Abstract
This study incorporates the Markov switching model with return jumps to depict the behavior of stock returns. Based on the daily Standard & Poor’s 500 index (hereafter SPX) and the daily closing price of the call option, we use the particle filtering algorithm to fit the parameter of the model. The joint log-likelihood evaluates the model performance: the weighted average log-likelihood with the rate of return of the SPX and the relative implied volatility root-mean-squared error for the SPX call options. The empirical results identify that the pricing model with jump risks improves the pricing performance to the median-term call options. According to the sensitivity analysis, option prices increase with the probability of remaining in the recession state but decrease with the probability of remaining in the expansion state. Moreover, the call option prices are positively associated with the volatility in each market state and the factors of jump risk.
Similar content being viewed by others
References
Bakshi G, Madan D (2000) Spanning and derivative-security valuation. J Financ Econ 55:205–238
Bakshi G, Cao C, Chen Z (1997) Empirical performance of alternative option pricing models. J Finance 52:2003–2049
Bates DS (1996) Jump and stochastic volatility: exchange rate processes implicit in Deutsche Mark options. Rev Financ Stud 9:69–107
Bates DS (2000) Post-87 crash fears in S&P 500 futures options. J Econom 94:181–238
Bates DS (2012) U.S. stock market crash risk, 1926–2010. J Financ Econ 105:229–259
Black F, Scholes M (1973) The pricing of options and corporate liabilities. J Polit Econ 81:637–654
Chang C, Cheng HW, Fuh CD (2018) Ensuring more is better: on the simultaneous application of stock and options data to estimate the GARCH options pricing model. J Deriv 26:7–25
Chen SS (2007) Does monetary policy have asymmetric effects on stock returns? J Money Credit Bank 39:667–688
Christoffersen P, Jacobs K, Ornthanalai C (2012) Dynamic jump intensities and risk premiums: evidence from S&P500 returns and options. J Financ Econ 106:447–472
Duan JC, Popova I, Ritchken P (2002) Option pricing under regime switching. Quant Finance 2:116–132
Eisenberg L, Jarrow R (1994) Option pricing with random volatilities in complete markets. Rev Quant Financ Acc 4:5–17
Elliott R, Chan L, Siu T (2005) Option pricing and Esscher transform under regime switching. Ann Finance 1:423–432
Eraker B (2004) Do stock prices and volatility jump? reconciling evidence from spot and option prices. J Finance 59:1367–1403
Eraker B, Johannes M, Polson N (2003) The impact of jumps in volatility and returns. J Finance 58:1269–1300
Fleming J, Ostdiek B, Whaley RE (1995) Predicting stock market volatility: a new measure. J Futures Mark 15:265–302
Gerber H, Shiu E (1994) Option pricing by Esscher transforms. Trans Soc Actuar 46:99–140
Hamilton JD (1989) A new approach to the economic analysis of nonstationary time series and the business cycle. Econometrica 57:357–384
Hamilton JD (1990) Analysis of time series subject to changes in regime. J Econom 45:39–70
Hamilton JD, Susmel R (1994) Autoregressive conditional heteroskedasticity and changes in regime. J Econom 64:307–333
Hardy MR (2001) A regime-switching model of long-term stock returns. N Am Actuar J 5:41–53
Harrison JM, Pliska SR (1981) Martingales and stochastic integrals in the theory of continuous trading. Stoch Process Their Appl 11:215–260
Harrison JM, Pliska SR (1983) A stochastic calculus model of continuous trading: complete markets. Stoch Process Their Appl 15:313–316
Harvey CR, Whaley RE (1992) Market volatility prediction and the efficiency of the S&P 100 index option market. J Financ Econ 31:43–73
Heston S (1993) A closed-form solution for options with stochastic volatility with applications to bond and currency options. Rev Financ Stud 6:327–343
Heston S, Nandi S (2000) A closed-form GARCH option pricing model. Rev Financ Stud 13:585–626
