Fundamentals and Exchange Rate Prediction Revisited

Abstract

This paper measures latent fundamental exchange rates with independent component‐based rates constructed from a cross‐section of exchange rates and then uses their deviations from exchange rates to forecast. Empirical results indicate that the independent component‐based model and its Taylor rule and purchasing power parity augmented models are superior to the random walk in predicting exchange rates. These results are robust to several scenarios and are likely to be observed if the U.S. sources and the recursive scheme are applied. Our results reveal that information regarding the third moment of exchange rate changes is helpful to explain exchange rate movements.

Simplified optimal design of MDOF structures with negative stiffness amplifying dampers based on effective damping

“…The proposed passive pseudo‐negative stiffness isolator works similarly to the friction pendulum sliding isolator but has a convex friction interface, achieving a parallelogram hysteresis curve with negative stiffness. In addition to using a ratchet–pawl mechanism, Wang et al also proposed another novel type of pseudo‐NSD with zero‐centered force that exhibited triangular‐shape hysteretic curves as the clutch inerter damper . The NSD proposed by Nagarajaiah et al was further developed by Sarlis et al and Pasala et al, which consists of a precompressed spring, a pivot plate, a gap spring, a self‐containing system, and additional VDs.…”

Liquidity premiums, interest rate differentials, and nominal exchange rate prediction

“…This is the well‐known “exchange rate disconnect puzzle” (Obstfeld & Rogoff, 2000) and is a hot research topic in empirical international finance. Several articles work on this puzzle and find evidence to beat the random walk in out‐of‐sample contests of exchange rates by focusing on linear long‐horizon prediction equations at long forecast horizons (Mark, 1995), by applying nonlinear long‐horizon prediction equations at medium forecast horizons (Kilian & Taylor, 2003), by using an unconstrained Taylor‐rule‐based model at short forecast horizons (Molodtsova & Papell, 2009), by adopting a factor‐based model at medium and long forecast horizons (Engel et al, 2015; Wang & Wu, 2015), and by employing macro‐financial fundamentals (Lilley et al, 2020) at short forecast horizons.…”

“…However, the principal component analysis (PCA) constructs orthogonal factors using information only up to the second moment. The independent component analysis (ICA) estimates latent factors by maximizing non‐Gaussianity via a measure of the distance to normality, such as skewness, excess kurtosis, or negentropy, which utilizes information about higher‐order moments of exchange rates (Wang & Wu, 2015). We, therefore, apply ICA to measure the latent fundamental exchange rate.…”

“…There are several differences between our paper and their papers. First, Engel et al (2015) and Wang and Wu (2015) apply quarterly panel data from 1973 to 2007 and 1973 to 2011, respectively. We use daily panel data from 1999 to 2018 2 .…”

Spurious Factor Analysis

“…Both strategies are used in practice. For example, Moon and Perron (2007), Eickmeier (2009), Wang and Wu (2015), von Borstel, Eickmeier, and Krippner (2016), and Barigozzi, Lippi, and Luciani (2018) fitted factor models to nonstationary data after first‐differencing them. Stock and Watson (2016) not only first‐differenced most of the series entering their dynamic factor model of the U.S. economy, but also locally demeaned the variables to minimize problems associated with low‐frequency variability.…”