24 research outputs found
Why better models do not always lead to better decisions
Get PDFIt was the environmental academic Jerome Ravetz who said: “We believe in numbers, just because they are numbers.” Other science historians, including Theodore Porter, have long held the same view. This statement could not be more relevant as the global economy recovers from the biggest financial crisis since 1929. Many in management privately agree that we use numerical decision models with near religiousness, to convince that our decisions are scientifically robust, as few will doubt a well-founded, widely accepted mathematical model: we call on the legitimacy of arithmetic to persuade rather than prove
U.K. banking experts as decision-makers: a historical view on banking technologies
No full textThis paper looks into Technological Investment Decision-Making (henceforth TIDM), and aims to surface the influence of the historical evolution of expertise in the banking sector on such decisions. In so doing, the paper underlines that, in each era, the technological decision-making activity is strongly linked to the views of dominant expert
groups in the industry. The paper proposes that TIDM, as viewed historically, has been highly contingent to both technological developments in banking and subsequent
developments in banking expertise.To illustrate this, it explores the relationship between (1)the technological developments in UK banking, (2) the evolution and distribution of expertise in the sector and (3) the respective evolution of TIDM. The paper adopts a historical
perspective spanning from the end of the 19th Century until today, and identifies patterns that connect the 3 historical processes through direct, albeit hidden, relationships. It illustrates that, counter to popular belief, TIDM is a socially constructed process as opposed to a normative exercise. Ultimately, this paper shows that history recognises no optimal method for TIDM. Rigour and accuracy of execution do not appear to define success. On the contrary, in each era, the "right way" to perform TIDM has always been underpinned by the standpoints and beliefs of specialist practitioners who dominated the UK banking industry and by the received wisdom of a community of expert professionals, administrators and think tanks, dictating "realities" on the state of the economy, the role of banks and the value of technologies
Technology investment decision making: an integrated analysis in UK Internet Banking
Get PDFThe research addresses the problem of technological investment decision making (TIDM) in UK Banks. It focuses on Internet Banking technologies and uses interviews with bank executives and industry practitioners to form a coherent understanding of how technological decisions are practically made and what, in that process, is the role of evaluation techniques. The aims of the research are (1) to identify and explain the discord between formal and practical evaluations of technologies, (2) to review the roles of expert professional groups in defining the norms of evaluation, and (3) to develop a model to reflect the reality of TIDM in UK banking. The ultimate aim is to contribute to reducing the ambiguity that notoriously characterises the evaluation of new technology.According to the theoretical framework the TIDM problem is socially constructed by expert groups (actors) who either participate in decision-making or assume roles in developing methodologies for facilitating it. Its ultimate shape is the outcome of negotiations between these viewpoints, in light of expert power positions and political advocacy Three classes of such "actors" are identified: (1) Practitioners, namely experts in Financial Institutions, (2) Observers, academic researchers, consultants and government bodies, and (3) the Community of Received Wisdom, comprising the commonly understood views on what TIDM is and how it should be made.A novel methodological approach is introduced as a variant of Grounded Theory. Called Informed Grounded Theory (IGT), it proposes that viewpoints are by default informed by individuals' academic and professional training; thus, past theory should not be considered as a contaminating factor for the data and their interpretation, (as Grounded Theory proposes) but as integral part of it.The key findings of the research concern (1) the unconventional usage of financial and other formal methodologies in TIDM practice, (2) the highly political role of dominant expert groups and the resulting dynamics of their development, (3) the influence of the wider economic cycles on how technological value is perceived and (4) the changing role of the Finance function in technological investment justification. The core conclusion from these points is that TIDM in UK banks is an act of justification and advocacy, far more than it is an assessment process; valuation techniques play an ancillary role in ascertaining views often founded on purely strategic or political grounds.The research recommends an interdisciplinary approach to improving TIDM methodologies. Unlike the traditional paradigm which might be characterised as improvable measurement, where measurement precision is sought as the solution to the valuation ambiguity, it is proposed that we seek improvement by taking explicit account of the perceptions of expert groups, as these are encoded into existing formal methodologies, and thus offer only partial evaluations. By mobilising these partialities, newer approaches may provide for including socio-political as well as economic factors in technological valuation processes.The research recommends an interdisciplinary approach to improving TIDM methodologies. Unlike the traditional paradigm which might be characterised as improvable measurement, where measurement precision is sought as the solution to the valuation ambiguity, it is proposed that we seek improvement by taking explicit account of the perceptions of expert groups, as these are encoded into existing formal methodologies, and thus offer only partial evaluations. By mobilising these partialities, newer approaches may provide for including socio-political as well as economic factors in technological valuation processes
