The determinants of inflation in South Africa: 
1997-2011 
Name: Tlamelo Phramencia Phetlhu 
This dissertation is submitted in partial fulfilment of the requirement for the degree 
Master of Business Administration at North West University, Mafikeng Campus 
Supervisor : Dr Mike Sekwila 
October 2012 
LIBRARY 
MAFIKENG CAMPUS 
CALL NO,: 
Page 1 of 112 2021 .. 02- 1 1 
Abstract 
This study targeted the a cause-effect relationship pertaining to the CPI inflation in 
South Africa and the core variables causing it. The result is a set of scenarios of effects 
on CPI inflation due to the core variables, the corresponding transitional probability 
matrices and the resulting Markov's chains for forecasting CPI inflation. The core 
variables have been identified using the findings of an in-depth literature review. The 
core variables are: total government expenditures, total private expenditures, exchange 
rate (USO to ZAR) and taxation. A five year sample has been taken from RBSA and 
OECD sources. First of all , a Pearson's correlation testing has been conducted to 
estimate the approximate transitional probabilities . Thereafter, four scenarios were 
defined to arrive at the final Markov's chains. Based on the findings , the current 
scenario of CPI inflation in South Africa has been discussed, compared with the past, 
and a method of forecasting CPI inflation using the Markov's chains is proposed. In 
addition, a discussion of how the Markov's chains can be modified by moving the five 
year (or larger sample) ahead has been presented. In the future, it is proposed that a 
benchmark be created by conducting a similar analysis on the comparable countries 
such that a better view of national performance of South Africa can be gauged. 
Page 2 of 112 
Acknowledgements 
Firstly, I would like to dedicate this research to my mother Keneilwe Gladys Phetlhu, 
and my sisters for the support over the of couple years. Secondly, Thanks to all the 
people who have helped and inspired me during my research . 
A special thanks go to my supervisor, Dr Mike Sekwila, for his guidance during my 
research. His perpetual energy and enthusiasm motivated all his advisees, including 
me. In addition, he was always accessible and willing to help his students with their 
research. As a result, research life become smooth and rewarding for me. 
Last but not least, thanks be to God for my life through all tests in the past years. You 
have made my life more bountiful. May your name be exalted , honoured , and glorified. 
Page 3 of 112 
Declaration of Authenticity 
I declare that all material presented in my research is my own work and I've 
acknowledged whatever has been adapted from other sources. I understand that if at 
any time it is shown that I have significantly misrepresented any material presented to 
University of North West, any action could be taken agaist me. 
[ ......... ........... .... .]  [. ........................]  
Tlamelo Phetlhu Dr Mike Sekwila 
Date: Date: 
Page 4 of 112 
Table of Contents 
List of Figures: 2-1-0 
List of Tables: ll++ 
Chapter I: Introduction ll-1-2 
1.1 I TRODUCTION llll 
1.2 BACKGROUND !lll 
1.2.1.GoVER ME T SPENDING 12-1-J 
1.2.2 PROBLEMS ASSOCIATED W ITH I FLATION 1314 
1.3 PROBLEM STATEME T 15¼ 
1.4 PURPOSE 161--1 
1.5 RESEARCH AIM 161--1 
1.6 OBJECTIVES 161--1 
1.7 RESEARCH QUESTIONS 17-1--+ 
1.7.1 PRIMARY RESEARCH QUESTION 1718 
1.7.2 SECO DARY RESEARCH QUESTIO 1718 
1.8 I TENDED OUTCOMES 1718 
1.9 SIGNIFICANCE OF THE STUDY 181-9 
1.10 RESEARCH METHODOLOGY 192-0 
1.10.1 DATA COLLECTIO 202-0 
1.10.2 MARKOV CHAI 2()2.1 
1.11 STUDY LAYOUT 21U 
1.12 SUMMARY OF CHAPTER 1 21ii 
Page 5 of 112 
Chapter 2: Literature Review 
2.1 INTRODUCTION TO INFLATION AND ITS ENABLERS 
2.2 THE IMPACT OF INFLATIO ON NATIONAL ECO OMIC GROWTH 
2.3 A REVIEW OF MACRO-ECONOMIC VARIABLES AND POLICIES THAT AFFECT PRICES A D 
WAGES 26i-1 
2.3.1 INTEREST RATES 26i-1 
2.3.2 EXCHANGE RATES 293() 
2.3.3 TAXATIO 31~ 
2.3.4 GoVER ME T SPENDl G 33J4 
2.3.5 CENTRAL BANK MONETARY POLICIES 3@+ 
2.3.6 EMPLOYME T 4041-
2.4 CRITICAL DISCUSSIO : RESEARCHER'S OWN VIEWS ON POLICIES TO CONTROL INFLATIO 
2.5 INFLATIO I SOUTH AFRICA 
2.6 SUMMARY OF CHAPTER 2 
Chapter 3: Research Methodology 
3.1 INTRODUCTION TO THE RESEARCH ED.9 
3.1.1 CONTEXT 519 
3.2 RESEARCH PHILOSOPHY AND STRATEGIES 52g 
3.2.1 A REVIEW OF RESEARCH PHILOSOPHIES AND STRATEGIES 52g 
3.2.2 PHILOSOPHY A D STRATEGY CHOSE FOR THIS RESEARCH 53M 
3.3 RESEARCH METHODOLOGY 54~ 
3.3.1 QUALITATIVE VERSUS QUANTITATIVE METHODOLOGY 54~ 
Page 6 of 112 
3.3.2 CHOICE OF METHODOLOGY IN THIS RESEARCH 55S<t 
3.4 SUMMARY OF CHAPTER 3 56S+ 
Chapter 4: Research methods, data collection arrl description of the tool 57~ 
4.1 RESEARCH METHODS 57S8 
4.1.1 A REVlEW OF RESEARCH METHODS 57S8 
4.1.2 CHOICE OF METHODS FOR THIS RESEARCH 5960 
4.2 SAMPLING, DATA COLLECTION AND ANALYSIS 6061-
4.2.1 REVIEW OF SAMPLING, DATA COLLECTION AND ANALYSIS METHODS 6061-
4.2.2 A REVIEW OF MARKOV CHAIN ANALYSIS 62~ 
4.2.3 APPROACH FOLLOWED IN THIS RESEARCH 6568 
4.3 ETHICAL CONSIDERATIONS 67~ 
4.4 SUMMARY OF CHAPTER 4 67~ 
Chapter 5: Data organisation and analysis 6869 
5.1 TABULATION AND STATISTICAL ANALYSIS OF DATA 6869 
5.2 MARKOV'S CHAIN ANALYSIS 808-1-
5.3 DISCUSSIONS 868+ 
5.4 SUMMARY OF CHAPTER 5 919i 
Chapter 6: Conclusions and Recommendations 929J 
6.1 CONCLUSIONS . 929J 
6.2 RECOMME DATIONS 9899 
6.3 SUGGESTIO S FOR FUTURE STUDIES 99100 
6.4 SUMMARY OF CHAPTER 6 100100 
Page 7 of 112 
7. References 100-1-0-l-
Appendix A ill-U-l-
Page 8 of 112 
List of Tables: 
Table 1: Differences between qualitative and quantitative methodologies ... ..... .. .... 54~ 
Table 2: Historical data of private and government total expenditures, exchange rate 
and CPI inflation rate ..... .. ......... ....... ............ ..... ........... ......... ....... ....... ... ............. 6800 
Table 3: Correlation analysis between the variables stated in Table 2 (private and 
government expenditures and exchange rate) and CPI inflation rate ........ ........ . 70+4 
Page 9 of 112 
List of Figures: 
Figure 1: Regression coefficients on regression lines plotting between growth rate of 
excess money and inflation with excess money growth being less than 10% and 
less than 20% (Dwyer and Fisher, 2009:1224) ......................................... ......... 37~ 
Figure 2: Regression coefficients on regression lines plotting between growth rate of 
excess money and inflation with excess money growth being less than 50% and 
less than 100% (Dwyer and Fisher, 2009: 1225) .. ..... .... .... .... ... .. ....... ................. 38~ 
Figure 3: Monetary transmission mechanism in South Africa (Smal and de Jager, 
2001 :5)4749 
Figure 4: Mathematical modeling chart (Joubert and Steyn, 2003:92) .................. 59@ 
Figure 5: Time series plotting of the four variables: private and government spending, 
exchange rate and CPI inflation .......................................... ... .. ..... ........ ... .......... 73-74 
Figure 6: Time series plotting of exchange rate and CPI inflation ...... ...... ............. 74-7-a 
Figure 7: Direct Tax slabs of past five years in South Africa 
(http://www. taxconsulting .co .za/tax-guide-2007. php) .......... ........ ..... ...... ............ 751--e 
Figure 8: Markov's chain for scenario A ............................ .. .... .. ...... ...................... 81~ 
Figure 9: Markov's chain for scenario 8 .... ....................... .. .. .. ... .. ........... ............... 8384 
Figure 9: Markov's chain for scenario C ........ ... .... ................................ .... ... ... ...... 848a 
Figure 11: Markov's chain for scenario D .. ...... .. .. ..................... .......... .. .... .. ............ 8500 
Page 10 of 112 
CHAPTER 1 
INTRODUCTION 
1.1 INTRODUCTION 
Inflation is defined as an upward movement of the general price level of most of the 
essential goods and services needed by the natives of an economy to sustain their well-
being (Harriott, 2000:6). 
Inflation is a very critical variable that affects the socio-economic lifestyles and overall 
well-being of the people of a nation (Salmanpour and Bahloli , 2011: 1226). It is a 
complex variable and is highly unpredictable due to uncertainties in the factors that 
cause inflation (Salmanpour and Bahloli, 2011 :1226). In general , inflation is viewed as a 
hindrance to the overall economic growth of a nation given that it affects the purchasing 
power of people and entities (Salmanpour and Bahloli , 2011 :1227). 
Inflation can have either a positive or a negative effect on the economy including loss of 
stability in the real value of money and other monetary items over time, with positive 
effects including mitigation of economic recessions and debt relief by reducing the real 
level of debt. 
1.2 BACKGROUND 
More often, the South African economy experiences a short-term disturbance in inflation 
caused by the exchange rate , investment , government spending and tax. As a result, 
inflation becomes a major concern in South Africa. The focus of this study is to find 
Page 11 of 112 
ways, means and solutions for predicting the probability of inflation, together with 
variables causing it, weather they would increase or decrease. 
Songezo Zibi (City Press,2011) reported that the National Treasury had deemed it fit to 
put out a draft discussion document to pave the way for the government to pass 
legislation making it difficult for people to withdraw their pension benefits when changing 
jobs. The intended legislation is further aimed at assisting with and focussing on the 
curbing of inflation as far as the subjects are concerned. 
For the purpose of enhancing the economy, there is also a need to introduce tax 
legislation that will support saving, in the sense that the interest earned on savings must 
be tax free. In a way, this will assist the community and the country in dealing with a 
high rate of inflation in South Africa. It is also imperative for the proposed legislation to 
allow contributors to have access to their benefits. 
1.2.1. Government Spending 
High and stable GDP is associated with economic prosperity and enhanced economic 
growth. High GDP may lead to higher income for consumers and could facilitate 
investment by firms and government. High GDP may lead to economic instability and 
increased cost to the economy. Lower inflation may facilitate economic stability and 
investment planning by firms , leading to higher rates of economic growth. 
Unemployment reflects an underutilised resource, but labour market concerns are now 
switching towards productivity. The balance of payments reflects a country's trading 
position with the rest of the world . Just like individuals, an economy has to be 
concerned about running up a long-term deficit. It is important to recognize that a 
Page 12 of 112 
government may not be capable of targeting all macro-economic variables. For 
example, higher GDP may lead to higher inflation (Begg & Ward , 2004). 
According to Begg & Ward (2004), conflicts exit because of a misalignment of interests. 
Without government control , firms can pollute the environment freely. The cost of 
pollution is, instead, carried by society. By making the firm bear the responsibility and 
cost of pollution, the firm 's and society's interests are aligned. The way to achieve this is 
to tax the firm if it pollutes, and perhaps even provide it with subsidies if it tries to 
operate pollution-free. More generally, solutions to misalignment of interests are the use 
of financial incentives to change the behaviour of one of the parties. 
Sorensen & Whitta-Jacobsen (2005) explain that the co-movements in different 
economic variables over the business cycle may be measured by their coefficients of 
correlation with GDP. Private investment, consumption and imports, all have a strong 
positive correlation with GDP. Employment also displays a clear positive correlation with 
GDP, but it is a lagging variable, being more strongly correlated with the GDP of the two 
previous quarters than with current GDP. Labour productivity tends to vary positively 
with GDP, and so does inflation in most countries, although the latter correlation is not 
very strong. The nominal short-term interest rate tends to go up in the two quarters, 
following a rise in GDP, reflecting a tightening of monetary policy. 
