Quarterly national accounts using forecasting methods

Benchmarking techniques are often employed to arrive at Quarterly National Accounts (QNA) using Annual National Accounts (ANA) as a benchmark to arrive at a consistent high frequency time series. In this context the basic Denton proportional difference (PD) method is very popular in estimating quarterly national accounts series whose movement matches the movement of a quarterly preliminary series as close as possible. This method is very efficient in distributing the annual series when the annual benchmark is available but cannot be applied when the most recent annual value is not available. In this backdrop this paper tries to address the issue of absence of benchmark value by estimating it using a combination of non-parametric regression technique and dynamic linear model with the help of past annual data. The paper also gives a simple matrix formulation to arrive at estimates of quarterly series employing proportional Denton method. Net savings of private corporate sector has been chosen as the annual benchmark, to compile quarterly net savings estimate using quarterly financial statements of listed Non-Government Non-Financial (NGNF) companies. Preliminary series has been built using quarterly growth rates of net profits adjusted for non-operating surplus/deficit for the NGNF companies. As the annual figure of net savings for private corporate sector is published after a considerable lag in national accounts statistics; forecast of the same has been attempted using historical net savings yearly growth rate data. Hence the paper is useful in devising a scheme of employing the PD method when the preliminary series of most recent quarters are available but the annual benchmark is not yet published without resorting to subjective extrapolation.