The history of banking technologies in the UK: patterns of technological investment decision-making and expertise
Get PDFThe central role of banking in the 2008 credit crisis has been the source of much controversy about the quality and robustness of decision-making in the Financial Services sector. This paper aims to surface the influence of the historical evolution of expertise in the banking sector, on such decisions and, in so doing, to underline that the decision-making activity is strongly linked to the views of dominant expert groups in the industry in each era. The paper proposes that Technological Investment Decision Making (TIDM), as viewed historically, has been highly contingent to both technological developments in banking and the subsequent developments in banking expertise that provides the pool for decision-makers in the industry.The paper adopts an historical perspective to illustrate that, counter to popular belief, TIDM is a socially constructed process rather than the outcome of any normative exercise. History demonstrates that there is no optimal method for TIDM with rigour and accuracy of execution determining successful outcomes. On the contrary, in each era, the "right way" to perform TIDM has always been underpinned by the standpoints and beliefs of specialised practitioners who dominated the UK banking industry and by the received wisdom of a community of expert professionals, administrators and think tanks, dictating "realities" on the state of the economy, the role of banks and the value of technologies
An event-driven serverless ETL pipeline on AWS
Get PDFThis work presents an event-driven Extract, Transform, and Load (ETL) pipeline serverless architecture and provides an evaluation of its performance over a range of dataflow tasks of varying frequency, velocity, and payload size. We design an experiment while using generated tabular data throughout varying data volumes, event frequencies, and processing power in order to measure: (i) the consistency of pipeline executions; (ii) reliability on data delivery; (iii) maximum payload size per pipeline; and, (iv) economic scalability (cost of chargeable tasks). We run 92 parameterised experiments on a simple AWS architecture, thus avoiding any AWS-enhanced platform features, in order to allow for unbiased assessment of our model’s performance. Our results indicate that our reference architecture can achieve time-consistent data processing of event payloads of more than 100 MB, with a throughput of 750 KB/s across four event frequencies. It is also observed that, although the utilisation of an SQS queue for data transfer enables easy concurrency control and data slicing, it becomes a bottleneck on large sized event payloads. Finally, we develop and discuss a candidate pricing model for our reference architecture usage
Computer-aided financial fraud detection: Promise and applicability in monitoring financial transaction fraud
Get PDFAnti-money Laundering (AML) and Financial Fraud Detection (FFD) have been receiving increasing attention in the past few years, especially in light of the global financial crisis. Closer systems integration and a number of latest steep technological developments in areas like Big Data; High Frequency Trading; e-payments; and mobile payment systems, to name a few, are now promising enhanced risk management through superior decision support for the global financial industry. At the same time, however, resident regulatory frameworks, national and international, appear to lack the connectivity and flexibility required to support integrated AML and FFD approaches. This is strongly testified by the disparate technological approaches to FFD across different Financial Institutions and their reluctance to share practice within the industry.
Focusing on Financial Transaction Fraud, this paper draws on the authors’ past research work which presented a prototype system that uses a workflow approach to identify abnormal financial transactions and applies Artificial Intelligence for classification. That work has shown successful applicability at short scale experiments, limited by the wide concern that information sharing should be achieved within the broader sector in order to achieve improved results. Drawing from there, this paper proposes that extending that approach across transaction infrastructure will deliver higher quality intelligent monitoring against Financial Transaction Fraud.
Following from that, we argue that the necessary technological maturity does exist to support full-scale operable FFD systems working on large disparate datasets. We then discuss the evidence in favour of the view that such systems can only be realised in the presence of wider regulatory consensus. There is, therefore, the need for a framework within which the technical infrastructure, business architecture and regulatory rules will harness that technological capability to deliver superior fraud prevention.