1.2.2 Problems associated with inflation 
Uncertainty about future inflation has two costs: firstly it makes planning more complex, 
raising the real resources society uses to make plans and do busin·ess. Secondly, 
Page 13 of 112 
people dislike risks. When people make nominal contracts, uncertainty about inflation 
means uncertainty about the eventual real value of the nominal bargains currently 
made. This is true cost of inflation, if a lower average level of inflation also reduces 
uncertainty about inflation. This may be a reason to aim for a low inflation rate. The 
institutions that commit the government to low inflation may also reduce the scope for 
uncertainty about inflation. If so, lower average inflation has a real benefit because it 
also diminishes uncertainty (Begg et al., 2005). 
According to Lapin (1976), the main distinguishing feature of a Markov process is that it 
is concerned with the probabilities for being in various states at any given time and 
moving from one state to another. A further characteristic of a Markov process is that in 
the long-run, the probability of being in a particular state will be constant over time and 
will hold , regardless of the state probability that applied in the beginning . 
A key property of a Markov process is that it has no memory in the usual sense; yielding 
some useful mathematical properties. A Markov process serves as a convenient model 
for many important business applications, especially in marketing and includes some 
inventory and queuing situations and various operations. Inflation is a very critical 
variable that affects the socio-economic lifestyles and overall well-being of the people of 
a nation (Salmanpour and Bahloli, 2011: 1226). It is a complex variable which is highly 
unpredictable due to uncertainties in the factors that cause inflation (Salmanpour and 
Bahloli , 2011 :1226). In general , inflation is viewed as a hindrance to the overall 
economic growth of a nation, given that it affects the purchasing power of people and 
Page 14 of 112 
entities (Salmanpour and Bahloli, 2011 :1227). Hence, when inflation is low in an 
economy, the central bank tries to maintain it as long as possible and when inflation 
rises, the central bank faces uncertainty about which monetary policy will bring it down 
to acceptable levels (Salmanpour and Bahloli , 2011: 1227). 
1.3 PROBLEM STATEMENT 
South Africa is experiencing a sudden increase in inflation caused by the exchange 
rate, investment spending, government spending and tax. High inflation is a major 
concern to South Africa .The focus of the study is to establish what the probability is that 
inflation, together with variables causing it, will increase or decrease using a Markov 
chain and available historical data. 
The biggest immediate threat to South Africa's continued economic growth is the 
capacity constraint that has increased because of the country's strong economic 
performance. A weak demand contains inflation pressure and inflation expectations 
should be more forward-looking . Inflation in South Africa remains persistently high, 
although pressures can be contained by weak demand and a relatively strong rand in 
order to reduce volatility of inflation expectations and to make the rand more 
sustainable. Progress has to be made to attain the inflation target and significant 
improvements have to be made to the previous government's policy by making the 
Reserve Bank more accountable for its actions and to increase the transparency and 
predictability of monetary policy. 
Page 15 of 112 
When dealing with this subject, it is important to look at the close relationship between 
fluctuations and output fluctuations and a clear tendency for stock price movements to 
lead in output. 
1.4 PURPOSE 
The purpose of this dissertation is to make an academic presentation on the causes of 
high inflation in South Africa, as well as the recommendations to reduce it without 
compromising on growth and development. This dissertation is expected to add value to 
the existing empirical knowledge on inflation in South Africa. 
1.5 RESEARCH AIM 
The aim of this study is to establish the probability that inflation, together with variables 
causing it ,will increase or decrease in South Africa and the strategies beneficial to the 
nation to optimise outputs and reduce inflation. 
1.6 OBJECTIVES 
The study seeks to achieve the following objectives: 
(a) To investigate the projected path of inflation rate using available historical data. 
(b) To indentify the relevant variables causing inflation in South Africa ., and 
(c) To recommend appropriate strategies to optimise output and reduce inflation so 
that growth and prosperity keep improving while inflation is contained or reduced. 
Page 16 of 112 
1. 7 RESEARCH questions 
The following are the primary and secondary research questions of this study. 
1.7.1 Primary research question 
The primary research question is stated as below: 
What is the probability that inflation, together with variables causing it, will increase or 
decrease in South Africa? 
1. 7 .2 Secondary research question 
The secondary research questions are stated below: 
(a) Why is South Africa experiencing a sudden increase in inflation? 
(b) Why is managing GDP, inflation , unemployment and the balance of payment 
important? 
(c) How are the various macro-economic variables affecting inflation and output in 
South Africa? 
1.8 INTENDED OUTCOMES 
The researcher firstly wants to establish a correlation between the four chosen variables 
and inflation in South Africa, using linear regression analysis to evolve the Pearson 
coefficients . The Pearson coefficients vary from Oto -1 for a negative correlation and 0 
to +1 for a positive correlation. Based on the values of the Pearson coefficients derived 
Page 17 of 112 
from the data set (monthly values taken from Reserve Bank website), the Markov 
transition matrix is formed and the Markov chain is completed. The Markov chain does 
not incorporate historical data and hence the resulting Pearson coefficients have been 
treated as indicators of present state correlations to apply in the Markov chain, and the 
future (expected) correlations are computed. This method has its own limitation 
because the Pearson coefficient values between the causal variables (exchange rate, 
investment spending, government spending and taxes) and the output variable 
(inflation), in the data set, are treated as present values in the Markov chain. However, 
the technique used, is expected to prove or disprove the hypotheses with a reasonable 
level of validity and reliability, given that one of the most trusted data sources of South 
Africa (on the Reserve Bank website) has been used. 
1.9 SIGNIFICANCE OF THE STUDY 
The research is expected to ensure an empirical generalisation about the causal factors 
of inflation in South Africa. The generalisation may be useful for existing research 
aspirants interested in this subject matter as an academic reference. The researcher 
wants to disseminate the resulting knowledge through conference participation and 
journal papers. To achieve this, the researcher will try to prepare the study results for 
publication in a journal dedicated to South African affairs. The researcher is also keen to 
participate in conferences and is looking forward for an opportunity to present the 
knowledge gained. lri this way, it is expected that the dissertation will ensure a sound 
academic contribution in South Africa pertaining to the nation's macro-economic affairs . 
Page 18 of 112 
On the personal front, it is expected that the efforts applied in finishing this report will 
enable the researcher to acquire skills of proving hypotheses employing advanced 
statistical techniques. The researcher expects to use these skills for conducting 
management, finance and economics related studies in future and to make useful 
contributions to the analysts and decision-making think tanks that may be the end-users 
of such reports. 
1.10 RESEARCH METHODOLOGY 
Quantitative research is a form of conclusive research involving large representative 
samples and fairly structured data collection procedures, and a primary role of 
quantitative research is to test hypotheses (Struwig and Stead 2001 :4). According to 
Denzin & Lincoln (in Welman et al. ,2005) quantitative research methods do not involve 
the investigation of processes but emphasise the measurement and analysis of causal 
relationships between variables within a value-free context. A hypothesis is a 
proposition or statement regarding the relationship between two or more variables 
(phenomena) and a hypothesis can be tested. As a result of dealing with numbers, the 
quantitative researcher uses a process of analysis that is based on complex structured 
methods to confirm or disprove hypotheses. Quantitative researchers try to keep the 
research process as stable as possible. They also control the investigation and 
structure of the research situation in order to identify and isolate variables. Specific 
measurement instruments are used to collect data (Welman et al., 2005:9). 
Page 19 of 112 
1.10 .1 Data Collection 
Analysis of data depends on the type of study and type of data used. Since the study 
involves the Markov chain, to find the future state of the inflation rate using the present 
values of the inflation rate, secondary data is used in the analysis. The time series plots 
for both original and transformed data were used and calculations were also be done. 
1.10.2 Markov's Chain 
The Markov chain is named after Andrey Markov and is a random process where all 
information about the future is contained in the present state. One does not need to 
examine the past to determine the future to be more exact. The process has Markov 
properties, meaning that future states depend only on the present state and are 
independent of past states. The main distinguishing feature of a Markov process is that 
it is concerned with the probabilities of being in various states at any given time, and 
moving from one state to another. A further characteristic of a Markov process is that 
the long term probability of being in particular states will be constant over time and will 
hold regardless of the state's probability that applied in the beginning (Lapin , 2002). 
A key property of the Markov process is that it has no memory in the usual sense, 
yielding some useful mathematical properties. The Markov process serves as a 
convenient model for many important business applications, especially in marketing, 
some inventory and queuing situations, and various operations (Lapin, 2002). 
Page 20 of 112 
1.11 STUDY LAYOUT 
In the first Chapter the literature on the subject of inflation and other related topics were 
reviewed. Chapter two reflects the research methodology to be used in this research. 
Chapter three also reflects the collection of data, primarily on providing help with the 
tools and techniques used in the process data collection. The discussion of the results 
from the model is in Chapter four. Chapter five constitute the,outline the conclusion and 
recommendation of the study. 
1.12 SUMMARY OF CHAPTER 1 
In this study the data will be collected from the South African Reserve Bank and be 
analysed , using a technique called Markov analysis. Graphs will also illustrate the 
investigation of the exchange rate, investment spending, government spending and tax, 
to establish whether they have effects of inflation. The steady-state revealed that 
exchange rate, investment spending, government spending and tax do not affect 
inflation. 
Page 21 of 112 
CHAPTER 2: 
LITERATURE REVIEW 
2.1 INTRODUCTION TO INFLATION AND ITS ENABLERS 
Inflation is defined as an upward movement of the general price level of most of the 
essential goods and services needed by the natives of an economy to sustain their well-
being (Harriott, 2000:6). Inflation is measured by the change in Consumer Price Index 
(CPI) that incorporates the prices of most of the essential commodities in an economy 
(Harriott, 2000:7). The noticeable effect of inflation on people is the reduction of demand 
for certain commodities because people can purchase lesser products with the same 
amount of money (Harriott, 2000:8). 
Inflation is a very critical variable that affects the socio-economic lifestyles and overall 
well-being of the people of a nation (Salmanpour and Bah Ioli, 2011: 1226). It is a 
complex variable and is highly unpredictable due to uncertainties of the factors that 
encourage inflation (Salmanpour and Bahloli, 2011 :1226). In general, inflation is viewed 
as a hindrance to the overall economic growth of a nation, given that it affects the 
purchasing power of people and entities (Salmanpour and Bahloli, 2011 : 1227). Hence, 
when inflation is low in an economy, the central bank tries to maintain it as long as 
possible and when it inflates, the central bank faces an uncertainty about which 
monetary policy will bring it down to acceptable levels (Salmanpour and Bahloli, 
2011: 1227). Some experts have noted that targeting inflation may cause financial 
Page 22 of 112 
recession if there are errors in adopting the right monetary policy (Salmanpour and 
Bahloli, 2011 :1228). This may be the effect of high rate of change of inflation if it is high 
in the country and is being actively targeted by the central bank and the government 
(Salmanpour and Bahloli , 2011 :1225). 
A report by HM Treasury (n. d.) stated the following factors that enable inflation in an 
economy: 
(a) The projections of future price changes 
(b) The demand pressures 
(c) The macro-economic variables and policies that directly or indirectly affect prices 
and wages 
These are dependent variables that are affected by a number of independent variables . 
The forward looking models for price changes itself cause inflation and they can cause 
change in demand pressures as well. For example, if there is an expectation of increase 
in prices of essential commodities (like, food , oil , household consumables , etc.) by a 
forecasting model published by the government, people I tend to purchase more than 
their daily needs, thus causing inflation in the markets. Demand may change due to the 
creation of an "output gap" with respect to the current demand, by either reduction of 
output or increase in demand. For example, reduction in unemployment or increase in 
general wages tends to increase demand in the marketplace. Hence, all the factors that 
cause increase in employment and wages tend to increase demand and hence the 
inflation if an "output gap" is evident(HM Treasury, n. d.). 
Page 23 of 112 
Comparing the contents presented by Salmanpour and Bahloli (2011 :1226-1227) and 
H.M. Treasury (n . d.), inflation in an economy may be caused by an overall social 
uplifting of people by the virtue of increase in employment and wages. Does this mean 
that inflation is positively related to economic growth of a country? This aspect is 
discussed in the next Section. 
2.2 THE IMPACT OF INFLATION ON NATIONAL ECONOMIC GROWTH 
Gylfason and Herbertsson (2001 :405-410) argued that the relationship between inflation 
and economic growth cannot be empirically generalised because it depends upon the 
underlying macro-economic factors that have caused the inflation. They reviewed many 
reports that have concluded that the reduction in per-capita growth of an economy is 
much lesser than the increase in inflation from 5 to 50% per year (ranging from - 0.14% 
to - 4.5%, depending upon the period of a study and the economies included in a 
study). However, there is some threshold beyond which the per capita growth of a 
country reduces significantly. Gylfason and Herbertsson (2001 :407) marked the 
threshold at 110%. A general analysis by Gylfason and Herbertsson (2001 :424) 
supports the view of central bankers that inflation has a systematic and lasting long-
term effect on economic growth and hence, its stability is essential to ensure sustained 
economic growth. 
