Abstract
This paper studies a multi-period investment–consumption optimization problem with a stochastic discount rate and a time-varying utility function, which are governed by a Markov-modulated regime switching model. The investment is dynamically reallocated between one risk-free asset and one risky asset. The problem is time inconsistent due to the stochastic discount rate. An analytical equilibrium solution is established by resorting to a game theoretical framework. Numerous sensitivity analyses and numerical examples are provided to demonstrate the effects of the stochastic discount rate and time-varying utility coefficients on the decision-maker’s investment–consumption behavior. Our results show that many properties which are satisfied in the classical models do not hold any more due to either the stochastic discount rate or the time-varying utility function.
Similar content being viewed by others
Notes
For a dynamic optimization problem, if the optimal strategy \(\pi (n)=\left( \pi _n,\pi _{n+1},\ldots ,\pi _{T-1}\right) \) at time \(n(n=0,1,\ldots ,T-1)\) is consistent with the optimal strategy \(\pi (k)\) determined at any time \(k>n\), then it is called a time-consistent strategy. Otherwise, if there exists some \(k>n\) such that the truncated part of \(\pi (n)\), i.e., \(\left( \pi _{k},\pi _{k+1},\ldots ,\pi _{T-1}\right) \) is not equal to \(\pi (k)\), then the strategy is called a time-inconsistent strategy. When the strategy is time inconsistent, the strategy \(\pi (n)\) previously decided at time n will not be implemented at time \(k>n\) unless some commitment mechanism exists or the decision-maker is self control (Hsiaw 2013).
References
Ainslie GW (1992) Picoeconomics. Cambridge University Press, Cambridge
Barro RJ (1999) Ramsey meets Laibson in the neoclasical growth model. Q J Econ 114:1125–1152
Björk T, Murgoci A (2014) A theory of Markovian time-inconsistent stochastic control in discrete time. Finance Stoch 18:545–592
Björk T, Murgoci A, Zhou XY (2014) Mean variance portfolio optimization with state dependent risk aversion. Math Finance 24:1–24
Campbell JY, Viceira LM (2001) Strategic asset allocation: portfolio choice for long-term investors. Oxford University Press, New York
Çanakoğlu E, Özekici S (2009) Portfolio selection in stochastic markets with exponential utility functions. Ann Oper Res 166:281–297
Çanakoğlu E, Özekici S (2010) Portfolio selection with stochastic markets with HARA utility functions. Eur J Oper Res 201:520–536
Çanakoğlu E, Özekici S (2012) HARA frontiers of optimal portfolios in stochastic markets. Eur J Oper Res 221:129–137
Chen ZP (2005) Multiperiod consumption and portfolio decisions under the multivariate GARCH model with transaction costs and CVaR-based risk control. OR Spectr 27:603–632
Cheung KC, Yang HL (2004) Asset allocation with regime-switching: discrete-time case. ASTIN Bull 34:247–257
Cheung KC, Yang HL (2007) Optimal investment–consumption strategy in discrete-time model with regime switching. Discrete Contin Dyn Syst 8:315–332
Cocco JF, Gomes FJ, Maenhout PJ (2005) Consumption and portfolio choice over the life cycle. Rev Financ Stud 18:491–533
Ekeland I, Lazrak A (2006) Being serious about non-commitment: subgame perfect equilibrium in continuous time. Working paper. http://arxiv.org/abs/math/0604264
Ekeland I, Pirvu TA (2008) Investment and consumption without commitment. Math Financ Econ 2:57–86
Ekeland I, Mbodji O, Pirvu TA (2012) Time-consistent portfolio management. SIAM J Financ Math 3:1–32
Frederick S, Loewenstein G, O’Donoghue T (2002) Time discounting and time preference: a critical review. J Econ Lit 40:351–401
Goldman SM (1980) Consistent plans. Rev Econ Stud 47:533–537
Grenadier SR, Wang N (2007) Investment under uncertainty and time-inconsistent preferences. J Financ Econ 84(1):2–39
Hamilton JD (1989) A new approach to the economic analysis of non-stationary time series. Econometrica 57:357–384
Harris C, Laibson D (2013) Instantaneous gratification. Q J Econ 128(1):205–248
Harrison GW, Lau MI, Williams MB (2002) Estimating individual discount rates in Denmark: a field experiment. Am Econ Rev 92:1606–1617
Hsiaw A (2013) Goal-setting and self-control. J Econ Theory 148(2):601–626
