This paper provides a theoretical justification for the Deviance Information Criterion (DIC) as a Bayesian model selection tool using MCMC output. Unlike Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), which balance model adequacy against complexity without considering prior information, DIC incorporates priors into this trade-off. The contributions of this paper are two-fold. First, it demonstrates that when a plug-in predictive distribution - obtained by substituting parameter values with their optimal estimates to yield the plug-in estimated sampling distribution - is used under a set of regularity conditions, the DIC serves as an asymptotically unbiased estimator of the expected Kullback-Leibler divergence between the data-generating process and the plug-in predictive distribution. Second, it develops higher-order expansions for DIC and the effective number of parameters, highlighting the effect of the priors. We employ DIC to compare discrete-choice models, stochastic frontier models, and copula models in three empirical applications; the results align with theoretical expectations, showing the utility of DIC as a versatile tool outperforming the traditional model selection criteria. It is found that the logit model is better than the probit model for investigating the marginal effects of parents' education on children's completion of high school. Additionally, the stochastic frontier model with an exponential distribution better fits electricity utility data than the normal distribution. Finally, the chosen copula models for S&P index returns exhibit heavy tails and strong tail dependence. By modelling the effect of priors through higher order expansions, we also find the above empirical models outperforming their benchmark counterparts in terms of predictive accuracy.

A multivariate fractional Brownian motion (mfBm) with component-wise Hurst exponents is used to model and forecast realized volatility (RV). We investigate the interplay between correlation coefficients and Hurst exponents and propose a novel method to estimate model parameters, establishing its consistency and asymptotic normality. Additionally, we develop a time-reversibility test, which is typically not rejected by RV data. When the data generating process is a time-reversible mfBm, we derive optimal forecasting formulae and analyze their properties. A key insight is that an mfBm with different Hurst exponents and non-zero correlations can reduce forecasting errors compared to a one-dimensional model. Consistent with this theory, out-of-sample forecasts using the time-reversible mfBm show improvements over univariate fBm, particularly when the estimated Hurst exponents differ significantly. Empirical results demonstrate that mfBm outperforms HAR and its variants in terms of out-of-sample forecast.

This note revisits the standard derivation of the asymptotic unbiasedness of Akaike information criterion (AIC). The existing literature has shown that AIC is an asymptotically unbiased estimator of the predictive risk. However, there is a technical issue in these justifications. In this note, we bridge the gap and establish the asymptotic unbiasedness of AIC under certain regularity conditions, which are applicable in various scenarios. Our result suggests that AIC should be used with caution if the moment conditions on the parameter estimators may not hold. When model misspecification is allowed, our result can be extended to Takeuchi information criterion (TIC).

This paper explores implications of weak identification in common 'long memory' and recent 'rough' approaches to modeling volatility dynamics of financial assets. We unveil an asymptotic near-observational equivalence between a long memory model with weak autoregressive dynamics and a rough model with a near-unit autoregressive root. Standard methods struggle to distinguish them, and conventional asymptotics are invalid. We propose an identification-robust approach to construct confidence sets that reveal the uncertainty and aid inference. Empirical studies based on realized volatility and trading volume often fail to statistically reject either model, thereby providing evidence of their potential coexistence.

Modeling multivariate stochastic volatility (MSV) can pose significant challenges, particularly when both variances and covariances are time-varying. In this study, we tackle these complexities by introducing novel MSV models based on the generalized Fisher transformation (GFT) proposed by Archakov and Hansen (2021). Our model exhibits remarkable flexibility, ensuring the positive-definiteness of the variance-covariance matrix, and disentangling the driving forces of volatilities and correlations. To conduct Bayesian analysis of the models, we employ a Particle Gibbs Ancestor Sampling (PGAS) method, facilitating efficient Bayesian model comparisons. Furthermore, we extend our MSV model to cover leverage effects and incorporate realized measures. Our simulation studies demonstrate that the proposed method performs well for our GFT-based MSV model. Furthermore, empirical studies based on equity returns show that the MSV models outperform alternative specifications in both in-sample and out-of-sample performances.

