A Demand System Approach to Asset Pricing

Abstract

The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research. NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies official NBER publications.

Global Portfolio Rebalancing and Exchange Rates

“…Importantly, this allows for substitution e¤ect across assets to a¤ect the exchange rate. Building on Koijen and Yogo (2019), they estimate an entire demand system using cross-country aggregate holdings for 36 countries and decompose asset prices in three sources of variation: policy variables, macroeconomic factors and latent demand. We focus instead on the interaction between equity portfolios at the fund level and di¤erential in equity returns and use a granular instrumental variable approach to estimate the e¤ect of proved di¢ cult (see Frankel (1982a); Frankel (1982b); Rogo¤ (1984)).…”

“…4 Important progress has been made in that direction: see, for example,Gabaix and Maggiori (2015),Coimbra and Rey (2019),Koijen and Yogo (2019) andKoijen and Yogo (2020).…”

Dirichlet Policies for Reinforced Factor Portfolios

“…Now, let us make the extreme simplification, as in our empirical section, thatx = [1 0 K ], so that firm characteristics have zero sample mean -apart for the first one. This is not uncommon in the recent literature as long as the data is preprocessed (see Freyberger et al (2020), Gu et al (2020b), Kelly et al (2019) and Koijen and Yogo (2019)). Then, X X = N diag(σ 2 X ), where the modified vector of variancesσ 2 X is simply σ 2 X with the first element equal to one (instead of zero).…”

“…The restriction to positive weights is realistic since most asset managers have long-only constraints. This is typically the case of institutional investors (see Koijen and Yogo, 2019). Therefore w t must be in the N − 1 simplex ∆ (we omit the dimension superscript to lighten notations), which is then the action space:…”

“…The features (predictors) are cross-sectionally processed so that for a fixed t and given predictor j, x (j) t is uniformly distributed (over the [-0.5,0.5] interval) across firms. Scaling predictors is standard practice both in the machine learning literature and in some recent asset pricing models (e.g., Koijen and Yogo (2019), Kelly et al (2019) and Freyberger et al (2020)). In characteristicsbased investing (e.g., in Brandt et al (2009)), the indicators are for instance also demeaned and standardized.…”