Abstract
We prove global existence of a continuous-time Nash equilibrium with endogenous persistent and exogenous temporary price impact. Relative to the analogous Radner and Pareto-efficient equilibria, the Nash equilibrium has a lower interest rate but has similar Sharpe ratios and stock-return volatility.
Similar content being viewed by others
References
Almgren, R.: Optimal execution with nonlinear impact functions and trading-enhanced risk. Appl. Math. Finance 10, 1–18 (2003)
Almgren, R., Thum, C., Hauptmann, E., Li, E.: Direct estimation of equity market impact. Risk 18, 58–62 (2005)
Amihud, Y., Mendelson, H., Pedersen, L.H.: Liquidity and asset prices. Found. Trends Finance 1, 269–364 (2006)
Back, K.: Insider trading in continuous time. Rev. Financ. Stud. 5, 387–409 (1992)
Back, K., Cao, H., Willard, G.: Imperfect competition among informed traders. J. Finance 55, 2117–2155 (2000)
Baker, M., Wurgler, J.: Trader sentiment and the cross-section of stock returns. J. Finance 61, 1645–1680 (2006)
Bansal, R., Yaron, A.: Risks for the long run: a potential resolution of asset pricing puzzles. J. Finance 59, 1481–1509 (2004)
Barro, R.J.: Rare disasters and asset markets in the twentieth century. Q. J. Econ. 121, 823–866 (2006)
Basak, S.: Dynamic consumption-portfolio choice and asset pricing with non-price-taking agents. Working paper (1994). Available online at https://rodneywhitecenter.wharton.upenn.edu/wp-content/uploads/2014/04/9408.pdf
Basak, S.: Consumption choice and asset pricing with a non-price-taking agent. Econ. Theory 10, 437–462 (1997)
Basak, S., Cuoco, D.: An equilibrium model with restricted stock market participation. Rev. Financ. Stud. 11, 309–341 (1998)
Bouchard, B., Fukasawa, M., Herdegen, M., Muhle-Karbe, J.: Equilibrium returns with transaction costs. Finance Stoch. 22, 569–601 (2018)
Breeden, D.: An intertemporal asset pricing model with stochastic consumption and investment opportunities. J. Financ. Econ. 7, 265–296 (1979)
Brunnermeier, M.K., Pedersen, L.H.: Predatory trading. J. Finance 60, 1825–1863 (2005)
Calvet, L.E.: Incomplete markets and volatility. J. Econ. Theory 98, 295–338 (2001)
Campbell, J.Y., Cochrane, J.H.: By force of habit: a consumption-based explanation of aggregate stock market behavior. J. Polit. Econ. 107, 205–251 (1999)
Campbell, J.Y., Grossman, S.J., Wang, J.: Trading volume and serial correlation in stock returns. Q. J. Econ. 108, 905–939 (1993)
Chen, X., Choi, J.H., Larsen, K., Seppi, D.J.: Learning about latent dynamic trading demand. Math. Financ. Econ. 16, 615–658 (2022)
Chien, Y., Cole, H., Lustig, H.: A multiplier approach to understanding the macro implications of household finance. Rev. Econ. Stud. 78, 199–234 (2011)
Choi, J.H., Larsen, K.: Taylor approximation of incomplete Radner equilibrium models. Finance Stoch. 19, 653–679 (2015)
Choi, J.H., Larsen, K., Seppi, D.J.: Equilibrium effects of intraday order-splitting benchmarks. Math. Financ. Econ. 15, 315–352 (2021)
Christensen, P.O., Graversen, S.E., Miltersen, K.R.: Dynamic spanning in the consumption-based capital asset pricing model. Eur. Finance Rev. 4, 129–156 (2000)
Christensen, P.O., Larsen, K., Munk, C.: Equilibrium in securities markets with heterogeneous traders and unspanned income risk. J. Econ. Theory 147, 1035–1063 (2012)
Christensen, P.O., Larsen, K.: Incomplete continuous-time securities markets with stochastic income volatility. Rev. Asset Pricing Stud. 4, 247–285 (2014)
Cochrane, J.H.: Financial markets and the real economy. In: Mehra, R. (ed.) Handbook of the Equity Risk Premium, pp. 237–325. Elsevier, Amsterdam (2008)
Constantinides, G.M.: Capital market equilibrium with transaction costs. J. Polit. Econ. 94, 842–862 (1986)
Constantinides, G.M.: Habit formation: a resolution of the equity premium puzzle. J. Polit. Econ. 98, 519–543 (1990)
Constantinides, G.M.: Understanding the equity risk premium puzzle. In: Mehra, R. (ed.) Handbook of the Equity Risk Premium, pp. 331–376. Elsevier, Amsterdam (2008)