Karatzas I, Shreve SE (1998) Methods of mathematical finance. Springer, New York
Kronborg MT, Steffensen M (2015) Inconsistent investment and consumption problems. Appl Math Optim 71:473–515
Kryger EM, Steffensen M (2010) Some solvable portfolio problems with quadratic and collective objectives. Working paper. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1577265
Li ZF, Tan KS, Yang HL (2008) Multiperiod optimal investment–consumption strategies with mortality risk and environment uncertainty. North Am Actuar J 12:47–64
Loewenstein G, Prelec D (1992) Anomalies in intertemporal choice: evidence and an interpretation. Q J Econ 107:573–597
Marín-Solano J, Navas J (2009) Non-constant discounting in finite horizon: the free terminal time case. J Econ Dyn Control 33:666–675
Marín-Solano J, Navas J (2010) Consumption and portfolio rules for time-inconsistent investors. Eur J Oper Res 201:860–872
Merton RC (1969) Lifetime portfolio selection under uncertainty: the continuous-time case. Rev Econ Stat 51:247–257
Merton RC (1971) Optimal consumption and portfolio rules in a continuous-time model. J Econ Theory 3:373–413
Pirvu TA, Zhang HY (2014) Investment–consumption with regime-switching discount rates. Math Soc Sci 71:142–150
Pollak RA (1968) Consistent planning. Rev Econ Stud 35:201–208
Read D, Read NL (2004) Time discounting over the lifespan. Organ Behav Hum Decis Process 94:22–32
Richard S (1975) Optimal consumption, portfolio and life insurance rules for an uncertain lived individual in a continuous time model. J Financ Econ 2:187–203
Samuelson PA (1969) Lifetime portfolio selection by dynamic stochastic programming. Rev Econ Stat 51:239–246
Strotz RH (1956) Myopia and inconsistency in dynamic utility maximization. Rev Econ Stud 23:165–180
Thaler R (1981) Some empirical evidence on dynamic inconsistency. Econ Lett 8:201–207
Wang J, Forsyth PA (2011) Continuous time mean-variance asset allocation: a time-consistent strategy. Eur J Oper Res 209:184–201
Wei J, Wong KC, Yam SCP, Yung SP (2013) Markowitz’s mean-variance asset-liability management with regime switching: a time-consistent approach. Insurance Math Econ 53:281–291
Weng C (2013) Constant proportion portfolio insurance under regime switching exponential Lévy process. Insurance Math Econ 52(3):508–521
Weng C (2014) Discrete-time CPPI under transaction cost and regime switching. Working paper. http://ssrn.com/abstract=2432233
Wu HL, Li ZF (2012) Multi-period mean-variance portfolio selection with regime switching and a stochastic cash flow. Insurance Math Econ 50:371–384
Zeng Y, Li ZF (2011) Optimal time-consistent investment and reinsurance policies for mean-variance insurers. Insurance Math Econ 49:145–154
Zou Z, Chen S, Wedge L (2014) Finite horizon consumption and portfolio decisions with stochastic hyperbolic discounting. J Math Econ 52:70–80
Acknowledgements
Wu acknowledges the funding support from the National Natural Science Foundation of China (Nos. 11301562, 11671411), the 111 Project (No. B17050) and the Program for Innovation Research in Central University of Finance and Economics. Weng thanks funding support from the Natural Sciences and Engineering Research Council of Canada (RGPIN-2016-04001), the Society of Actuaries Centers of Actuarial Excellence Research Grant, the National Natural Science Foundation of China (No. 71671104) Zeng acknowledges the funding support from the National Natural Science Foundation of China (Nos. 71571195, 71771220, 71721001), Fok Ying Tung Education Foundation for Young Teachers in the Higher Education Institutions of China (No. 151081), Guangdong Natural Science Funds for Distinguished Young Scholar (No. 2015A030306040) and Guangdong Natural Science for Research Team (2014A030312003).
Author information
Authors and Affiliations
Corresponding author
Appendices
The proof of Lemma 3.2
Proof
We only give the proof of (16) for \(\gamma <0\), as (15) can be proved in a similar way. When \(\gamma <0\),
Consider the following maximization problem:
where H is a positive constant. Since
the minimal value of \({c^\gamma } + {(w - c)^\gamma }H\) is achieved at
As a consequence,
which is a decreasing function of H, and thus we have (16).