However, Guerrero (2006:72-77) presented a detailed analysis to prove that countries 
that have faced hyperinflation and a slump in economic growth should look at other 
Page 24 of 112 
r NWU \ 
factors that might have caused such a scenario . Poor economic condition may be due 
to many other factors, like - poor governmental policies, inadequate industrial 
productivity and performance, high corruption and high unemployment. Inflation may be 
just an outcome of such a scenario. Guerrero (2006:77) sampled the hyperinflation 
periods of a number of countries that were facing numerous problems that affected their 
growth. Guerrero (2006:77-78) demonstrated that the inflation levels reduced 
considerably in these countries once the economic uncertainty was over. Hence, it 
should be noted that both inflation and economic growth are complex secondary 
variables that may or may not be related to each other through a one-way causal 
relationship. Does this mean that inflation targeting is an effective method to produce 
good results pertaining to economic growth in a country? Mallick, Cabral and Carneiro 
(2011 :537) partially agree with this , stating that "inflation targeting has resulted in 
positive outcomes when dynamic models have been employed but have resulted in not-
so-positive outcomes when static models have been employed". As explained by 
Mallick, Cabral and Carneiro (2011 :543-546), static modelling of inflation targeting 
comprises announcing a numerical target that the government and central bank try to 
achieve by closely monitoring and throttling certain underlying indices. The indices vary 
by country. For example, emerging economies monitor the population growth index as 
one of the indices, whereas economies that are highly active in global markets monitor 
the globalisation index as one of the indices. On the contrary, as explained by Mallick, 
Cabral and Carneiro (2011 :546-549), a number of proxy variables are also monitored 
that cause dynamic effects on the indices. For example, if an economy is monitoring 
globalisation index, they may consider certain proxy variables, like - assets related 
Page 25 of 112 
globalisation, equity related globalisation and trade openness. Inflation targeting in such 
economies is highly complex but yields better per capita growth results than static 
modelling, as concluded by Mallick, Cabral and Carneiro (2011 :547-548). 
In the next Section, the researcher has reviewed the variables that affect prices of 
commodities and services in an economy, and hence impact inflation. 
2.3 A REVIEW OF MACRO-ECONOMIC VARIABLES AND POLICIES 
THAT AFFECT PRICES AND WAGES 
In Section 1.1 , the researcher reviewed the article by HM Treasury, which revealed that 
there can be three factors that enable inflation in an economy - future price projections, 
demand pressures and the macro-economic variables and policies that affect prices 
and wages. In this Section, the researcher has conducted a focussed review of the third 
factor stated as "macro-economic variables and policies that affect prices and wages". 
2.3.1 Interest rates 
The relationship between interest rates and expected inflation rate is most popularly 
known by the Fisher equation, as stated below: 
Nominal interest rate= Real interest rate X Expected inflation rate(lreland , 1996:23) 
The nominal interest rate in this equation is defined as the return on investment on a 
nominal bond in a given period. The real interest rate in this equation is defined as the 
Page 26 of 112 
return on investment on a real bond in the same period. In Fisher equation, the nominal 
bond is similar to a US treasury bond. A real bond in the Fisher equation is similar to 
investing in the cost of one unit of a commodity product at the beginning of the period 
and getting "N" units of the commodity product at the end of the period at the maturity 
value of the bond (N may be a fraction or a whole number)(lreland, 1996:23-24) 
Crowder and Hoffman (1996:115-116) tested the Fisher equation after tax-adjusting it 
and found that long-term relationship between nominal and real interest rates holds 
better than short-term relationships because the shocks in real interest rates are 
effectively absorbed by the nominal interest rates. Hence, inflation predictions are more 
accurate when a long-term relationship between nominal and real interest rates are 
considered. 
Modern research studies, however, have empirically proved that both short-term and 
long-term interest rates affect inflation in an economy. Tillmann (2008:2725-2726) 
explained the concept of cost channel that directly affects the supply side and hence 
contributes to the dynamics of inflation (Tillmann , 2008:2725). The concept of cost 
channel is linked with holding of working capital for bearing the costs of production 
(Tillmann, 2008:2726). The key agent that plays a role in this dynamics is the financial 
intermediary (a lending institution) (Tillmann, 2008:2725). Supposing that the 
companies borrow their working capitals from financial intermediaries, a short-term 
increase of interest directly inflates the cost of working capital holding by the companies 
(Tillmann, 2008:2726). This may result in the supply side productivity declining 
(Tillmann, 2008:2726), and hence causing an "output gap" (a term coined by HM 
Page 27 of 112 
Treasury n. d. reviewed in Section 1.1) . Hence, governing of interest rates can be used 
by central banks as a tool to govern the supply side outputs by changing their marginal 
costs and hence the inflation (Tillmann, 2008:2727). This process is called "policy-
controlled interest rates" to drive the retail lending rates, whereby the latter is always 
higher as a result of the lending policies of the financial intermediaries (Tillmann, 
2008:2727). The discussion by Tillmann (2008:2724-2727) revealed that even short-
term interest rates can also contribute effectively to inflation dynamics. Sometimes, the 
short-term impact on interest rates can be adverse due to the sudden rise in the prices 
of most essential commodities. Wu and Ni (2011 :4158-4160) believe that sudden 
shocks in oil prices have shaken even the most stable economies of the world (like the 
great depression in US) because the monetary policy is tightened by the congress as a 
response to such adverse oil price shocks. The most natural outcome is the rise of 
interest rates and the fall of industrial output, thus causing steep increases in inflation 
(Wu and Ni , 2011 :4159). Historically, turbulent oil prices have caused significant 
impacts on consumer price indices, economic activities, and the average inflation rises 
even if oil prices subside temporarily during a period (Wu and Ni , 2011 :4159-4160). 
Reynard (2007:1441 ). presented weaknesses in monetary policy modelling when 
including short-term interests and real economic activities. They argued that the 
monetary developments and price movements are related in such a way that outputs of 
monetary policies lags real monetary developments in an economy (output gaps) 
(Reynard , 2007:1464). In fact, price increases are not always mathematically linked with 
quantity adjustments in the outputs (Reynard , 2007:1464). Sometimes, prices keep on 
Page 28 of 112 
increasing due to market sentiments even if the output gaps show a reducing trend 
(Reynard , 2007:1464). The price increase stops only when the output gaps are 
completely gone and a surplus output is visible in the market (Reynard , 2007:1465). 
This phenomenon is very difficult to incorporate in inflation projection models if only 
aggregate monetary demands are included in the models (Reynard , 2007:1465). 
Reynard (2007:1466) recommended that inflation targeting should not be based on the 
data of aggregate monetary demand, rather should be wisely based on a number of 
indicators and models. In such models, governing short-term interest rates may not 
produce the desired results in the economy. The central banks will have to wait for 
prolonged periods to witness any tangible effects on inflation, and monitor a number of 
indicators and models. 
2.3.2 Exchange rates 
Impact of exchange rates on inflation is higher in dollarized economies than the 
economies where majority of debt is issued in local currency (Carranza, Galdon-
Sanchez and Gomez-Biscarri , 2009:98). The balance sheets in such economies are 
maintained in US Dollars and hence there is a negative effect on net worth of the 
companies whenever the value of US Dollar depreciates (Carranza, Galdon-Sanchez 
and Gomez-Biscarri , 2009:99). The impacts can be on revenues and profitability, as 
well if the depreciation of US Dollar is with respect to the currencies of countries with 
which these companies have trading relationships (Carranza, Galdon-Sanchez and 
Gomez-Biscarri, 2009:99). 
Page 29 of 112 
Given that there is an indirect impact on the capital budgeting of companies, the output 
gaps may occur, thus causing a finite impact on inflation (Carranza, Galdon-Sanchez 
and Gomez-Biscarri, 2009: 102). The impact of exchange-rate fluctuations are, however, 
small on inflation having substantial local production (Garcia, Restrepo and Roger, 
2011: 1590). But it is always beneficial to incorporate exchange rate fluctuations in 
inflation targeting , because it is advantageous for handling risk premium shocks and 
thus avoiding any adverse effect on output performance and inflation (Garcia, Restrepo 
and Roger, 2011 :1591 ). In general , inflation targeting process takes into account 
exchange-rate risks in a standard framework (Garcia, Restrepo and Roger, 2011: 1591 ). 
Exchange rate shocks affect the importers and exporters differently (Aizenman, 
Hutchison and Noy, 2011 :717-719). If a country is facing higher costs in importing US 
Dollars, the commodity exporters earn better revenues but the commodity importers 
have to pay higher prices in local currency (Aizenman, Hutchison and Noy, 2011 :717). 
This scenario increases local inflation of imported commodities especially when they are 
essential and are largely imported as in the case of emerging economies with limited 
indigenous production (Garcia, Restrepo and Roger, 2011 :1591 ). If the country faces 
such inflation in most essential commodities (example, oil , food , ) then there can be a 
widespread cascading effect on overall inflation of the economy (Aizenman, Hutchison 
and Noy, 2011 :718). Arize, Malindretos and Nippani (2004:240-241) carried out an 
analysis of exchange rate variability and inflation of 82 countries and found that small 
countries face the significant impact of inflation due to nominal exchange rate 
fluctuations. 
Page 30 of 112 
Inflation targeters work upon those control variables that are related to the internal 
effects of the economy, because external variables largely remain out of control 
(Aizenman, Hutchison and Noy, 2011 :719). For example, many inflation targeters 
throttle the international currency reserves to control the exchange rates and their 
impact on inflation (Aizenman, Hutchison and Noy, 2011 :719). The most vulnerable to 
exchange rate shocks are the emerging economies that hardly have any indigenous 
variables to control (Aizenman, Hutchison and Noy, 2011 :718). However, as proved by 
Alfaro (2005:231 ), more open economies suffer lesser fluctuations in real interest rates 
and hence suffer lesser impacts on the inflation. Alfaro (2005:229) also found that the 
fixed exchange-rate regimes are shielded from exchange rate shocks and their impact 
on inflation, whereas transparent economies enjoy better results in containing inflation 
against exchange rate fluctuations over a longer term (they do get affected in shorter 
runs). To conclude, whether caused by internal or external factors or whether inflation 
can be lowered or not against exchange rate fluctuations, central bank policy makers 
should keep a close eye on the variation in real exchange rates because it is a good 
indicator of future inflation rates in the economy (Aizenman, Hutchison and Noy, 
2011:719). 
2.3.3 Taxation 
In any economy, investors calculate their investments and savings after deducting the 
taxations. Hence, the Fisher equation needs to be modified to include after tax effects. 
Assuming that the interest payments in an economy are tax deductable and interest 
Page 31 of 112 
receipts are taxable, Darby (cited by Gandolfi , 1982:797) has modified the Fisher 
equation as follows: 
After tax real rate of interest= (1 - tax rate) X nominal rate of interest - expected rate of 
inflation (Darby's equation cited by Gandolfi, 1982:797). In fact, the following rearranged 
presentation of this equation is called Darby's effect: 
Nominal rate of interest= After tax real rate of interest + expected rate of inflation / (1 -
tax rate) (Gandolfi , 1982:797) 
Differentiating this equation with respect to expected rate of inflation, we get: 
Change in nominal rate of interest with respect to change in expected rate of inflation = 
1 / (1 - tax rate) (Gandolfi , 1982:797) 
The taxation in this equation can be a corporate tax on companies or income tax on 
individuals (Gandolfi, 1982:798). Hence, when taxation is considered , the nominal rate 
of interest increases by a multiple of expectations in inflation increase (Gandolfi , 
1982:798). 
Governments levy taxes under their budgetary constraints (Poterba and Rotemberg , 
1990: 1) . Higher taxes can cause higher rate of change of nominal interest rate with 
respect to inflationary expectations as explained by Darby's effect above. The 
Page 32 of 112 
government needs consider the inefficiencies caused by the two effects , whether 
empirically related or not (Poterba and Rotemberg , 1990:2). High inflation can cause 
economic slowdown and high taxes may reduce investment decisions, savings and 
labour supplies (Poterba and Rotemberg , 1990:2). If there are higher revenue pressures 
on the government amidst large scale spending , the government will tend to increase 
tax rates and will also tend to increase lending interest rates (Poterba and Rotemberg , 
1990:3). In this way, the inflation will increase, but by virtue of a consolidated 
government policy and not just by one of the variables (Poterba and Rotemberg , 
1990:3). However, there is a drop in social security of an income tax payer if the 
government policy makers do not adjust the tax rates for inflation (lmmervoll , 2000:13). 
The low income groups suffer a lot because of the dual impact of increased taxation and 
inflation (lmmervoll , 2000:13). There may be an increase in the price of every essential 
commodity if direct taxes are levied without adjusting for inflation (lmmervoll , 2000:14). 
The bottom line is that in the end consumers suffer a larger inflation than what is 
targeted by the government (lmmervoll , 2000:14;Koreshkova,2003:2) 
2.3.4 Government spending 
Government spending is an important and one of the largest components of the GDP 
(calculated using the expenditure approach) of a nation (Campillo and Miron, 1997:341 ). 
However, it was found that public debt to GDP ratio has a correlation with inflation 
(Campillo and Miron, 1997:341 ). The level of government debt indirectly affects inflation 
because it causes a distorting effect on the national output (Campillo and Miron , 
1997:341 ). However, if the government cares for social welfare (Campillo and Miron , 
Page 33 of 112 
1997:339) and invests in technologies and infrastructure (Campillo and Miron, 
1997:349), it helps in reducing inflation. Hence, the research by Campillo and Miron 
( 1997) discovered that debt to GDP ratio is inversely related to inflation in highly 
developed nations (p. 344) and directly related to inflation in less developed nations (p. 