Constantinides, G.M., Gosh, A.: Asset pricing with countercyclical household consumption risk. J. Finance 72, 415–460 (2017)
Constantinides, G.M., Duffie, D.: Asset pricing with heterogeneous consumers. J. Polit. Econ. 104, 219–240 (1996)
Coval, J., Stafford, E.: Asset fire sales (and purchases) in equity markets. J. Financ. Econ. 86, 479–512 (2007)
Cvitanić, J., Cuoco, D.: Optimal consumption choices for a large investor. J. Econ. Dyn. Control 22, 401–436 (1998)
Duffie, D., Gârleanu, N., Pedersen, L.H.: Over the counter markets. Econometrica 73, 1815–1847 (2005)
Easley, D., O’Hara, M.: Microstructure and asset pricing. In: Constantinides, G., et al. (eds.) Handbook of the Economics of Finance 1B, pp. 1021–1051. Elsevier, Amsterdam (2003)
Gârleanu, N., Pedersen, L.H.: Dynamic portfolio choice with frictions. J. Econ. Theory 165, 487–516 (2016)
Gonon, L., Muhle-Karbe, J., Shi, X.: Asset pricing with general transaction costs: theory and numerics. Math. Finance 31, 595–648 (2021)
Grossman, S.J., Stiglitz, J.E.: On the impossibility of informationally efficient markets. Am. Econ. Rev. 70, 393–408 (1980)
Hartman, P.: Ordinary Differential Equations, 2nd edn. SIAM, Philadelphia (2002)
Hasbrouck, J.: Measuring the information content of stock trades. J. Finance 46, 179–207 (1991)
Heaton, J., Lucas, D.: The effects of incomplete insurance markets and trading costs in a consumption-based asset pricing model. J. Econ. Dyn. Control 16, 601–620 (1992)
Hellwig, M.F.: On the aggregation of information in competitive markets. J. Econ. Theory 22, 477–498 (1980)
Hendershott, T., Menkveld, A.J.: Price pressures. J. Financ. Econ. 114, 405–423 (2014)
Hugonnier, J.: Rational asset pricing bubbles and portfolio constraints. J. Econ. Theory 147, 2260–2302 (2012)
Kardaras, C., Xing, H., Žitković, G.: Incomplete stochastic equilibria with exponential utilities close to Pareto optimality. In: Yin, G., Zariphopoulou, T. (eds.) Stochastic Analysis, Filtering, and Stochastic Optimization: A Commemorative Volume to Honor Mark H.A. Davis’s Contributions, pp. 267–292. Springer, Berlin (2022)
Krueger, D., Lustig, H.: When is market incompleteness irrelevant for the price of aggregate risk (and when is it not)? J. Econ. Theory 145, 1–41 (2010)
Kyle, A.: Continuous auctions and insider trading. Econometrica 53, 1315–1336 (1985)
Larsen, K., Sae-Sue, T.: Radner equilibrium in incomplete Lévy models. Math. Financ. Econ. 10, 321–337 (2016)
LeRoy, S.F., Porter, R.D.: The present value relation: Tests based on implied variance bounds. Econometrica 49, 555–574 (1981)
Mehra, R., Prescott, E.C.: The equity premium: a puzzle. J. Monet. Econ. 15, 145–162 (1985)
Noh, E., Weston, K.: Price impact equilibrium with transaction costs and TWAP trading. Math. Financ. Econ. 16, 187–204 (2022)
O’Hara, M.: High frequency market microstructure. J. Financ. Econ. 116, 257–270 (2015)
Pritsker, M.: Large traders: implications for equilibrium asset returns, shock absorption, and liquidity. Working paper (2009). Available online at https://w4.stern.nyu.edu/finance/docs/pdfs/Seminars/091f-pritsker.pdf
Sannikov, Y., Skrzypacz, A.: Dynamic trading: price inertia and front-running. Working paper (2016). Available online at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2882809
Schied, A., Schöneborn, T.: Risk aversion and the dynamics of optimal liquidation strategies in illiquid markets. Finance Stoch. 13, 181–204 (2009)
Shiller, R.J.: Do stock prices move too much to be justified by subsequent changes in dividends? Am. Econ. Rev. 71, 421–436 (1981)
Storesletten, K., Telmer, C., Yaron, A.: Asset pricing with idiosyncratic risk and overlapping generations. Rev. Econ. Dyn. 10, 519–548 (2007)
Storesletten, K., Telmer, C., Yaron, A.: Asset prices and intergenerational risk sharing: the role of idiosyncratic earnings shocks. In: Mehra, R. (ed.) Handbook of the Equity Risk Premium, pp. 565–590. Elsevier, Amsterdam (2008)
Vayanos, D.: Strategic trading and welfare in a dynamic market. Rev. Econ. Stud. 66, 219–254 (1999)
