betaFeedback

WorksEconomics, Econometrics and Finance

A Semiparametric Real-Time GARCH Model with Nonstationary Variance

Abstract

We propose a semiparametric real-time GARCH (SRT-GARCH) model that combines a smooth deterministic long-run volatility component with a stationary short-run process driven by contemporaneous return innovations. The model accommodates gradual changes in unconditional variance while allowing short-run volatility to respond directly to current information rather than relying solely on lagged observations. We derive the variance normalization, conditional moments, and closed-form conditional density of the model. For estimation, we develop a two-step procedure that combines kernel smoothing with conditional Gaussian maximum likelihood. Under Gaussian innovations and mild regularity conditions, we establish consistency and asymptotic normality of the short-run parameter estimator. The analysis exploits geometric moment contraction to control serial dependence and shows that estimating the long-run component adds a positive semidefinite rank-one adjustment to the oracle covariance matrix obtained when the long-run component is known. The resulting asymptotic covariance matrix admits a plug-in estimator that requires neither lag truncation nor additional tuning parameters. Monte Carlo simulations show good finite-sample performance of the estimator and the associated inference procedure. Applications to FTSE~100 returns and U.S. natural gas futures show that allowing for a time-varying long-run component improves forecasting relative to the parametric RT-GARCH benchmark, while the additional real-time feedback mechanism provides more pronounced gains in the natural gas application than in the FTSE application. The proposed model also delivers competitive one-step-ahead Value-at-Risk forecasts.

JEL classification: C12; C14; C58

Introduction

Financial market volatility fluctuates over multiple horizons. Daily return innovations contain immediate information about current volatility, while structural changes in macroeconomic and market conditions can generate slow-moving variation in unconditional variance Hong et al., [21]. Capturing both sources of variation is important for volatility measurement, forecasting, and risk management. In the seminal generalized autoregressive conditional heteroskedasticity (GARCH) frameworks of Engle [12] and Bollerslev [4], short-run volatility clustering is modeled through a conditional variance driven by lagged returns and past volatility. As emphasized by Politis [31], however, conditioning exclusively on past information may overlook the contemporaneous signal contained in the current return. A related insight underlies the realized-volatility literature, which demonstrates the value of within-period information for tracking rapid changes in volatility [3, 19].

To incorporate contemporaneous information directly into daily return models, Smetanina [33] introduced the real-time GARCH (RT-GARCH) model, in which the current innovation enters both the return equation and the volatility process. Unlike stochastic volatility models, which introduce a separate latent volatility shock, RT-GARCH allows the volatility scale to respond immediately to the return innovation while retaining a tractable closed-form likelihood. Smetanina and Wu [34] subsequently established the large-sample theory for maximum likelihood estimation of the stationary RT-GARCH model. More recent extensions have incorporated joint nowcasting and forecasting, low-frequency macroeconomic covariates, heavy-tailed conditional distributions, realized volatility measures, and intraday range information; see, for example, Ding [11]; Wu et al. (2023); Kruse-Becher and Liu [25]; Wu and Cai (2025); Xu and Wu [42]. Despite their flexibility in describing short-run volatility dynamics, these models maintain a constant unconditional variance and may therefore be misspecified over long spans of financial data.

A complementary literature addresses nonstationarity in unconditional volatility. Mikosch and Stărică [28] show that unmodeled changes in baseline volatility can generate spurious integrated-GARCH behavior and distort estimates of volatility persistence when constant unconditional variance is imposed. This observation has motivated models with smooth deterministic trends or structural breaks [35, 13, 6, 36], as well as locally stationary ARCH processes with time-varying parameters [9]. Within this literature, multiplicative component models provide an especially convenient framework by decomposing returns into a smooth deterministic scale and a stationary GARCH process [14, 18, 23, 8]. Related work studies adaptive estimation, specification testing, outlier robustness, and simultaneous inference in time-varying volatility models [24, 5, 1, 2]. Existing multiplicative GARCH specifications, however, continue to rely on lagged return shocks and therefore do not exploit contemporaneous return information in the short-run volatility recursion.

