This paper proposes a second-order jump diffusion model to study the jump dynamics of stock market returns via adding a jump term to traditional diffusion model. We develop an appropriate maximum likelihood approach to estimate model parameters. A simulation study is conducted to evaluate the performance of the estimation method in finite samples. Furthermore, we consider a likelihood ratio test to identify the statistically significant presence of jump factor. The empirical analysis of stock market data from North America, Asia, and Europe is provided for illustration.
National Natural Science Foundation of China1160140911701286Natural Science Foundation of Jiangsu ProvinceBK20171073University Natural Science Foundation of Education Department of Jiangsu Province of China17KJB110006University Philosophy and Social Science Foundation of Education Department of Jiangsu Province of China2017SJB03501. Introduction
Continuous-time stochastic processes have been widely used to model securities prices for option valuation. Nicolau considered a second-order diffusion process which is defined by(1)dYt=Xtdt,dXt=μXtdt+σXtdWt,where μ(Xt) and σ(Xt) are the drift and diffusion functions, respectively [1]. Wt is a standard one-dimensional Brownian motion. Yt is directly observable and differentiable. In this model, Yt can also be expressed as the integrated process(2)Yt=Y0+∫0tXudu.For model (1), a nonparametric approach which is based on the infinitesimal generator and Taylor series expansion has been developed to estimate the drift and diffusion functions [1]. Thereafter, Wang and Lin [2] presented the local linear estimation of the diffusion and drift functions and proved that the estimators are weakly consistent. Wang et al. [3] proposed the reweighted estimation of the diffusion function and investigated the consistency of the estimator. Furthermore, Hanif [4] studied the nonparametric estimation of the drift and diffusion functions using an asymmetric kernel and proved that the estimators are consistent and asymptotically normal.
As pointed out by Nicolau [1, 5], model (1) is especially useful in empirical finance. First, the model accommodates nonstationary integrated stochastic process Yt that can be made stationary by differencing. Second, in the context of stock prices, Xu represents stock return and Yt indicates the cumulation of Xu. The model suggests directly modeling return rather than stock price and meets many general properties of stock returns such as stationarity in the mean, nonnormality of the distribution and weak autocorrelation.
In financial markets, the correct specification of drift and volatility is essential and instructive among practitioners in obtaining valid conclusions. Unfortunately, the existing economic theory generally provides little guidance about the precise specification of them. Model misspecification may lead to misleading conclusions in estimation and hypothesis testing. Therefore, much attention has been paid to the issue of specifying the functions forms for continuous-time diffusion models. On the other hand, it has the advantage that the problem of its estimation can be reduced to the determination of some low-dimensional parameters by applying more efficient statistical methods (see [6, 7] for details). In particular, Nicolau [1, 5] pointed out that(3)dYt=Xtdt,dXt=α0+α1Xtdt+β+τXt-θ2dWtis a promising model for stock returns and possesses some interesting properties. In Nicolau’s empirical studies on the American stock markets, a regular pattern in all estimators of drift and diffusion can be observed: the drift is clearly linear, the volatility is a quadratic function with a minimum in the neighborhood of zero, and the specification σ2(Xt)=β+τ(Xt-θ)2 fits the nonparametric estimators very well. Furthermore, Yan and Mei [8] developed the generalized likelihood ratio test to check the empirical finding of Nicolau. The empirical analysis of real-world data sets demonstrates that it is in general reasonable to suppose that the volatility is a quadratic function in stock markets.
The empirical distribution of stock returns typically exhibits skewness and excess kurtosis, which could be induced by various macroeconomic shocks, such as the unemployment announcement, the Gulf war, and the oil crisis. Unfortunately, the standard second-order diffusion framework (3) is not tailored to capture these stylized facts. A more appropriate specification is to modify the aforementioned standard second-order diffusion model to allow for discontinuity. This is easily obtained by combining the general diffusion model with a jump factor. The popularity of the jump diffusion models stems from at least two facts. First, as distinguished from pure diffusion processes, the jump processes can affect and match high levels of kurtosis and skewness. Second, they are economically attractive because they admit that stock prices change by sudden jumps in a short time, which is a reasonable assumption for an efficient stock market. Recently, jump noises have been also applied in population models to describe the abrupt changes of population sizes (see [9, 10], etc.). For example, Zhou et al. [9] introduced a two-population mortality model with transitory jump effects and applied it to pricing catastrophic mortality securitizations.
