Title: Temporal disaggregation of short time series with structural breaks: Estimating quarterly data for emerging economies
Authors: Jerome Trinh - THEMA - CREST (France) [presenting]
Abstract: Official data disclosed by emerging national administrations are sparse, especially time series such as national accounts which are mostly published once a year with from three to seven quarters of latency. Higher-frequency indicators or components can help disaggregating annual data into up-to-date quarterly data by using a well known method of temporal distribution. However, the small sample size and the instability of emerging countries official data prevent a good fit of the target time series and its indicator. Incorporating a procedure of structural break detection, such as a test of cointegration with structural breaks, into the method of temporal disaggregation method can improve the fit, hence the prediction of higher-frequency estimations. We detail disaggregation formulas that considers different types of structural break in the parameters of the model linking the target time series and its higher-frequency indicator. Small sample null distribution of the statistics for the test of cointegration with structural breaks are also computed. Then the predictive performance of the method is assessed by using empirical advanced countries data and simulated time series. Finally, an example of an application to macroeconomic business cycles analysis is initiated by studying the response of the Chinese macroeconomic fluctuations to aggregate supply and aggregate demand shocks.