345). 
Inflation is also positively related to a fiscal deficit imbalance, as discovered by Budina 
and van Wijnbergen (2000:3). By definition, fiscal deficit encompasses all deficits by the 
government of a nation and not just the deficits due to interest paid on public debts 
(Budina and van Wijnbergen, 2000:7). However, prolonged fiscal deficit and its 
imbalance with the monetary policies result in a continuously mounting public debt, 
which is not good for a nation (Budina and van Wijnbergen, 2000:30). Prolonged fiscal 
deficit indicates that a government is running at a loss because the total expenditures 
made by the government exceed the total revenues earned by the government (Budina 
and van Wijnbergen, 2000:8). It needs to be analysed by a government year-on-year as 
the percentage of GDP in a given quarter (Budina and van Wijnbergen, 2000:6). In 
macroeconomics, it is already well established that prolonged and persistent fiscal 
deficits force the government of a nation to finance it through the creation of money 
(Catao and Terrones, 2005:531 ). This phenomenon is called seigniorage and is widely 
established as an inflation enabler (Catao and Terrones, 2005:531 ). It is represented by 
the following formula: 
Page 34 of 112 
S = Rate of change of money with respect to Residual Money Balance X Residual 
Money Balance / Commodity Price Level 
(Serban, 2002:8) 
The commodity price level is dependent upon consumer price index at time "t", which is 
represented as: 
Consumer price index at time "t" = Budget deficit at time "t" I Money balance at the end 
of time "t" (Serban, 2002:7) 
If we divide the numerator and denominator by GDP at time "t", the result is: 
Consumer price index at time "t" = Ratio of Budget deficit at time "t" to GDP at time "t" / 
ratio of Money balance at the end of time "t" to GDP at time "t" (Serban, 2002:7) 
This gives an equation relating the inflation level with the size of budget deficit (Serban, 
2002:9). 
However, Catao and Terrones (2005:549-550) proved that the causal relationship 
between fiscal deficit and inflation is not always linear. They demonstrated that the 
linearity increases over sufficiently longer timelines of continued fiscal deficit, and is 
more persistent in developing countries having high inflation rates (Catao and Terrones, 
2005:550). This means that developed countries possess non-linear causal relationship 
between fiscal deficit and inflation (p. 552). This correlates well with the discussion by 
Campillo and Miron (1997:339, 349), that governments can reduce the impact of fiscal 
deficit on inflation by spending in infrastructure development, technology and 
innovations that cause economic growth in the nation, rather than employing faulty 
monetary policies. In this context, Talvi and Vegh (2005: 156-159) described that the 
Page 35 of 112 
influence of fiscal deficit on inflation will be minimal if the government expenditures and 
output are positively correlated. Moreover, they also stated that "fiscal and monetary 
policy of the country should be stable throughout the assessment period unless some 
significant deviations in output are evident" (p. 158 ). To judge this fact empirically, the 
next sub-section is dedicated to a review of influence of central bank monetary policies 
on inflation. 
2.3.5 Central bank monetary policies 
Monetary policies matter in a country because they cause adaptive expectations among 
the people, confidence (or non-confidence) in the local currency, tendency to spend or 
save, resulting in variations in spending patterns in the society, and hence an increase 
or decrease of inflation (Bernasconi and Kirchkamp, 2000:315-316). Brumm (2006:189) 
described that central bank independence as having a strong negative correlation with 
inflation in developing nations, whereas Cukierman, Miller and Neyapti (2002:237) 
found that such a powerful negative correlation emerged only after the liberalisation in 
developing countries gained significant momentum after the banking reforms. In many 
developing nations, the central bank has been given independent authority to target 
inflation (reduction) through appropriate monetary policies, while in developed countries 
inflation targeting has been persistent since the 1990s (Brimmer, 2002:93). As per 
studies made by Brimmer (2002:94-97), inflation targeting by central banks had 
returned positive results in developed as well as developing nations. A recent study by 
Lin and Ye (2009:118) on thirteen developing countries also revealed that inflation 
targeting by central banks has returned excellent results in reducing inflation as well as 
Page 36 of 112 
r • ■ •all I 
reducing inflation variability. Their study was mostly in countries that adopted a policy of 
central bank independence and their monetary policies of inflation targeting in or after 
2004. 
The central bank monetary policies are linked with money growth in a nation with an 
understanding that excess money growth has a direct correlation with inflation (Dwyer 
and Fisher, 2009:1223). The following Figures (1 and 2) show the correlation coefficient 
(Pearson coefficient) on the regression lines plotted between growth rate of excess 
money and inflation. 
Excess money growth Excess money growth 
less than 20 percent I , I less than 10 percent 
correlation 0.853 correlation 0.465 
0 
30 slope 0.883 10 slope 0.411 
0 
, , 
, 
0 0 , , I 
0 ,' 'Cc> I 
I I 
I 8 , 
, 0 Q,' 
- 0 ,' 
I 
Q) ,' 0 0 ' 
.n.. 20 l  cl ,' 0 0 I 6 
I 
C ,c;t,o 
0 ,0 
.:: 0 0 
nl 0 
:;:: 
.E 4 
10 O O O o O 0 
2 0 
, 00 'e) 0 
,' oo 
, , 0 
, 
0 , I 0 
I 
0 10 20 30 0 2 4 6 8 10 
Growth Rate of Excess Money Growth Rate of Excess Money 
Figure 1: Regression coefficients on regression lines plotting between growth rate of 
excess money and inflation with excess money growth being less than 10% 
and less than 20% (Dwyer and Fisher, 2009:1 224) 
Page 37 of 112 
In all the four plots, the solid line is the regression line and the dotted line is a line 
starting from the origin with a slope of 1.0. The regression correlation is insignificant 
(0.465) in the plot of excess money growth less than 10%. In this plot, the data points 
are highly scattered and the slope of the regression line is significantly different from the 
slope of the dotted line. The regression coefficient has improved (0.853) when the plot 
is created with excess money growth less than 20%. The data points tend to align with 
the regression line that has a slope closer to the dotted line. (Dwyer and Fisher, 
2009:1224) 
Excess money growth Excess money growth , 
less than 100 percent less than 50 percent 0 ' 
corre lation 0.979 correlation 0.951 
100 
slope 1.006 50 slope 0.992 
80 40 
-Q) cu 30 
0:: 60 
C: 
0 
;: 
cu 0 20 
'E 40 
10 0 
20 0 
0 
0 
0 20 40 60 80 100 0 10 20 30 40 50 
Growth rate of Growth rate of 
Page 38 of 112 
Figure 2: Regression coeffici~nts on regression lines plotting between growth rate of 
excess money and inflation with excess money growth being less than 50% 
and less than 100% (Dwyer and Fisher, 2009: 1225) 
When the regression line plotting is carried out with excess money growth less than 
50%, the data points are densely populated along the regression line, barring a few 
scattered data points drifted away from the line. The regression line is only slightly away 
from the dotted line. The correlation coefficient has increased to 0.951 in this case. 
Finally, when the regression line plotting is carried out with excess money growth less 
than 100%, the data points are densely populated and closely aligned with the 
regression line that is now almost along the dotted lineDwyer and Fisher, 2009:1225). 
The above analysis shows that the monetary policy of a central bank is positively 
correlated with inflation if it results in moderate to high excess money growth in the 
society. Hence, the central bank should tune the monetary policy in such a way that the 
excess money in the market should be curbed as much as possible, without causing 
hindrances in achieving the growth targets. This interpretation of the above data is 
supported by the research by Mallick, Cabral and Carneiro (2011 :537). Their research 
revealed that inflation targeting in under-developed countries results in longer lags in 
increasing the per capita income output, and shorter lags in fully developed 
industrialised economies (Mallick, Cabral and Carneiro, 2011 :549). Hence, inflation 
targeting is a good strategy keeping the growth targets in mind (Mallick, Cabral and 
Carneiro, 2011 :548). For example, a developing country cannot afford to curb the 
Page 39 of 112 
determinants of its growth, like financial globalisation and trade liberalisation (Mallick, 
Cabral and Carneiro, 2011 :537). As further reiterated by Baltensperger, Fischer and 
Jordan (2007:88), a central bank cannot play the sole role of inflation targeter It should 
have a sophisticated toolkit of monetary policies to fuel growth as well as curb inflation 
(Baltensperger, Fischer and Jordan, 2007:104). In this context, Hung (2003:45-48) 
advised that central banks should target institutional support liberalisation and inflation 
targeting in parallel , keeping the growth targets in consideration. Liberal institutional 
support can release excess money in the markets thus increasing the demand (Hung, 
2003:46). If the output does not increase proportionately, there is a high chance that 
inflation can increase (Hung , 2003:48). However, it should also be noted that growth 
can generate higher employment and influence increase in wages. This phenomenon 
can also cause excess money in the markets causing a direct correlation with inflation. 
This phenomenon may not be in the control of the central banks and is reviewed in the 
next sub-section. 
2.3.6 Employment 
Inflation has an indirect relationship with employment, as revealed by (Huang et al, 
2010:229). They further revealed that persistent inflation is associated with uncertainty 
and hence it has a distortive effect on investment projects and corresponding resource 
allocation (p. 230-231 ). For example, the public-private partnership scenario in the 
country may be distorted (p. 231 ). These effects slow down growth of businesses and 
hence result in a reduction of employment rate in the country (Huang .et al , 2010:229). 
They sampled a large longitudinal data set from 71 countries, tested for threshold 
Page 40 of 112 
effects in linear regression results and found that an increase of inflation beyond 8% of 
GDP causes significant negative effects on productivity and growth (p. 234 ). Reduction 
of employment rates is an outcome of negative productivity and growth (Huang et al., 
2010:229). However, when a central bank's monetary policies result in the reduction of 
inflation, this phenomenon reverses and the industry witnesses growth as well as an 
increased rate of employment (Gali et al., 2001 :1239). Increased growth results the in 
increased incentives/bonuses of employees that may result in excess money in the 
market (Gali et al., 2001 :1239). This may result in increase of inflation if a certain 
threshold is crossed (like, the 10% threshold shown by Dwyer and Fisher, 2009:1224). 
Hence, employment rate, growth rate, resource allocation and capacity utilisation are 
cyclical indicators that are linked with inflation through the traditional Phillips curve (Gali 
et al. , 2001 :1240). However, it should also be noted that increase in wages may not 
always result in higher purchasing power and hence inflation (Gali et al. , 2001: 1263). 
The increase in wages may also be delinked with development and growth (example: 
due to high pressures by the labour unions) (Gali. , 2001 :1263). Such a phenomenon 
may result in employment stagnation influencing a reduction of rate of employment and 
hence low inflationary pressures (Gali et al., 2001: 1264 ). This effect was evident in 
Europe in 1980s that caused a severe recession (Gali et al. , 2001 :1266). 
As per discussion in previous sub-section, central banks can maintain a balance 
between inflation targeting and meeting growth (and hence employment) targets. 
However, Surico (2007:306) revealed that central banks normally take the route of the 
political mandate of the nation. If the mandate is to increase employment, they prefer to 
Page 41 of 112 
ease the monetary controls by invoking growth friendly policies (Surico, 2007:306). 
Such a mandate may cause inflationary pressures but if the political mandate allows to 
ignore it for the time being , they will prefer to do the same (Surico, 2007:306). On the 
other hand, if the political mandate is to reduce inflationary pressures on the society, the 
central banks will attempt to tighten the monetary controls (Surico, 2007:307). The 
economic scenario of the nation, and the political mandate to solve the problems, drives 
the decision-making by the central banks (Surico, 2007:307). Overall , as revealed by 
Tillmann (2008:2727), the output of a country should increase along with employment 
(or wage) growth such that the increase in supply is able to cater to the increase in 
demand due to excess disbursable money in the society. 
Having studied the empirical theories that presented the factors that cause inflation, the 
researcher, in the next Section, presents a critical discussion on how inflation can be 
controlled . Thereafter, the researcher reviews the empirical literature on inflation 
dynamics of South Africa in Section 1.5. 
2.4 CRITICAL DISCUSSION: RESEARCHER'S OWN VIEWS ON 
POLICIES TO CONTROL INFLATION 
Inflation is a general increase in consumer price index comprising the most essential 
commodities purchased by the citizens. It is expressed as a percentage of the GDP of 
the country. Fundamentally, inflation rises or falls due to a gap in supply and demand. If 
the output of a country is insufficient to meet the demand , the prices of essential 
commodities may go up. The prices also go up when the government tries to bridge the 
Page 42 of 112 
gap through expensive means (like importing essential commodities). The logistics and 
supply chain costs incurred by the government in importing goods may be higher than 
local production and hence the prices are raised. The attempts by the government in 
reducing the fiscal deficit and in reducing the balance of payments may push the prices 
upwards. The steps that a government may take are: increase taxes to pay debt, 
stimulate economic growth to generate taxes to pay debt and print more local currency 
notes to pay debt. The first option will be highly undesirable for the citizens of a country 
and the third option may result in high inflation because the problem of low output is not 
solved by simply printing additional currency notes. Hence, the second option makes 
more sense, because increased output to meet demand makes a country self sufficient. 
