CyberSec Research

This study investigates adaptive experimental design for treatment choice, also known as fixed-budget best-arm identification. We consider an adaptive procedure consisting of a treatment-allocation phase followed by a treatment-choice phase, and we design an adaptive experiment for this setup to efficiently identify the best treatment arm, defined as the one with the highest expected outcome. In our designed experiment, the treatment-allocation phase consists of two stages. The first stage is a pilot phase, where we allocate each treatment arm uniformly with equal proportions to eliminate clearly suboptimal arms and estimate outcome variances. In the second stage, we allocate treatment arms in proportion to the variances estimated in the first stage. After the treatment-allocation phase, the procedure enters the treatment-choice phase, where we choose the treatment arm with the highest sample mean as our estimate of the best treatment arm. We prove that this single design is simultaneously asymptotically minimax and Bayes optimal for the simple regret, with upper bounds that match our lower bounds up to exact constants. Therefore, our designed experiment achieves the sharp efficiency limits without requiring separate tuning for minimax and Bayesian objectives.

We propose a weak-identification-robust test for linear instrumental variable (IV) regressions with high-dimensional instruments, whose number is allowed to exceed the sample size. In addition, our test is robust to general error dependence, such as network dependence and spatial dependence. The test statistic takes a self-normalized form and the asymptotic validity of the test is established by using random matrix theory. Simulation studies are conducted to assess the numerical performance of the test, confirming good size control and satisfactory testing power across a range of various error dependence structures.

We develop misspecification tests for building additive time-varying (ATV-)GARCH models. In the model, the volatility equation of the GARCH model is augmented by a deterministic time-varying intercept modeled as a linear combination of logistic transition functions. The intercept is specified by a sequence of tests, moving from specific to general. The first test is the test of the standard stationary GARCH model against an ATV-GARCH model with one transition. The alternative model is unidentified under the null hypothesis, which makes the usual LM test invalid. To overcome this problem, we use the standard method of approximating the transition function by a Taylor expansion around the null hypothesis. Testing proceeds until the first non-rejection. We investigate the small-sample properties of the tests in a comprehensive simulation study. An application to the VIX index indicates that the volatility of the index is not constant over time but begins a slow increase around the 2007-2008 financial crisis.

Researchers often report empirical results that are based on low-dimensional IVs, such as the shift-share IV, together with many IVs. Could we combine these results in an efficient way and take advantage of the information from both sides? In this paper, we propose a combination inference procedure to solve the problem. Specifically, we consider a linear combination of three test statistics: a standard cluster-robust Wald statistic based on the low-dimensional IVs, a leave-one-cluster-out Lagrangian Multiplier (LM) statistic, and a leave-one-cluster-out Anderson-Rubin (AR) statistic. We first establish the joint asymptotic normality of the Wald, LM, and AR statistics and derive the corresponding limit experiment under local alternatives. Then, under the assumption that at least the low-dimensional IVs can strongly identify the parameter of interest, we derive the optimal combination test based on the three statistics and establish that our procedure leads to the uniformly most powerful (UMP) unbiased test among the class of tests considered. In particular, the efficiency gain from the combined test is of ``free lunch" in the sense that it is always at least as powerful as the test that is only based on the low-dimensional IVs or many IVs.

This paper investigates the impact of posterior drift on out-of-sample forecasting accuracy in overparametrized machine learning models. We document the loss in performance when the loadings of the data generating process change between the training and testing samples. This matters crucially in settings in which regime changes are likely to occur, for instance, in financial markets. Applied to equity premium forecasting, our results underline the sensitivity of a market timing strategy to sub-periods and to the bandwidth parameters that control the complexity of the model. For the average investor, we find that focusing on holding periods of 15 years can generate very heterogeneous returns, especially for small bandwidths. Large bandwidths yield much more consistent outcomes, but are far less appealing from a risk-adjusted return standpoint. All in all, our findings tend to recommend cautiousness when resorting to large linear models for stock market predictions.

This paper introduces a novel Proxy-Enhanced Correlated Random Effects Double Machine Learning (P-CRE-DML) framework to estimate causal effects in panel data with non-linearities and unobserved heterogeneity. Combining Double Machine Learning (DML, Chernozhukov et al., 2018), Correlated Random Effects (CRE, Mundlak, 1978), and lagged variables (Arellano & Bond, 1991) and innovating within the CRE-DML framework (Chernozhukov et al., 2022; Clarke & Polselli, 2025; Fuhr & Papies, 2024), we apply P-CRE-DML to investigate the effect of social trust on GDP growth across 89 countries (2010-2020). We find positive and statistically significant relationship between social trust and economic growth. This aligns with prior findings on trust-growth relationship (e.g., Knack & Keefer, 1997). Furthermore, a Monte Carlo simulation demonstrates P-CRE-DML's advantage in terms of lower bias over CRE-DML and System GMM. P-CRE-DML offers a robust and flexible alternative for panel data causal inference, with applications beyond economic growth.