In view of the abovementioned facts, we extend the work of Nicolau [1, 5] and then concentrate on a new second-order jump diffusion model:(4)dYt=Xtdt,dXt=α0+α1Xtdt+β+τXt-θ2dWt+Jtdπtλ,where the arrival of jumps Jt is governed by the continuous-time Poisson process πt with frequency parameter λ, which denotes the average number of jumps per year. The jump size may be a constant or may be drawn from a probability distribution. The diffusion and Poisson process are independent of each other, and each of them is independent of jump Jt as well. Consequently, the stock return is the sum of three components. The component (α0+α1Xt)dt represents the instantaneous expected return on the stock. The β+τ(Xt-θ)2dWt part describes the instantaneous variance of the stock return due to the arrival of “normal” information, and the Jtdπt(λ) part describes the total instantaneous stock return owing to the arrival of “abnormal” information. Next, we develop parameter estimation methodology to estimate the coefficients in the drift, diffusion, and jump factor.
2. Model Estimation
For model (4), the integrated process Yt is generally observable at the time points {ti,i=0,1,⋯,n+1}, while Xt is a nonobservable process. In fact, for the fixed sampling interval Δ=ti-ti-1, the exact distribution of {Yti}i=0n+1 is generally not explicit. An exception is the case where Xt follows an Ornstein-Uhlenbeck process [11]. In practical applications, the time points are equally spaced. For example, when the time unit is a year, weekly data are sampled at ti=t0+i/52(i=0,1,⋯,n+1) for a given initial time point t0. Based on Yt=Y0+∫0tXsds and the discrete-time observations {Yti}i=0n+1, we have(5)Yti-Yti-1=∫0tiXsds-∫0ti-1Xsds=∫ti-1tiXsds,i=1,2,⋯,n.Then Xti can be approximated by(6)X¯ti=Yti-Yti-1Δ,i=1,2,⋯,n.Naturally, the smaller the time span Δ is, the closer X¯ti is to Xti. In fact, stock prices are usually observed daily or higher frequency.
Let ΔUti=X¯ti+1-X¯ti and Vti=Wti+1-Wti. According to the independent increment property of the standard Brownian motion, Vti(i=1,2,⋯,n) are independently and normally distributed with mean zero and variance Δ. Therefore, we can write Vti as Vti=Δεti(i=1,2,⋯,n), where {εti}i=1n are independently and identically distributed as the standard normal distribution. Then the Euler discretization of model (4) can be expressed as(7)ΔUti=X¯ti+1-X¯ti≈α0+α1X¯tiΔ+β+τX¯ti-θ2Δεti+JtiμJ,σJΔπtiλ,i=1,2,⋯,n,
where Jti is normally distributed with mean μJ and variance σJ2. Δπti(λ) is the discrete-time Poisson increment. We set the mean jump size equal to zero, μJ = 0, guaranteeing a symmetric stock return distribution according to [12, 13]. It is well recognized that discretization of continuous-time diffusion models for estimation does introduce an estimation bias, but this is relatively small (see [14]). In this case the discrete-time approach (7) allows us to estimate the proposed model where the jump is normally distributed. In this model, the density functions, a mixture of Poisson−Gaussian distribution is generally used to define the jump diffusion model. Here, we employ a Bernoulli approximation, first introduced in [8], to approximates Poisson−Gaussian distribution underlying the discretized (7). We assume that in each time interval either only one jump occurs or no jump occurs. No other information arrivals over this period of time are allowed. This is tenable for short frequency data, e.g., daily stock returns, and may be debatable for data at higher frequencies. As Ball and Torous demonstrate, it provides an approximation procedure which is highly tractable, stable, and convergent [8]. Hence, the discretized (7) can be rewritten as(8)X¯ti+1-X¯ti≈α0+α1X¯tiΔ+β+τX¯ti-θ2Δiεti+JtiIti,i=1,2,⋯,n.where Iti=1, if there is a jump with probability q=λΔ+O(Δ) and 0 otherwise with probability 1-q. All other notations have been defined previously. Since the limit of the Bernoulli process is governed by a Poisson distribution (see the Appendix for some technique details), we can approximate the likelihood function for the Poisson−Gaussian model using a Bernoulli mixture of the normal distributions. Let Φ=(α0, α1, β, τ, θ, σJ, q). Then, the transition probabilities for the stock returns following a Poisson−Gaussian process are written as f(X¯t+Δ∣X¯t;Φ):(9)fX¯t+Δ∣X¯t;Φ=qexp-X¯t+Δ-X¯t-α0+α1X¯tΔ22β+τX¯t-θ2Δ+σJ2×12πβ+τX¯t-θ2Δ+σJ2+1-qexp-X¯t+Δ-X¯t-α0+α1X¯tΔ22β+τX¯t-θ2Δ×12πβ+τX¯t-θ2Δ,which