Vayanos, D.: Strategic trading in a dynamic noisy market. J. Finance 56, 131–171 (2001)
Vayanos, D., Vila, J.-L.: Equilibrium interest rate and liquidity premium with transaction costs. Econ. Theory 13, 509–539 (1999)
Wachter, J.A.: Can time-varying risk of rare disasters explain aggregate stock market volatility? J. Finance 68, 987–1035 (2013)
Wang, N.: Caballero meets Bewley: the permanent-income hypothesis in general equilibrium. Am. Econ. Rev. 93, 927–936 (2003)
Weil, P.: The equity premium puzzle and the risk-free rate puzzle. J. Monet. Econ. 24, 401–421 (1989)
Weston, K.: Existence of a Radner equilibrium in a model with transaction costs. Math. Financ. Econ. 12, 517–539 (2018)
Weston, K., Žitković, G.: An incomplete equilibrium with a stochastic annuity. Finance Stoch. 24, 359–382 (2020)
Žitković, G.: An example of a stochastic equilibrium with incomplete markets. Finance Stoch. 16, 177–206 (2012)
Acknowledgements
The authors have benefited from helpful comments from two anonymous referees, Peter Bank, Yashar Barardehi, Suleyman Basak, René Carmona, George Constantinides, Ibrahim Ekren, Paolo Gasoni (the Editor), Joel Hasbrouck, Burton Hollifield, Ulrich Horst, Ioannis Karatzas, Lars Kuehn, Bryan Routledge, Mete Soner, Chris Telmer, Nicolae Gârleanu (AFA discussant), Hao Xing and seminar participants at the Tepper School of Business (Carnegie Mellon), ORFE (Princeton), FSU, Intech, TU Berlin and the 2021 AFA meetings.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing Interests
The authors declare no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The second author was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2020R1C1C1A01014142 and No. 2021R1A4A1032924). The third author has been supported by the National Science Foundation under Grant No. DMS 1812679 (2018–2022). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF).
Appendices
Appendix A: Auxiliary ODE result
In the following ODE existence proof, there are no restrictions on the finite time horizon \(T\in (0,\infty )\) and the constant \(C_{0}\in {\mathbb{R}}\). We note that the ODE (A.3) is quadratic in \(g(t)\) and that the squared term \(-\frac{2}{\alpha}\) is negative given \(\alpha >0\).
Proposition A.1
For \(I\in {\mathbb{N}}\), \(C_{0}\in {\mathbb{R}}\) and positive constants \(T,a,\sigma _{D},\alpha ,k>0\), there exists a unique constant \(\hat {h}_{0} \in (0,k)\) such that the ODE system
with initial condition \(h_{0}:=\hat {h}_{0}\) has a unique solution for \(t\in [0,T]\) such that \(h(T)=k\).
Proof
The proof consists of three steps.
Step 1 (range of \(h\)). Let \(h_{0}\in (0,k)\) be given. We evolve the ODEs (A.1)–(A.3) from \(t=0\) to the right (\(t>0\)). The local Lipschitz property of the ODEs ensures that there exists a maximal interval of existence \([0, \tau )\) with \(\tau \in (0,\infty ]\) by the Picard–Lindelöf theorem (see e.g. Hartman [38, Theorem II.1.1]).
For a constant \(c\), let \(T_{f=c}\in [0,\tau ]\) be defined as
where as usual \(\inf \emptyset := +\infty \). Define \(T_{g=c}\) and \(T_{h=c}\) similarly. Suppose that \(T_{f=0}<\tau \). Then \(f(0)=1\) and the continuity of \(f\) imply that \(f(t)>0\) for \(t \in [0,T_{f=0})\). Since \(f(T_{f=0})=0\), we have \(f'(T_{f=0})\leq 0\), but (A.2) implies \(f'(T_{f=0})=1>0\). Therefore we conclude that
Because \(g(0)=0\) and \(g'(0)=a \sigma _{D}^{2}>0\), we have \(T_{g=0}>0\) and \(g(t)>0\) for \(t\in (0,T_{g=0})\). The ODE (A.1) with \(h(0)=h_{0}>0\) implies that \(t \mapsto h(t)\) increases on the interval \([0, T_{g=0})\). Therefore, the ODE (A.2) and the positivity of \(f,h\) produce
Then Gronwall’s inequality yields
This inequality implies that
Suppose that \(T_{g=0}<\tau \). Since \(g(t)>0\) for \(t\in (0,T_{g=0})\) and \(g(T_{g=0})=0\), we have \(g'(T_{g=0})\leq 0\). However, this is a contradiction because (A.3) and (A.4) imply that \(g'(T_{g=0})\geq a \sigma _{D}^{2} C_{1} >0\), where the positive constant \(C_{1}\) is defined in (A.4).