This paper introduces a novel Bayesian reverse unrestricted mixed-frequency model applied to a panel of nine European electricity markets. Our model analyzes the impact of daily fossil fuel prices and hourly renewable energy generation on hourly electricity prices, employing a hierarchical structure to capture cross-country interdependencies and idiosyncratic factors. The inclusion of random effects demonstrates that electricity market integration both mitigates and amplifies shocks. Our results highlight that while renewable energy sources consistently reduce electricity prices across all countries, gas prices remain a dominant driver of cross-country electricity price disparities and instability. This finding underscores the critical importance of energy diversification, above all on renewable energy sources, and coordinated fossil fuel supply strategies for bolstering European energy security.

OLS estimators are widely used in network experiments to estimate spillover effects via regressions on exposure mappings that summarize treatment and network structure. We study the causal interpretation and inference of such OLS estimators when both design-based uncertainty in treatment assignment and sampling-based uncertainty in network links are present. We show that correlations among elements of the exposure mapping can contaminate the OLS estimand, preventing it from aggregating heterogeneous spillover effects for clear causal interpretation. We derive the estimator's asymptotic distribution and propose a network-robust variance estimator. Simulations and an empirical application reveal sizable contamination bias and inflated spillover estimates.

In this paper, we study causal inference when the treatment variable is an aggregation of multiple sub-treatment variables. Researchers often report marginal causal effects for the aggregated treatment, implicitly assuming that the target parameter corresponds to a well-defined average of sub-treatment effects. We show that, even in an ideal scenario for causal inference such as random assignment, the weights underlying this average have some key undesirable properties: they are not unique, they can be negative, and, holding all else constant, these issues become exponentially more likely to occur as the number of sub-treatments increases and the support of each sub-treatment grows. We propose approaches to avoid these problems, depending on whether or not the sub-treatment variables are observed.

Causal inference is central to statistics and scientific discovery, enabling researchers to identify cause-and-effect relationships beyond associations. While traditionally studied within Euclidean spaces, contemporary applications increasingly involve complex, non-Euclidean data structures that reside in abstract metric spaces, known as random objects, such as images, shapes, networks, and distributions. This paper introduces a novel framework for causal inference with continuous treatments applied to non-Euclidean data. To address the challenges posed by the lack of linear structures, we leverage Hilbert space embeddings of the metric spaces to facilitate Fr\'echet mean estimation and causal effect mapping. Motivated by a study on the impact of exposure to fine particulate matter on age-at-death distributions across U.S. counties, we propose a nonparametric, doubly-debiased causal inference approach for outcomes as random objects with continuous treatments. Our framework can accommodate moderately high-dimensional vector-valued confounders and derive efficient influence functions for estimation to ensure both robustness and interpretability. We establish rigorous asymptotic properties of the cross-fitted estimators and employ conformal inference techniques for counterfactual outcome prediction. Validated through numerical experiments and applied to real-world environmental data, our framework extends causal inference methodologies to complex data structures, broadening its applicability across scientific disciplines.

We propose an empirical Bayes estimator for two-way effects in linked data sets based on a novel prior that leverages patterns of assortative matching observed in the data. To capture limited mobility we model the bipartite graph associated with the matched data in an asymptotic framework where its Laplacian matrix has small eigenvalues that converge to zero. The prior hyperparameters that control the shrinkage are determined by minimizing an unbiased risk estimate. We show the proposed empirical Bayes estimator is asymptotically optimal in compound loss, despite the weak connectivity of the bipartite graph and the potential misspecification of the prior. We estimate teacher values-added from a linked North Carolina Education Research Data Center student-teacher data set.

We propose a method for multilevel decomposition of generalized entropy (GE) measures that explicitly accounts for nested population structures such as national, regional, and subregional levels. Standard approaches that estimate GE separately at each level do not guarantee compatibility with multilevel decomposition. Our method constrains lower-level GE estimates to match higher-level benchmarks while preserving hierarchical relationships across layers. We apply the method to Japanese income data to estimate GE at the national, prefectural, and municipal levels, decomposing national inequality into between-prefecture and within-prefecture inequality, and further decomposing prefectural GE into between-municipality and within-municipality inequality.

Econometric identification generally relies on orthogonality conditions, which usually state that the random error term is uncorrelated with the explanatory variables. In convex regression, the orthogonality conditions for identification are unknown. Applying Lagrangian duality theory, we establish the sample orthogonality conditions for convex regression, including additive and multiplicative formulations of the regression model, with and without monotonicity and homogeneity constraints. We then propose a hybrid instrumental variable control function approach to mitigate the impact of potential endogeneity in convex regression. The superiority of the proposed approach is shown in a Monte Carlo study and examined in an empirical application to Chilean manufacturing data.