The paper first reviews computer-aided techniques and approaches for FFD available to the financial sector and discusses the business value of their application. It then addresses the main impediments for their full-scale applicability and uses an analytical framework for assessing their significance, in technological, business-specific and regulatory terms. A brief account of the authors’ workflow-based approach is then provided and its capabilities are outlined
Practitioners, observers and the community of received wisdom: The actor-based approach to technological investment decisions
Get PDFThis paper addresses the problem of technology valuation in UK financial institutions, specifically concerning the introduction of Internet Banking. The research looked into the prescribed processes and the respective established practices for Technological Investment Decision-Making (TIDM) in banks. Significant disparity between process and practice was found, on the grounds that the actual decisions are determined by experts’ perceptions and are less about the normative assessment of economic value, as defined in academic literature and corporate handbooks. The research suggests that the TIDM problem is socially constructed (rather than externally addressed) by experts who either participate directly in decision-making or, alternatively, contribute to developing relevant methodologies. The TIDM problem is ultimately defined by the disparate perceptions of the problem that these different interested parties, or “actors”, assume. Three classes of actors were identified: (1) Practitioners, namely expert professionals in Financial Institutions, (2) Observers, primarily academic researchers, consultants and government bodies, and (3) the Community of Received Wisdom, reflecting commonly understood views on what TIDM is and how it should be made. According to the Actor-based approach, the shape of the TIDM problem results from continuous negotiations between actors’ viewpoints, in light of expert power positions, political advocacy and fitness to the prevailing TIDM paradigms. These viewpoints are by default informed by experts’ academic and professional backgrounds, which strongly influence both the received understanding of the TIDM problem, and the perceptions of practitioner and research experts. The paper recommends that the Actor-based approach may contribute to improving TIDM: instead of seeking measurement precision as the solution to valuation ambiguities, notoriously characterising technological investment, it is suggested that we take explicit account of the differently-informed perceptions of expert groups, as these are encoded into existing formal methodologies. By mobilising these disparities, newer approaches can combine the socio-political together with the economic factors for technological valuation
Using BiLSTM networks for context-aware deep sensitivity labelling on conversational data
Get PDFInformation privacy is a critical design feature for any exchange system, with privacy-preserving applications requiring, most of the time, the identification and labeling of sensitive information. However, privacy and the concept of “sensitive information” are extremely elusive terms, as they are heavily dependent upon the context they are conveyed in. To accommodate such specificity, we first introduce a taxonomy of four context classes to categorise relationships of terms with their textual surroundings by meaning, interaction, precedence, and preference. We then propose a predictive context-aware model based on a Bidirectional Long Short Term Memory network with Conditional Random Fields (BiLSTM + CRF) to identify and label sensitive information in conversational data (multi-class sensitivity labelling). We train our model on a synthetic annotated dataset of real-world conversational data categorised in 13 sensitivity classes that we derive from the P3P standard. We parameterise and run a series of experiments featuring word and character embeddings and introduce a set of auxiliary features to improve model performance. Our results demonstrate that the BiLSTM + CRF model architecture with BERT embeddings and WordShape features is the most effective (F1 score 96.73%). Evaluation of the model is conducted under both temporal and semantic contexts, achieving a 76.33% F1 score on unseen data and outperforms Google’s Data Loss Prevention (DLP) system on sensitivity labeling tasks
Chapter 4. Regtech frontiers: innovations, trends, and insights redefining compliance
Get PDFThis chapter discusses the contributions and challenges involving regulatory technology (regtech) in financial services. It explores the salient areas where regtech can and should focus, observing existing and forthcoming industry, technology, and legal developments. This chapter outlines regtech use cases to clarify the shaping of that industry sector. It draws on developments in industry and academia, where significant research sets the tone and direction of technological solutions and regulatory drivers. A brief critical account of the benefits and challenges in regtech is offered. This chapter presents potential future directions, focusing on the salient areas of environmental, social, and governance (ESG), cryptocurrency, and decentralized compliance
A systematic local fork management framework for blockchain sandbox environments
No full textBlockchain technology presently permeates multiple industries, resulting in an increasing number of agents (primarily end-users and engineers) interacting with it in a variety of contexts. This, in turn, has introduced the practical need for a sufficient tooling ecosystem for blockchain solutions testing and evaluation under secure environment conditions. To that end, this study presents a robust framework for the creation and management of customisable, persistent, private, scalable blockchain environments that fulfil precisely the need for live sandbox platforms with on-chain interaction capabilities. We extend the concept of local chain forking, emphasising the limitations of existing tooling and methodologies and propose how our framework mitigates identified weaknesses and bridges some of the resulting gaps in desirable non-functional attributes. We offer a reference implementation for our framework and discuss how it can be applicable to a broad array of DevOps and security-oriented Use Cases. Finally, emerging challenges are discussed and potential directions for further research are drawn
