Accurately fitting the term structure of interest rates is critical to central banks and other market participants. The Nelson-Siegel and Nelson-Siegel-Svensson models are probably the best-known models for this purpose due to their intuitive appeal and simple representation. However, this simplicity comes at a price. The difficulty in calibrating these models is twofold. Firstly, the objective function being minimized during the calibration procedure is nonlinear and has multiple local optima. Secondly, there is strong co-dependence among the model parameters. As a result, their estimated values behave erratically over time. To avoid these problems, we apply a heuristic optimization method, specifically the Genetic Algorithm approach, and show that it is able to construct reliable interest rate curves and stable model parameters over time, regardless of the shape of the curves.

Deep Reinforcement learning is a branch of unsupervised learning in which an agent learns to act based on environment state in order to maximize its total reward. Deep reinforcement learning provides good opportunity to model the complexity of portfolio choice in high-dimensional and data-driven environment by leveraging the powerful representation of deep neural networks. In this paper, we build a portfolio management system using direct deep reinforcement learning to make optimal portfolio choice periodically among S\&P500 underlying stocks by learning a good factor representation (as input). The result shows that an effective learning of market conditions and optimal portfolio allocations can significantly outperform the average market.

We study a bivariate latent factor model for the pricing of commodity fu- tures. The two unobservable state variables representing the short and long term fac- tors are modelled as Ornstein-Uhlenbeck (OU) processes. The Kalman Filter (KF) algorithm has been implemented to estimate the unobservable factors as well as unknown model parameters. The estimates of model parameters were obtained by maximising a Gaussian likelihood function. The algorithm has been applied to WTI Crude Oil NYMEX futures data.

Motivated by recent developments in risk management based on the U.S. bankruptcy code, we revisit De Finetti optimal dividend problems by incorporating the reorganization process and regulator's intervention documented in Chapter 11 bankruptcy. The resulting surplus process, bearing financial stress towards the more subtle concept of bankruptcy, corresponds to non-standard spectrally negative Levy processes with endogenous regime switching. In both models without and with fixed transaction costs, some explicit expressions of the expected net present values under a barrier strategy, new to the literature, are established in terms of scale functions. With the help of these expressions, when the tail of the Levy measure is log-convex, the optimal dividend control in each problem is verified to be of the barrier type and the associated optimal barrier can be obtained in analytical form.

Rough Volterra volatility models are a progressive and promising field of research in derivative pricing. Although rough fractional stochastic volatility models already proved to be superior in real market data fitting, techniques used in simulation of these models are still inefficient in terms of speed and accuracy. This paper aims to present the accurate tools and techniques that could be used also in nowadays largely emerging pricing methods based on machine learning. In particular, we compare three widely used simulation methods: the Cholesky method, the Hybrid scheme, and the rDonsker scheme. We also comment on implementation of variance reduction techniques. In particular, we show the obstacles of the so-called turbocharging technique whose performance is sometimes contra productive. To overcome these obstacles, we suggest several modifications.

The two unobservable state variables representing the short and long term factors introduced by Schwartz and Smith in [16] for risk-neutral pricing of futures contracts are modelled as two correlated Ornstein-Uhlenbeck processes. The Kalman Filter (KF) method has been implemented to estimate the short and long term factors jointly with un- known model parameters. The parameter identification problem arising within the likelihood function in the KF has been addressed by introduc- ing an additional constraint. The obtained model parameter estimates are the conditional Maximum Likelihood Estimators (MLEs) evaluated within the KF. Consistency of the conditional MLEs is studied. The methodology has been tested on simulated data.

Social scientists have become increasingly interested in how narratives -- the stories in fiction, politics, and life -- shape beliefs, behavior, and government policies. This paper provides a novel, unsupervised method to quantify latent narrative structures in text, drawing on computational linguistics tools for clustering coherent entity groups and capturing relations between them. After validating the method, we provide an application to the U.S. Congressional Record to analyze political and economic narratives in recent decades. Our analysis highlights the dynamics, sentiment, polarization, and interconnectedness of narratives in political discourse.

Despite the key role of multinational enterprises (MNEs) in both international markets and domestic economies, there is no consensus on their impact on their host economy. In particular, do MNEs stimulate new domestic firms through knowledge spillovers? Here, we look at the impact of MNEs on the entry and exit of domestic industries in Irish regions before, during, and after the 2008 Financial Crisis. Specifically, we are interested in whether the presence of MNEs in a region results in knowledge spillovers and the creation of new domestic industries in related sectors. To quantify how related an industry is to a region's industry basket we propose two cohesion measures, weighted closeness and strategic closeness, which capture direct linkages and the complex connectivity structure between industries in a region respectively. We use a dataset of government-supported firms in Ireland (covering 90% of manufacturing and exporting) between 2006-2019. We find that domestic industries are both more likely to enter and less likely to leave a region if they are related to so-called 'overlapping' industries containing both domestic and MNE firms. In contrast, we find a negative impact on domestic entry and survival from cohesion to 'exclusive MNE' industries, suggesting that domestic firms are unable to 'leap' and thrive in MNE-proximate industries likely due to a technology or know-how gap. This dynamic was broken, with domestic firms entering MNE exclusive sectors, by a large injection of Brexit diversification funds in 2017-18. Finally, the type of cohesion matters. For example, strategic rather than weighted closeness to exclusive domestic sectors matters for both entries and exits.