Proof of Theorem 3.1
Proof
The proof will be shown only for \(0<\gamma <1\), as the results for \(\gamma <0\) can be proved in a parallel way. Throughout the proof, we always keep the constraints \(0\le c_n \le w_n\) and \(0\le \alpha _n \le 1\) for all \(n=0,1,\ldots ,T-1\). The proof will be achieved by an induction argument. To proceed, let
Then, according to Lemma 3.1, there exists a unique solution \({\hat{\alpha }}(i)\) such that
and according to (17), \(Y(i)=\mathrm{E}\left[ \zeta _n(\hat{\alpha }(i), i)\right] \). It should be noted that both \(\hat{\alpha }(i)\) and Y(i) depend on the market regime i only and are independent of time n. When \(n=T-1\), it follows from (13) and (21) that
where, obviously, \(\mathrm{E}\left[ \zeta _{T-1}(\alpha _{T-1}, i)\right] >0\). Consequently, applying Lemmas 3.1 and 3.2 to (B.1), we obtain
and
Substituting \({\hat{c}}_{T-1}(i,w_{T - 1} )\) into \(V_{T-1}\) yields
This implies that (22)–(24) hold for \(n=T-1\). Generally, we assume that (22)–(24) hold for \(T-1,T-2,\ldots ,n+1\), and we shall prove that they hold for n as well. From the given assumptions, we obtain
for \(k=T-1,T-2,\ldots ,n+1\). This recursive equation implies
which along with (23) implies
From (1), it is clear that \(W_{n+1}^{\pi _n}\) depends on the random return \(R_n(i)\) at time n. Moreover, the term \(\prod \nolimits _{m=n+1}^{k-1}{\zeta _m({\hat{\alpha }}_m,\xi _m)}\) depends on the random return \(R_{n+1}(\xi _{n+1}),\ldots ,R_{k-1}(\xi _{k-1})\), and \(\varpi _k(\xi _k) \vartheta _k(\xi _k)\) and \(A_k(\xi _k)\) are relevant to \(\xi _k\). These variables are not independent of each other at time n because the market states \(\xi _{n+1},\ldots ,\xi _k\) are interdependent. However, we recall that \(\left\{ R_0(\cdot ),R_1(\cdot ),\ldots ,R_{T-1}(\cdot )\right\} \) are conditionally independent given the market states; hence, \(W_{n+1}^{\pi _n}\), \(\zeta _{n+1}({\hat{\alpha }}_{n+1},\xi _{n+1}),\ldots ,\zeta _{k-1}({\hat{\alpha }}_{k-1},\xi _{k-1})\) are independent for given values of \(\xi _n,\xi _{n+1},\ldots ,\xi _{k-1}\). Moreover, by definition we conclude that \(\prod \nolimits _{m=n+1}^{k-1}\varpi _{m}(\xi _m)\), \(\vartheta _k(\xi _k)\) and \(A_k(\xi _k)\) are all constant for given values of \(\xi _{n+1},\ldots ,\xi _k\). Consequently, an application of the tower property of the expectation operator yields
where
Therefore, we have
Substituting (B.2) and (B.4) into (13) implies
where
and
According to the expressions of \(I_1(n,i,w_n)\) and \(I_2(n,i,w_n)\), we have
In view of
we first obtain
Further referring to (21), we have
which leads to \(I_1(n,i,w_n)-I_2(n,i,w_n)=0\). Consequently, (B.5) is equivalent to
Since \(\mathrm{E}(\zeta _n(i))H_n(i)>0\) and is a function of \(\alpha _n\), we can apply Lemma 3.2 to obtain
where the supremum of the value function is attained at
Substituting \({\hat{c}}_n(i,w_n)\) into \(V_n(i,w_n)\) and rearranging yield
This implies that the desired results also hold for n, and thus, by induction principle, the proof is complete.
Proof of Proposition 4.1
Proof
It follows from (21) that \(H_n(i)\) does not depend on \(\vartheta _n(i)\), which along with (23) immediately yields that the equilibrium consumption proportion \({\hat{\phi }}_n(i)\) is strictly increasing in \(\vartheta _n(i)\). Further note that \({\hat{\eta }}_n(i)=(1-{\hat{\phi }}_n(i)){\hat{\alpha }}_n(i)\) and \({\hat{\alpha }}_n(i)\) is irrelevant to \(\vartheta _n(i)\). Thus, \({\hat{\eta }}_n(i)\) is strictly decreasing with \(\vartheta _n(i)\).
Proof of Proposition 4.2
Proof
When the discount rates are a constant, (21) can be written as
By (18), \(H_T(\xi _T)=0\) and thus (D.1) implies
which is obviously increasing in \({\varvec{\vartheta }}_T\) for each \(i=1,\ldots , L\). Consequently, it follows from (20) and (27) that \({\hat{\phi }}_{T-1}(i)\) is decreasing in \({\varvec{\vartheta }}_T\) for each \(i=1,\ldots , L\).
