Actually, we are having very good uni-variate time series algorithms like PSF (https://www.researchgate.net/publication/304131481_PSF_Introduction_to_R_Package_for_Pattern_Sequence_Based_Forecasting_Algorithm) which are able to forecast much accurate than many multivariate predictions, which indirectly leads to less computation and data storage requirements. Is univariate time series prediction not enough to forecast accurately as that of multivariate TS prediction?
Article PSF: Introduction to R Package for Pattern Sequence Based Fo...