approximates the true Poisson−Gaussian density with a mixture of normal distributions. Then Φ can be estimated via maximizing the profile pseudo likelihood function(10)lnLΦ=∑i=1nlnfX¯ti+1∣X¯ti;Φ,which yields estimators of Φ. The maximization problem can be carried out in many scientific computing packages. Note that the likelihood function (10) is complex nonlinear function based on transition probability density; hence the analytical formula for the estimators could not derived and the large sample property of estimators becomes challenging. We will not discuss this aspect any further. For more details, we refer the reader to Xu and Wang [15] or Zheng and Lin [16]. The numerical technique, sequential quadratic programming algorithm (see [17, 18]) is considered to solve this maximization problem. In our implementation, the likelihood function (10) is numerically maximized with the “fmincon” routine embedded in the “Optimization Toolbox” of MATLAB using the sequential quadratic programming algorithm. Furthermore, in [19], maximum likelihood estimation allows the construction of approximate confidence intervals for the parameters of interest and these confidence intervals are asymptotically optimal.
Remark 1.
The sequential quadratic programming algorithm has been proved to be an excellent nonlinear programming method for solving constrained optimization problems in a variety of statistical models, such as linear models with longitudinal data under inequality restrictions [15], semiparametric regression models with censored data [16], autologistic models and exponential family models for dependent data [20], Cox-Ingersoll-Ross model for the interest rate [21], integer-valued GARCH models [22], and Gaussian stochastic process (GASP) models [23]. Our experience shows that the algorithm performs very well for moderate sample sizes. To gain more confidence in the estimates, we also try “Global Optimization Toolbox” of MATLAB based on the “fmincon” routine. We find that choosing different initial values yields identical results.
Furthermore, we also provide a formal statistical test for the presence of jumps in the stock and stock index returns. Since a pure diffusion model is nested within a combined diffusion and jump model, a likelihood ratio test can be used to test the null hypothesis H0: stock and stock index returns are normally distributed. The corresponding likelihood ratio test statistic is(11)Ω=2lnLX¯;Φ∗-lnLX¯;Φ0,where Φ∗ denotes the maximum likelihood estimates under jump diffusion model and Φ0 is the maximum likelihood estimates corresponding to the situation when no jump structure occurs (i.e., λ=0). Under the null hypothesis, the stock returns are consistent with a log-normal diffusion process without jump factor and Ω is asymptotically distributed χ2 with d-5 degrees of freedom, where d denotes the number of parameters to be estimated. More details about asymptotically distribution of the likelihood ratio test statistic can be found in ([24, 25]).
3. Simulation Study
In this section, we conduct a simulation study for second-order jump diffusion model (4) aimed at examining the performance of the estimation methodology. We consider a special case of (4):(12)dYt=Xtdt,dXt=0.2-20Xtdt+0.1+10Xt-0.052dWt+Jtdπtλ,where Jt~N(0,0.12), probability of jump q=0.5, and initial values X0=0.01 and Y0=0. The values of the parameters except jump term are given in [1]. We generate 100 replications (paths) of Yt and Xt according to the above design with each replication consisting of n (n=1260, 2520, 3780, 5040) daily observations and a time step between two consecutive observation equal to Δ=1/252. Then the coefficients α0 and α1 in the drift functions, β, τ, and θ, in the diffusion function, and σJ and q, in the jump factor, can be estimated by the proposed estimation methodology. The empirical sample mean (MEAN), empirical standard deviation (SD), and root mean square error (RMSE) of the coefficient estimators α^0, α^1, β^, τ^, θ^, σ^J, and q^ are reported (Table 1). We can see the mean of the estimators becomes more accurate; meanwhile the empirical standard deviation and root mean square error become smaller as the observations n increases from 1260 to 5040.
Parameters estimate, standard error, and root of mean square error for model (12), where α0=0.2, α1=−20, β=0.1, τ=10, θ=0.05, σJ=0.1, and q=0.5.