In this paper, we bridge these two strands of the literature by proposing a semiparametric real-time GARCH (SRT-GARCH) model. Returns are decomposed into a smooth deterministic scale function τ(t/T)\tau(t/T) and a stationary RT-GARCH process. A second-moment normalization of the short-run component identifies τ(t/T)\tau(t/T) directly as the time-varying unconditional variance of returns. The proposed specification nests the semiparametric GARCH (S-GARCH) model of Jiang et al. [23] when the contemporaneous feedback parameter is zero and a variance-targeted version of the RT-GARCH model of Smetanina [33] when the long-run scale is constant. We characterize the conditional moments and conditional density of the model. The long-run component determines the overall volatility, whereas the real-time feedback parameter introduces additional short-run distributional flexibility, including time-dependent conditional kurtosis and, under Gaussian innovations, exponential-type conditional heavy tails.

Combining an unknown long-run volatility component with contemporaneous short-run feedback creates nontrivial challenges for estimation and inference. The asymptotic theory cannot be obtained by simply combining existing results for S-GARCH and RT-GARCH models. First, because short-run volatility depends contemporaneously on the current innovation, recovering the latent volatility scale from observed returns requires inversion of a nonlinear quadratic equation at each time point. The resulting observed-data volatility recursion and its parameter derivatives therefore follow coupled nonlinear dynamics rather than the linear recursions familiar from conventional GARCH models. To control temporal dependence without imposing high-level α\alpha- or β\beta-mixing assumptions, we exploit a random-coefficient representation of the true short-run state to establish the geometric moment contraction property [37, 32]. The resulting coupling bounds are then extended to the nonlinear observed-data volatility process and its higher-order parameter derivatives.

Second, because the deterministic variance function τ(⋅)\tau(\cdot) is unknown, we estimate the model in two steps. We first estimate τ(⋅)\tau(\cdot) nonparametrically from squared returns using a boundary-reflected Nadaraya–Watson kernel estimator τ^t\widehat{\tau}_t for τt=τ(t/T)\tau_t=\tau(t/T). We then estimate the short-run parameters by Gaussian conditional maximum likelihood based on the standardized returns u^t=rt/τ^t\widehat{u}_t=r_t/\sqrt{\widehat{\tau}_t}, where {rt}\{r_t\} is the observed return process, and utu_t denotes the corresponding oracle RT-GARCH process. Since u^t2\widehat{u}_t^2 enters both the contemporaneous quadratic root and all subsequent volatility states, the first-step estimation error propagates nonlinearly through the likelihood recursion and has a non-vanishing first-order effect on the limiting distribution of the second-step estimator. By projecting this infinite-horizon perturbation onto the first-step innovations, we establish consistency and asymptotic normality of the two-step estimator. Under the variance normalization, we further show that the second-step score is contemporaneously orthogonal to the martingale innovation driving the first-step error and that the infinite-series sensitivity vector admits a closed-form contemporaneous representation. These two structural properties reduce the contribution of first-step estimation to a positive semidefinite rank-one adjustment of the oracle covariance matrix and yield a tuning-free plug-in asymptotic covariance estimator that requires neither lag truncation nor additional tuning parameters.

We also derive closed-form expressions for multi-step-ahead volatility forecasts and one-step-ahead Value-at-Risk (VaR) forecasts. Monte Carlo experiments indicate that the finite-sample behavior of the estimator and the associated inference procedure is broadly consistent with the asymptotic theory. We apply the model to daily FTSE 100 returns and U.S. natural gas futures. In both applications, covariance stationarity is rejected, providing evidence consistent with a time-varying long-run volatility component. The forecasting results show that incorporating a time-varying long-run component improves substantially on the RT-GARCH benchmark, while the additional real-time feedback mechanism delivers more pronounced gains in the natural gas application than in the FTSE application. The VaR results are also broadly supportive of the proposed specification.

