Dynamic Asset Allocation Research Articles

Life-time portfolio selection model

In the field of financial economics, many researchers have developed optimal portfolio selection models. Unfortunately, those models cannot be implemented directly by investors since the coefficient of risk aversion is an exogenous variable. However, even if investors have no idea about their attitudes toward risk, investors might specify a maximum probability of failing to reach a specific portfolio threshold. According to this argument, we regard stop-loss level as portfolio threshold and penetrate the attitude towards risk through the stop-loss level. Then, we can achieve investor’s optimal life-time portfolio selection model. Key words: Dynamic asset allocation, portfolio optimization, value-at-risk, stop-loss.

Correlations between stock returns and bond returns: income and substitution effects

We attempt to better understand the varying correlations between stock and bond returns across countries and over sample periods using international data. The observation is that there are two forces that affect the correlation between stock and bond returns. The force that drives a positive correlation is identified as the income effect. The force that drives a negative correlation is identified as the substitution effect. In combination, the two effects help determine the actual correlation between stock and bond returns. We contribute to the literature by proposing an empirical method, the structural vector autoregression (VAR) identification method, to identify the two—income and substitution—effects and to measure the relative importance of the two effects that determine the actual net relation between the two asset returns. We further provide some evidence that the income and substitution effects are related to, among other things, the size of the financial market, the growth and volatility (risk) of the economy, and the business cycle over time. In addition, the framework of the income and substitution effects helps us better understand the automatic stabilizing effects of the dynamic optimal asset allocation during business cycles.

Risk Management for Institutional Funds

Many investment managers agree that the traditional approach to asset allocation and risk management is “dead.” No consensus has emerged on a new approach, but a risk-driven approach is gaining popularity. Many institutions are exploring the use of a more dynamic asset allocation process. Any new approach should be adopted on the basis of the investment beliefs of the organization, the governance setup of the organization, and the skill set of the staff.

Dynamic Asset Allocation with Learning

This paper shows that when an investor optimally rebalances her portfolio, learning about the parameter of the return process still induces a large negative hedging demand even after observing 83 years of asset market data. An investor with a 5-year investment horizon decreases the percentage of wealth she allocates to the stock index by over 20 percent when she takes learning into account. Furthermore, this paper shows that the initial estimation sample length needs to be at least 500 years in order for the effect of learning to vanish.

Asset Allocation in Markets with Contagion: The Interplay between Volatilities, Jump Intensities, and Correlations

We study the impact of financial contagion on the dynamic asset allocation problem of a CRRA investor facing an incomplete market with two risky assets. We apply a Markov chain regime-switching framework with state-dependent jump intensities, diffusion volatilities and diffusion correlations. The key model feature that a switch to the bad contagion regime is triggered by a loss in one of the risky assets allows for the implementation of a hedging demand against contagion risk. Moreover, a state-dependent diffusion correlation combined with heterogeneity in jump intensities and volatilities can, e.g., generate a flight to quality effect upon a systemic jump.

Time Consistent vs. Time Inconsistent Dynamic Asset Allocation: Some Utility Cost Calculations for Mean Variance Preferences

We solve for the time consistent dynamic asset allocation of an investor with a mean variance objective function in a multiple assets affine setting. We use as a benchmark the pre-commitment strategy widely used in the literature and assess the potential welfare gains from pre-commitment by comparing the time consistent strategy to the pre-commitment, time inconsistent, strategy. The gains from pre-commitment are simply considerable since, in some cases, at the 5 years horizon the yearly certainty equivalent of the pre-commitment strategy is 48% compared with 9% for the time consistent strategy. However, these welfare gains result from huge and unrealistic positions in the risky assets; in some cases, the pre-commitment strategy is more than 60 times the time consistent strategy. We thus looked for alternative time inconsistent strategies that improve relative to the time consistent strategy while still involving reasonable risky asset positions. To identify these strategies, we explore an original aspect of the time consistent mean variance strategy: the presence of intertemporal hedging in such a strategy reflects welfare degradation. Therefore, a natural candidate is the time consistent strategy without the intertemporal hedging component. The second component of the time consistent strategy is the traditional myopic component discounted. We show that this component could be seen as a standard myopic strategy which is marked to market and the discount factor acts as a tailing factor. This marked to market myopic (MMM) strategy is shown to yield reasonable risky assets positions and substantial welfare gains at long horizons relative to the time consistent strategy. We also show that it dominates the standard myopic strategy as well as the equally weighted strategy.

The costs of suboptimal dynamic asset allocation: General results and applications to interest rate risk, stock volatility risk, and growth/value tilts

The recent theoretical asset allocation literature has derived optimal dynamic investment strategies in various advanced models of asset returns. But how sensitive is investor welfare to deviations from the theoretically optimal strategy? Will unsophisticated investors do almost as well as sophisticated investors? This paper develops a general theoretical framework for answering such questions and applies it to three specific models of interest rate risk, stochastic stock volatility, and mean reversion and growth/value tilts of stock portfolios. Among other things, we find that growth/value tilts are highly valuable, but the hedging of time-varying stock risk premia is less important.