α^0
α^1
β^
τ^
θ^
σ^J
q^
n=1260
MEAN
0.2746
−20.7403
0.0960
10.2741
0.0538
0.1086
0.4663
SD
0.2708
2.6095
0.0264
1.4731
0.0167
0.0069
0.0473
RMSE
0.2856
2.5985
0.0261
1.4763
0.0166
0.0067
0.0471
n=2520
MEAN
0.2372
−20.3026
0.1014
9.8620
0.0496
0.1024
0.4986
SD
0.2250
2.0568
0.0132
1.2086
0.0082
0.0028
0.0222
RMSE
0.2269
2.1638
0.0132
1.2105
0.0082
0.0028
0.0221
n=3780
MEAN
0.2151
−20.1906
0.0994
10.0149
0.0498
0.1006
0.5040
SD
0.2017
1.4665
0.0114
0.8613
0.0069
0.0023
0.0204
RMSE
0.2041
1.4679
0.0114
0.8570
0.0068
0.0023
0.0203
n=5040
MEAN
0.2112
−20.1625
0.0992
9.9446
0.0504
0.1001
0.5026
SD
0.1502
1.4916
0.0093
0.7038
0.0062
0.0023
0.0181
RMSE
0.1498
1.5089
0.0092
0.7098
0.0061
0.0023
0.0184
4. Empirical Analysis4.1. Data Source
In this section, we analyze stock market data based on jump diffusion model (4). The data used in this study consists of daily closing stock prices of US companies (Alcoa, American Express, Boeing, Coca Cola, Disney, General Electric, Microsoft, Johnson & Johnson and JP Morgan) between January 25, 2000 and January 25, 2005 and daily stock market indexes between January 1998 and September 2008 from North America, Asia, and Europe: USA (S&P500, NASDAQ, DJIA), UK (FTSE100), Germany (DAX), France (CAC40), Hong Kong (HANG SENG), Australia (SPASX200), Switzerland (SMI), and Taiwan (TWII). The datasets are available at the Yahoo Finance website (https://finance.yahoo.com/). Following the conventional practice in stock market research, we take the logarithmic transformation of the selected data.
4.2. Descriptive Statistics
Microsoft Corporation’s daily stock prices from January 25, 2000 to January 25, 2005 are plotted in Figure 1. The most striking feature in the figure is the existence of conspicuous spikes in the stock prices and returns. The descriptive statistics are reported (Table 2). The change in stock prices is clear, positively skewed, and has excess kurtosis indicating that leptokurtosis is undeniable, which also predicates the use of a jump diffusion model. The Jarque-Bera test of both dataset rejects the hypothesis of normality. Other stock prices and indexes have similar statistical characterization. We do not enumerate them here.
Summary statistics.
Statistic
PanelA:Yt
PanelB:ΔYt
Mean
3.7957
0.0011
Standard deviation
0.4370
0.0314
Skewness
−0.1045
7.4655
Excess kurtosis
−1.4564
155.9618
Minimum
3.1268
−0.1788
Maximum
4.7173
0.6602
Jarque-Bera
113.2∗∗∗
1283600∗∗∗
∗∗∗ Indicates significance at 1% level.
Note: The statistics reported are for log-price (Yt) and change (ΔYt). The Jarque-Bera statistic tests for normality distribution. If the data are normally distributed, the skewness is 0, whereas the kurtosis is equal to 3.
Daily closing stock price of Microsoft Corporation from January 25, 2000 to January 25, 2005.
Daily closing stock prices of Microsoft Corporation
Change in daily closing stock price of Microsoft Corporation
4.3. Parameter Estimation and Testing Results
We shall summarize and compare the parameter estimates between second-order diffusion model (3) and second-order jump diffusion model (4) for the stocks and stock indexes (Tables 3 and 4). In addition to the 7 parameters to be estimated (α0 and α1 in instantaneous mean and β, τ, and θ in the volatility component, the variance σJ2 and probability q of jump factor), the standard error, the log-likelihood value (ln L), and the likelihood ratio test statistic Ω are also reported. For each sampled stock and stock index, the first row and the second row present maximum likelihood estimates and corresponding standard errors of model (4), respectively. The third row and the fourth row present maximum likelihood estimates and corresponding standard errors of model (3). Note that the estimates of standard errors of Φ∗ and Φ0 for the maximum likelihood estimation are obtained from the main diagonal of the inverse of the Hessian matrix, respectively.
Parameter estimates for stocks.
