We evaluate the applicability of the generic Vickrey-Clarke-Groves (VCG) mechanism as an antimonopoly measure against a profit-maximizing producer with market power operating a portfolio of generating units at the centralized two-settlement energy market. The producer may indicate in its bid not only the altered cost function but also the distorted values of the technical parameters of its generating units, which enter the system-wide constraints of the centralized dispatch optimization problem. To ensure the applicability of the VCG method in this setting, we identify an additional assumption on the changes of the feasible set of the centralized dispatch optimization problem induced by variations of the producer's technical parameters. In the framework of the generic VCG mechanism, we propose an antimonopoly regulation method based on a regulator estimate of the producer's truthful bid. If this estimate is exact, the producer's maximum profit coincides with that in the case of the truthful bidding when no antimonopoly measure is applied. If the estimate is not exact, the error affects neither the producer's (weakly) dominant bid nor its optimal nodal output but manifests itself in the total uplift payment. This ensures an efficient allocation in the form of the optimal output/consumption schedule and shields the (pre-uplift) market prices from the producer's market power. We compare the suggested method with the alternative antimonopoly regulation approach based on the replacement of the producer's bid by a bid composed by the regulator.

Combination and aggregation techniques can significantly improve forecast accuracy. This also holds for probabilistic forecasting methods where predictive distributions are combined. There are several time-varying and adaptive weighting schemes such as Bayesian model averaging (BMA). However, the quality of different forecasts may vary not only over time but also within the distribution. For example, some distribution forecasts may be more accurate in the center of the distributions, while others are better at predicting the tails. Therefore, we introduce a new weighting method that considers the differences in performance over time and within the distribution. We discuss pointwise combination algorithms based on aggregation across quantiles that optimize with respect to the continuous ranked probability score (CRPS). After analyzing the theoretical properties of pointwise CRPS learning, we discuss B- and P-Spline-based estimation techniques for batch and online learning, based on quantile regression and prediction with expert advice. We prove that the proposed fully adaptive Bernstein online aggregation (BOA) method for pointwise CRPS online learning has optimal convergence properties. They are confirmed in simulations and a probabilistic forecasting study for European emission allowance (EUA) prices.

In this research we perform hedonic regression model to examine the residential property price determinants in the city of Boulder in the state of Colorado, USA. The urban housing markets are too compounded to be considered as homogeneous markets. The heterogeneity of an urban property market requires creation of market segmentation. To test whether residential properties in the real estate market in the city of Boulder are analyzed and predicted in the disaggregate level or at an aggregate level we stratify the housing market based on both property types and location and estimate separate hedonic price models for each submarket. The results indicate that the implicit values of the property characteristics are not identical across property types and locations in the city of Boulder and market segmentation exists.

We describe three independent implementations of a new agent-based model (ABM) that simulates a contemporary sports-betting exchange, such as those offered commercially by companies including Betfair, Smarkets, and Betdaq. The motivation for constructing this ABM, which is known as the Bristol Betting Exchange (BBE), is so that it can serve as a synthetic data generator, producing large volumes of data that can be used to develop and test new betting strategies via advanced data analytics and machine learning techniques. Betting exchanges act as online platforms on which bettors can find willing counterparties to a bet, and they do this in a way that is directly comparable to the manner in which electronic financial exchanges, such as major stock markets, act as platforms that allow traders to find willing counterparties to buy from or sell to: the platform aggregates and anonymises orders from multiple participants, showing a summary of the market that is updated in real-time. In the first instance, BBE is aimed primarily at producing synthetic data for in-play betting (also known as in-race or in-game betting) where bettors can place bets on the outcome of a track-race event, such as a horse race, after the race has started and for as long as the race is underway, with betting only ceasing when the race ends. The rationale for, and design of, BBE has been described in detail in a previous paper that we summarise here, before discussing our comparative results which contrast a single-threaded implementation in Python, a multi-threaded implementation in Python, and an implementation where Python header-code calls simulations of the track-racing events written in OpenCL that execute on a 640-core GPU -- this runs approximately 1000 times faster than the single-threaded Python. Our source-code for BBE is freely available on GitHub.

We design a novel framework to examine market efficiency through out-of-sample (OOS) predictability. We frame the asset pricing problem as a machine learning classification problem and construct classification models to predict return states. The prediction-based portfolios beat the market with significant OOS economic gains. We measure prediction accuracies directly. For each model, we introduce a novel application of binomial test to test the accuracy of 3.34 million return state predictions. The tests show that our models can extract useful contents from historical information to predict future return states. We provide unique economic insights about OOS predictability and machine learning models.

This paper studies an optimal consumption problem for a loss-averse agent with reference to past consumption maximum. To account for loss aversion on relative consumption, an S-shaped utility is adopted that measures the difference between the non-negative consumption rate and a fraction of the historical spending peak. We consider the concave envelope of the realization utility with respect to consumption, allowing us to focus on an auxiliary HJB variational inequality on the strength of concavification principle and dynamic programming arguments. By applying the dual transform and smooth-fit conditions, the auxiliary HJB variational inequality is solved in closed-form piecewisely and some thresholds of the wealth variable are obtained. The optimal consumption and investment control of the original problem can be derived analytically in the piecewise feedback form. The rigorous verification proofs on optimality and concavification principle are provided.

