Nowcasting indian gdp with mixed-frequency singular spectrum analysis

Nowcasting GDP is crucial for policymakers, especially due to the significant publication lag of GDP numbers. This study employs mixed-frequency singular spectrum analysis, a data-driven nonparametric approach, to nowcast the real GDP of the Indian economy at various nowcasting time points. Furthermore, monthly GDP and its major components are estimated with this approach. The technique is tailored with an innovative method for determining data-driven window length, noise control, and handling missing values at the start or end of the series. The nowcasting results using this technique outperformed ARIMA and dynamic factor models in terms of root mean square error and mean absolute error, even during the pandemic. The data-driven technique described in this study may be useful in broad real-life applications in nowcasting or forecasting, considering its usefulness in handling both stationary and nonstationary, as well as linear and nonlinear data.