Stock
N
α0
α1
β
τ
θ
σJ
q
ln L
Ω
Alcoa
1257
0.1713
−257.9342
0.6380
23.9529
-0.0156
0.0972
0.1135
1718.1
56.18∗∗∗
(0.3494)
(7.7857)
(0.0588)
(3.9444)
(0.0167)
(0.0164)
(0.0540)
0.0504
−258.9853
0.9154
23.2125
−0.0056
1690.0
(0.3294)
(8.6151)
(0.0465)
(8.7870)
(0.0165)
American
1251
−0.2182
−256.5039
0.1810
75.6013
-0.0038
0.0646
0.4051
1873.8
87.21∗∗∗
Express
(0.1785)
(8.6131)
(0.0514)
(16.0989)
(0.0054)
(0.0051)
(0.1017)
−0.3313
−258.0656
0.5884
83.6826
0.00458
1830.2
( 0.4116)
(9.2299)
(0.0341)
(12.5241)
(0.0048)
Boeing
1257
−0.2355
−265.9633
0.4276
52.2001
-0.0069
0.0885
0.1011
1899.7
48.24∗∗∗
(0.4224)
(7.7640)
(0.0655)
(5.9056)
(0.0067)
(0.0257)
(0.0760)
0.0139
−263.8557
0.5745
72.6873
−0.0086
1875.6
(0.3656)
(8.8101)
(0.0303)
(12.8031)
(0.0056)
Coca Cola
1257
0.0434
−249.5004
0.1210
48.5000
-0.0062
0.0580
0.3086
2263.1
135.99∗∗∗
(0.2459)
(7.9359)
(0.0210)
(6.1083)
(0.0050)
(0.0048)
(0.0611)
−0.0280
−248.9476
0.3378
79.3847
−0.0103
2195.1
(0.1851)
(9.1881)
(0.0197)
(12.1064)
(0.0042)
Disney
1257
−0.2449
−260.6646
0.5405
57.6146
0.0107
0.1658
0.0312
1812.0
140.31∗∗∗
(0.9548)
(7.7072)
(0.0377)
(10.0298)
(0.0064)
(0.0326)
(0.0141)
−0.2363
−254.5632
0.7399
50.7316
0.0287
1741.8
(1.0434)
(7.8464)
(0.0664)
(20.1801)
(0.0091)
General
1257
0.0495
−257.2053
0.2935
47.9844
0.0058
0.0779
0.1955
1969.8
110.54∗∗∗
Electric
(0.1662)
(8.1046)
(0.0431)
(7.8468)
(0.0070)
(0.0092)
(0.0664)
−0.2494
−260.3366
0.5816
52.8035
0.0048
1914.5
( 2.0980)
(117.4492)
(0.1926)
(13.7811)
(0.0130)
Johnson
1257
−0.3607
−257.3816
0.2195
69.4946
0.0032
0.1207
0.0244
2368.4
106.95∗∗∗
&Johnson
(0.0753)
(7.2543)
(0.0176)
(4.0190)
(0.0041)
(0.0356)
(0.0161)
−0.4890
−255.5323
0.2457
112.3180
0.0110
2314.9
(0.0743)
(8.9071)
(0.0162)
(23.3878)
(0.0029)
JP Morgan
1256
0.3209
−257.6896
0.4855
42.1887
0.0162
0.1640
0.0892
1722.0
335.67∗∗∗
(0.2897)
(6.8796)
(0.0443)
(7.8642)
(0.0088)
(0.0180)
(0.0252)
0.1524
−260.0101
1.0547
42.8909
−0.0307
1554.2
(1.1554)
(21.8999)
(0.0803)
(92.4838)
(0.0503)
Microsoft
1257
0.4073
−267.0635
0.5217
52.5552
0.0041
0.3059
0.0370
1761.8
570.82∗∗∗
(0.3472)
(8.1481)
(0.0452)
(11.3793)
(0.0069)
(0.0491)
(0.0119)
0.6737
−267.9022
0.6690
264.2013
−0.0316
1476.4
(0.1681)
(11.8308)
(0.0409)
(31.7660)
(0.0036)
∗∗∗ Indicates significance at 1% level.
a Standard errors of maximum likelihood estimates are in parentheses.
Parameter estimates for stock indexes.