Kim CJ, Morley JC, Nelson CR (2004) Is there a positive relationship between stock market volatility and the equity premium? J Money Credit Bank 36:336–360
Kim CJ, Morley JC, Nelson CR (2005) The structural break in the equity premium. J Bus Econ Stat 23:181–191
Li B (2019) Option-implied filtering: evidence from the GARCH option pricing model. Rev Quant Financ Acc. https://doi.org/10.1007/s11156-019-00816-5
Lin CH, Lin SK, Wu AC (2015) Foreign exchange option pricing in the currency cycle with jump risks. Rev Quant Financ Acc 44:755–789
Maheu JM, McCurdy TH (2000) Identifying bull and bear markets in stock returns. J Bus Econ Stat 18:100–112
Merton R (1976) Option pricing when underlying stock returns are discontinuous. J Financ Econ 3:125–144
Ornthanalai C (2014) Lévy jump risk: evidence from options and returns. J Financ Econ 112:69–90
Schwert GW (1989) Business cycles, financial crises, and stock volatility. Carnegie Rochester Conf Ser Public Policy 31:83–126
Stentoft L (2008) American option pricing using GARCH models and the normal inverse Gaussian distribution. J Financ Econom 6:540–582
Turner CM, Startz R, Nelson CR (1989) A Markov model of heteroscedasticity, risk and learning in the stock market. J Financ Econ 25:3–22
Wu CC (2006) The GARCH option pricing model: a modification of lattice approach. Rev Quant Financ Acc 26:55–66
Funding
Funding was provided by Ministry of Science and Technology, Taiwan (Grant No. NSC 105-2410-H-004-087-).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix A: European call option pricing formula
Assume that the jump risk premiums are the same in the market state 1 and 2, i.e. \(h_{2,1} = h_{2,2} = h_{2}\). Given the filtration \({\mathbb{F}}\left( t \right)\), the days belonging to the market state 1 (denoted by \(m\)), the stock return between time \(t\) and \(T\), \(\ln S\left( T \right) - \ln S\left( t \right) = \mathop \sum \limits_{u = t + 1}^{T} R\left( u \right)\), under the \({\mathbb{Q}}\) measure is
where \(V_{m} = \sigma_{1}^{2} k + \sigma_{2}^{2} \left( {T - t - m} \right)\) is the variance with \(m\) days visiting the market state 1 between time \(t\) and \(T\). And, the random variables under the \({\mathbb{Q}}\) measure: \(\varepsilon^{{\mathbb{Q}}} \left( t \right) \sim {\text{Normal}}\left( {0,1} \right)\) is the Gaussian noise process, \(N\left( {T - t} \right) \sim {\text{Poisson}}\left[ {\lambda^{{\mathbb{Q}}} \left( {T - t} \right)} \right]\) is the number of jumps during \(\left( {t,T} \right]\), and \(Y_{n} \mathop \sim\limits^{\text{IID}} f_{Y}^{{\mathbb{Q}}} \left( {y;h_{2} } \right)\) is the jump size. Let the random variable \(X\) be the log-price of stock at time \(T\), i.e. \(X = \ln S\left( T \right) = \ln S\left( t \right) + \mathop \sum \limits_{u = t + 1}^{T} R\left( u \right)\), and \({\mathbb{G}}\left( t \right) = {\mathbb{F}}\left( t \right) \vee m\) be enlarged filtration. Then, we can derive its conditional characteristic function as follow:
Consider a \(T\)-maturity European call option with the strike price \(K\). Given the filtration \({\mathbb{F}}\left( t \right)\), the option pricing formula is
where the second equation is derived by the law of iterated expectation. By the inverse Fourier transform of the characteristic function, we can rewrite Eq. (22) as follows:
The second equations in Eqs. (23) and (24) are computed by the Fubini theorem, and the third equations in Eqs. (23) and (24) are derived by the property of complex conjugate. A similar proof can be shown in Bakshi and Madan (2000). Therefore, the pricing formula in Eq. (22) is rewritten by
where \(\psi_{T,m|q\left( t \right) = j}\) presents the probability that the stock market belongs to the state 1 for \(k\) days while the time-to-maturity is \(T - t\) days given the initial market state \(q\left( t \right) = j\). It is proved in Duan, Popova, and Ritchken (2002) with a hidden Markov Chain.
