that LiFT enables effective adaptation through selective tuning of transformer blocks and regularization parameters, yielding consistent improvements in test perplexity across multiple layer configurations and regularization settings, with particularly pronounced gains in overfitting-prone scenarios. Beyond empirical performance, LiFT establishes a principled connection between transformer fine-tuning, bilevel optimization, local search, and regularization theory.