Up to this point, we have shown that
To proceed, the positive constant
satisfies
Because \(0\leq g(t) < C_{2}\) for \(t\in [0, T_{g=C_{2}})\), we can bound (A.3) from below using (A.5) and (A.6) to see that for \(t\in [0, T_{g=C_{2}})\),
By integrating (A.7) and using the initial condition \(g(0)=0\), we get \(g(t)\ge \frac{1}{2}a \sigma _{D}^{2} C_{1}t\) for \(t\in [0, T_{g=C_{2}})\). Therefore,
Suppose that \(T_{g=C_{2}}=\tau \). Then for \(t\in [0,\tau )\), we have \(0 \leq g(t)< C_{2}\) and the ODE (A.1) produces
where the second inequality uses Gronwall’s inequality. Similarly, for \(t\in [0,\tau )\), the ODE (A.2) and Gronwall’s inequality imply
The boundedness properties \(g(t)< C_{2}\), (A.9) and (A.10) imply that \(h,f\) and \(g\) do not blow up for finite \(t\). Then [38, Theorem II.3.1] ensures \(\tau =\infty \) which contradicts (A.8). Consequently, we cannot have \(T_{g=C_{2}}=\tau \) and it must be the case that
Let \(\hat {T}_{g=C_{2}}\) be defined as the first time \(g\) reaches \(C_{2}\) strictly after time \(t=T_{g=C_{2}}\), that is,
Because \(g'(T_{g=C_{2}})\geq \frac{a \sigma _{D}^{2} C_{1}}{2}>0\) by (A.7), we have
Suppose that \(\hat {T}_{g=C_{2}}<\tau \). Then \(g(\hat {T}_{g=C_{2}})=C_{2}\) and (A.11) imply that \(g'(\hat {T}_{g=C_{2}})\leq 0\), but (A.3), (A.5) and (A.6) produce the contradiction
Therefore, it must be the case that \(\hat {T}_{g=C_{2}}=\tau \), which implies the lower bound
Combining (A.7) and (A.12) gives the global lower bound
In turn, using the ODE (A.1), the bound (A.13) produces via Gronwall’s inequality the global lower bound
Next, we suppose \(T_{h=k}=\tau \). Then for \(t\in [0,\tau )\), we have \(0 \leq h(t)< k\), and the ODEs (A.2), (A.3) and Gronwall’s inequality imply that
where \(C_{3}:= \frac{a^{2} \sigma _{D}^{2}}{2I}k +|C_{0}| \). The inequalities in (A.15) and \(0 \leq h(t)\le k\) imply that \(h,f\) and \(g\) do not blow up for finite \(t\). Then [38, Theorem II.3.1] ensures \(\tau =T_{h=k}=\infty \). This is a contradiction because (A.14) implies that \(h(t)\) reaches \(k\) in finite time. Therefore, it must be the case that
Step 2 (Monotonicity). Let \(0< h_{0}<\tilde{h}_{0}<k\), and denote the solution of the ODE system (A.1)–(A.3) with initial condition \(h(0)=\tilde{h}_{0}\) by \(\tilde{f},\tilde{h}\) and \(\tilde{g}\). The corresponding maximal existence interval is denoted by \((0,\tilde{\tau})\). We define \(T_{g=\tilde{g}}\) as
Because we have \(g(0)=\tilde{g}(0)=0\), the ODEs (A.1)–(A.3) have the properties that \(g'(0)=\tilde{g}'(0)=a \sigma _{D}^{2}\) and
Therefore,
Suppose that \(T_{g=\tilde{g}}<\tau \wedge \tilde{\tau}\). The inequality (A.17) and the ODEs (A.1) and (A.2) imply that
Also, (A.17) and \(g(T_{g=\tilde{g}})=\tilde{g}(T_{g=\tilde{g}})\) produce \(g'(T_{g=\tilde{g}})\geq \tilde{g}'(T_{g=\tilde{g}})\). However, this contradicts the fact that
where we used (A.3) and (A.18). Therefore we conclude that \(T_{g=\tilde{g}}=\tau \wedge \tilde{\tau}\) and
Step 3 (Existence). In order to emphasise the dependence on the initial condition \(h(0)=h_{0}\), we write \(\tau (h_{0})\) and \(T_{h=k}(h_{0})\). For example,
The inequality (A.16) in Step 1 implies that \(T_{h=k}(h_{0})<\infty \) for \(h_{0}\in (0,k)\). Step 2 implies that the map \((0,k)\ni h_{0}\mapsto T_{h=k}(h_{0})\) is strictly decreasing. Therefore the following three statements and the intermediate value theorem complete the proof in the sense that we can choose a unique \(\hat{h}_{0}\in (0,k)\) such that \(T_{h=k}(\hat{h}_{0})=T\) (recall that \(T\in (0,\infty )\) is the model time horizon):
-
(i)
\(\lim _{h_{0} \uparrow k}T_{h=k}(h_{0})=0\).