Dynamic Versus Static Asset Allocation: From Theory (Halfway) to Practice

In a static optimisation process, the structure of the portfolio in which the investor’s wealth is invested is chosen once for all at the beginning of the period. In a dynamic optimisation process, the structure of the portfolio is continuously (at least theoretically) adjusted. There is no doubt that the dynamic optimisation process is superior to the static one, as one of the dynamic strategies among many is to choose not to adjust the portfolio structure, i.e. to choose a static strategy. The aim of this paper is to assess, with parameter values as reasonable as possible, the order of magnitude of the extent by which the dynamic optimisation process dominates the static optimisation process; and to gauge whether the existence of transaction costs changes significantly the assessment. That assessment is made considering that the values of the relevant parameters are known at the beginning of the investment period. The paper thus does not to try to assess the benefits that dynamic optimisation could reap by incorporating new information on those parameters. The main results are that for high risk adverse investors, the improvement provided by dynamic versus static optimisation is moderate; it may be huge for low risk adverse investors, but this comes principally from the fact that dynamic optimisation leads to a huge leverage, which is normally prohibited in a static framework. Reasonable values for transaction costs do not jeopardise the superiority of dynamic asset allocation.

Infrastructure Investments: New Evidence from Dynamic Asset Allocation

Direct infrastructure has emerged as a separate asset class in recent years and found its way into investment portfolios. Employing a US transaction-based index which covers the Q2 1990 to Q2 2010 period, this paper investigates the role of infrastructure in a multi-asset portfolio. We determine time-varying asset allocations using a mean-variance, as well as a mean-downside risk optimization algorithm. The results show that infrastructure plays an important role in asset allocation, particularly in mean-downside-risk efficient portfolios. The share of infrastructure amounts to a maximum of almost 37% of total wealth and it is allocated consistently over time. Allocations in mean-variance efficient portfolios are significantly lower and infrastructure is not consistently allocated over time. Independently of the optimization algorithm employed, infrastructure is allocated predominantly to low-risk low-return portfolios. Furthermore, infrastructure is an attractive asset for investors with investment horizons exceeding two years. Moreover, the allocation to infrastructure is highly dependent on whether large cap stocks are allocated to the portfolio. Due to the similar underlying long-term characteristics of large cap stocks and infrastructure, both assets exhibit significant positive correlations over longer investment horizons.

Dynamic portfolio frontier in a mean–variance framework

The dynamic portfolio frontier theory in a mean–variance framework previously developed by scholars suffers some limitations. Specifically, the theory assumes the use of the martingale approach, the assumption of a complete market and particular probability distribution of asset returns. Accordingly, under relaxing these limitations, this study develops a calculation process for explicitly deriving the dynamic portfolio frontier and the corresponding dynamic asset allocation. Finally, for comparison, this study provides a numerical example and then draws the dynamic and static portfolio frontiers on the same graph.

Are There Benefits from Dynamic Asset Allocation Strategies Across Hedge Funds?

The performance of various asset allocation strategies across hedge fund indices is compared using both static and dynamic methods based on forecasts of conditional volatility. Daily rebalanced dynamic portfolios are examined for the three main subindices of Standard & Poor’s Hedge Fund Index. Out-of-sample results are also reported for nine Credit Suisse First Boston/Tremont hedge fund indices. Time-varying volatility and volatility clustering characterizes most hedge fund indices. Using forecasts of next-period volatility based on a time-varying procedure generally improves the risk–return profile of the portfolio. All of the dynamic hedge fund index portfolios largely outperform the passive S&P 500 Index, both on an expected return and risk-adjusted return basis.

Dynamic strategic asset allocation: Risk and return across the business cycle

We propose a practical investment framework for dynamic asset allocation across different phases in the business cycle, which we illustrate using a sample of US data from 1948 to 2007. We identify four phases in the business cycle and find that these capture pronounced time variation in the risk and return properties of asset classes. Time variation is also observed in the risk of a traditional, static strategic asset mix. In order to stabilize risk across the business cycle, we propose a dynamic strategic asset allocation approach, which has the potential to enhance expected return as well. The proposed investment framework is found to be robust to variations in the variable composition of the business cycle indicator and can easily be extended with different economic variables and/or additional assets.

Market Timing with Option-Implied Distributions: A Forward-Looking Approach

We address the empirical implementation of the static asset allocation problem by developing a forward-looking approach that uses information from market option prices. To this end, constant maturity S&P 500 implied distributions are extracted and subsequently transformed to the corresponding risk-adjusted ones. Then, we form optimal portfolios consisting of a risky and a risk-free asset and evaluate their out-of-sample performance. We find that the use of risk-adjusted implied distributions times the market and makes the investor better off compared with the case where she uses historical returns’ distributions to calculate her optimal strategy. The results hold under a number of evaluation metrics and utility functions and carry through even when transaction costs are taken into account. Not surprisingly, the reported market timing ability deteriorated during the recent subprime crisis. An extension of the approach to a dynamic asset allocation setting is also presented.