The remainder of the paper is organized as follows. § 2 introduces the SRT-GARCH model and characterizes its moment and distributional properties. § 3 develops the two-step estimation procedure, establishes the large-sample theory, and derives the plug-in covariance estimator. § 4 presents the volatility forecasting framework and its multi-step extension. § 5 and § 6 report the Monte Carlo experiments and empirical applications, respectively. § 7 concludes. All technical proofs are provided in the Appendix.

The SRT-GARCH(1,1) Model

For an observed return series {rt}t=1T\{r_t\}_{t=1}^{T}, we consider the following semiparametric real-time GARCH (SRT-GARCH) model:

{rt=τtut,τt=τ(t/T),ut=ht εt,t=1,…,T,ht=ω+αut−12+βht−1+φεt2,(1)\begin{cases} r_t = \sqrt{\tau_t}u_t, & \tau_t=\tau(t/T),\\ u_t = \sqrt{h_t}\,\varepsilon_t, & t=1,\ldots,T,\\ h_t = \omega+\alpha u_{t-1}^{2}+\beta h_{t-1}+\varphi\varepsilon_t^{2}, \tag*{(1)} \end{cases}

where τ:[0,1]→(0,∞)\tau:[0,1]\to(0,\infty) is an unknown, smoothly time-varying deterministic function.

In model (1), τt\tau_t captures slow-moving changes in the unconditional variance, while utu_t governs stationary short-run volatility dynamics. The innovations {ϵt}\{\epsilon_t\} are assumed to be independent and identically distributed (i.i.d.) with zero mean and unit variance. The short-run parameters satisfy ω>0\omega> 0 and α,β,φ≥0\alpha,\beta,\varphi\ge0. Here, the coefficient φ\varphi measures the immediate feedback of the contemporaneous return shock on the short-run scale: setting φ=0\varphi= 0 recovers the semiparametric GARCH (S-GARCH) model of Jiang et al. [23], whereas restricting τt\tau_t to be constant yields a variance-normalized version of the real-time GARCH (RT-GARCH) model of Smetanina [33].

Unlike stochastic volatility models, model (1) does not introduce a separate latent volatility innovation. Because the same shock ϵt\epsilon_t drives both the return equation and the short-run volatility update, hth_t and ϵt\epsilon_t can be uniquely recovered from the observed return rtr_t given past information via a nonlinear transformation. This property yields a closed-form conditional density and makes conditional maximum likelihood estimation feasible.

Remark 1. Let Ft=σ(ϵs:s≤t)\mathcal{F}_t = \sigma(\epsilon_s : s \le t) denote the σ\sigma-field generated by the innovations up to time tt. Because the contemporaneous innovation ϵt\epsilon_t enters the recursion for hth_t, the quantity hth_t is not measurable with respect to the usual past information set Ft−1\mathcal{F}_{t-1} as in the standard GARCH models. It should therefore be interpreted as a real-time short-run scale rather than as the conventional one-step-ahead conditional variance.

We begin with the regularity conditions that ensure the short-run process is well defined and the long-run component is identifiable. Let θ=(α,β,φ)T\theta= (\alpha,\beta,\varphi)^\mathsf{T} and denote the true parameter as θ0=(α0,β0,φ0)T\theta_0 = (\alpha_0,\beta_0,\varphi_0)^\mathsf{T}.