-
(ii)
\(\lim _{h_{0} \downarrow 0} T_{h=k}(h_{0})= \infty \).
-
(iii)
The map \((0, k)\ni h_{0} \mapsto T_{h=k}(h_{0})\) is continuous.
The proofs of these three statements are as follows:
(i) Inequality (A.14) implies (i).
(ii) The inequalities in (A.15) and Gronwall’s inequality produce
Obviously, the function \([0,\infty )\ni t \to \exp ( \int _{0}^{t} \frac{2a \sigma _{D}^{2}e^{C_{3} s}(s+\frac{1}{2}s^{2})}{\alpha} ds )\) is increasing. Therefore, for any \(t_{0}>0\), we can choose \(h_{0}>0\) such that
and use (A.19) to see that \(T_{h=k}(h_{0})>t_{0}\). This shows (ii).
(iii) Let \(h_{0}\in (0,k)\) be fixed. To emphasise the dependence on the initial condition, we write \((h(t),g(t) )\) as \((h(t,h_{0}),g(t,h_{0}) )\). The local Lipschitz structure of the ODEs (A.1)–(A.3) gives us the continuous dependence of their solutions on the initial condition \(h_{0}\) (see e.g. [38, Theorem V.2.1]), that is,
For \(0< x< h_{0}\), we have \(T_{h=k}(h_{0})< T_{h=k}(x)\), and the ODE (A.1) and the fundamental theorem of calculus produce
where the second to last line uses the bounds (A.13) and (A.14) and \(C_{4}>0\) is an irrelevant constant independent of \(x\). Letting \(x\uparrow h_{0}\) and using (A.20) yields
The opposite inequality trivially holds because \(x \mapsto T_{h=k}(x)\) is strictly decreasing. Therefore (A.21) holds with equality. Similarly, for \(x\in (h_{0},\frac{k+h_{0}}{2})\), we have \(T_{h=k}( \frac{k+h_{0}}{2})< T_{h=k}(x)< T_{h=k}(h_{0})\) and
for a constant \(C_{5}\) independent of \(x\). Letting \(x\downarrow h_{0}\) and using (A.20) yields
Again, the opposite inequality trivially holds because \(x\mapsto T_{h=k}(x)\) is strictly decreasing. Therefore (A.22) holds with equality, and the continuity property follows. □
Proposition A.2
Let \(h_{0}=0\) in (A.1). Then the ODEs (A.1)–(A.3) have unique solutions on \(t\in [0,\infty )\) with \(h(t)=0\) for all \(t\geq 0\).
Proof
As in the proof of Proposition A.1, denote the maximal interval of existence by \((0,\tau )\) for \(\tau \in (0,\infty ]\). For \(t\in [0,\tau )\), the solutions to (A.1) and (A.2) are
As in the proof of Proposition A.1, we can check that
Then (A.3), (A.23) and (A.24) imply that for \(t\in [0,\tau )\),
Gronwall’s inequality implies that \(g\) cannot blow up in finite time. Therefore we conclude that \(\tau =\infty \). □
Appendix B: Proof of Lemma 3.2, Theorem 3.3 and Corollary 4.1
Proof of Lemma 3.2
We prove that the coupled ODEs (3.1)–(3.3) have unique solutions for \(t\in [0,T]\). We apply Propositions A.1 and A.2 with
where \(k\) is nonnegative by the Cauchy–Schwarz inequality. The functions
solve (3.1)–(3.3) for \(t\in [0,T]\).
From (A.4) in the proof of Proposition A.1, we know that \(f(t)\) is bounded away from zero for \(t\in [0,T]\). Therefore the solutions to the linear ODEs for \(Q(t)\) and \(Q_{2}(t)\) in (3.5) and (3.4) can be found by integration. □
Proof of Theorem 3.3
The proof consists of three steps.