The government can achieve a dual purpose by stimulating growth: reduce supplies 
from external sources and reduce fiscal deficit. For stimulation of growth, if the 
government creates a mandate of trade liberalisation with institutional support, 
businesses may flourish and result in higher outputs and employment. More growth may 
result in increased wages resulting in excess money left with the citizens, that may 
increase the demand, and hence the prices of essential commodities. The central bank 
may plan and execute the money supply/control policies to meet growth and 
employment targets. 
Hence, there are two facets resulting in higher inflation in a country: solving a deficit 
crisis or meeting ambitious growth targets. The key attribute that results in inflation is a 
reduction of output below the level of current demand. In both the facets , increasing 
output to meet demand should be the primary focus of a government. The monetary 
Page 43 of 112 
policies of the central bank should be balanced to take care of growth, as well as 
reduction in inflation. The government should promote many policies to ensure that best 
results can be achieved from the central bank's monetary policy. For example, if the 
central bank has liberalised institutional support for small and medium scale enterprises 
(SMEs), the government should create an environment in which the SME businesses 
flourish. Special schemes should be launched for entrepreneurship development, export 
promotions and foreign direct investments (FDls). The government should focus on 
each sector that contributes to the output of most essential commodities of the nation. 
However, output may not increase indefinitely at the pace of growth in demand. Hence, 
the government should also implement policies to curtail the demand. For example, if 
the citizens get excess money due to increase in wages and bonus payments, the 
government should offer certain tax saving investment schemes in which, the citizens 
can invest and avail dual benefits: tax saving and growth of money. With such schemes 
around , the citizens may be compelled to save money for future rather than spend all of 
them and fuel demand. This will also be good for the well-being of the citizens in the 
longer run. 
The expenditures made by the government should ensure growth and output 
optimisation results . Growth of businesses will lead to higher taxes that government can 
use to pay the debts. On the other hand, output optimisation will lead to the control of 
inflation in spite of increased demand due to the growth of businesses. Globalisation 
schemes (like liberalisation of foreign direct investments and opening the markets for 
multinational players) will lead to better accumulation of key fully convertible currencies 
Page 44 of 112 
thus strengthening the local currency in the global markets. If there is persistent inflation 
and inadequate growth of outputs , then the government may be spending money in the 
wrong schemes. 
There may be problems related to corruption or industrial cartels that may be pushing 
the prices upwards. For example, if there is a high demand for homes, there is the 
chance that cement and iron manufacturers may form a cartel and push the prices 
upwards while the regulatory authorities do not notice it because they are supporting 
such cartels. The government policies should strictly prohibit formation of cartels and 
should operate effective monitoring and control mechanisms. 
2.5 INFLATION IN SOUTH AFRICA 
Inflation in South Africa has not been researched adequately in the recent past. It has, 
however, appeared in group data treatments of many countries by scholars and the 
International Monetary Fund reports. It is currently a topic of interest due to its 
persistently rising trend in the recent past. As learnt from the literature review in this 
chapter, persistently rising inflation and rising fiscal deficit indicates that the growth and 
output targets are not met. Hence, it is important to study the underlying causes of 
inflation in South Africa and the policies that can help in reducing inflation. 
In South Africa, the Reserve Bank has been made autonomous after positive results of 
inflation targeting in the 1990s (van der Merwe, 2004: 1 ). South African Reserve Bank 
follows a formal and transparent inflation targeting strategy (van der Merwe, 2004:2). 
Page 45 of 112 
The inflation targeting in South Africa encompasses the following five basic elements 
executed and controlled by the central bank: 
(a) Medium term inflation targets (numerical) are announced to the public. 
(b) The monetary policy enforces an institutional commitment to price stability as the 
primary goal be achieved by the banking and financial institutions, whereby all 
other goals are treated as sub-ordinates of this primary goal. 
(c) The policy instruments are chosen to encompass underlying variables including 
exchange rate, monetary aggregates, industrial growth, GDP, balance of 
payments, interest rates and other such variables that affect inflation in medium 
term . 
(d) Monetary strategy and polices along with aims, objectives, plans and decisions 
are communicated to the public transparently. 
(e) Central bank is made independent and accountable to achieve the inflation 
targets. 
(van der Merwe, 2004:2-3) 
Three types of consumer price indices are measured in South Africa: headline 
consumer price index·, core consumer price index and overall index. The headline 
consumer price index includes prices of all the items that reflect the domestic cost of 
living in the nation and covers the most essential commodities (but does not include 
mortgage bonds). Core inflation excludes all such items that are affected by exogenous 
shocks not in the control of the central bank. In South Africa, the core inflation does not 
include mortgage bonds, certain food products and certain indirect taxes. It, however, 
Page 46 of 112 
includes oil prices. The overall index includes all items of headline and core indices plus 
the exclusions in them. These indices are published with versions covering 
metropolitan, urban and rural areas. The most significant index is the overall price index 
covering the entire nation (van der Merwe, 2004:4-9). 
The following Figure presents the monetary transmission system of South Africa (Smal 
and de Jager, 2001 :5). 
Repurchase rate 
OltlQ< asset p!IC8S 
Olllllr lntenlst rates Nornhll exchange rate Relative piices Money and credit (8CJJIY, lend, pq>erty) 
l 
EJcpenclture Net Expendlbn ln11atlon 
end Net el<ports and expectations W9lllltl 
lnfeslment ll"N8Stmenl 
l 
Demand ind supply of Demand anCI suwt; of Expa'du'a 
lmrx>rt prtces goods and services QOOOSand98M:9S and 
Investment 
Dell'&ld anCI st4)Ply cl 
Wat;ps goods and S8MC8S 
Inflation rate 
Figure 3: Monetary transmission mechanism in South Africa (Smal and de Jager, 
2001 :5) 
The South African Reserve Bank controls the change in repurchase (repo) rate through 
a series of economic events throughout the country. The events are presented in the 
Figure 3 above. The repo rate directly influences decisions on investments and 
spending, asset prices, money and credits, exchange rates and interest rates. Changes 
in repo rate affect both demand and supply of essential commodities. The domestic 
Page 4 7 of 112 
inflationary pressure is the result of pressure on supply by the demand. Other factors 
that influence inflationary pressures in South Africa are: the creation of new markets for 
goods and services, pressures originating from the labour markets and exchange rate 
fluctuations (also referred to as imported inflation). Money and credits, other asset 
prices (equity, land and property) and interest rates are three primary throttles used by 
the central bank in South Africa(Smal and de Jager, 2001 :5-6). 
Aron and Muellbauer (2001 :26) described that South African economy as being 
dependent upon its minerals (mining), and the local stock market is less liquid and 
important to raise capitals. Hence, the terms of trade in South Africa are different. The 
high interest rates and lack of liberal money policies were the key growth limiters and 
high inflation during 1990s (Aron and Muellbauer, 2001 :27). As per their analysis, the 
South African growth is fuelled by international capital flows and trade liberalisation 
(Aron and Muellbauer, 2001 :28). In a recent research by Aron and Muellbauer 
(2012:456), they found that inflation forecasting in South Africa has been very 
challenging given that sectoral inflationary pressures in South Africa are highly volatile 
due to volatility in food prices and exchange rates. Moreover, the responses to 
monetary changes (expansions/contractions) vary significantly in different regions of 
South Africa (Fielding and Shields, 2006:965). Hence, a single indicator may not 
represent the inflationary pressures across the entire South Africa (Fielding and Shields, 
2006:978). It is also observed that monetary policy changes result in persistent and 
significant changes in relative prices of various provinces of South Africa (Fielding and 
Shields, 2006:979). In addition, Naraidoo and Paya (2012:446) discovered that there 
Page 48 of 112 
are nonlinearities in the short-term impacts of monetary policy rules on inflation that can 
be reduced by lengthening the forecasting horizon. They recommended semi-
parametric modelling for forecasting interest rates (Naraidoo and Paya, 2012:454 ). 
Gupta and Kabundi (2011 :1076) also supported long range time series forecasting of 
nominal interest rate, inflation and per capita growth rate in South Africa employing 
large factor models. 
These recent efforts on forecasting macro-economic variables in South Africa have 
returned commendable results. However, this researcher believes that more empirical 
results are needed by testing additional models. In this study, the Markov chain has 
been tested by using Pearson coefficients of linear regression of four causal variables 
correlated with inflation as the elements of Markov transition matrix. 
2.6 SUMMARY OF CHAPTER 2 
Inflation, output and growth are related to underlying microeconomic variables in a very 
complex manner. The complexity and the relationship itself vary by country, and the 
type of country (rich , poor, developed and developing). The literature reviewed in this 
Chapter has presented many perspectives of such relationships . However, not much 
emphasis has been given by academicians on the specific case study of South Africa, 
especially in the recent past when high development and growth rates have resulted in 
persistently increasing inflation. Some recent studies by scholars have returned useful 
models and their results. However, it is important that more models should be tested 
such that better empirical generalisations can be evolved. The next Chapter comprises 
Page 49 of 112 
a review of research philosophies and methodologies and the researcher's choices 
made for conducting this study. 
Page 50 of 112 
CHAPTER 3: 
RESEARCH METHODOLOGY 
3.1 INTRODUCTION TO THE RESEARCH 
In this Chapter, the researcher has introduced the research details, the research 
philosophy and its related research methodology. Based on these fundamentals, the 
researcher has created a research design in Chapter 1 incorporating the conclusions in 
this Chapter and additional reviews carried out in Chapter 4. 
3.1.1 Context 
Inflation in South Africa has been increasing considerably, as per the published Reserve 
Bank data. In this research , an organised study of inflation in South Africa and the 
variables causing it is conducted. Based on the learning from the literature, the following 
underlying variables have been chosen in this study: 
(a) Exchange rate 
(b) Investment spending 
(c) Government spending 
(d) Taxes 
The empirical support of correlation between these variables and inflation is reviewed in 
Chapter 1. The correlations have been reviewed in the empirical literature by taking 
data from many Western and European countries. However, there is insufficient 
empirical support for the causes of inflation in South Africa. Hence, the researcher has 
Page 51 of 112 
studied the correlations between these variables and inflation in South Africa to 
evaluate how they are related. As reviewed in Chapter 1, the key attribute to contain 
inflation is a rise in output to ensure that the rise in demand can be fulfilled . Hence, the 
researcher has also studied a correlation of government and investment spending with 
the GDP to determine the chances that inflation will not increase as a result of growth. 
Based on the findings and a comparison with the literature, the researcher has 
presented recommendations on how inflation can be reduced in South Africa without 
compromising on growth and development. 
3.2 RESEARCH PHILOSOPHY AND STRATEGIES 
3.2.1 A review of research philosophies and strategies 
Research philosophy is linked with the motivation and purpose of the research (Bryman 
and Bell , 2007:12). If the motivation is to generate new theories , the philosophy followed 
is interpretivism and if the motivation is to prove theories (hypotheses), the philosophy 
followed is positivism (Bryman and Bell, 2007:13). Saunders, Lewis and Thornhill 
(2011 :113) stated that "Positivism is a way of working in the traditions followed by 
natural scientists" and "interpretivism is a way to distinguish between intrinsic 
understanding of social actors" (p. 114). Both positivism and interpretivism belong to the 
framework of epistemology that evolves the knowledge that is acceptable to the 
interested parties in a particular field of study (Saunders, Lewis and Thornhill , 
2011 :112). Hence, the underlying methodologies to make a study credible are branches 
of epistemology (Saunders, Lewis and Thornhill, 2011 :112). Epistemology demands 
that each researcher should describe and justify the research methodologies and the 
Page 52 of 112 
linked techniques and instruments that have been employed to arrive at the results 
(Saunders, Lewis and Thornhill, 2011 :120). The other two philosophies under 
epistemology are: pragmatism and realism. Pragmatism is a philosophy in which, a 
researcher tests different positions (Saunders, Lewis and Thornhill, 2011 :109), and 
realism is the philosophy that raises a question about existence of objects of 
phenomena independent of our knowledge about their existence (Saunders, Lewis and 
Thornhill , 2011 : 114 ). In addition to epistemology, there is another framework called 
ontology that deals with studies on social structures (Bryman and Bell, 2007:20). It has 
two philosophies: objectivism and social constructivism (Bryman and Bell , 2007:21 ). 
Given that this is not a research on social structures, a discussion on these philosophies 
has been skipped . 
Positivists employ scientific methods to deduce knowledge from data and information by 
discovering correlations between independent and dependent variables, whereas 
interpreters employ hermeneutics (an art of human interpretation skills) to induct 
knowledge from data and information (Bryman and Bell, 2007:15). Hence, positivisms 
employ deductive strategy and interpreters employ inductive strategy of study. 
