Dynamic asset allocation and latent variables

Carsten Sørensen, Anders Bjerre Trolle

Research output: Working paperResearch

1 Downloads (Pure)

Abstract

We derive an explicit solution to the portfolio problem of a power utility investorwith preferences for wealth at a ¯nite investment horizon. The investor can investin assets with return dynamics described as part of a general multivariate model.The modeling framework encompasses discrete-time VAR-models where some ofthe state-variables (e.g. expected excess returns) may not be directly observable.A realistic multivariate model is estimated and applied to analyze the portfolioimplications of investment horizon and return predictability when real interest ratesand expected excess returns on stock and bonds are not directly observed but mustbe estimated as part of the problem faced by the investor. The solution exhibitssmall variability in portfolio allocations over time compared to the case when excessreturns are assumed observable.JEL Classification: G11Keywords: Portfolio choice, predictability, VAR, unobserved state-variables, hedging demands
Original languageEnglish
Place of PublicationKøbenhavn
PublisherCopenhagen Business School [wp]
Number of pages55
ISBN (Print)8790705874
Publication statusPublished - 2004

Cite this

Sørensen, C., & Trolle, A. B. (2004). Dynamic asset allocation and latent variables. København: Copenhagen Business School [wp].
Sørensen, Carsten ; Trolle, Anders Bjerre. / Dynamic asset allocation and latent variables. København : Copenhagen Business School [wp], 2004.
@techreport{ecb6efe0c02011db9769000ea68e967b,
title = "Dynamic asset allocation and latent variables",
abstract = "We derive an explicit solution to the portfolio problem of a power utility investorwith preferences for wealth at a ¯nite investment horizon. The investor can investin assets with return dynamics described as part of a general multivariate model.The modeling framework encompasses discrete-time VAR-models where some ofthe state-variables (e.g. expected excess returns) may not be directly observable.A realistic multivariate model is estimated and applied to analyze the portfolioimplications of investment horizon and return predictability when real interest ratesand expected excess returns on stock and bonds are not directly observed but mustbe estimated as part of the problem faced by the investor. The solution exhibitssmall variability in portfolio allocations over time compared to the case when excessreturns are assumed observable.JEL Classification: G11Keywords: Portfolio choice, predictability, VAR, unobserved state-variables, hedging demands",
keywords = "Investeringsteori, Portef{\o}ljestyring, Asset allocation, Value at risk, Portef{\o}ljeteori",
author = "Carsten S{\o}rensen and Trolle, {Anders Bjerre}",
year = "2004",
language = "English",
isbn = "8790705874",
publisher = "Copenhagen Business School [wp]",
address = "Denmark",
type = "WorkingPaper",
institution = "Copenhagen Business School [wp]",

}

Sørensen, C & Trolle, AB 2004 'Dynamic asset allocation and latent variables' Copenhagen Business School [wp], København.

Dynamic asset allocation and latent variables. / Sørensen, Carsten; Trolle, Anders Bjerre.

København : Copenhagen Business School [wp], 2004.

Research output: Working paperResearch

TY - UNPB

T1 - Dynamic asset allocation and latent variables

AU - Sørensen, Carsten

AU - Trolle, Anders Bjerre

PY - 2004

Y1 - 2004

N2 - We derive an explicit solution to the portfolio problem of a power utility investorwith preferences for wealth at a ¯nite investment horizon. The investor can investin assets with return dynamics described as part of a general multivariate model.The modeling framework encompasses discrete-time VAR-models where some ofthe state-variables (e.g. expected excess returns) may not be directly observable.A realistic multivariate model is estimated and applied to analyze the portfolioimplications of investment horizon and return predictability when real interest ratesand expected excess returns on stock and bonds are not directly observed but mustbe estimated as part of the problem faced by the investor. The solution exhibitssmall variability in portfolio allocations over time compared to the case when excessreturns are assumed observable.JEL Classification: G11Keywords: Portfolio choice, predictability, VAR, unobserved state-variables, hedging demands

AB - We derive an explicit solution to the portfolio problem of a power utility investorwith preferences for wealth at a ¯nite investment horizon. The investor can investin assets with return dynamics described as part of a general multivariate model.The modeling framework encompasses discrete-time VAR-models where some ofthe state-variables (e.g. expected excess returns) may not be directly observable.A realistic multivariate model is estimated and applied to analyze the portfolioimplications of investment horizon and return predictability when real interest ratesand expected excess returns on stock and bonds are not directly observed but mustbe estimated as part of the problem faced by the investor. The solution exhibitssmall variability in portfolio allocations over time compared to the case when excessreturns are assumed observable.JEL Classification: G11Keywords: Portfolio choice, predictability, VAR, unobserved state-variables, hedging demands

KW - Investeringsteori

KW - Porteføljestyring

KW - Asset allocation

KW - Value at risk

KW - Porteføljeteori

M3 - Working paper

SN - 8790705874

BT - Dynamic asset allocation and latent variables

PB - Copenhagen Business School [wp]

CY - København

ER -

Sørensen C, Trolle AB. Dynamic asset allocation and latent variables. København: Copenhagen Business School [wp]. 2004.