This paper analyzes realized return behavior across a broad set of crypto assets by estimating heterogeneous exposures to idiosyncratic and systematic risk. A key challenge arises from the latent nature of broader economy-wide risk sources: macro-financial proxies are unavailable at high-frequencies, while the abundance of low-frequency candidates offers limited guidance on empirical relevance. To address this, we develop a two-stage ``divide-and-conquer'' approach. The first stage estimates exposures to high-frequency idiosyncratic and market risk only, using asset-level IV regressions. The second stage identifies latent economy-wide factors by extracting the leading principal component from the model residuals and mapping it to lower-frequency macro-financial uncertainty and sentiment-based indicators via high-dimensional variable selection. Structured patterns of heterogeneity in exposures are uncovered using Mean Group estimators across asset categories. The method is applied to a broad sample of crypto assets, covering more than 80% of total market capitalization. We document short-term mean reversion and significant average exposures to idiosyncratic volatility and illiquidity. Green and DeFi assets are, on average, more exposed to market-level and economy-wide risk than their non-Green and non-DeFi counterparts. By contrast, stablecoins are less exposed to idiosyncratic, market-level, and economy-wide risk factors relative to non-stablecoins. At a conceptual level, our study develops a coherent framework for isolating distinct layers of risk in crypto markets. Empirically, it sheds light on how return sensitivities vary across digital asset categories -- insights that are important for both portfolio design and regulatory oversight.

We propose a simple modification to the wild bootstrap procedure and establish its asymptotic validity for linear regression models with many covariates and heteroskedastic errors. Monte Carlo simulations show that the modified wild bootstrap has excellent finite sample performance compared with alternative methods that are based on standard normal critical values, especially when the sample size is small and/or the number of controls is of the same order of magnitude as the sample size.

This paper studies analytic inference along two dimensions of clustering. In such setups, the commonly used approach has two drawbacks. First, the corresponding variance estimator is not necessarily positive. Second, inference is invalid in non-Gaussian regimes, namely when the estimator of the parameter of interest is not asymptotically Gaussian. We consider a simple fix that addresses both issues. In Gaussian regimes, the corresponding tests are asymptotically exact and equivalent to usual ones. Otherwise, the new tests are asymptotically conservative. We also establish their uniform validity over a certain class of data generating processes. Independently of our tests, we highlight potential issues with multiple testing and nonlinear estimators under two-way clustering. Finally, we compare our approach with existing ones through simulations.

Adaptive experiments such as multi-armed bandits offer efficiency gains over traditional randomized experiments but pose two major challenges: invalid inference on the Average Treatment Effect (ATE) due to adaptive sampling and low statistical power for sub-optimal treatments. We address both issues by extending the Mixture Adaptive Design framework (arXiv:2311.05794). First, we propose MADCovar, a covariate-adjusted ATE estimator that is unbiased and preserves anytime-valid inference guarantees while substantially improving ATE precision. Second, we introduce MADMod, which dynamically reallocates samples to underpowered arms, enabling more balanced statistical power across treatments without sacrificing valid inference. Both methods retain MAD's core advantage of constructing asymptotic confidence sequences (CSs) that allow researchers to continuously monitor ATE estimates and stop data collection once a desired precision or significance criterion is met. Empirically, we validate both methods using simulations and real-world data. In simulations, MADCovar reduces CS width by up to $60\%$ relative to MAD. In a large-scale political RCT with $\approx32,000$ participants, MADCovar achieves similar precision gains. MADMod improves statistical power and inferential precision across all treatment arms, particularly for suboptimal treatments. Simulations show that MADMod sharply reduces Type II error while preserving the efficiency benefits of adaptive allocation. Together, MADCovar and MADMod make adaptive experiments more practical, reliable, and efficient for applied researchers across many domains. Our proposed methods are implemented through an open-source software package.

This paper examines the effects of daily temperature fluctuations on subnational economic growth in Thailand. Using annual gross provincial product (GPP) per capita data from 1982 to 2022 and high-resolution reanalysis weather data, I estimate fixed-effects panel regressions that isolate plausibly exogenous within-province year-to-year variation in temperature. The results indicate a statistically significant inverted-U relationship between temperature and annual growth in GPP per capita, with adverse effects concentrated in the agricultural sector. Industrial and service outputs appear insensitive to short-term weather variation. Distributed lag models suggest that temperature shocks have persistent effects on growth trajectories, particularly in lower-income provinces with higher average temperatures. I combine these estimates with climate projections under RCP4.5 and RCP8.5 emission scenarios to evaluate province-level economic impacts through 2090. Without adjustments for biases in climate projections or lagged temperature effects, climate change is projected to reduce per capita output for 63-86% of Thai population, with median GDP per capita impacts ranging from -4% to +56% for RCP4.5 and from -52% to -15% for RCP8.5. When correcting for projected warming biases - but omitting lagged dynamics - median losses increase to 57-63% (RCP4.5) and 80-86% (RCP8.5). Accounting for delayed temperature effects further raises the upper-bound estimates to near-total loss. These results highlight the importance of accounting for model uncertainty and temperature dynamics in subnational climate impact assessments. All projections should be interpreted with appropriate caution.