This paper proposes a novel exact maximum likelihood (ML) estimation method for general Gaussian processes, where all parameters are estimated jointly. The exact ML estimator (MLE) is consistent and asymptotically normally distributed. We prove the local asymptotic normality (LAN) property of the sequence of statistical experiments for general Gaussian processes in the sense of Le Cam, thereby enabling optimal estimation and facilitating statistical inference. The results rely solely on the asymptotic behavior of the spectral density near zero, allowing them to be widely applied. The established optimality not only addresses the gap left by Adenstedt(1974), who proposed an efficient but infeasible estimator for the long-run mean μ, but also enables us to evaluate the finite-sample performance of the existing method – the commonly used plug-in MLE, in which the sample mean is substituted into the likelihood. Our simulation results show that the plug-in MLE performs nearly as well as the exact MLE, alleviating concerns that inefficient estimation of μ would compromise the efficiency of the remaining parameter estimates.

This paper develops a framework to investigate the time-varying asymmetric and zone-like preferences of a central bank. We derive the optimal forward-looking monetary policy rule and explore four distinct types of loss related to inflation, output, and leverage, resulting in 64 distinct models. We estimate these models and utilize the Akaike information criterion (AIC) to select the most suitable one. Using data from 1996 to 2022, we find that the People's Bank of China (PBoC): (1) reacts to inflation gaps with slight asymmetry, featuring a no-intervention zone between-1% and 1%; (2) intervenes asymmetrically in output gaps, displaying stronger inclination toward averting overheating than downturns; (3) applies symmetric regulation to leverage gaps. These findings underscore the PBoC's adaptability and responsiveness to economic fluctuations and its preference for risk management.

Explainable machine learning aims to strike a balance between prediction accuracy and model transparency, particularly in settings where black-box predictive models, such as deep neural networks or kernel-based methods, achieve strong empirical performance but remain difficult to interpret. This work introduces a mixture of generalized additive models (GAMs) in which random Fourier feature (RFF) representations are leveraged to uncover locally adaptive structure in the data. In the proposed method, an RFF-based embedding is first learned and then compressed via principal component analysis. The resulting low-dimensional representations are used to perform soft clustering of the data through a Gaussian mixture model. These cluster assignments are then applied to construct a mixture-of-GAMs framework, where each local GAM captures nonlinear effects through interpretable univariate smooth functions. Numerical experiments on real-world regression benchmarks, including the California Housing, NASA Airfoil Self-Noise, and Bike Sharing datasets, demonstrate improved predictive performance relative to classical interpretable models. Overall, this construction provides a principled approach for integrating representation learning with transparent statistical modeling.

A heteroskedasticity-autocorrelation robust (HAR) test statistic is proposed to test for the presence of explosive roots in financial or real asset prices when the equation errors are strongly dependent. Limit theory for the test statistic is developed and extended to heteroskedastic models. The new test has stable size properties unlike conventional test statistics that typically lead to size distortion and inconsistency in the presence of strongly dependent equation errors. The new procedure can be used to consistently time-stamp the origination and termination of an explosive episode under similar conditions of long memory errors. Simulations are conducted to assess the finite sample performance of the proposed test and estimators. An empirical application to the S&P 500 index highlights the usefulness of the proposed procedures in practical work.

Based on a continuous-time stochastic volatility model with a linear drift, we develop a test for explosive behavior in financial asset prices at a low frequency when prices are sampled at a higher frequency. The test exploits the volatility information in the high-frequency data. The method consists of devolatizing log-asset price increments with realized volatility measures and performing a supremum-type recursive Dickey-Fuller test on the devolatized sample. The proposed test has a nuisance-parameter-free asymptotic distribution and is easy to implement. We study the size and power properties of the test in Monte Carlo simulations. A real-time date-stamping strategy based on the devolatized sample is proposed for the origination and conclusion dates of the explosive regime. Conditions under which the real-time date-stamping strategy is consistent are established. The test and the date-stamping strategy are applied to study explosive behavior in cryptocurrency and stock markets.