Stock index
N
α0
α1
β
τ
θ
σJ
q
ln L
Ω
CAC40
2798
0.2383
−257.6798
0.1330
54.8009
0.0116
0.0627
0.1645
5376.7
344.59∗∗∗
( 0.1155)
(4.9044)
(0.0103)
(5.0972)
(0.0029)
(0.0043)
(0.0285)
−0.0615
−257.9421
0.2687
82.9046
0.0093
5204.4
( 0.0978)
(6.3565)
(0.0104)
(7.5923)
(0.0024)
DAX
2785
0.2784
−255.8185
0.1555
57.6903
0.0127
0.0645
0.1652
5164.5
280.85∗∗∗
( 0.2202)
(24.7216)
(0.0167)
(17.3980)
(0.0031)
(0.0039)
(0.0300)
−0.0118
−252.6514
0.2919
86.8931
0.0124
5024.1
( 0.1394)
(6.1700)
(0.0110)
(9.5216)
(0.0023)
DJIA
2703
0.1067
−210.7317
0.0590
47.0208
0.0082
0.0430
0.1481
6373.6
317.71∗∗∗
( 0.0605)
(5.4284)
(0.0048)
(8.5147)
(0.0021)
(0.0035)
(0.0263)
0.0699
−211.6276
0.1050
66.2194
0.0168
6214.8
( 0.1136)
(5.6852)
(0.0041)
(8.1033)
(0.0022)
FTSE100
2713
−0.1870
−260.5673
0.0649
61.1198
−0.0018
0.0379
0.3053
5841.7
222.60∗∗∗
(0.3935)
(9.5712)
(0.0387)
(31.0472)
(0.0030)
(0.0084)
(0.2009)
−0.1749
−264.5221
0.1641
79.7363
−0.0028
5730.4
( 0.4693)
(18.1376)
(0.0456)
(65.8072)
(0.0018)
HANGSENG
2702
0.1166
−245.2607
0.1369
41.1688
0.0186
0.0655
0.2412
4863.4
366.22∗∗∗
( 1.5002)
(7.6386)
(0.1075)
(8.8111)
(0.0167)
(0.0125)
(0.1701)
−0.0341
−238.4197
0.3550
81.9699
0.0035
4683.4
(0.1558)
(6.8947)
(0.0132)
(12.5799)
(0.0028)
NASDAQ
2767
0.4242
−253.6221
0.1810
66.7837
0.0047
0.0658
0.2204
4740.6
238.63∗∗∗
(0.1197)
(6.3600)
(0.0143)
(9.0416)
(0.0033)
(0.0043)
(0.0312)
−0.3926
−255.9448
0.3824
91.8324
0.0070
4621.3
( 0.0814)
(7.9904)
(0.0177)
(19.2416)
(0.0028)
SP500
2804
0.1679
−261.8900
0.0441
24.8916
0.0235
0.0360
0.4098
6122.3
278.23∗∗∗
(0.1935)
(5.3790)
(0.0108)
(4.8198)
(0.0050)
(0.0022)
(0.0706)
−0.0194
−261.4696
0.1745
43.5322
0.0093
5983.2
(0.1217)
(5.5881)
(0.0072)
(4.1993)
(0.0031)
SPASX200
2870
0.2102
−257.0073
0.0570
71.0473
0.0061
0.0514
0.0785
6951.4
389.95∗∗∗
(0.1332)
(6.5243)
(0.0035)
(6.9430)
(0.0014)
(0.0052)
(0.0180)
0.0966
−253.5373
0.0908
112.8947
0.0093
6756.5
(0.0567)
(6.6217)
(0.0034)
(8.9553)
(0.0011)
SMI
2704
0.1878
−255.8562
0.0000
0.1737
0.8342
0.0526
0.1877
5684.6
39.95∗∗∗
(0.1344)
(4.5107)
(0.0386)
(0.0315)
(0.1503)
(0.0037)
(0.0266)
0.0394
−247.5534
0.1492
113.4965
0.0079
5664.6
(0.1479)
(6.3341)
(0.0062)
(12.8102)
(0.0014)
TWII
2639
−0.0621
−240.4190
0.1094
21.5185
0.0008
0.0508
0.4276
4836.2
238.07∗∗∗
(0.1073)
(5.0817)
(0.0153)
(3.9170)
(0.0065)
(0.0023)
(0.0457)
−0.0273
−245.3496
0.3676
31.7291
0.0037
4717.1
(0.4595)
(5.6751)
(0.0111)
(2.9732)
(0.0056)
∗∗∗ Indicates significance at 1% level.
a Standard errors of maximum likelihood estimates are in parentheses.