Assumption 1. (i) The parameter space Θ\Theta is a compact subset of

{(α,β,φ)T∈[0,∞)3:α+β≤ρ<1, ω(θ)>0}\left\{(\alpha,\beta,\varphi)^\mathsf{T} \in[0,\infty)^3 : \alpha+ \beta\le\rho< 1,\ \omega(\theta) > 0\right\}

where ρ∈(0,1)\rho\in(0,1) is a fixed constant, and

ω(θ)=1−α−β−βφ+φ(β−1)μ4,μ4=E(ϵt4).(2)\omega(\theta) = 1 - \alpha- \beta- \beta\varphi+ \varphi(\beta- 1)\mu_4,\qquad\mu_4 = \mathbb{E}(\epsilon_t^4). \tag*{(2)}

(ii) The true parameter satisfies θ0∈Θ\theta_0 \in\Theta.

(iii) The innovations {ϵt}\{\epsilon_t\} are independently and identically distributed, symmetric about zero, and satisfy E(ϵt)=0\mathbb{E}(\epsilon_t) = 0, E(ϵt2)=1\mathbb{E}(\epsilon_t^2) = 1, and μ4=E(ϵt4)<∞\mu_4 = \mathbb{E}(\epsilon_t^4) < \infty.

Normalization and Second-Order Properties

Assumption 1(i) defines a compact parameter space. The nonnegativity restrictions together with ω(θ)>0\omega(\theta) > 0, ensure the positivity of the real-time scale hth_t. The persistence condition α+β≤ρ<1\alpha+\beta\le\rho< 1 guarantees that the short-run dynamic admits a strictly stationary and ergodic solution with finite second-order moments, as formalized in the following proposition.

Proposition 1. Under Assumption 1, the short-run recursion has a unique causal, strictly stationary, and ergodic solution satisfying E(ht)<∞\mathbb{E}(h_t) < \infty and E(ut2)<∞\mathbb{E}(u_t^2) < \infty.

To separately identify the nonparametric scale τt\tau_t and the short-run process utu_t in (1), Assumption 1 also helps normalize the unconditional variance of utu_t to one so that τt\tau_t can be interpreted as the unconditional variance.

Proposition 2. Under Assumption 1, we have E(ut2)=1\mathbb{E}(u_t^2) = 1, and E(ht)=1−φ(μ4−1)\mathbb{E}(h_t) = 1 - \varphi(\mu_4 - 1). Moreover, E(ut∣Ft−1)=0\mathbb{E}(u_t \mid\mathcal{F}_{t-1}) = 0 and E(utut−j)=0\mathbb{E}(u_tu_{t-j}) = 0 for every j≥1j \ge1, i.e. {ut}\{u_t\} is a martingale difference sequence with unit unconditional variance.

Unlike standard GARCH models, where E(ut2)=E(ht)\mathbb{E}(u_t^2) = \mathbb{E}(h_t), the normalization in (2) must account for the contemporaneous covariance between hth_t and ϵt2\epsilon_t^2. Writing mu=E(ut2)m_u = \mathbb{E}(u_t^2) and mh=E(ht)m_h = \mathbb{E}(h_t) and taking unconditional expectations in the short-run equations ht=ω+αut−12+βht−1+φϵt2h_t = \omega+ \alpha u_{t-1}^2 + \beta h_{t-1} + \varphi\epsilon_t^2 and ut2=(ω+αut−12+βht−1)ϵt2+φϵt4u_t^2 = (\omega+ \alpha u_{t-1}^2 + \beta h_{t-1})\epsilon_t^2 + \varphi\epsilon_t^4 yields that

mh=ω+αmu+βmh+φ,mu=ω+αmu+βmh+φμ4.\begin{aligned} m_h &= \omega+ \alpha m_u + \beta m_h + \varphi,\\ m_u &= \omega+ \alpha m_u + \beta m_h + \varphi\mu_4. \end{aligned}

Together with (2) and α+β<1\alpha+ \beta< 1, we obtain that mu=1m_u = 1 and mh=1−φ(μ4−1)m_h = 1 - \varphi(\mu_4 - 1). Consequently, E(rt2)=τt\mathbb{E}(r_t^2) = \tau_t, so the long-run component has the direct interpretation of unconditional return variance. When φ=0\varphi= 0, the normalization reduces to ω=1−α−β\omega= 1 - \alpha- \beta as in Francq et al. [15] and Jiang et al. [23]; under standard Gaussian innovations, it becomes ω=1−α−β−3φ+2βφ\omega= 1 - \alpha- \beta- 3\varphi+ 2\beta\varphi.

