Explore the vast range of titles available.

Sinonasal malignancies such as lymphoepithelial carcinoma (LEC) and esthesioneuroblastoma (ENB) are rare but aggressive neoplasms that frequently necessitate extensive resection of anterior skull base structures, creating complex defects. This case series discusses two patients with recurrent sinonasal malignancies who underwent anterior skull base reconstruction using microvascular free flaps. The first case involved a recurrent LEC managed with a radial forearm free flap, and the second case involved a recurrent ENB with cerebrospinal fluid (CSF) leaks managed via an anterolateral thigh free flap. Both cases highlight the role of free flap reconstruction in managing large, high-risk defects while addressing complications such as CSF leaks, infections, and prior radiation-induced tissue damage. This paper underscores a reconstructive algorithm for managing extensive anterior skull base defects using microvascular free tissue transfer.

Do firms benefit more from agglomeration-based spillovers than the technical know-how obtained through inter-firm collaboration? Quantifying the relative value of the industrial policy of cluster development vis-à-vis firm’s internal decision of collaboration can be valuable for policy-makers and entrepreneurs. I observe the universe of Indian MSMEs inside an industrial cluster (Treatment Group 1), those in collaboration for technical know-how (Treatment Group 2) and those outside clusters with no collaboration (Control Group). Conventional econometric methods to identify the treatment effects would suffer from selection bias and misspecification of the model. I use two data-driven, model-selection methods, developed by (Belloni, A., Chernozhukov, V., and Hansen, C. (2013). Inference on treatment e ects after selection among high-dimensional controls. Review of Economic Studies, 81(2):608 650.) and (Chernozhukov, V., Hansen, C., and Spindler, M. (2015). Post selection and post regulariza- tion inference in linear models with many controls and instruments. American Economic Review, 105(5):486 490.), to estimate the causal impact of the treatments on GVA of firms. The results suggest that ATE of cluster and collaboration is nearly equal at 30%. I conclude by offering policy implications.

PurposeFinancial access is key to achieving several economic goals in developing countries. This paper aims to construct a longitudinal village-level measure of financial access in India and understand the role of RBI's policies and village characteristics in influencing the access.Design/methodology/approachThe authors adopt a spatial approach in developing a metric of financial access. In particular, they measure the distance of each unbanked village in India to the nearest banked-centre from 1951 to 2019. The authors use this measure to conduct two exercises. First, a descriptive study is undertaken to assess how RBI's policies on bank branch expansion from 1951 to 2019 influenced the proximity to bank branches. Second, the authors conduct regression analyses to investigate how socio-economic and demographic characteristics of villages influence their proximity to bank branches.FindingsThe average distance of an unbanked village to the nearest banked-centre has declined from 43.5 km in 1951 to 4.2 km in 2019. The gain in bank access has varied geographically and over time. In 2001, bank branches were relatively distant from villages with under-privileged caste groups and proximate to areas with better infrastructure. This relationship worsened after 2005 when RBI introduced liberalized branch expansion policies. By 2019, proximity responds much more adversely to the presence of underprivileged groups. At the same time, banks have moved closer to economically better-off villages and villages with workforce in non-farm enterprises rather than agriculture.Originality/valueFirst, studies in the Indian context focus on state-level determinants of bank branching, this is the first study to develop a longitudinal measure of financial access at the village level. This helps to understand spatial heterogeneity in bank branch access within states, which other studies are unable to do. Second, the paper analyses the role of village-level socio-economic and demographic characteristics in proximity to bank branches. This analysis helps in discovering micro-foundations of growth of bank branch network. The granularity of the approach adopted here overcomes the confoundedness problems that the studies at a more aggregate level face.

Despite several advantages associated with borrowing from micro-finance institutions, such as self-help groups (SHGs), many enterprises in developing countries continue to rely on informal lenders. Using machine learning techniques on a novel village-enterprise matched dataset from India, we predict an enterprise’s choice of credit source as a function of three key mechanisms: supply-side factors, infrastructural facilities and socio-demographic characteristics. Proximity to markets and social norms of the village, proxied by high literacy rates and sex ratios, play important roles in credit uptake from SHGs. However, the absence of financial access points, such as commercial or cooperative bank branches, is not prohibitive.

