Multivariate time series econometric performance of divisia monetary aggregates for the euro area

Bissoondeeal, R.K., 2005. Multivariate time series econometric performance of divisia monetary aggregates for the euro area. PhD, Nottingham Trent University.

10290196.pdf - Published version

Download (20MB) | Preview


Many economists believe that the amount of money in the economy affects either real variables like national output or monetary variables like the price level or both. Developments in monetary aggregates therefore can provide useful information about future price developments. Such a belief has led the European Central Bank to use broad monetary aggregate M3 as a compass for monetary policy strategy, which is aimed at maintaining price stability in the region.

A few decades ago many countries had put similar faith in monetary aggregates to guide their monetary policy strategies. However, a few years later empirical evidence began to emerge showing that monetary aggregates were no longer reliable as a tool for conducting monetary policy and consequently many countries abandoned monetary targeting. A possible reason for the monetary aggregates not being as reliable as previously thought is argued, by researchers such as Barnett, to be the simple summation technique of constructing official monetary aggregates. In this kind of aggregation assets as different as cash and interest bearing time deposits are weighted equally. Clearly one hundred pounds in interest bearing time deposits do not provide the same level of monetary services as the equivalent amount in currency. Therefore, the simple summation aggregation technique produces an unsatisfactory definition of the amount of money in the economy.

Divisia aggregates derived from microeconomic theory, aggregation theory and index number theory are considered to be a viable alternative to Simple Sum aggregates as in their construction assets are given weights according to the level of monetary services they provide. Since the derivation of Divisia aggregates a number of studies from around the world have compared their empirical performance to their Simple Sum counterparts. The results are found to be mixed but leaning slightly in favour of Divisia aggregates. Since the Euro area has come into existence only recently not many studies exist that compare the relative performance of Simple Sum and Divisia aggregates for the Euro area. Hence it is the main objective in this thesis to provide new empirical evidence on the relative performance of Simple Sum and Divisia aggregates for the Euro area with a view to adding to the literature on the appropriate method of monetary aggregation.

The monetary aggregates are compared in three different frameworks, namely, cointegrated VAR money demand framework, composite leading indicator of inflation turning point framework and inflation forecasting framework. Prior to constructing monetary aggregates, however, weak separability tests are carried out to identify assets that can be reliably included in a monetary aggregate. Weak Separability tests are carried out using the Fleissig and Whitney's Linear Programming test. The evaluation of monetary aggregates in cointegrated VAR money demand framework consists of the following steps. Firstly, graphical analysis and unit root tests are carried out to investigate the stationarity properties of the series entering the VAR models. Secondly, given most of the series were found to be nonstationary, Johansen maximum likelihood tests were used for testing for cointegrating relationships. Finally, the long run stability of the parameters of the different cointegrating vectors was investigated. The evaluation of monetary aggregates in the composite leading indicators of inflation consisted of the following steps. Firstly, the cycles of the inflation series and the indicator series are extracted and their turning points identified. Fourier analysis is then used to model the cycles of the series and lead time of the indicator series over the inflation series are identified for constructing a set of short leading indicators and a set of long indicators. The individual leading indicators series are then aggregated to form composite leading indicators of inflation turning point. Kalman filters are the used to filter out false turning points in the composite leading indicators. Evaluation in the inflation forecasting framework consists of constructing linear and nonlinear forecasting models. Linear models are represented by univariate time series models and multivariate cointegrated VAR models. Nonlinear models are represented by neural networks, so called because their creation was inspired by the functioning of the brain.

To increase the relevance of this study a few other issues of interest to policymakers are also investigated. These additional issues are: (1) whether or not the UK should join the Euro area, (2) whether or not central banks should use nonlinear models for macroeconomic forecasting, and (3) whether or not Divisia aggregates are disadvantaged compared to Simple Sum aggregates when they are tested in a linear framework, given the presence of nonlinear' structures in Divisia aggregates.

The main findings of the thesis are as follows: (1) As has been found in many previous studies, findings regarding the relative performance of Divisia and Simple Sum aggregates are mixed, however leaning slightly in favour of weighted Divisia aggregates (2) under present circumstances the UK should not join the Euro area, (3) nonlinear models provide more accurate forecasts of inflation, (4) Divisia aggregates are better modelled in a nonlinear framework. Further work to incorporate the construction of a risk-adjusted Euro Divisia and to optimise the weights of Euro Divisia aggregate using neural networks.

Item Type: Thesis
Creators: Bissoondeeal, R.K.
Date: 2005
ISBN: 9781369324457
Divisions: Schools > Nottingham Business School
Record created by: Linda Sullivan
Date Added: 12 Nov 2020 12:48
Last Modified: 12 Oct 2023 09:47

Actions (login required)

Edit View Edit View


Views per month over past year


Downloads per month over past year