Conditional Moments

While the deterministic function τt\tau_t represents the overall level of return variance, the real-time feedback parameter φ0\varphi_0 can change the shape of the conditional return distribution. Following Smetanina [33], we characterize their different roles through the second and fourth conditional moments of rtr_t.

References

  1. [1]Ahlgren, N., A. Back, and T. Teräsvirta (2024). A new GARCH model with a deterministic time-varying intercept. arXiv preprint arXiv:2410.03239.arxiv.org/abs/2410.03239
  2. [2]Amado, C. (2025). Outlier robust specification of multiplicative time-varying volatility models. Computational Economics 66(5), 4107–4135.DOI
  3. [3]Andersen, T. G., T. Bollerslev, F. X. Diebold, and P. Labys (2003). Modeling and forecasting realized volatility. Econometrica 71(2), 579–625.
  4. [4]Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics 31(3), 307–327.
  5. [5]Campos-Martins, S. and G. Sucarrat (2024). Modeling nonstationary financial volatility with the R package tvgarch. Journal of Statistical Software 108(9), 1–38.
  6. [6]Chen, B. and Y. Hong (2016). Detecting for smooth structural changes in GARCH models. Econometric Theory 32(3), 740–791.DOI
  7. [7]Christoffersen, P. F. (1998). Evaluating interval forecasts. International Economic Review 39(4), 841–862.DOI
  8. [8]Chronopoulos, I., L. Giraitis, and G. Kapetanios (2022). Choosing between persistent and stationary volatility. The Annals of Statistics 50(6), 3466–3483.DOI
  9. [9]Dahlhaus, R. and S. S. Rao (2006). Statistical inference for time-varying ARCH processes. The Annals of Statistics 34(3), 1075–1114.DOI
  10. [10]Diebold, F. X. and R. S. Mariano (1995). Comparing predictive accuracy. Journal of Business & Economic Statistics 13(3), 253–263.
  11. [11]Ding, Y. D. (2023). A simple joint model for returns, volatility and volatility of volatility. Journal of Econometrics 232(2), 521–543.
  12. [12]Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica 50(4), 987–1007.
  13. [13]Engle, R. F. and J. G. Rangel (2008). The spline-GARCH model for low-frequency volatility and its global macroeconomic causes. The Review of Financial Studies 21(3), 1187–1222.
  14. [14]Feng, Y. (2004). Simultaneously modeling conditional heteroskedasticity and scale change. Econometric Theory 20(3), 563–596.
  15. [15]Francq, C., L. Horvath, and J.-M. Zakoïan (2011). Merits and drawbacks of variance targeting in garch models. Journal of Financial Econometrics 9(4), 619–656.DOI
  16. [16]Francq, C. and J.-M. Zakoïan (2009). Testing the nullity of garch coefficients: correction of the standard tests and relative efficiency comparisons. Journal of the American Statistical Association 104(485), 313–324.DOI
  17. [17]Francq, C. and J.-M. Zakoïan (2020). Virtual historical simulation for estimating the conditional VaR of large portfolios. Journal of Econometrics 217(2), 356–380.arxiv.org/abs/1909.04661
  18. [18]Hafner, C. M. and O. Linton (2010). Efficient estimation of a multivariate multiplicative volatility model. Journal of Econometrics 159(1), 55–73.DOI
  19. [19]Hansen, P. R., Z. Huang, and H. H. Shek (2012). Realized GARCH: a joint model for returns and realized measures of volatility. Journal of Applied Econometrics 27(6), 877–906.
  20. [20]Harvey, D., S. Leybourne, and P. Newbold (1997). Testing the equality of prediction mean squared errors. International Journal of Forecasting 13(2), 281–291.DOI