3.2.2 Philosophy and strategy chosen for this research 
This study is based on statistical and probability models that are widely used in natural 
science research . Moreover, the researcher has worked on correlations between 
independent and dependent variables chosen in this research using deductive strategy 
and has evolved a forecasting model. Hence, the researcher has chosen positivism 
Page 53 of 112 
philosophy. This philosophy and strategy is suitable for this study because the research 
is based on numerical data archives administered and maintained by the central bank of 
South Africa. An interpretive philosophy for inducting knowledge from such large data 
sets and evolving visible trends may not be feasible . There may be some experts that 
have developed advanced skills to extract information from such data sets maintained 
by central bank of a nation. But such skills are developed after an extensive experience 
in macroeconomics analytics and working on various models. The researcher is a 
student and hence lacks such an experience. Hence, the positivism philosophy with 
deductive strategy is more feasible for the researcher to conduct this study. 
3.3 RESEARCH METHODOLOGY 
A research methodology is an umbrella classification for grouping various underlying 
research methods (Bryman and Bell , 2007:24). It is of two types: qualitative and 
quantitative (Bryman and Bell, 2007:24), with differences as described in the following 
sub-section. 
3.3.1 Qualitative versus quantitative methodology 
Bryman and Bell (2007:25) tabulated the following differences between qualitative and 
quantitative methodologies: 
Table 1: Differences between qualitative and quantitative methodologies 
I Qualitative I Quantitative 
Page 54 of 112 
Based on interpretivism in epistemology Based on positivism in epistemology 
Based on social constructivism in ontology Based on objectivism in ontology 
Accepts texts , graphics and videos as Accepts only numerical inputs 
inputs 
Applies methods of organised human Applies mathematical and statistical 
interpretations modelling 
Employs inductive strategy Employs deductive strategy 
Used to build new theories Used to prove/reject theories 
Based on the above table, the researcher's choice is mentioned in the next sub-section. 
3.3.2 Choice of methodology in this research 
Based on the differentiation in Table 1, it is revealed that quantitative methodology is 
the natural choice for this study. This is because positivism and deductive strategy, 
chosen by the researcher, are presented as characteristics of quantitative research as 
indicated by Bryman and Bell (2007:25). In fact , Bryman and Bell (2007:26) argued that 
it is not always true that positivism cannot be adopted by qualitative researchers. 
However, as stated by them , "it is a question of epistemological compliance, and 
traditionally the interested parties in most fields of study are more comfortable with 
quantitative methodology for proving hypotheses (positivistic approach)" . Hence, 
quantitative methodology is the natural choice of the researcher for conducting this 
study. 
Page 55 of 112 
3.4 SUMMARY OF CHAPTER 3 
In this Chapter, the researcher has presented her intent to choose positivism 
philosophy, deductive strategy and quantitative methodology after some literature was 
reviewed. In the next Chapter, the researcher has built upon these fundamental choices 
to evolve a complete research design for this study. 
Page 56 of 112 
CHAPTER 4: 
RESEARCH METHODS, DATA COLLECTION AND 
DESCRIPTION OF THE TOOL 
4.1.lntroduction 
In this Chapter, the researcher has reviewed the research methods, data collection and 
instruments used under the umbrella of quantitative methodology to determine the 
choices for this study. 
4.2 RESEARCH METHODS 
4.2.1 A review of research methods 
In this sub-section, the researcher has reviewed only quantitative methods because it 
has been shortlisted in the previous Chapter. Bryman (2004:78) identified two 
approaches for conducting a quantitative research: statistical modelling and 
mathematical modelling. In statistical modelling, the concepts are carefully mapped with 
their indicators and measures of the chosen indicators (Bryman, 2004:82). The 
sampling is carried out in such a way that sufficient data about the indicators is collected 
from the population (Bryman, 2004:82). Two most common methods of collecting 
quantitative data are: experimentation (includes field experiments and simulations in 
laboratory environment) and surveys (Bryman, 2004:121). The researcher should 
conduct the sampling very carefully to ensure that accurate results about the indicators 
and the underlying concepts can be obtained to ensure validity of results (Bryman , 
Page 57 of 112 
2004:83). The data is then organised, tabulated/charted and applied in statistical 
analysis models. Some of the key statistical data analysis methods (described in 
Section 3.2.1) are the following: (Saunders, Lewis and Thornhill, 2011 :450-466) 
(a) Significance testing 
(b) Association testing (chi-square) 
(c) Differential testing (Kolmogorov-Smirnov, independent groups t-test, paired t-test, 
one-way analysis of variance) 
(d) Strength of correlation between pairs of variables (Pearson product moment 
correlation coefficient test, Spearman's rank correlation test, Kendall 's correlation 
coefficient) 
(e) Cause-and - effect relationship between variables (linear regression analysis) 
(f) Trend analysis (descriptive statistics) and forecasting (time series forecasting 
and Markov chain). 
The other method for quantitative research is mathematical modelling (Joubert and 
Steyn, 2003:91 ). In this method , the researcher denotes the variables of the 
phenomena as mathematical symbols, applies mathematical theories and formulae to 
create theorems, applies mathematical theorems, solves them, tests in simulation or 
laboratory environment and verifies the results (Joubert and Steyn, 2003:93). The 
workflow of mathematical modelling is presented in Figure 4 below (Joubert and Steyn, 
2003:92). It is preferred by scientists testing a natural phenomenon in an experimental 
research environment (Joubert and Steyn, 2003:94). It is also highly preferred in 
Page 58 of 112 
operations research and operational management sciences (Joubert and Steyn, 
2003:91 ). 
Figure 4: Mathematical modelling chart (Joubert and Steyn, 2003:92) 
4.2.2 Choice of methods for this research 
The researcher has chosen statistical modelling in this study. The study is on inflation in 
South Africa and its underlying macro-economic variables. The data is collected from 
the website of Reserve Bank of South Africa. Given that this study is based on the 
archives of a government administered database, an experimental approach with 
mathematical modelling cannot be used in this research. Hence, the researcher's 
choice is statistical modelling. The study comprises three secondary research questions 
stated in subsection 2.1.5.2. The first two questions have been addressed with the help 
of an in-depth literature review. The third question has been addressed by conducting a 
Page 59 of 112 
quantitative analysis of data collected from the Reserve Bank of South Africa for proving 
the hypotheses stated in subsection 2.1.6. The analysis comprises two steps: 
(a) A linear correlation analysis (Pearson coefficient) between the variables stated in 
the hypotheses points. 
(b) A Markov chain analysis to present a method of predicting inflation based on the 
transitional probabilities of the variables affecting inflation. 
The linear correlation analysis is sufficient to accept or reject the hypotheses. However, 
it is not sufficient to make a prediction of how inflation is expected to vary in future. The 
correlation outcomes do not form a direct input to Markov chain, because the only input 
to it is the present state and the estimated transitional probabilities. However, the 
correlation coefficient values have given an idea on the probability values to be 
assumed to create the Markov transition matrix. The two methods are discussed in the 
next Section. 
4.3 SAMPLING, DATA COLLECTION AND ANALYSIS 
In this Section, the sampl ing method, the data collection procedure and analysis 
procedure chosen for this research have been reviewed and discussed. The choice for 
this research is presented after carrying out a brief review of the literature, detailing the 
possible techniques in quantitative research. 
4.3.1 Review of sampling, data collection and analysis methods 
Collis and Hussey (2009:147-152) explained the following sampling methods in 
quantitative research: 
Page 60 of 112 
(a) A simple random sampling is executed when each member of the population has 
an equal probability of getting chosen in the sample. 
(b) A restricted sampling is executed when each member of the population has a 
unique probability of getting chosen in the sample. 
(c) A systematic sampling is executed following a fixed pattern when each member 
of the population belongs to an organised pattern or array. 
(d) A stratified random sampling is executed when multiple members of the 
population possess identical attributes and hence can be grouped in strata. 
(e) Cluster sampling is executed when heterogeneous members of a population can 
be clustered together to achieve a common purpose. 
(f) Area sampling is executed when members of the sample can be chosen from 
different geographical areas of the population. 
(g) Double sampling is executed when a large scale data set (already existing) 
needs to be organised and reduced in the form of a sample of secondary data 
such that it can be analysed. 
Based on the above analysis , it is evident that the sampling method used in this sample 
is double sampling. A sample data of five years (Table 2) has been collected after 
analysing the existing data of about 40 years in the OECD and RBSA websites , an 
existing data of 25 years in Liberta.co.za, and a data set of 10 years existing the 
exchange rates.co.uk. The data is collected from secondary sources and hence this 
study is technically a secondary research. 
Page 61 of 112 
The data analysis is carried out in two steps. The first step is carried out employing 
Pearson correlation and a basic analysis of time series plotting. Pearson correlation is a 
coefficient represented by the following formula , which represents the linearity of 
correlation between two variables (University of West England , 2007): 
r = n(l:,xy) - {l:.x)(l:y} 
✓ [ nrx2 - a:x>2) [ nl:y2 - (l:y>2 J 
Here, 'x' and 'y' are the values of the variables and 'n' is the sample size. The value of 
'r' varies from - 1 to + 1. A negative correlation is indicated by a value from - 1 and zero 
and a positive correlation is indicated by a value between zero and +1 . A value below 
0.5 is viewed as low correlation, a value of 0.5 to 0.8 is viewed as moderate correlation 
and a value above 0.8 is viewed as significant correlation. (University of West England , 
2007) 
In this study, the correlations between the independent variables (government and 
private spending, exchange rates and direct tax rates) and inflation have been 
analysed. In addition, an interpretive analysis of the time series plotting of the data in 
the sample has been carried out to understand the correlations more deeply. The 
correlation coefficients have been used as inputs to the Markov transitional matrices. 
The rationale for following this strategy is discussed in the next sub-section. 
4.2.2 A review of Markov chain analysis 
A Markov chain may be viewed as a sequence of random variables that have Markov 
properties, stated as the following : 
Page 62 of 112 
If the value of a variable at a time (t) is known, then the past values of the random 
process may not influence the future conditional probabilities of the variable. Hence, a 
Markov chain has no memory. However, Markov chain works for a finite state space 
represented by a matrix of transition probabilities presented as: 
Poo Pi 1 P 
PlO Pll Pl 
P 20 P21 JJ2 
P O P 1 P 
For example, with reference to the following Markov chain : 
.,\ 
1- A cG--,-, - -8Ji-µ 
The matrix of transition probabilities is the following: 
Po Pl ) (l -,\ /\ ) 
- ( P10 Pll - j.l l - µ 
(Henkel , Martin and Nardari , 2008:4-10; Oxy.edu, 2005:5-9) 
In the matrix of transitional probabilities, the sum of a row is equal to 1 and the matrix is 
always of a M X M type (square matrix). The transitional probabilities can be calculated 
after conducting a set of experiments and observing the transitions. It is represented by 
the total number of times a transition has reached a state (say, j) divided by the number 
of trials. If experiments are not feasible , then the transitions need to be taken from 
historical records (if available). (Oxy.edu, 2005:9) 
Page 63 of 112 
However, this simple formulation cannot be applicable if the transitions are dependent 
on some other variable or variables. In this case, the transition probabilities can be 
obtained by correlating the transitions of the other variables (assumed as independent 
variables) with the transitions of the studied variable (assumed as dependent variable) 
(Walsh, 2004:4). If the data set is very large, then a statistical correlation analysis may 
be needed (like, autocorrelation, cross-correlation and Pearson correlation) (Walsh, 
2004:5-8). As discussed by Henkel, Martin and Nardari (2008: 2-5), there may be a 
number of overlapping series of discrete data values in a sample having a finite 
correlation. Hence, on a finite time horizon having inter-related variables, the probability 
of the next state of a data value of a variable in a sample depends upon its correlation 
with the other variables on the horizon (Henkel, Martin and Nardari, 2008:14-15). In 
such a case, the matrix of transitional probabilities is said to have a stochastic property 
as indicated by the following equations (Coolen, 2009:16): 
V i j E 2 . PiJ E [O 1] 
L Pki = 1 
i 
L Pln + l = L Pi(n 
i i 
Hence, the transitional probabilities can assume a value between 0 and 1 and the sum 
of all transitional probabilities is unity. Moreover, the sum of transitional probabilities at 
state 'n' is equal to the sum of transitional probabilities at state 'n + 1' , which is equal to 
unity. This implies that if the probability of an increase in variable 'A' due to increase in 
variable 'B' is 'p', then the probability of an increase in variable 'A' due to reduction in 
Page 64 of 112 
variable 'B' is '1 - p'. However, there is a third event possible that 'A' might increase 
without any change in variable 'B'. Hence, if the probability of increase in 'A' due to an 
increase in 'B' is ascertained as 'p', then '(1 - p)' can be split into two probabilities: a 
probability of increase in 'A' due to reduction in 'B', and a probability of increase in 'A' 
without any change in 'B'. A similar decision should be taken if the probability of 
increase in 'A' due to a reduction in 'B' is ascertained (say, p'). In this case, the 
probability '(1 - p') ' can be split into two probabilities. This process may be viewed as 
normalising a probability when the Markov matrix exhibits a stochastic property. 
