|
on Econometrics |
| By: | Muris, Chris; Wacker, Konstantin M. |
| Abstract: | This paper analyzes how interaction effects can be consistently estimated un- der economically plausible assumptions in linear panel models with a fixed T - dimension. We advocate for a correlated interaction term estimator (CITE) and show that it is consistent under conditions that are not sufficient for consistency of the interaction term estimator that is most common in applied econometric work. Our paper discusses the empirical content of these conditions, shows that standard inference procedures can be applied to CITE, and analyzes consistency, relative efficiency, inference, and their finite sample properties in a simulation study. In an empirical application, we test whether labor displacement effects of robots are stronger in countries at higher income levels. The results are in line with our theoretical and simulation results and indicate that standard interaction term estimation underestimates the importance of a country's income level in the relationship between robots and employment and may prematurely reject a null hypothesis about interaction effects in the presence of misspecification. |
| Keywords: | panel data, interaction effects, correlated random coefficients, robots |
| JEL: | C13 C23 C33 J23 O30 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:glodps:1583 |
| By: | Cardenas Hurtado, Camilo; Moustaki, Irini; Chen, Yunxiao; Marra, Giampiero |
| Abstract: | We introduce a general framework for latent variable modeling, named Generalized Latent Variable Models for Location, Scale, and Shape parameters (GLVM-LSS). This framework extends the generalized linear latent variable model beyond the exponential family distributional assumption and enables the modeling of distributional parameters other than the mean (location parameter), such as scale and shape parameters, as functions of latent variables. Model parameters are estimated via maximum likelihood. We present two real-world applications on public opinion research and educational testing, and evaluate the model’s performance in terms of parameter recovery through extensive simulation studies. Our results suggest that the GLVM-LSS is a valuable tool in applications where modeling higher-order moments of the observed variables through latent variables is of substantive interest. The proposed model is implemented in the R package glvmlss, available online. |
| Keywords: | latent variable models; distributional regression; GAMLSS; EM algorithm; heteroscedasticity |
| JEL: | C1 |
| Date: | 2025–03–06 |
| URL: | https://d.repec.org/n?u=RePEc:ehl:lserod:127387 |
| By: | Demetrescu, Matei; Frondel, Manuel; Tomberg, Lukas; Vance, Colin |
| Abstract: | We investigate a bracketing property that purports to yield upper- and lower bounds on the treatment effects obtained from a fixed effects- and lagged dependent variable model. Referencing both analytical results and a Monte Carlo simulation, we explore the conditions under which the bracketing property holds, confirming this to be the case when the data generating process (DGP) is characterized by either unobserved heterogeneity or feedback effects from a lagged dependent variable (LDV). However, when the DGP is characterized by both features simultaneously, we find that bracketing of the treatment effect only holds under certain conditions - but not in general. Practitioners can nevertheless obtain the lower bound estimate by referencing a model that includes both fixed effects and a LDV. While the Nickell bias in the coefficient of the LDV is known to be of order 1/T, we show that the Nickell-type bias in the estimator of the treatment effect is of order 1/T squared. |
| Abstract: | Wir untersuchen eine Klammereigenschaft, die obere und untere Grenzen für die mit einem Fixed-Effects-Modell und einem Lagged-Dependent-Variable-Modell geschätzten Treatmenteffekte liefern soll. Unter Bezugnahme auf analytische Ergebnisse und eine Monte-Carlo-Simulation untersuchen wir die Bedingungen, unter denen die Klammerungseigenschaft gilt, und bestätigen, dass dies der Fall ist, wenn der datengenerierende Prozess (DGP) entweder durch unbeobachtete Heterogenität oder durch Rückkopplungseffekte von einer verzögerten abhängigen Variable (LDV) gekennzeichnet ist. Wenn der DGP jedoch durch beide Merkmale gleichzeitig charakterisiert ist, stellen wir fest, dass die Klammerung des Treatmenteffekts nur unter bestimmten Bedingungen gilt - aber nicht im Allgemeinen. Anwender können jedoch eine untere Schätzgrenze erhalten, indem sie ein Modell verwenden, das sowohl Fixed-Effects als auch eine LDV enthält. Während die Nickell-Verzerrung im Koeffizienten der LDV bekanntlich die Ordnung 1/T hat, zeigen wir, dass die Nickell-Verzerrung im Schätzer des Behandlungseffekts die Ordnung 1/T zum Quadrat hat. |
| Keywords: | Monte Carlo simulation, treatment effect, bounds, Nickell bias |
| JEL: | C18 C23 C52 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:rwirep:315488 |
| By: | Felix Reichel |
| Abstract: | Bessel’s correction adjusts the denominator in the sample variance formula from n to n −1 to produce an unbiased estimator for the population variance. This paper includes rigorous derivations, geometric interpretations, and visualizations. It then introduces the concept of “bariance, †an alternative pairwise distances intuition of sample dispersion without an arithmetic mean. Finally, we address practical concerns raised in Rosenthal’s article [1] advocating the use of n-based estimates from a more holistic MSE-based viewpoint for pedagogical reasons and in certain practical contexts. Finally, the empirical part using simulation reveals that the run-time of estimating population variance can be shortened when using an algebraically optimized “bariance“ approach to estimate an unbiased variance. |
| Keywords: | Unbiased sample variance, Runtime-optimized linear unbiased sample variance estimators |
| JEL: | C10 C80 |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:jku:econwp:2025-06 |
| By: | João A. Bastos |
| Abstract: | A deep learning binary classifier is proposed to test if asset returns follow martingale difference sequences. The Neyman-Pearson classification paradigm is applied to control the type I error of the test. In Monte Carlo simulations, I find that this approach has better power properties than variance ratio and portmanteau tests against several alternative processes. I apply this procedure to a large set of exchange rate returns and find that it detects several potential deviations from the martingale difference hypothesis that the conventional statistical tests fail to capture. |
| Keywords: | Martingale difference hypothesis; Convolutional network; Variance ratio test; Portmanteau test; Exchange rates. |
| Date: | 2025–03 |
| URL: | https://d.repec.org/n?u=RePEc:ise:remwps:wp03742025 |