The fractional Brownian motion (fBm) process, governed by a fractional parameter , is a continuous‐time Gaussian process with its increment being the fractional Gaussian noise (fGn). This article first provides a computationally feasible expression for the spectral density of fGn. This expression enables us to assess the accuracy of a range of approximation methods, including the truncation method, Paxson's approximation, and the Taylor series expansion at the near‐zero frequency. Next, we conduct an extensive Monte Carlo study comparing the finite sample performance and computational cost of alternative estimation methods for under the fGn specification. These methods include two semi‐parametric methods (based on the Taylor series expansion), two versions of the Whittle method (utilising either the computationally feasible expression or Paxson's approximation of the spectral density), a time‐domain maximum likelihood (ML) method (employing a recursive approach for its likelihood calculation), and a change‐of‐frequency method. Special attention is paid to highly anti‐persistent processes with close to zero, which are of empirical relevance to financial volatility modelling. Considering the trade‐off between statistical and computational efficiency, we recommend using either the Whittle ML method based on Paxson's approximation or the time‐domain ML method. We model the log realized volatility dynamics of 40 financial assets in the US market from 2012 to 2019 with fBm. Although all estimation methods suggest rough volatility, the implied degree of roughness varies substantially with the estimation methods, highlighting the importance of understanding the finite sample performance of various estimation methods.

This paper proposes estimators for the parameters of an explosive fractional Ornstein-Uhlenbeck process. The asymptotic properties for the diffusion estimators are developed under the in-fill asymptotic scheme, while the asymptotic properties for the drift estimators are developed under the double asymptotic scheme for the full range of the Hurst parameter. The double asymptotic distribution of the estimator of the persistency parameter explicitly depends on the initial condition. Simulation results demonstrate the effectiveness of the proposed estimators, and the asymptotic distributions provide a good approximation in finite samples. An empirical application is presented to demonstrate the model's usefulness and the practical value of the asymptotic theory.

This article introduces a discrete‐time fractional stochastic volatility model (FSV) based on fractional Gaussian noise. The new model includes the standard stochastic volatility model as a special case and has the same limit as the fractional integrated stochastic volatility (FISV) model, which is the continuous‐time fractional Ornstein–Uhlenbeck process. A simulated maximum likelihood method, which maximizes the time‐domain log‐likelihood function calculated by the importance sampling technique, and a frequency‐domain quasi maximum likelihood method (or quasi Whittle) are employed to estimate the model parameters. Simulation studies suggest that, while both estimation methods can accurately estimate the model, the simulated maximum likelihood method outperforms the quasi Whittle method. As an illustration, we fit the FSV and FISV models with the proposed estimation techniques to the S&P 500 composite index over a sample period spanning 45 years.

This paper investigates the performance of different forecasting formulas with fractional Brownian motion based on discrete and finite samples. Existing literature presents two formulas for generating optimal forecasts when continuous records are available. One formula relies on a history over an infinite past, while the other is designed for a record limited to a finite past. In reality, only observations at discrete time points over a finite past are available. In this case, the forecasting formula, which has been widely used in the literature, is the one obtained by Gatheral et al. (Volatility is rough. Quant. Finance, 2018, 18(6), 933-949) that truncates and discretizes the formula based on continuous records over an infinite past. The present paper advocates an alternative forecasting formula, which is the conditional expectation based on finite past discrete-time observations. The findings suggest that the conditional expectation approach produces more accurate forecasts than the existing method, as demonstrated by both simulated data and actual daily realized volatility (RV) observations. Moreover, we also provide empirical evidence showing that the conditional expectation approach can lead to larger economic values than the existing method.

This paper introduces a novel and easy-to-implement method for accurately approximating the spectral density of discretely sampled fractional Ornstein-Uhlenbeck (fOU) processes. The method offers a substantial reduction in approximation error, particularly within the rough region of the fractional parameter H is an element of (0, 0.5). This approximate spectral density has the potential to enhance the performance of estimation methods and hypothesis testing that make use of spectral densities. We introduce the approximate Whittle maximum likelihood (AWML) method for discretely sampled fOU processes, utilizing the approximate spectral density, and demonstrate that the AWML estimator exhibits properties of consistency and asymptotic normality when H is an element of (0, 1), akin to the conventional Whittle maximum likelihood method. Through extensive simulation studies, we show that AWML outperforms existing methods in terms of estimation accuracy in finite samples. We then apply the AWML method to the trading volume of 40 financial assets. Our empirical findings reveal that the estimated Hurst parameters for these assets fall within the range of 0.10 to 0.21, indicating a rough dynamic.