We develop a quasi-likelihood analysis procedure for a general class of multivariate marked point processes. As a by-product of the general method, we establish under stability and ergodicity conditions the local asymptotic normality of the quasi-log likelihood, along with the convergence of moments of quasi-likelihood and quasi-Bayesian estimators. To illustrate the general approach, we then turn our attention to a class of multivariate marked Hawkes processes with generalized exponential kernels, comprising among others the so-called Erlang kernels. We provide explicit conditions on the kernel functions and the mark dynamics under which a certain transformation of the original process is Markovian and $V$-geometrically ergodic. We finally prove that the latter result, which is of interest in its own right, constitutes the key ingredient to show that the generalized exponential Hawkes process falls under the scope of application of the quasi-likelihood analysis.

What causes countercyclicality of industry--level productivity dispersion in the U.S.? Empirically, we construct an index of negative profit shocks and show that both productivity dispersion and R&D intensity dispersion enlarge at the onset of the shock and gradually dissipate. Theoretically, we build a duopolistic technology--ladder model in which heterogeneous R&D costs determine firms' post--shock optimal behaviors and equilibrium technology gap. Quantitatively, we calibrate a parameterized model, simulate firms' post--shock responses and predict that productivity dispersion is due to the low--cost firm increasing R&D efforts and the high--cost firm doing the opposite. We provide two empirical tests for this mechanism.

Empirical studies have emphasized that the equity implied volatility is characterized by a negative skew inversely proportional to the square root of the time-to-maturity. We examine the short-time-to-maturity behavior of the implied volatility smile for pure jump exponential additive processes. An excellent calibration of the equity volatility surfaces has been achieved by a class of these additive processes with power-law scaling. The two power-law scaling parameters are $\beta$, related to the variance of jumps, and $\delta$, related to the smile asymmetry. It has been observed, in option market data, that $\beta=1$ and $\delta=-1/2$. In this paper, we prove that the implied volatility of these additive processes is consistent, in the short-time, with the equity market empirical characteristics if and only if $\beta=1$ and $\delta=-1/2$.

AI and reinforcement learning (RL) have improved many areas, but are not yet widely adopted in economic policy design, mechanism design, or economics at large. At the same time, current economic methodology is limited by a lack of counterfactual data, simplistic behavioral models, and limited opportunities to experiment with policies and evaluate behavioral responses. Here we show that machine-learning-based economic simulation is a powerful policy and mechanism design framework to overcome these limitations. The AI Economist is a two-level, deep RL framework that trains both agents and a social planner who co-adapt, providing a tractable solution to the highly unstable and novel two-level RL challenge. From a simple specification of an economy, we learn rational agent behaviors that adapt to learned planner policies and vice versa. We demonstrate the efficacy of the AI Economist on the problem of optimal taxation. In simple one-step economies, the AI Economist recovers the optimal tax policy of economic theory. In complex, dynamic economies, the AI Economist substantially improves both utilitarian social welfare and the trade-off between equality and productivity over baselines. It does so despite emergent tax-gaming strategies, while accounting for agent interactions and behavioral change more accurately than economic theory. These results demonstrate for the first time that two-level, deep RL can be used for understanding and as a complement to theory for economic design, unlocking a new computational learning-based approach to understanding economic policy.

We attempt to reconcile Gabaix and Koijen's (GK) recent Inelastic Market Hypothesis with the order-driven view of markets that emerged within the microstructure literature in the past 20 years. We review the most salient empirical facts and arguments that give credence to the idea that market price fluctuations are mostly due to order flow, whether informed or non-informed. We show that the Latent Liquidity Theory of price impact makes a precise prediction for GK's multiplier $M$, which measures by how many dollars, on average, the market value of a company goes up if one buys one dollar worth of its stocks. Our central result is that $M$ increases with the volatility of the stock and decreases with the fraction of the market cap. that is traded daily. We discuss several empirical results suggesting that the lion's share of volatility is due to trading activity.

We quantify model risk of a financial portfolio whereby a multi-period mean-standard-deviation criterion is used as a selection criterion. In this work, model risk is defined as the loss due to uncertainty of the underlying distribution of the returns of the assets in the portfolio. The uncertainty is measured by the Kullback-Leibler divergence, i.e., the relative entropy. In the worst case scenario, the optimal robust strategy can be obtained in a semi-analytical form as a solution of a system of nonlinear equations. Several numerical results are presented which allow us to compare the performance of this robust strategy with the optimal non-robust strategy. For illustration, we also quantify the model risk associated with an empirical dataset.

In the face of the upcoming 30th anniversary of econophysics, we review our contributions and other related works on the modeling of the long-range memory phenomenon in physical, economic, and other social complex systems. Our group has shown that the long-range memory phenomenon can be reproduced using various Markov processes, such as point processes, stochastic differential equations and agent-based models. Reproduced well enough to match other statistical properties of the financial markets, such as return and trading activity distributions and first-passage time distributions. Research has lead us to question whether the observed long-range memory is a result of actual long-range memory process or just a consequence of non-linearity of Markov processes. As our most recent result we discuss the long-range memory of the order flow data in the financial markets and other social systems from the perspective of the fractional L\`{e}vy stable motion. We test widely used long-range memory estimators on discrete fractional L\`{e}vy stable motion represented by the ARFIMA sample series. Our newly obtained results seem indicate that new estimators of self-similarity and long-range memory for analyzing systems with non-Gaussian distributions have to be developed.