Thesis Update

I have completed the final version of the thesis and have also had it proof read. The final copy of the thesis includes relevant results which were obtained throughout the duration of the thesis.

The problem which this paper attempts to analyse is that of the impact of IT upon industry structure. This problem is of particular importance to the Australian Economy as a whole, as Australia becomes more and more IT dependent. This problem has not been addressed sufficiently in previously literature, and as such, this paper will contribute significantly to the work of researchers and policymakers in this field of study.

This paper has analysed the changes which IT has brought about upon organizations, and on an aggregate level, industry. This paper has employed generally accepted methods in order to carry out this study, using regressions, structural breakpoint tests, and fuzzy c-means clustering in order to achieve this goal.

This paper has proven Langdons’ (2003) conclusion that the Software aspect of IT is rapidly becoming more dominant than the hardware aspect of IT, and thusly, this confirmation will allow decision makers to design policies which augment and support this finding in order to promote additional growth in the Australian Economy.

The results obtained for the study indicate that there is a shift in industry structure in the form of coalescence. Based on this, policies have been recommended in order to take advantage of the changes which IT has brought about. Additionally, the resulting structural change may be of significance in predicting the future structure of industry.

In summation, the author of this paper wishes to thank Dr Simon Poon for his ongoing support and help throughout the entire process. Without his support and advice, this paper would not have been possible. It is his patience and understanding that has been of the greatest help to me in the writing of this paper.

Additionally, this paper wishes to thank Dr Rafael Calvo for the administrative work which was performed during the execution of this paper.

This paper also wishes to acknowledge the Australian Bureau of Statistics, MIT GmBH and the Open R Archive for their advice and assistance in the provision of resources used in the provision of this project.

Thesis Update

Have finished the write up of the thesis based on the latest results which were discussed with Dr Poon, which were obtained for two cycles which were based on a structural breakpoint test.

The results indicate that IT has had an effect upon the shift in industry, as can be seen from the regression results obtained for the regression with dummy variables. The regression results indicate that of the IT aspects (Hardware and Software), industry has been increasingly dependent upon Software, and this has had the most effect upon Value added. Comparing this with the clustering results also indicates that Software has had the most impact, since there is coaelscence across the cycles.

One should note that the effects of IT, as discussed by David (1990), and various papers by Brynjolfsson, were able to be illustrated by the results obtained, as the time-lag effect was clearly shown, and disproves the productivity paradox.



Reference:

David, P, 1990, "The Dynamo and The Computer: An historical perspective on the modern productivity paradox", The American Economic Review, Vol. 80, No. 2, Papers and Proceedings of the Hundred and Second Annual Meeting of the American Economic Association. (May, 1990), pp. 355-361.

Thesis Update

Have completed all the major components of the analysis required, and am now in the process of writing up.

The analysis components included performing a Quandt Test, a Chow Test, a Switching Regression, and a cluster analysis, using both fixed and industry effects to account for the different industries and their effects upon the overall function which this thesis project uses. The Chow test was selected in order to determine breakpoints, which varied from the original thesis scope in that the original breakpoints for the cycles to perform the analysis were set arbitrarily in order to evenly distribute the data across the cycles. Furthermore, the clustering analysis, when performed upon 19090-1995, 1996-1999, and 2000-2006, reveals that there are originally 3 clusters, which diverges into 5 during 1996-1999, and then settles down to 4 clusters. The clustering analysis, done via setting the fuzziness exponenet m to 2, using the partition coefficient method 1/c, therefore indicates that there is a divergence, and then convergence, across the industries during the cycles. The divergence between 1990-1995 and 1996-1999 can be explained by the fact that the middle cycle, which was determined by a Chow test, can be called a transitional period (such as that in Dr Poon's paper), and also reveals intra- and inter- differences between the clusters.

The other major component, that of switching regression, indicates that the model used is an endogenous model, and not exogenous, as the rhos for both cycles of the switching regression significantly differ from 0, which indicates the aforementioned (Bertschek et al, 2005).

Thesis Update

The fuzzy clustering has been completed, with results showing that there has been consistently 4 clusters from 1990 - 2006. This is in line with Dr Poons findings in 2002, and thus, validates the results obtained by Dr Poon in 2002.

Additionally, the switching regression part of the analysis is near completion, with results obtained for switching between Cycle 1 (1990-95), Cycle 2 (1996-2000), and Cycle 3 (2001-2006). However, a problem was encountered for the switching regression between cycle2 and cycle3, with a null value produced after the 15th iteration of the switching regression function.

Using another method, I have also performed linear regressions on all three cycles, and performed the Goldfeld-Quandt test, as well as compared the coefficient of IT in each cycle, in order to determine the effect of IT upon VA as a measure of impact on industry. I have attempted to add fixed effects of industry into the model, but so far, the model rejects fixed effects, which I shall attempt to repair before concluding.

Thesis Update

This week in my meeting with Dr Poon, we encountered a problem about the validity of the use of switching regression in the analysis of the impact of IT upon industry structure, as industries, based on cluster analysis, do not elect to move from cluster to cluster. It was therefore decided in the meeting that there were several options to pursue for this case, listed as below:

1) Continue with Switching Regression
2) Perform a Chou Test
3) Implement Random Coefficient Analysis

Furthermore, a new, raw data set has been compiled for a final fuzzy clustering analysis, the results of which will be posted in the next two days. Additionally, the switching regression problem, if implemented, will use modified equations from Berstchek's 2005 paper analysing organisational change using switching regression. By following this, it is hoped that the results of the switching regression will show that the use of IT leads to a higher productivity, and hence, a change in industry structure caused by IT, which would be found be analysis the outcomes of the switching regression equations, which uses a TRUE/FALSE vector to determine which strategy is the best for each data point.

Thesis Update

For the last two weeks, I have been attempting to perform a switching regression analysis upon the data provided by Dr Poon. To this end, I have read various papers by Bezdek and Hathaway, Chau, as well as examples of switching regression. On Wednesday, I located an online tutorial which explained how to perform this analysis in R, and as such, I will attempt to do so over the weekend, and post the results up.

Additionally, fuzzy clustering is now being performed by data engine. The final results of this analysis will be posted up later. However, preliminary indications match that of Dr Poon's findings now, as compared to the ambiguous results derived from R.

Thesis Update

Last Week I completed the partial draft, which consisted of introduction, literary review, methodology and results. However, these parts were not totally completed, and require editing.

Additionally, I performed the fuzzy clustering analysis over three cycles using R, as stated before, and discussed the results with Dr Poon in our weekly meeting. It was decided that the use of R and the determination of clusters in this manner was not suitable, and as such, we will be switching to Data Engine (a fuzzy clustering tool) for this part of the project.

I will also be taking a look at the use and development of switching regression, and will also implement this in R on Dr Poons given data for now. I will also need to develop a presentation for the overall progress for this wednesdays meeting with Dr Poon, which I am currently in the process of doing.

I will post results up as soon as they become available.