In recent years, significant advancements have been made in the field of identifying financial asset price bubbles, particularly through the development of time‐series unit‐root tests featuring fractionally integrated errors and panel unit‐root tests. This study introduces an innovative approach for assessing the sign of the persistence parameter () within a panel fractional Ornstein‐Uhlenbeck process, based on the least squares estimator of . This method incorporates three distinct test statistics based on the Hurst parameter (), which can take values in the range of , be equal to , or fall within the interval of . The null hypothesis corresponds to . Based on a panel of continuous records of observations, the null asymptotic distributions are obtained when the time span () is fixed and the number of cross sections () goes to infinity. The power function of the tests is obtained under the local alternative where is close to zero in the order of . This alternative covers the departure from the unit root hypothesis from the explosive side, enabling the calculation of lower power in bubble tests. The hypothesis testing problem and the local power function are also considered when a panel of discrete‐sampled observations is available under a sequential limit, that is, the sampling interval shrinks to zero followed by the goes to infinity.

The sales of fashion products are influenced by uncertain and heterogeneous demands, necessitating predictive analytics to consider multiple explanatory variables and address the challenge of model uncertainty, which has been overlooked in prior research. To illustrate our solution, we first propose a novel forecasting estimator, which is characterized by provable optimal weighted forecasts and a collection of sub-model forecasts with various model specifications. We then validate our method empirically with store-level sales observations of a well-known international footwear brand, as an example of how a retailer can enhance its sales forecasts and improve promotion decisions after controlling model uncertainty. In a predictive analysis, the results show that controlling for model uncertainty between various predictors and store sales can produce more accurate forecasts of sales. With our proposed estimator, we also demonstrate the heterogeneity of promotion strategy importance for store with high and low previous sales. The additional analysis reveals that combo promotions have the most significant impact, and suggests adjusting the frequencies of gift and combo promotions to boost store sales. However, caution is advised when implementing both gift and combo promotions together to mitigate cannibalization effects.

This chapter derives asymptotic properties of the least squares (LS) estimator of the autoregressive (AR) parameter in local to unity processes with errors being fractional Gaussian noise (FGN) with the Hurst parameter H∈(0,1). It is shown that the estimator is consistent for all values of H∈(0,1). Moreover, the rate of convergence is n−1 when H∈[0.5,1). The rate of convergence is n−2H when H∈(0,0.5). Furthermore, the limiting distribution of the centered LS estimator depends on H. When H=0.5, the limiting distribution is the same as that obtained in Phillips (1987a) for the local to unity model with errors for which the standard functional central limit theorem is applicable. When H > 0.5 or when H < 0.5, the limiting distributions are new to the literature. The asymptotic properties of the LS estimator with fitted intercept are also derived. Simulation studies are performed to check the reliability of the asymptotic approximation for different values of sample size.

A model of financial asset price determination is proposed that incorporates flat trading features into an efficient price process. The model involves the superposition of a Brownian semimartingale process for the efficient price and a Bernoulli process that determines the extent of flat price trading. The approach is related to sticky price modeling and the Calvo pricing mechanism in macroeconomic dynamics. A limit theory for the conventional realized volatility (RV) measure of integrated volatility is developed. The results show that RV is still consistent but has an inflated asymptotic variance that depends on the probability of flat trading. Estimated quarticity is similarly affected, so that both the feasible central limit theorem and the inferential framework suggested in Barndorff-Nielsen and Shephard (J Royal Stat Soc Ser B (Stat Methodol) 64:253–280, 2002) remain valid under flat price trading even though there is information loss due to flat trading effects. The results are related to work by Jacod (J Financ Econom 16:526–569, 2018) and Mykland and Zhang (Ann Stat 34:1931–1963, 2006) on realized volatility measures with random and intermittent sampling, and to ACD models for irregularly spaced transactions data. Extensions are given to include models with microstructure noise. Some simulation results are reported. Empirical evaluations with tick-by-tick data indicate that the effect of flat trading on the limit theory under microstructure noise is likely to be minor in most cases, thereby affirming the relevance of existing approaches.