(Coolen, 2009:16-17) 
4.3.3 Approach followed in this research 
In this research, there are five variables overlapped on the same time horizon: private 
expenditure, government expenditure, exchange rate, tax rate, and CPI (consumer price 
index) inflation (increase in general price level). The variations of these variables have 
been captured over a five year horizon (2007 to 2011 ; quarterly data) from RBSA and 
OECD databases. The CPI inflation is analysed as the studied (observed) variable and 
other variables have been analysed as independent variables. The variations in CPI 
inflation due to tax rate has been interpreted manually because the tax rate has not 
changed in South Africa during this period albeit the taxable slabs have been adjusted. 
In the first step, the 2-tailed Pearson correlation is calculated between the independent 
variables (private expenditure, government expenditure, exchange rate) and CPI 
inflation rate. The variables have been analysed further by plotting them on a time 
Page 65 of 112 
series. The Pearson correlation coefficient has been taken as the transitional probability 
based on the outcomes of the following two studies: 
(a) Ratto and Stokke (2012:9) used the Pearson test to estimate transition 
probabilities over a fixed time horizon. To test its validity, they divided the time 
horizon into four sub-periods and compared the transition matrices of the sub-
periods (dividing number of positive transitions with the total number of 
transitions) with the full period matrix (comprising Pearson coefficient values). 
The differences between transition probabilities were minor, at 95% confidence 
level. Hence, with a reasonable accuracy, the Pearson correlation coefficient 
values can be included as transitional probabilities in the Markov matrix. 
(b) Fallahi and Rodriguez (2007:20) argued that the Pearson correlation coefficient 
is a measure of the probability of two variables remaining within a common 
regime. For example, Pearson correlation of 0.8 implies that two variables on a 
time horizon will remain in the same regime 80% of the time. It may be viewed in 
another way that, both variables have a high probability of varying in the same 
direction 80% of the time. Hence, 0.8 can also be assumed to be the transition 
probability, with a reasonable level of accuracy. 
However, the researcher observed two characteristics of both the case studies: 
(a) They are based on a sample taken at a significantly long time horizon. 
(b) The transitions of the variables are stochastic. 
Page 66 of 112 
Hence, in this study we follow the process by Ratto and Stokke (2012) and Fallahi and 
Rodriguez (2007). The sample has been taken with a reasonably long time horizon (five 
years), and it is assumed that all the variables exhibit stochastic transitions . 
4.4 ETHICAL CONSIDERATIONS 
As per Collis and Hussey (2009:45), a researcher should conduct the study in such a 
way that no deception or harm is caused to anyone in any form. In this study, there is no 
chance of deception or harm because the study has been carried out using secondary 
data collected from highly reliable sources. There is no primary research conducted, 
and hence no respondents are interviewed or surveyed in this study. Moreover, the 
analysis is general in nature for academic theory building on economic variables and is 
not expected to cause any harm to an individual, community or an organisation. 
4.5 SUMMARY OF CHAPTER 4 
In this Chapter, the researcher has studied the quantitative research methods and 
chosen the approach of this study. The data is collected on a time horizon of 5 years , 
from 2007 to 2011 , on a quarterly basis. The approach chosen in this study comprises 
two steps: a Pearson correlation analysis between the economic variables (stated in 
hypotheses) and CPI inflation, and a Markov chain analysis of the CPI inflation. In the 
next Chapter, the primary data has been tabulated and the approach has been applied 
to generate the results for interpretations. 
Page 67 of 112 
CHAPTER 5: 
DATA ORGANISATION AND ANALYSIS 
5.1 INTRODUCTION 
In this Chapter, the data collected from Reserve Bank of South Africa (RBSA) pertaining 
to the variables stated in the hypotheses are presented and analysed . In the next 
Section, the data is tabulated and the results of a correlation analysis , and a Markov 
chain analysis are presented. 
5.2 TABULATION AND STATISTICAL ANALYSIS OF DATA 
The following Table presents quarterly data from Q1 2007 to Q4 2011 pertaining to 
private and government total expenditures, exchange rates, and CPI inflation rate. The 
data sources are the following: 
• RBSA ( expenditures - current) 
• OECD (expenditures - archives) 
• Liberta.co.za (CPI inflation history) 
• Exchange rates.co.uk (historical exchange rates) 
Table 2: Historical data of private and government total expenditures, exchange rate 
and CPI inflation rate 
Page 68 of 112 
Private total Government Exchange rate CPI Inflation 
expenditure (R total (USO to SA rate on the last 
Billions) expenditure (R Rand) on the date of each 
Billions) last day of the quarter(%) 
quarter 
Q4 2011 461 .16 170.75 8.0763 6.1 
Q3 2011 453.35 169.25 8.0975 5.7 
Q2 2011 437.56 162.96 6.7717 5.0 
Q1 2011 428.20 153.60 6.7801 4.1 
Q4 2010 418.15 150.62 6.6026 3.5 
Q3 2010 407.66 149.72 6.9711 3.2 
Q2 2010 396.32 142.70 7.6755 4.2 
Q1 2010 388.04 141 .38 7.2855 5.1 
Q4 2009 383.89 139.65 7.4100 6.3 
Q3 2009 375.06 137.75 7.5222 6.1 
Q2 2009 367.46 124.87 7.7005 6.9 
Q1 2009 360.48 118.86 9.4402 8.5 
Q4 2008 357.89 120.99 9.5160 9.5 
Q3 2008 357.95 110.97 8.2732 13.1 
Q2 2008 354.04 104.61 7.8950 12.2 
Q1 2008 344.24 106.39 7.9803 10.6 
Q4 2007 336.73 104.45 6.8062 9.0 
Page 69 of 112 
Private total Government Exchange rate CPI Inflation 
expenditure (R total (USD to SA rate on the last 
Billions) expenditure (R Rand) on the date of each 
Billions) last day of the quarter(%) 
quarter 
Q3 2007 332.12 96.61 6.8900 7.2 
Q2 2007 321.71 95.53 7.0798 7.1 
Q1 2007 310.36 94.62 7.2230 6.1 
In the following table(Table3), the Pearson correlation between the variables stated in 
2nd , 3rd and 4th columns and CPI inflation has been obtained from SPSS. As a ballpark 
figure, the correlation coefficient values above 0.5 are accepted as moderate and the 
values above 0.8 are accepted as significant. As noticed in Table3 below, CPI inflation 
has a moderate negative correlation with total private expenditure ( - 0.543), a 
moderate negative correlation with total government expenditure ( - 0.635), and a 
moderate positive correlation with exchange rate from USD to ZAR ( + 0.526). However, 
the government and private expenditure are significantly positively correlated ( + 0.978), 
while the exchange rate is almost not correlated with government and private 
expenditures. 
Table 3: Correlation analysis between the variables stated in Table 2 (private and 
government expenditures and exchange rate) and CPI inflation rate 
Page 70 of 112 
7 
Correlations 
Private Govt. Exchange Inflation 
Expenditure Expenditure Rate Rate CPI 
- . Private Pearson 1 .978 -.089 -.543 
Expenditure Correlation 
Sig. (2-tailed) .000 .710 .013 
N 20 20 20 20 
Govt. Pearson .978 -- 1 -.097 -.635 --
Expenditure Correlation 
Sig. (2-tailed) .000 .684 .003 
N 20 20 20 20 
. 
Exchange Rate Pearson -.089 -.097 1 .526 
Correlation 
Sig. (2-tailed) .710 .684 .017 
N 20 20 20 20 
. - . Inflation Rate Pearson -.543 -.635 .526 1 
CPI Correlation 
Sig. (2-tailed) .013 .003 .017 
N 20 20 20 20 
**. Correlation is significant at the 0.01 level (2-tailed). 
*. Correlation is significant at the 0.05 level (2-tailed). 
Page 71 of 112 
These correlations may be investigated further by looking into their time series charts 
(created on SPSS). The following Figure presents how the four variables have varied in 
five years from 2007 to 2011 on a quarterly basis. For better clarity, the time series 
plotting of exchange rate and CPI inflation is presented in a separate plot in Figure 6. It 
is very interesting to observe that the government and private spending follow an yearly 
incremental pattern with the values almost connected. The inflation variations follow a 
reducing pattern and quite consistently follow the government and private spending in 
the opposite direction, without a lag. Hence, it appears that the effects of government 
and private spending on inflation are visible within the same quarter. 
Page 72 of 112 
- flallo Rate CPI 
- Govt.Expeodit re 
A'ivateEx pentiture 
, 
..... _,., 
-
i-- I-v 
~ 
v ,_ l,-" V 
Date 
Figure 5: Time series plotting of the four variables: private and government spending , 
exchange rate and CPI inflation 
The exchange rates and CPI inflation follow reasonably similar patterns, with the crests 
and troughs well connected to some points. Hence, it may reasonably be concluded that 
the effects of exchange rates on CPI inflation is almost like government and private 
spending, although in the same direction. Moreover, unlike the government and private 
spending, both exchange rate and CPI inflation vary in a cyclic pattern. 
Page 73 of 112 
- f;xchan eRale 
- nf lion teCA 
12.5 
10, 
7.5 
b 6 
Date 
Figure 6: Time series plotting of exchange rate and CPI inflation 
Tentatively, the variations in CPI inflation as per the correlation values reflect within the 
same quarter (of the entire model and not just one variable). However, the individual 
correlation pairs will be needed to assume the transition probabilities in the Markov 
chain. 
To estimate the variation of CPI inflation with direct tax rates, an interpretive method 
has been employed. This is because the tax rates prevail in multiple slabs and hence 
Page 74 of 112 
entering them in SPSS will be cumbersome and prone to errors. The tax slabs of the 
five years are analysed as following : 
'fax Culdt 2007 Tax Guldt 2001 
► ltl7 TIX IIIIH for 01 llfrtfl 2007 to 2t flllna,y- ► 2008TuR11_uf,o,.t 1_111_ttftl_Nlto1 lt.f.-...Y.,.1,'.1 11 
SoulllAfnc,n~ r_,.rs 
l'leon\f But Income 8ut Tu lu 
tll<ffdlng - Below tXCffd!nO - - Tax Guldt 2011 RC R lllSOC 1111 RO R 122.COO 1£1\ 
R20.250 · 2!'Hfl"t • • R 122001 R ,~oce li2tiec:•2$1.c• -t st t bo,e Al 12!CI A1 80000 ► 2011 Tu Rates lor01 Mardi 2011 to 29 februlry 2012 
112!00 lu. 
q3i 12S • »-c~~ • «: I• :l1C·!t'H1 e •bCJt SouttlAfncan lndll1dull TU!)IYers R ,eo001 R:!50000 R 19S.OOI R27C 
l&0000 I~ 
Income But 
~ .. m-~• oi . tbOvt Tu 
A2!~COI AJ~.000 R58.12!•3$liofl"<• '1fbo. R270.00I R? SO CCO - exceeding ... Below 2!0.000 27e. 
RO R 150.000 Ir.I 
Al!O COI R93.12! • 3811 Of I"<• Rla0.001 R1 Ct 210 • !!'1'o1 l"I01o:"'ltbc1e R•50000 ~0 000 * ·• A•~~ !i:_~ 
R 150.001 R23S.OOO R27000•2!'\lof et'!IOu<l abo,e 
It C!OC OO l "'d t bc•t RIJI 12!••~ofh 6bo.t 
150000 
450000 R'50 001 Wtbo,e 
Rl•!CI • ~ t ' ••""IO ~ubo,e 
.. l•i! -.. ;msooo R '8 250 • 30'4 of u,e a'!IO.nt abo,e R 2'..S .001 
TH Guldt 2009 Tix Culdt 2010 235000 
R 325,001 R S.250 • 35% ofllle l'!IO.•ta~e ► 2'09 TU Altes for ti llfrtll 2'0t to 2t febMry 2010 ► 2010Tu-fa,'1111ttft-l021februlry:lt11 R 45$.000 325000 
- Soull1 African lnchicul TlllPl)'Ort louthAfnca_.,.f_ - R 120 iO • 3811 of Ille IJ!"Ount abo, e R4SS001 RSS0.000 But lnton\f 8ut ,ssooo TU Tu excNOng - Below Htff009 - R 168.~ • <Oll of l "'IOu"1 abo,e ~e R sao.001 and abo\ e Rl32000 18'. RC R ,,ecec It'<. ~000 
A2 3.760 • 2$% 0111,ea ,a.bo,e n,<tC-~ H!2COI R2IOOOO AI L0.001 R221.CICC ••• • »-•• 
132000 l<C~ 
R• 3.260 • la'ilo'll't ,-6bo t A.l! !: .. !:~ c ~ ! t)C~e R21C OCI R290.000 R221COI R?C! eet 210000 22uc, 
a -x• • ?!:i\ e1 .,.,. 
RlSO!Ot R<tOOOO R&7lEO • lSII o•!!'t yf' l~ t Rl05e01 P •lt GCC 
290000 ?C!e: 
R<I0.001 R 109~• 3.l'Jlof .1~:>c, t R•90000 . A•J1001 A!!lll'JC 
Rll• ~ • !K<! • . l!>o ,• 
'10 000 ,:f.: ... 