Step 1 (Individual optimality). In this step, we define the function
for \(t\in [0,T]\) and \(M_{i},D,\theta _{i},Y_{i}\in {\mathbb{R}}\). In (B.2), the deterministic functions are defined in (3.2) and (3.3). We note that the terminal ODE conditions produce
Consequently, because \(S_{i,T}=D_{T}\), we have
which is the terminal condition in (2.7). Next, we show that the function \(e^{-\delta t} v\) with \(v\) defined in (B.2) is the value function for (2.7). Let \((\theta '_{i},c_{i})\in {\mathcal {A}}\) be arbitrary. Itô’s lemma shows that the process \((e^{-\delta t}v+\int _{0}^{t} e^{-a c_{i,u} - \delta u}du)\), with \(v\) being shorthand notation for \(v(t,M_{i,t},D_{t},\theta _{i,t},Y_{i,t})\), has dynamics
where we have used the ODEs (3.1)–(3.3) and the conjecture for the interest rate in (3.8), and have inserted the perceived price-impact model (3.7) into the money market account balance process \(dM_{i,t}\) in (2.5). The local martingale in the last line in (B.3) can be upgraded to a martingale. To see this, we note that \(\theta _{i}\) is bounded and \(v\) is square-integrable by (2.8), so that we can use the Cauchy–Schwarz inequality to obtain the needed integrability. Furthermore, to see that the drift in (B.3) is nonnegative, we note that the second-order conditions for the HJB equation are (there are no cross terms)
The first inequality in (B.4) holds because \(F(t)\) in (3.2) is the annuity (\(>0\)). Consequently, the drift in (B.3) is minimised to zero by the controls (3.10) and (3.11). This implies that \((e^{-\delta t}v+\int _{0}^{t} e^{-a c_{i,u} - \delta u}du)\) is a submartingale for all admissible order-rate and consumption processes \(\theta '_{i}\) and \(c_{i}\).
It remains to verify admissibility of the controls (3.10) and (3.11). The explicit solution (3.14) is deterministic and uniformly bounded. Inserting the controls (3.10) and (3.11) into the money market account balance dynamics (2.5) produces
The linear SDE (B.5) has a unique well-defined (Gaussian) solution that satisfies (2.8). All in all, this shows that the admissibility requirements in Definition 2.1 are satisfied, and hence optimality of (3.10) and (3.11) follows from the martingale property of \((e^{-\delta t}v+\int _{0}^{t} e^{-a \hat{c}_{i,u} - \delta u}du)\).
Step 2 (Clearing). Clearly, summing the orders in (3.10) and using \(\sum _{i=1}^{I} \theta _{i,0}=L\) shows that the stock market clears for all \(t\in [0,T]\). Summing (3.11) gives
Because \(\psi (0) = \sum _{i=1}^{I}\theta _{i,0}^{2}\) and \(t \mapsto \sum _{i=1}^{I}\hat{\theta}_{i,t}^{2}\) satisfies the ODE (3.1), we have \(\psi (t)= \sum _{i=1}^{I}\hat{\theta}_{i,t}^{2}\) for all \(t\in [0,T]\). Therefore, the real-goods market clears if and only if
The terminal conditions in the ODEs (3.3)–(3.5) ensure that clearing holds at time \(t=T\). By computing time derivatives in (B.6) and using \(r(t)\) defined in (3.8), we see that clearing holds for all \(t\in [0,T]\).
Step 3 (Consistency and terminal conditions). By replacing \(\theta _{i,t}\) in (3.7) with \(\hat{\theta}'_{i,t}\) from (3.10), we verify that trader \(i\)’s perceived stock-price \(S_{i,t}\) becomes \(\hat{S}_{t}\) in (3.9) (identical for all traders \(i\in \{1,\dots ,I\}\)). Also, direct computations show that (3.6), (3.9) and (3.10) produce the consistency condition (2.14). Finally, the terminal stock-price condition \(\hat{S}_{T}=D_{T}\) for the equilibrium stock-price process \(\hat{S}\) in (3.9) holds by the terminal conditions in the ODEs (3.2), (3.4) and (3.3). □
Proof of Corollary 4.1
(i) The expressions in (3.8) and (C.1) imply that the inequality \(r(t)\leq r^{\mathrm{Radner}}\) is equivalent to the inequality \(\psi (t) \geq \frac{L^{2}}{I}\). From the proofs of Propositions A.1 and A.2, we know that the function \(h\) is either strictly positive or identically equal to zero on the interval \([0,T]\). Therefore, the expression for \(\psi \) in (B.1) implies that \(\psi (t)\geq \frac{L^{2}}{I}\). Furthermore, \(\psi (t)= \frac{L^{2}}{I}\) if and only if \(\sum _{i=1}^{I} \theta _{i,0}^{2}=\psi (0)=\frac{L^{2}}{I}\), which by the Cauchy–Schwarz inequality is equivalent to \(\theta _{i,0}=\frac{L}{I}\) for all \(1\leq i\leq I\).
When \(r(t)=r^{\mathrm{Radner}}\), the corresponding annuities \(F(t)\) and \(F^{\mathrm{Radner}}(t)\) also agree. Therefore, from (3.9) and (C.2), we see that \(\hat{S} = S^{\mathrm{Radner}}\) if and only if
To see (B.7), we first note that for \(t=T\), both sides are equal to zero. Second, by computing \(t\)-derivatives of both sides of (B.7) and using the ODEs (3.4) and (3.3), we see that these derivatives match.
By using the expressions in (C.1) and (D.2), we obtain
When \(r^{\mathrm{Pareto}}=r^{\mathrm{Radner}}\), we see that \(S^{\mathrm{Radner}}\) in (C.2) agrees with \(S^{\mathrm{Pareto}}\) in (D.3).