In this thesis we develop optimization-based methods to deal with uncertainty arising from data, first in the context of robust multi-class prediction and second for prescriptive analytics for medical resource allocation.In the first part, we make progress on training robust multi-class classifiers using error-correcting output codes (ECOC). We propose linear and non-linear integer programming (IP) formulations for the codebook design problem. By making connections with graph-theory such as edge-clique covering and graph-coloring, we develop tractable solutions approaches to both linear and non-linear IP formulations while maintaining low-optimality gaps, estimated using Plotkin's bound.We provide extensive computational experiments on small class datasets including MNIST and CIFAR10. In the nominal setting, our IP-generated compact codebooks outperform commonly used large codebooks. Furthermore, in the adversarial setting, our IP-generated codebooks achieve non-trivial robustness. This is surprising due to three reasons: (1) We do not employ any {adversarial training}; (2) Most other codebooks (except Dense) do not exhibit any robustness even when they use more than twice the number of columns; (3) The robustness that we obtain is not simply because of the large network capacity. On large class datasets such as CIFAR100, Caltech-101 and Caltech-256, we leverage transfer-learning to overcome the large computational expense associated. We provide experiments under two different settings, first when the source classifier is nominally trained and second when it is adversarially trained. ECOC-based classifiers achieve better classification performance in comparison to multiclass CNNs in both settings. These experiments indicate that our large-scale discrete optimization approaches for designing ECOC-based classifiers can be extremely useful for robust operation of modern urban-systems.In second part of this thesis we shift our focus from robust prediction to developing a new approach for prescriptive analytics. We make progress on the problem of uncertainty informed medical resource (vaccine) allocation to a set of different sub-populations to control the spread of a pandemic such as Covid-19. Here, we tackle two major challenges: (1) To develop a principled data-driven approach to model and estimate uncertainty in the parameters of a system of ordinary differential equations (ODE) based compartmentalized epidemiological model. (2) To develop tools to solve a large-scale, non-linear optimization problem which is constrained by ODE dynamics with uncertain parameters.We provide a data-driven approach to generate a tractable scenario set by estimating the posterior-distribution on the model parameters using Bayesian inference with Gaussian processes. Using the scenario set, we provide the nominal and stochastic (i.e. uncertainty informed) formulations for optimal vaccine allocation. We develop a parallelized solution algorithm to efficiently solve both nominal and stochastic optimization problems. Importantly, our scenario-set estimation procedure, optimization formulations and solution approach are all flexible in that they are not limited to any particular class of ODE models. We provide experiments with two different non-linear epidemiological ODE models under different setups. Our computational experiments indicate that accounting for uncertainty in key epidemiological parameters can improve the efficacy of time-critical allocation decisions by 4-8%.

Multi-Armed bandit (MAB) framework is a widely used sequential decision making framework in which a decision-maker needs to select one of the available K actions in each round, with the objective of maximizing their long-term reward. This framework has been used in practice for several applications including web advertising, medical testing by viewing the use of different ads/treatments as the arms in the MAB problem. The user's response corresponding to these different actions generates a reward for the decision-maker. Under the classical MAB framework, it is implicitly assumed that the rewards corresponding to different actions are independent of each other. But, this may not be the case in practice as rewards corresponding to different actions (i.e., ads/drugs) are likely to be correlated. Motivated by this, we study the structured and correlated MAB problem in this thesis. First, we study the structured MAB problem where the mean rewards corresponding to different actions are a known function of a hidden parameter, thereby imposing a structured on the mean rewards of different actions. We study this problem in the most general form by imposing no restriction on the form of mean reward functions and as a result subsume the setting of several previously studied structured bandit frameworks where the mean reward function is assumed to be of a specific form. While mean rewards of different actions may be related to one another in the structured bandit setup, the reward realizations may not necessarily be correlated. Motivated by this, we propose a novel correlated MAB framework which explicitly captures the correlation in reward across different actions. For both these frameworks, we design novel algorithms that allow us to extend any classical bandit algorithm to the structured and correlated bandit settings. Through rigorous analysis, we show that our proposed algorithms sample certain sub-optimal actions, termed as non-competitive actions, only O(1) times as opposed to the typical O(log T) samples required by classical algorithms such as Upper Confidence Bound (UCB), Thompson sampling. These significant theoretical performance gains are reflected in our experiments performed on real-world recommendation system datasets such as Movielens, Goodreads. For our proposed correlated bandit framework, we also design best-arm identification algorithms where the task is to identify the best action in as few samples as possible. We demonstrate the achieved performance gains theoretically through sample complexity analysis and empirically through experiments on recommendation system datasets. To further demonstrate the utility of our proposed correlated bandit framework, we show how the framework can be employed to solve online resource allocation problems, which frequently arise in tasks such as power allocation in wireless systems, financial optimization and multi-server scheduling. This is done by extending our correlated bandit framework and algorithms to the setting of online resource allocation. The performance gains are demonstrated theoretically through regret analysis and empirically through synthetic experiments on the task of online power allocation in wireless systems, job scheduling in multi-server systems and channel assignment under the ALOHA protocol.