It can be seen that the standard errors of parameter estimates for jump diffusion model (4) are extremely small, confirming the preferability of the estimation methodology, and are mostly smaller than that of pure diffusion model (3) (Tables 3 and 4). For all the stocks and stock indexes, the corresponding log-likelihood values, ln L, of model (4) are also higher than that of model (3). Furthermore, the likelihood ratio test demonstrates the presence of jumps at the 1 percent significance level for each stock and stock index. These empirical results show that the second-order jump diffusion model (4) is more suitable for modeling the stock market returns from North America, Asia, and Europe than second-order pure diffusion model (3).
5. Conclusion and Discussion
In this paper, a new second-order jump diffusion model is proposed to investigate the dynamic characteristic of stock market returns from North America, Asia, and Europe. The maximum likelihood approach is employed to estimate parameters of the model. The simulation study demonstrates that the estimation approach works satisfactorily. By analyzing stock market data from North America, Asia, and Europe, we find that the second-order jump diffusion model can capture high levels of kurtosis and skewness of stock return distributions, while the second-order diffusion model that ignores the jump factor is misspecified. The likelihood ratio test also confirms the statistically significant presence of jump factor in the selected stock market returns.
It should be pointed out that there are some topics that can be further considered. This paper mainly focuses on the Poisson jump. In fact, there are some random perturbations, such as the telephone noise which can be further studied [26]. Besides, stochastic differential equations have been recently applied into the SIS and SIRS epidemic models with Markov-switching [27–29], the budworm growth model with regime switching [30], and stochastic regime switching predator-prey model [31, 32]. It is very promising to incorporate the jump factors into these stochastic models.
AppendixTechnical Details
The Bernoulli approximation here can be achieved as follows. Let the random variable N denote the number of events that occur in a time interval of length t. Define Δ=t/n for an arbitrary integer n and subdivide the interval (0,t) into n equal subintervals each of length Δ. Let Yi denote the number of events that occur in subinterval i and by the stationary independent increment assumption,(A.1)N=∑i=1nYiis the sum of n independent identically distributed random variables such that(A.2)ProbYi=0=1-λΔ+OΔ,ProbYi=1=λΔ+OΔ,fori=1,2,⋯,nProbYi>1=OΔ.For large n, each Yi has approximately the Bernoulli distribution with parameter λΔ. Consequently, being the sum of n independent identically distributed Bernoulli random variables, N has the binomial distribution. For k occurrences,(A.3)ProbN=k=NCkλtnk1-λtnn-k,k=0,1,2,⋯,nthen,limn→∞ProbN=k=e-λtλtkk!.k=0,1,2,⋯,n
This is a standard construction of the Poisson process. When Δ is very small, it follows then that we can satisfactorily approximate N by the Bernoulli variate Y defined by (A.4)PY=0=1-λΔ,PY=1=λΔ.The distinguishing characteristic of the Bernoulli approximation is that over a short period of time, Δ, either no piece of information impacts the interest rate or one relevant piece of information occurs with probability λΔ. No other information arrivals over the period of time are allowed.
Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.
Conflicts of Interest
The authors declare that there are no conflicts of interest regarding the publication of this paper.
Acknowledgments
The research work is supported by the National Natural Science Foundation of China under Grants nos. 11601409 and 11701286, the Natural Science Foundation of Jiangsu Province of China under Grant no. BK20171073, the University Natural Science Foundation of Education Department of Jiangsu Province of China under Grant no. 17KJB110006, and the University Philosophy and Social Science Foundation of Education Department of Jiangsu Province of China under Grants no. 2017SJB0350.
