The properties of cycles in South African financial variables and their relation to the business cycle
Boshoff, Willem Hendrik
Stellenbosch : University of Stellenbosch
The goal of this thesis is twofold: it aims, firstly, at a description of cycles in South African financial variables and, secondly, at the evaluation of the relationship between cycles in financial variables and the South African business cycle. The study is based on the original business cycle framework of Arthur Burns and Wesley Mitchell, but incorporates recent contributions by Australian economists Don Harding and Adrian Pagan, as well as the work of the Economic Cycle Research Institute in New York. Part I of the thesis is concerned with the characteristics of cycles in financial variables within the South African context. The first chapter presents a taxonomy of the concepts of classical, deviation and growth rate cycles in order to establish a simple reference framework for cycle concepts. At this point the concept of a ‘turning point cycle’ is introduced, with particular focus on the non-parametric method of turning point identification, following Harding and Pagan’s recent translation of the original work of Burns and Mitchell into a modern version with a sound statistical basis. With the turning points identified the dissertation proceeds to an exposition of descriptive measures of expansion and contraction phases. The second chapter entails an empirical report on descriptive results for amplitude and duration characteristics of cycle phases in the different financial variables, with separate reports for classical cycles and growth rate cycles. Chapter two concludes with a series of tables in which the behaviour of cycle phases are compared for different financial variables. Part II considers financial variables as potential leading indicators of the business cycle in South Africa. Chapter 3 introduces the concept ‘leading indicator’ to this end and distinguishes the original concept from modern, econometric versions. The chapter then introduces a framework for evaluating potential leading indicators, which emphasises two requirements: firstly, broad co-movement between cycles in the proposed leading indicator and the business cycle and, secondly, stability in the number of months between turning points in cycles of the proposed indicator and business cycle turning points. The capacity of potential indicators to meet these criteria is measured via the concordance statistic and the ‘lead profile’ respectively. Chapter four provides the statistical basis for the concordance statistic, after which the empirical results (presented separately for classical and growth rate cycles) are presented. The fifth chapter presents the statistical test for the stability of the interval by which cyclical turning points in the potential indicator lead turning points in the business cycle. Empirical results are presented in both tabular form (the ‘lead profile’) and graphical form (the ‘lead profile chart’). As far as can be determined, this analysis represents the first application of the ‘lead profile’ evaluation to financial variables. Chapter six concludes by presenting a summary of the results and a brief comparison with findings from an econometric study of leading indicators for South Africa.
Thesis (MComm (Economics)--University of Stellenbosch, 2006.
Dissertations -- Economics, Theses -- Economics