This dissertation studies internal and external factors affecting firm outcomes. The first two chapters explore the sources of variation in managerial skill within an Indian life insurance firm. The existing literature has investigated the association between managerial productivity and management practices across firms, but has largely overlooked how individual traits and skills affect managerial performance. Intra-firm variation in managerial productivity allows us to study managerial skill without the confounded effects of variation in management practices. The third chapter models how external technological change affects competition between media firms, and what that implies for information availability in a society. For the first two chapters, I use a novel dataset drawn from a life insurance firm in India, with 211 managers, each leading a sales team of insurance agents. Chapter 1 studies the sources of large variation in performance across teams. I find that the performance of newly recruited agents is positively correlated with the managers' past team productivity index. I also observe that when agents move across teams in the firm's internal labor market, there is no change in the output of such agents, except when they join the team of a high performing manager (in the top decile of team performance). This allows me to infer that most managers differ from along their recruiting skill, whereas the high performers are able to provide some form of managerial contribution to productivity such as training, supervision or guidance. Chapter 2 examines the dynamics of managerial skills in this firm. I distinguish between internally-hired managers who were working previously as agents in the firm, and externally-hired managers, who joined the firm directly as managers. I find that the teams of internally-hired managers are 14% more productive, but that the teams of externally-hired managers catch up in a span of six to seven years. Among different mechanisms, I find evidence that the managers differ in the recruitment of good workers and also in the contribution to the output of their workers. Further, I find evidence that the externally-hired managers learn how to recruit good workers. This is the first study to show evidence supporting learning-by-doing on part of managers. The third chapter, co-authored with Benjamin Ogden, develops a model of endogenous media polarization – or, product differentiation among news sources – to study how this affects political outcomes. We show that under internet-based technology, where users provide additional values when they are served their preferred content, media firms would have an incentive to skew their content, leading to divergence. However, the degree of divergence will depend on the distribution of audience. Under reasonable restrictions on the distribution of voters, informed political choices are implemented. The model demonstrates why increasing media polarization does not necessarily lead to incorrect political outcomes and may in fact create conditions for correct policy choice.

Background: Basal cell carcinoma (BCC) can be categorized as one of the commonly occurring skin malignancies in the world, with several variations in treatment protocols. Sun exposure has been attributed to its causality; however, other factors such as gender, age, and occupation also affect its incidence. We aimed to characterize the patient population who underwent surgical management for facial BCC at a tertiary referral hospital. Further, we have described an algorithm that may aid in surgical decision-making based on the location of the lesions on the face. Materials and Methods: We performed a retrospective chart review of all patients who presented with a facial BCC to our institution between 2018 and 2019. Data regarding patients' demographic characteristics, skin phototype, average sun exposure, occupation, residence place (rural or urban), and surgical outcomes were recorded. Results: Sixty-eight patients underwent reconstructive procedures after oncologic resection of facial BCC: 41.2% were males and 58.8% were females. Forty-eight (70.6%) patients were from rural areas, and 20 patients (29.4%) from urban areas (P < 0.001). Twenty-six patients reported >2 h of sunlight exposure, 16 reported <2 h of continuous sun exposure, and 26 reported intermittent sun exposure. A significantly higher proportion of patients with facial BCC presented with a Fitzpatrick skin type 4 in comparison to types 3 and 5 (P < 0.001). The most common reconstructive technique was the V-Y advancement flap (n=22, 32.4%), followed by the forehead flap (n=12, 17.6%) and the Limberg flap (n=12, 17.6%). All the flaps were healthy post-operatively and none of them suffered from flap failure, infection, or suture line dehiscence. There was no recurrence at 1-year follow-up. Conclusion: This study gives a correlation between incidence of BCC and age, gender, and sun exposure in Indian population. In our experience, local flaps yield outstanding results and are the first choice for reconstruction of the face when composite defects are not present. Our algorithm aids in surgical decision-making.

The implementation of the COVID-19 national lockdown announced suddenly in March 2020 in India provided a unique opportunity to capture real-time changes in business sentiments during episodes of unexpected and sudden disruptions. Using a logit-probability model to analyse data of this natural experiment showed that firms’ 6-months ahead sentiments for its financial condition worsened drastically during lockdown compared to firms surveyed immediately prior to the announcement. Further, smaller firms showed a relatively higher impact. We also find that firms perceive this as a relatively higher demand shock in terms of falling domestic sales post-lockdown whereas supply shocks are perceived to be on the downside. Lastly the mitigation strategy of firms involved reducing employment for unskilled workers and wages for skilled workers. This unique study gives insights not only about firms and their strategies but regarding appropriate policy choices during lockdown. The lessons are applicable for governments which imposed local lockdowns during the second wave and potential disruption for the expected third wave.

Purpose: The aim of this research is to contrast transaction multiples of mergers and acquisitions backed by private equity and venture capital investors against all other types of transactions. Design: The researcher conducted an independent samples t-test on 20-year historical data of mergers and acquisitions gathered through Capital IQ database. Findings: The researcher found that there is a statistically significant difference in transaction multiples of mergers and acquisitions backed by private equity and venture capital firms as compared to those mergers and acquisitions that were not backed by a private equity or venture capital firm. Originality and Value: The findings of this study will provide evidence to suggest that industry group may play an important role in the transaction multiples of deals backed by venture capital and private equity investors.