For \(n=T-2\), (D.1) implies
Thus, for each \(i=1,\ldots , L\), \(H_{T-2}(i)\) is also increasing in \({\varvec{\vartheta }}_T\) as a property inherited from \(H_{T-1}(i)\), and consequently, \({\hat{\phi }}_{T-2}(i)\) is decreasing in \({\varvec{\vartheta }}_T\) due to (20) and (27).
In general, we assume that \(H_{n+1}(i)\) is increasing in \({\varvec{\vartheta }}_T\) for each \(i=1,\ldots , L\), and from (D.1) we can obtain a similar equation for \(H_n(\cdot )\) in terms of \(H_{n+1}(\cdot )\) similar to (D.2) to conclude that \(H_n(i)\) is indeed increasing in \({\varvec{\vartheta }}_T\). Then, by (20) and (27), \({\hat{\phi }}_n(i)\) is decreasing in \({\varvec{\vartheta }}_T\), and therefore, by the induction principle, we complete the proof w.r.t. the monotonic property of \({\hat{\phi }}_n\) in the general case.
The results for \({\hat{\eta }}_n(i)\) follow straightforward from the equation \({\hat{\eta }}_n(i)=(1-{\hat{\phi }}_n(i)){\hat{\alpha }}_n(i)\) and the fact that \({\hat{\alpha }}_n(i)\) is irrelevant to \(\vartheta _n(i)\). The proof for results with all elements of Q being positive can be obtained in a complete parallel way.
Proof of Proposition 4.3
Proof
When the discount rates are constant, (21) can be equivalently written as Eq. (D.1), from which it follows for \(k=n, \ldots , T-1\) that
where \(H_{k+1}(j)\) does not depend on \({\varvec{\vartheta }}_{k+1}\). Therefore, it is straightforward to verify that, for any \(i,j=1,\ldots ,L\) and \(k=n,\ldots , T-1\),
which implies that \(H_k(i)\) is increasing in \({\varvec{\vartheta }}_{k+1}\). On the other hand, by (20) and (27), \({\hat{\phi }}_n(i)\) admits the following expression
and hence it is decreasing in \({\varvec{\vartheta }}_{n+1}\). Moreover, (E.1) and (E.3) together imply that \(\hat{\phi }_n(i)\) depends on \({\varvec{\vartheta }}_{n+2}\) only via \(H_{n+1}(\cdot )\). Since (E.2) indicates that \(H_{n+1}(j)\) is increasing in \({\varvec{\vartheta }}_{n+2}\), \(\hat{\phi }_n(i)\) is also decreasing in \({\varvec{\vartheta }}_{n+2}\). Repeatedly using recursion (E.1) and the positiveness of the derivative in (E.2), we can conclude by the induction principle that the equilibrium consumption proportion \({\hat{\phi }}_n(i)\) is decreasing in \({{\varvec{\vartheta }}}_k\) for each \(k=n+1,n+2,\ldots ,T-1\).
The increasing property of \({\hat{\eta }}_n(i)\) with respect to \({\varvec{\vartheta }}_k\) for each \(k=n+1,n+2,\ldots ,T-1\) follows trivially from the equation \({\hat{\eta }}_n(i)=(1-{\hat{\phi }}_n(i)){\hat{\alpha }}_n(i)\) and the fact that \({\hat{\alpha }}_n(i)\) is irrelevant to \(\vartheta _n(i)\). The results with all elements Q being positive can be obtained in a complete parallel way.
Proof of Proposition 4.4
Proof
When the assumptions in Proposition 4.4 hold, by (18) and (21), \(H_n\) can be simplified as
where \(Y=\mathrm{E}\left[ \left( (1-{\hat{\alpha }}_n)r_f+{\hat{\alpha }}_nR_n\right) ^{\gamma }\right] >0\). According to (20), substituting \(Y{H_n} = \vartheta _n{A_n}\) into the above equation yields
Therefore, we have
with \(A_T=0\). Next, under the assumptions in Proposition 4.4, we show that \(A_n>A_{n+1}~(n=0,1,\ldots , T-1)\) by using induction principle. To this end, by (20), we first note that
In general, with the assumption \(A_{n+1}\ge A_{n+2}\), we obtain
If \(\frac{{{\vartheta _n}}}{{{\vartheta _{n + 1}}}} \le \frac{{{\vartheta _{n + 1}}}}{{{\vartheta _{n + 2}}}},~~n=0,1,\ldots ,T-3\), then \(A_n\ge A_{n+1}\). Consequently, by (29), we obtain
Rights and permissions
About this article
Cite this article
Wu, H., Weng, C. & Zeng, Y. Equilibrium consumption and portfolio decisions with stochastic discount rate and time-varying utility functions. OR Spectrum 40, 541–582 (2018). https://doi.org/10.1007/s00291-017-0502-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00291-017-0502-2