(ii) The equilibrium stock-price processes in (3.9), (C.2) and (D.3) have quadratic variation processes
where the Nash annuity is given in (3.2) and the Radner and Pareto annuities are defined as
The Radner and Pareto annuities have derivatives (4.7) and (4.8). Therefore, inserting (3.8) into the Nash annuity (3.2) and using the interest rate ordering (4.3) yields the ordering (4.6).
(iii) To see that (4.5) holds, we write the equity premium (4.1) as
where \(S^{(0)}_{t}:= e^{\int _{0}^{t} r(u)du}\). L’Hôpital’s rule and the ODEs (3.1)–(3.3) produce the limit
To compute the numerator in (4.2), we define the function
Then we apply Itô’s lemma to compute the dynamics of \(d (B_{t}G(t) )\), which produces the representation
Therefore, the Itô isometry gives the variance
Based on (B.8) and (B.9), the limit in (4.5) equals
where \(\lambda \) is the market price of risk defined in (3.13).
The Radner and Pareto instantaneous Sharpe ratios can be computed similarly. □
Appendix C: Competitive Radner equilibrium
This section briefly describes the analogous competitive Radner equilibrium where all traders are price takers who trade with no price impact. Christensen et al. [23, Theorem 2] show that there exists a competitive Radner equilibrium in which the equilibrium interest rate is given by
and the equilibrium stock-price process is given by
Itô’s lemma and (C.2) yield the competitive Radner equilibrium stock-price volatility coefficient of \(S^{\mathrm{Radner}}_{t}\) to be
Equivalently, we can write (C.3) as \(F^{\mathrm{Radner}}(t)\sigma _{D}\), where the Radner annuity \(F^{\mathrm{Radner}}(t)\) is given by (4.7). Furthermore, based on (C.2) and \(F^{\mathrm{Radner}}(t)\) in (4.7), there exists a deterministic function \(f^{\mathrm{Radner}}(t)\) such that the Radner stock-return volatility is
Appendix D: Pareto-efficient equilibrium
While Pareto-efficient markets can be incomplete, they must be effectively complete as defined in Christensen et al. [22]. This section briefly describes the competitive Pareto-efficient equilibrium where all traders are price takers who trade with no price impact, and the resulting equilibrium allocations are Pareto-efficient. To implement this equilibrium when \(\rho ^{2}<1\) in (2.3), it is necessary to add \(I\) additional non-dividend-paying securities (each in zero net supply). However, for our purposes, the specific implementation is irrelevant. The following analysis uses the CCAPM analysis from Breeden [13]. The utilities (2.7) produce the representative agent’s utility function as
Because the economy’s aggregate consumption is \(LD_{t} + \sum _{i=1}^{I} Y_{i,t}\), the Pareto-efficient equilibrium model’s unique state-price density \(\xi ^{\mathrm{Pareto}}=(\xi _{t}^{\mathrm{Pareto}})_{t\in [0,T]}\) is proportional to the process
Itô’s lemma produces the relative state-price dynamics to be
From (D.1), the Pareto-efficient equilibrium’s interest rate (i.e., the \(dt\)-term in the dynamics \(-\frac{d\xi _{t}^{\mathrm{Pareto}}}{\xi _{t}^{\mathrm{Pareto}}}\)) and the market price of risk related to the Brownian motion \(B\) (i.e., the \(dB_{t}\) volatility term in the dynamics \(-\frac{d\xi _{t}^{\mathrm{Pareto}}}{\xi _{t}^{\mathrm{Pareto}}}\)) are
In turn, (D.2) yields the stock-price process in the Pareto-efficient equilibrium as
Itô’s lemma and (D.3) produce the Pareto-efficient equilibrium stock-price volatility coefficient of \(S^{\mathrm{Pareto}}_{t}\) to be
Equivalently, we can write (D.4) as \(F^{\mathrm{Pareto}}(t)\sigma _{D}\), where the annuity \(F^{\mathrm{Pareto}}(t)\) is given by (4.8).
Similarly to (C.4), we can write the Pareto-efficient stock-return volatility as
for a deterministic function \(f^{\mathrm{Pareto}}(t)\).
Appendix E: Model parameters
This appendix describes the parameters used in Sect. 4.2 to illustrate the quantitative effects of price impact. In our numerics, time is measured on an annual basis (i.e., one year is \(t=1\)). We normalise the outstanding stock supply to \(L=100\). As noted in Remark 3.4, 3), the key quantity in explaining the asset pricing puzzles is the heterogeneity in traders’ initial endowments as measured by the difference \(\sum _{i=1}^{I} \theta _{i,0}^{2}- \frac{L^{2}}{I}\ge 0\) which is a metric for the distance of the initial stock endowments from Pareto-efficiency. To provide some intuition for this difference, we note that the cross-sectional average and standard deviation of a set of initial stock endowments \(\boldsymbol {\theta}_{0}:=\{\theta _{1,0},\dots ,\theta _{I,0}\}\) are
where \(\psi (t)\) is the function from (3.1).