NicolauJ.Nonparametric estimation of second-order stochastic differential equations200723588089810.1017/S0266466607070375MR2395838Zbl1237.621052-s2.0-34547939447WangH.LinZ.Local linear estimation of second-order diffusion models201140339440710.1080/03610920903391345MR2765836Zbl1208.62131WangY.ZhangL.TangM.Re-weighted functional estimation of second-order diffusion processes20127581129115110.1007/s00184-011-0372-6MR2989323Zbl1254.62090HanifM.Non parametric estimation of second-order diffusion equation by using asymmetric kernels20154491896191010.1080/03610926.2012.756915MR3348312Zbl1344.60056NicolauJ.Modeling financial time series through second-order stochastic differential equations200878162700270410.1016/j.spl.2008.03.024MR2542466DetteH.PodolskijM.Testing the parametric form of the volatility in continuous time diffusion models—a stochastic process approach20081431567310.1016/j.jeconom.2007.08.002MR2384433FanJ.JiangJ.ZhangC.ZhouZ.Time-dependent diffusion models for term structure dynamics2003134965992MR2026058Zbl1065.621772-s2.0-0348229233YanT.MeiC.A test for a parametric form of the volatility in second-order diffusion models20173241583159610.1007/s00180-016-0685-zMR3723754Zbl068355182-s2.0-84988575988ZhouR.LiJ. S.-H.TanK. S.Pricing standardized mortality securitizations: A two-population model with transitory jump effects20138037337742-s2.0-8488323028810.1111/j.1539-6975.2013.12015.xLiuY.LiJ. S.The age pattern of transitory mortality jumps and its impact on the pricing of catastrophic mortality bonds20156413515010.1016/j.insmatheco.2015.05.005MR3394640GloterA.Parameter estimation for a discrete sampling of an integrated Ornstein-Uhlenbeck process200135322524310.1080/02331880108802733MR1925514BallC. A.TorousW. N.A Simplified Jump Process for Common Stock Returns198318153652-s2.0-8497184805310.2307/2330804BeckersS.A note on estimating the parameters of the diffusion-jump model of stock returns19811611271402-s2.0-8497439997310.2307/2330670BergstromA. R.The history of continuous-time econometric models19884336538310.1017/S0266466600013359MR9851522-s2.0-84974313172XuJ.WangJ.Maximum likelihood estimation of linear models for longitudinal data with inequality constraints2008376-793194610.1080/03610920701762788MR2408287Zbl1135.62048ZengD.LinD. Y.Maximum likelihood estimation in semiparametric regression models with censored data200769450756410.1111/j.1369-7412.2007.00606.xMR2370068ColemanT.BranchM. A.GraceA.19993rdNatickGilatA.20042ndHobokenWileyWilksS. S.Shortest Average Confidence Intervals from Large Samples19389316617510.1214/aoms/1177732308GeyerC. J.ThompsonE. A.Constrained Monte Carlo maximum likelihood for dependent data1992543657699MR1185217KladvkoK.2007Technical Computing Prague10.1007/s10614-010-9208-0Zbl1231.91485ZhuF.A negative binomial integer-valued GARCH model2011321546710.1111/j.1467-9892.2010.00684.xMR2790672RanjanP.BinghamD.MichailidisG.Sequential experiment design for contour estimation from complex computer codes200850452754110.1198/004017008000000541MR2655651GouriérouxC.HollyA.MonfortA.Likelihood ratio test, Wald test, and Kuhn-Tucker test in linear models with inequality constraints on the regression parameters1982501638010.2307/1912529MR640166DickeyD. A.FullerW. A.Likelihood ratio statistics for autoregressive time series with a unit root19814941057107210.2307/1912517MR625773Zbl0471.62090StuckB. W.KleinerB.A statistical analysis of telephone noise19745371263132010.1002/j.1538-7305.1974.tb02791.x2-s2.0-0016100974GrayA.GreenhalghD.MaoX.PanJ.The {SIS} epidemic model with Markovian switching2012394249651610.1016/j.jmaa.2012.05.029MR2927473Zbl1271.920302-s2.0-84862005170LahrouzA.SettatiA.Asymptotic properties of switching diffusion epidemic model with varying population size201321924111341114810.1016/j.amc.2013.05.019MR3073267Zbl1304.921212-s2.0-84879371158LahrouzA.OmariL.Extinction and stationary distribution of a stochastic SIRS epidemic model with non-linear incidence201383496096810.1016/j.spl.2012.12.021MR3041364KempW. P.DennisB.BeckwithR. C.Stochastic Phenology Model for the Western Spruce Budworm (Lepidoptera: Tortricidae)198615354755410.1093/ee/15.3.547ZuL.JiangD.O'ReganD.Conditions for persistence and ergodicity of a stochastic Lotka-Volterra predator-prey model with regime switching2015291-311110.1016/j.cnsns.2015.04.008MR33680112-s2.0-84937430988LiuM.WangK.Global stability of a nonlinear stochastic predator-prey system with Beddington-DeAngelis functional response20111631114112110.1016/j.cnsns.2010.06.015MR2736618Zbl1221.341522-s2.0-77957351473