Appendix B: Benchmark models: stochastic volatility processes
Our model comprises a Markov switching dynamics for the volatility and it can be categorized in the stochastic volatility models. Under the discrete-time framework, Li (2019), Ornthanalai (2014), Christoffersen et al. (2012), Wu (2006) Heston and Nandi (2000) employ the affine GARCH models to depict the stock return and its variance. That is,
where \(\mu\) is the mean of stock return, \(V\left( t \right)\) is the conditional variance of stock return at time \(t\), \(\sigma^{2}\) is the intercept term of the variance equation, \(\gamma\) is the coefficient of leverage effect, \(\eta\) measures the level of asymmetric leverage effect between the good and bad events, \(\xi\) is the parameter of volatility clustering effect. The variance is s stationary process if and only if \(\gamma \eta^{2} + \xi < 1\) is satisfied. Also, note that \(V\left( t \right)/2\) and \(\lambda \left[ {g_{Y}^{{\mathbb{Q}}} \left( { - i} \right) - 1} \right]\) are the convexity adjustment terms, which make the stock return equal to \(\mu\) under the \({\mathbb{P}}\) measure, i.e. \({\text{E}}_{{{\mathbb{F}}\left( {t - 1} \right) \vee q\left( t \right)}}^{{\mathbb{P}}} \left[ {S\left( t \right)} \right] = S\left( {t - 1} \right)e^{{\mu_{q\left( t \right)} }}\). Equations (26) and (27) are called by the GARCH model with return jumps (GARCH-RJ). If there are no jumps, \(\lambda = 0\), Eqs. (26) and (27) reduce to Heston and Nandi (2000)’s GARCH model.
On the other hand, the stock return is also measured under the continuous-time framework like Bates (2012, 2000, 1996), Eraker (2004), Eraker et al. (2003), Bakshi et al. (1997), Eisenberg and Jarrow (1994), Heston (1993). The dynamic of stock log-price under the stochastic volatility model is
where \(\mu\) is the instantaneous mean, \(V\left( t \right)\) is the instantaneous variance at time \(t\), \(1 - \xi\) is the mean-reverting speed, \(\sigma^{2}\) is the long-run mean level, and \(\gamma\) is the volatility of variance. Moreover, \(W^{{\mathbb{P}}} \left( t \right) \sim {\text{Normal}}\left( {0,t} \right)\) and \(\hat{W}^{{\mathbb{P}}} \left( t \right) \sim {\text{Normal}}\left( {0,t} \right)\) are the correlated Brownian motions with the correlation coefficient \(\rho\). \(N\left( t \right) \sim {\text{Poisson}}\left( {\lambda t} \right)\) and \(\hat{N}\left( t \right) \sim {\text{Poisson}}\left( {\hat{\lambda }t} \right)\) are the independent number of jumps for the return and the variance with the time-homogeneous arrival rate. \(Y_{n} \mathop \sim \limits^{\text{IID}} {\text{Normal}}\left( {\theta ,\nu^{2} } \right)\) and \(\hat{Y}_{n} \mathop \sim \limits^{\text{IID}} {\text{Exponential}}\left( {\hat{\theta }} \right)\) are the independent jump sizes for the return- and the variance-jumps, respectively. Equations (28) and (29) are called by the stochastic volatility model with return-jumps and volatility-jumps (SV-RJ-VJ). If there are no jumps for volatility, \(\hat{\lambda } = 0\), the model reduces to the stochastic volatility model with return jumps (SV-RJ) like Bates (2012, 2000) and Bakshi et al. (1997). Next, if the stock market does not exist the jump risks for the return and volatility, \(\lambda = \hat{\lambda } = 0\), the dynamic of stock log-price degenerates to Heston (1993)’s SV model. For comparing under the same assumption, in this paper, we suppose that the jump frequencies of return and volatility are time-homogeneous under the model with jumps.
Rights and permissions
About this article
Cite this article
Wang, SY., Chuang, MC., Lin, SK. et al. Option pricing under stock market cycles with jump risks: evidence from the S&P 500 index. Rev Quant Finan Acc 56, 25–51 (2021). https://doi.org/10.1007/s11156-020-00885-x
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11156-020-00885-x