The utility parameters for (2.7) in our numerics are
The annual time-preference rate \(\delta \) is consistent with calibrated time preferences in Bansal and Yaron [7], and the level of absolute risk aversion \(a\) is from the numerics in Christensen et al. [23]. The coefficients for the arithmetic Brownian motion for the stock dividends in (2.2) are
The parametrisations of \(\mu _{D}\) and \(\sigma _{D}\) are the annualised mean and standard deviation of monthly percentage changes in aggregate real US stock market dividends from January 1970 through December 2019 (which were taken from Robert Shiller’s website http://www.econ.yale.edu/~shiller/data.htm). The starting annualised dividend rate is set to \(D_{0}=1\) in (E.2). The discount rate \(\delta \), dividend parameters \(\mu _{D}\) and \(\sigma _{D}\) and income parameters \(\mu _{Y}\) and \(\sigma _{Y}\) are all quoted in decimal form, where \(0.01 = 1\%\). The annualised income volatility and income-dividend correlation, from the numerics in Christensen et al. [23], are
The drift \(\mu _{Y}\) and number of traders \(I\in {\mathbb{N}}\) are found by calibrating the Radner equilibrium model so that
which produces the remaining coefficients
We set the model horizon to \(T = 3\) years. In our analysis, our numerics are relatively insensitive to \(T\) once \(T\) is sufficiently large.
The Nash model with price impact has two additional parameters relative to the competitive Radner model: the temporary price-impact coefficient \(\alpha \) in (3.7) and the difference \(\sum _{i=1}^{I}\hat{\theta}_{i,t}^{2} - L^{2}/I= \psi (t) - L^{2}/I\) for deviations of initial stock endowments from identical stock holdings \(L/I\), which is related to \(\text{SD}[\boldsymbol {\theta}_{0}]\) in (E.1). Figures 1 and 2 in Sect. 4.2 illustrate the sensitivity of the model to these two parameters.
The challenge in calibrating the temporary price-impact parameter \(\alpha \) in (3.7) is that \(\alpha \) in our model is a measure of the perceived price impact of fundamental trading imbalances for the aggregate stock market due to frictions in accessing asset-holding capacity from other natural end-counterparties (e.g. large pensions and mutual funds) and not transactional bid–ask bounce and market-maker inventory effects. In contrast, most empirical research measures temporary price effects for individual orders for individual stocks (e.g. as in Hasbrouck [39], Hendershott and Menkveld [42] and Almgren et al. [2]). The two concepts are related but there are some differences. First, \(\alpha \) represents the temporary price effects of orders in trading programs associated with underlying parent orders rather than with isolated child orders (see e.g. O’Hara [51]) and one-off single orders. Second, trading occurs in practice both via liquidity-making limit orders as well as via liquidity-taking market orders. From a transactional perspective, market and limit orders have opposite prices of liquidity since one is paying for liquidity and the other is being compensated for providing liquidity. However, limit buying and market buying both create fundamental asset-holding pressure on the available ultimate (i.e., non-market-maker) asset sellers. It is the latter that \(\alpha \) measures in our model. Third, stock in our model represents the aggregate stock market as an asset class and thus differs from individual stocks both in terms of its scale and as being a source of systematic risk rather than also including idiosyncratic stock-specific randomness.
Our numerics in Sect. 4.2 consider two temporary price-impact parameters of \(\alpha \in \{0.01,0.002\}\). The value \(\alpha =0.002\) is roughly consistent with temporary price-impact estimates in [2]. To put this in perspective, a price impact of \(\alpha = 0.002\) means that if a trader trades at a constant rate \(\theta '_{i}=265\) to sell \(\int _{0}^{\frac{1}{265}}\theta '_{i}dt=1\) unit of the stock over a day (i.e., a large daily parent trade of 1 percent of \(L = 100\) shares outstanding), the associated temporary price increase at each time \(t\) in the day would be \(0.002\times 265=0.53\). The stock (with \(\alpha =0.002\) and \(\text{SD}[ \boldsymbol {\theta}_{0}]=5\)) has an endogenous initial equilibrium price of \(\hat{S}_{0} = 3.5737\); so this corresponds to a sustained percentage temporary price impact of \(\frac{0.002\times 265}{3.5737}= 14.83\%\) over the day.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Chen, X., Choi, J.H., Larsen, K. et al. Price impact in Nash equilibria. Finance Stoch 27, 305–340 (2023). https://doi.org/10.1007/s00780-023-00499-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00780-023-00499-w