The Costs of Suboptimal Dynamic Asset Allocation: General Results and Applications to Interest Rate Risk, Stock Volatility Risk, and Growth/Value Tilts

The recent theoretical asset allocation literature has derived optimal dynamic investment strategies in various advanced models of asset returns. But how sensitive is investor welfare to deviations from the theoretically optimal strategy? Will unsophisticated investors do almost as well as sophisticated investors? This paper develops a general theoretical framework for answering such questions and applies it to three specific models of interest rate risk, stochastic stock volatility, and mean reversion and growth/value tilts of stock portfolios. Among other things, we find that growth/value tilts are highly valuable, but the hedging of time-varying stock risk premia is less important.

A NONLINEAR CONTROL POLICY USING KERNEL METHOD FOR DYNAMIC ASSET ALLOCATION(<Special Issue>SCOPE (Seminar on Computation and OPtimization for new Extensions))

A NONLINEAR CONTROL POLICY USING KERNEL METHOD FOR DYNAMIC ASSET ALLOCATION(<Special Issue>SCOPE (Seminar on Computation and OPtimization for new Extensions))

Glide path and dynamic asset allocation of target date funds

The glide path of typical target date funds is based on the relatively simple assumption of risk. If an explicit term structure of risk is present or risk is time-varying, the conventional glide path may not be adequate to fulfil the purpose of target date funds. We introduce a new approach to define the glide path of target date funds. Our starting point is to determine the level of risk budget for each target date. According to the pre-defined risk budget, we derive the asset allocation of target date funds by explicitly incorporating the current term structure of risk. Because risk does change through different market phases, we implement a dynamic asset allocation strategy for target date funds which considers simultaneously both the pre-defined risk budget and the prevailing market risks. The main difference is that at any given time our risk-controlled dynamically-rebalanced target date funds would not exceed the pre-defined risk budget regardless of market movements.

Market timing under multiple economic regimes

This article models the US equity premium as a regime-switching process where the regimes are dependent on economic variables. To characterise the economic regimes, we employ the dimension reduction technique of a principal components analysis to extract business cycle signals from a set of observed macroeconomic variables. We use these conditioning agents to infer the ex ante economic regime. We then test a dynamic asset allocation strategy, which invests in equity and cash on the basis of the predicted regimes. This timing strategy is shown to outperform a simple buy and hold strategy on a risk-adjusted basis.

Dynamic Asset Allocation under Mispricing, Predictability and Portable Alpha

The existing literature about portfolio management has investigated how to update a portfolio allocation, conditional on the information that possibly predicts asset returns and volatilities. We add several innovations to fill the lacuna of prior research in the contexts of global asset allocation. First, we suggest a simple method of how to rebalance portfolios automatically and dynamically in order to exploit potential market inefficiencies. The existing literature has not developed such a strategy. Out-of-sample tests demonstrate that our strategy dominates both static allocation and dynamic strategies that do not account for possible mispricing. Thus, our strategy can contribute not only to academia, but also to practical portfolio managers who endeavour to beat markets. Second, we elaborate portable alpha strategies using the new dynamic strategy. Once we add an alpha strategies using the new dynamic strategy. Once we add an alpha portfolio to existing portfolios, then they perform better in terms of mean and risk. Thus, it makes our alpha portfolio portable, i.e., we can apply the alpha portfolio to any fund and can enhance its performance. Third, our dynamic strategy implies a convenient method to estimate a conditional mean and covariance matrix as functions of predictive information matrix without consuming much computational risk managers and traders who need to control the risks of large target portfolios on a real time basis.

Out-of-sample Predictions of Bond Excess Returns and Forward Rates: An Asset-Allocation Perspective

This paper revisits the predictability of bond excess returns by means of long-term forward interest rates. We assess the economic value of out-of-sample forecasting ability of empirical models based on forward rates in a dynamic asset allocation strategy. Our results show that the information content of forward rates does not generate any systematic economic value to investors. The performance of the predictive models against the no-predictability benchmark worsens over time and the few positive performance fees recorded from dynamic portfolio strategies based on forward rates are generally small in size and do not offset realistic transaction costs.

Correlations between Stock Returns and Bond Returns: Income and Substitution Effects

We attempt to better understand varying correlations between stock and bond returns across countries and over sample periods using international data. The observation is that there are two forces that affect the correlation between stock and bond returns. The force that drives a positive correlation is identified as the income effect. The force that drives a negative correlation is identified as the substitution effect. In combination, the two effects help determine the actual correlation between stock and bond returns. We contribute to the literature by proposing an empirical method, structural vector autoregression (VAR) identification method, to identify the two – income and substitution – effects and to measure the relative importance of the two effects that determine the actual net relation between the two asset returns. We further provide some evidence that the income and substitution effects are related to, among other things, the size of the financial market, the growth and volatility (risk) of the economy, and the business cycle over time. In addition, the framework of the income and substitution effects helps us better understand the automatic stabilizing effects of the dynamic optimal asset allocation during business cycles.