  21. [21]Hong, Y., X. Wang, and S. Wang (2017). Testing strict stationarity with applications to macroeconomic time series. International Economic Review 58(4), 1227–1277.DOI
  22. [22]Jiang, F., D. Li, W. K. Li, and K. Zhu (2023). Testing and modelling for the structural change in covariance matrix time series with multiplicative form. Statistica Sinica 33(2), 787–818.
  23. [23]Jiang, F., D. Li, and K. Zhu (2021). Adaptive inference for a semiparametric generalized autoregressive conditional heteroskedasticity model. Journal of Econometrics 224(2), 306–329.arxiv.org/abs/1907.04147
  24. [24]Karmakar, S., S. Richter, and W. B. Wu (2022). Simultaneous inference for time-varying models. Journal of Econometrics 227(2), 408–428.arxiv.org/abs/2011.13157
  25. [25]Kruse-Becher, R. and Y. Liu (2024). Improving financial volatility nowcasts. The European Journal of Finance 30(2), 101–126.
  26. [26]Liu, W. and W. B. Wu (2010). Asymptotics of spectral density estimates. Econometric Theory 26(4), 1218–1245.DOI
  27. [27]Liu, W., H. Xiao, and W. B. Wu (2013). Probability and moment inequalities under dependence. Statistica Sinica 23(3), 1257–1272.DOI
  28. [28]Mikosch, T. and C. Stărică (2004). Nonstationarities in financial time series, the long-range dependence, and the IGARCH effects. Review of Economics and Statistics 86(1), 378–390.DOI
  29. [29]Newey, W. K. and D. McFadden (1994). Large sample estimation and hypothesis testing. In Handbook of Econometrics, Volume 4, pp. 2111–2245. Elsevier.
  30. [30]Patton, A. J. (2011). Volatility forecast comparison using imperfect volatility proxies. Journal of Econometrics 160(1), 246–256.DOI
  31. [31]Politis, D. N. (2007). Model-free versus model-based volatility prediction. Journal of Financial Econometrics 5(3), 358–359.DOI
  32. [32]Shao, X. and W. B. Wu (2007). Asymptotic spectral theory for nonlinear time series. The Annals of Statistics 35(4), 1773–1801.
  33. [33]Smetanina, E. (2017). Real-time GARCH. Journal of Financial Econometrics 15(4), 561–601.DOI
  34. [34]Smetanina, E. and W. Wu (2021). Asymptotic theory for QMLE for the real-time GARCH (1, 1) model. Journal of Time Series Analysis 42(5-6), 752–776.
  35. [35]Stărică, C. and C. Granger (2005). Nonstationarities in stock returns. Review of Economics and Statistics 87(3), 503–522.DOI
  36. [36]Truquet, L. (2017). Parameter stability and semiparametric inference in time varying auto-regressive conditional heteroscedasticity models. Journal of the Royal Statistical Society Series B: Statistical Methodology 79(5), 1391–1414.DOI
  37. [37]Wu, W. B. (2005). Nonlinear system theory: Another look at dependence. Proceedings of the National Academy of Sciences 102(40), 14150–14154.pubmed.ncbi.nlm.nih.gov/16179388
  38. [41]Xiao, Z. and L. R. Lima (2007). Testing covariance stationarity. Econometric Reviews 26(6), 643–667.
  39. [42]Xu, B. and Z. Wu (2025). Real-time GARCH@CARR: A joint model of returns, realized measure of volatility and current intraday information. The North American Journal of Economics and Finance 76, 102368.

Paper details

vlnc:W_b82xskmj
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 1

A view counts once per browser per day, a download each time the PDF is requested.

  • Economics, Econometrics and Finance

Not recorded for this paper

View statement
Looking for related papers…
Contents