RIS2 S., • <0'4of!!>t m ,e 
R! 2!-.001 a"id abGve R!-!2.0011nd 1bche 
i< T.C.i'A • 'foef e 
!25000 !:2.C-
Figure 7: Direct Tax slabs of past five years in South Africa 
(http://www.taxconsulting.co.za/tax-guide-2007.php) 
It may be observed that the tax rates have remained the same, but the slabs have been 
increased gradually. Hence, effectively it may be interpreted that the net taxes paid by 
individuals have been reduced in South Africa in the past five years. The inflation rates 
at the end of second quarter in each year in the sample (assuming that the effects are 
visible two quarters after the taxes are paid) are the following: 
Page 75 of 112 
Q2 2007: 7.1 
Q2 2008: 12.2 
Q2 2009: 6.9 
Q2 2010: 4.2 
Q2 2011: 5.0 
While the tax slabs have been increasing gradually, the inflation is following a cyclic 
pattern in this sample. Hence, it appears that direct tax rates and inflation are poorly 
correlated. Keeping this in view, variations in direct tax rates have not been taken as an 
input to the Markov chain analysis. 
The above observations may be used to estimate the transition probabilities and their 
values in Markov chain. It may be noted that this is the sole purpose of carrying out the 
above observations, because the Markov chain is a memory less framework and does 
not take into account the historical values. The transitional probabilities are 
approximately assumed as following : 
(a) CPI inflation variation (approximately at the beginning of next quarter) with 
respect to increase in government spending : 
a. No variation: 0.1 - it is assumed that there may be small probability that 
CPI inflation may remain the same. 
b. CPI inflation will reduce: 0.63 - a closest value of the correlation 
coefficient has been taken as an assumption. 
Page 76 of 112 
c. CPI inflation will increase: 0.27 - the remaining probability has been 
assumed to be of an increase in CPI inflation. 
The probability values of points b and c will swap if government spending 
reduces. 
(b) CPI inflation variation (approximately at the beginning of next quarter) with 
respect to increase in private spending: 
a. No variation: 0.1 - it is assumed that there may be small probability that 
CPI inflation may remain the same. 
b. CPI inflation will reduce: 0.54 - a closest value of the correlation 
coefficient has been taken as an assumption. 
c. CPI inflation will increase: 0.36 - the remaining probability has been 
assumed to be of an increase in CPI inflation. 
The probability values of points b and c will swap if private spending reduces. 
(c) CPI inflation variation (approximately at the beginning of next quarter) with 
respect to increase in exchange rates (USO to ZAR): 
a. No variation: 0.1 - it is assumed that there may be small probability that 
CPI inflation may remain the same. 
b. CPI inflation will increase: 0.53 - a closest value of the correlation 
coefficient has been taken as an assumption. 
c. CPI inflation will reduce: 0.37 - the remaining probability has been 
assumed to be of a reduction in CPI inflation. 
The probability values of points b and c will swap if exchange rate 
reduces. 
Page 77 of 112 
Theorem: If there are three probabilities of the same event due to three different 
stochastic variables in their respective samples (say, NA , N8 , Ne), they can be 
combined as following: 
P(E) = P(A) [ 1 - (Ns + Ne)/ (NA+ Ns + Ne)] + P(B) [ 1 - (NA+ Ne)/ (NA+ Ns + Ne)]+ 
P(C) [ 1 - (NA+ Ns) / (NA + Ns + Ne)] 
Proof: 
It is assumed that the three stochastic varying variables are part of three different 
samples causing the same event E. However, If the samples could be combined, then 
P(E) =(A+ B + C) /(NA+ Ns + Ne) 
= A I (NA + Ns + Ne) + B / (NA + Ns + Ne) + C / (NA + Ns + Ne) 
Where, A, B and C = no. of times the event occurs in samples NA , Ns and Ne , 
respectively. 
We can rewrite the equation as: 
P(E) = (A I NA) X [ 1 - (Ns + Ne )/ (NA + Ns + Ne)] + (B / Ns) X [ 1 - (NA+ Ne )/ (NA + 
Ns + Ne)] + (C / Ne) X [ 1 - (NA+ Ns) / (NA + Ns + Ne)] 
= P(A) [ 1 - (Ns + Ne)/ (NA + Ns + Ne)] + P(B) [ 1 - (NA+ Ne )/ (NA+ Ns + Ne)] + P(C) [ 
1 - (NA+ Ns) / (NA + Ns + Ne)] 
(concept taken from: Meyer and Nagpal , 2002) 
In this study, sample is N=20, for the three independent stochastic variables 
(government spending, private spending and exchange rate) . Hence, 
Page 78 of 112 
P(E) = P(A) (1 - 40/60) + P(B) (1 - 40/60) + P(C) (1 - 40/60) 
= [P(A) + P(B) + P(C)] X (1 - 40/60) = [P(A) + P(B) + P(C)] X 1 / 3 
Hence, in this study, the three probabilities of inflation caused by three events can be 
combined by simple averaging. With this approach, there can be three scenarios: 
[A] If the government spending and private spending increase, and exchange rate also 
increases (it may be recalled that government spending and private spending have a 
high Pearson correlation coefficient of 0.987 and exchange rate has almost no 
correlation with either of them and hence can vary independent of them), the following 
will be achieved: 
(a) CPI inflation remaining same: 0.1 
(b) CPI inflation will increase: 0.39 
(c) CPI inflation will reduce: 0.51 
[B] If the government spending and private spending reduce, and the exchange rate 
also reduces, the following will be achieved: 
(a) CPI inflation remaining same: 0.1 
(b) CPI inflation will increase: 0.51 
(c) CPI inflation will reduce: 0.39 
[C] If the government spending and private spending increase, and the exchange rate 
reduces, the following will be achieved: 
(a) CPI inflation remaining same: 0.1 
(b) CPI inflation will increase: 0.33 
(c) CPI inflation will reduce: 0.57 
Page 79 of 112 
[D] If the government spending and private spending reduce, and the exchange rate 
increases, the following will be achieved : 
(a) CPI inflation remaining same: 0.1 
(b) CPI inflation will increase: 0.57 
(c) CPI inflation will reduce: 0.33 
5.3 MARKOV CHAIN ANALYSIS 
Let us suppose that the chain comprises three states: State1 , State 2 and State3. The 
analysis of Markov chain for scenarios A, B, C and D stated in previous Section, is 
presented as following. 
The Markov transition matrix is given by the following: 
P11 P12 P13 
P21 P22 P23 
P31 P32 P33 
Here, 
P11 = Transitional probability of Inflation remaining at state 1 
P12 = Transitional probability of inflation transitioning from state 1 to state 2 
P13 = Transitional probability of inflation transitioning from state 1 to state 3 
P21 = Transitional probability of inflation transitioning from state 2 to state 1 
P22 = Transitional probability of inflation remaining at state 2 
P23 = Transitional probability of inflation transitioning from state 2 to state 3 
Page 80 of 112 
P31 = Transitional probability of inflation transitioning from state 3 to state 1 
P32 = Transitional probability of inflation transitioning from state 3 to state 2 
P33 = Transitional probability of inflation remaining at state 3 
Scenario [A]: Let state 1 be the current state , state 2 is the state of increased inflation 
and state 3 is the state of reduced inflation. When government and private expenditures 
increase and exchange rate also increases, the following will be the three state Markov 
chain : 
Figure 8: Markov's chain for scenario A 
The transitional probability matrix for scenario A will be 
Page 81 of 112 
0 .10 0.39 0.51 
0.51 0 .10 0 .39 
0.39 0 .51 0 .10 
It may be noticed that the sums of probabilities of each row is 1 to cater to the property 
of Markov's transitional probability matrix. Also, the matrix is in M X M form. The 
transition probabilities for transition from state 1 to state 2, and state 1 to state 3, has 
been entered in the matrix as per the calculations in previous Section. However, 
transitions from state 2 to state 3, and state 3 to state 2 are hidden Markov transitions 
and hence, their transition probabilities can be estimated by following the rule of a 
stochastic Markov transition matrix stated in Section 3.2.2 (the sum of probabilities in 
any row equals 1 ). 
This three state Markov chain gives a number of clues by analysing the probability 
values. For example, it shows that if Markov's transition matrix needs to be satisfied, 
then there is a hidden transition from state 2 (higher inflation than state 1) to state 3 ( 
lower inflation than state 1) , which will occur at a lower probability at 0.39 (obtained with 
the help of Markov's transition matrix). On the other hand , the vice versa (state 3 to 
state 2) is expected to occur at a high probability (0.51) 
Page 82 of 112 
Keeping the same philosophy, the Markov chain and the corresponding transition matrix 
of the other three scenarios are shown as following: 
Figure 9: Markov's chain for scenario B 
The transitional probability matrix for scenario B will be 
0.10 0.51 0.39 
0 .39 0 .10 0.51 
0.51 0.39 0 .10 
Page 83 of 112 
Figure 10: Markov's chain for scenario C 
The transitional probability matrix for scenario C will be 
Page 84 of 112 
0.10 0.33 0.57 
0.57 0 .10 0 .33 
0.33 0.57 0.10 
Figure 11 : Markov's chain for scenario D 
The transitional probability matrix for scenario D will be 
Page 85 of 112 
0.10 0.57 0 .33 
0.33 0 .10 0.57 
0.57 0.33 0 .10 
5.4 DISCUSSIONS 
In the Markov chains, it can be observed that the transitions are inversely proportional 
to the government and private spending , and directly proportional to exchange rates . 
However, from Table 3, it can be observed that government and private expenditures 
are highly correlated (Pearson coefficient at 0.978). Hence, it may be concluded that 
both of them will vary proportionately and may almost certainly push inflation 
downwards. The exceptions may occur due to change in exchange rate (USO to ZAR). 
Hence, the probability of reduction of inflation due to government and private spending 
may happen if there is an increase in exchange rates . This is the reflection from the four 
scenarios of Markov's chain presented in previous Section. 
It should be observed that the transitions from state 1 to state 2, state 1 to state 3, state 
2 to state 1, and state 3 to state 1 have been included in the transition probability matrix 
using the probabilities calculated in the four scenarios (A, B, C and D). However, the 
transitions from state 2 to state 3, and from state 3 to state 2, in the four scenarios are 
Page 86 of 112 
hidden and has been evolved due to the stochastic property of Markov's transition 
matrix. This is the unique feature of Markov's matrix. It can evolve the hidden state 
transitions not apparent from the transitional probability calculations. 
The Markov chain analysis of stochastic variables having a high correlation with each 
other in a sample taken on a common time horizon can be carried out using correlation 
analysis. The Pearson correlation between two variables can be taken as a transitional 
probability with a reasonable level of accuracy (up to 95 percentile). Hence, in this study 
the Pearson correlation coefficient has been taken as the estimated probability value 
(say, P) for a transition. The probability value for the opposite transition can be taken as 
(1 - P). However, the researcher has split this probability (1 - P) into two parts to 
include a "no transition" scenario, which can very much happen in case of the variable 
studied in this research (i.e. , CPI inflation not varying at all, in spite of variations in the 
independent variables). This approach has been used in this study in all the four 
scenarios. 
Based on the current state of inflation, one can analyse the probability of the next state 
by investigating the scenario (change in government and private expenditures, and the 
change in exchange rate) . The Markov chain is dynamic and hence the probabilities 
assumed based on correlation analysis, may change as the sample window is 
expanded or moved ahead. This is because the Pearson correlation values between the 
independent variables and CPI inflation might change. Hence, the Markov chains may 
Page 87 of 112 
have to be prepared again with slight variations of transition probabilities in each 
preparation cycle. 
Before concluding the dissertation, it is essential to analyse which hypotheses have 
been accepted . Based on the Pearson correlation analysis and mapping with the 
literature review, the following hypotheses have been accepted: 
(a) The hypothesis no. HAo stating that exchange rate affects inflation in South Africa 
is accepted with a moderate confidence level. The Pearson test returned a direct 
correlation between CPI inflation and exchange rate with a moderate value of 
0.526. Compared with the literature review, this finding matches with the findings 
by Carranza, Galdon-Sanchez and Gomez-Biscarri (2009), Garcia, Restrepo and 
Roger (2011 ), Aizenman, Hutchison and Noy (2011 ), Arize, Malindretos and 
Nippani (2004) and Alfaro (2005). However, their discussion revealed two key 
facts about this correlation. The first fact is that the correlation between exchange 
rate and CPI inflation is significant in highly Dollarized economies. The second 
fact is that the correlation is significant in economies significantly dependent 
upon imports. The Pearson correlation test reveals that both the facts are 
applicable in South Africa, albeit at moderate levels. Hence, South Africa is better 
placed in terms of local sustainability and openness of its economy. To improve 
further, the government should focus on better output growths and reduce the 
dependency on imports. Moreover, the local markets should be developed to a 
large extent such that the economical dependency on imports for the growth of 
Page 88 of 112 
the nation can be reduced. Both these factors require significant local 
expenditures by government and private companies to create local opportunities 
and employment. 
(b) The hypothesis no. H80 stating that investment spending affects inflation in South 
Africa is accepted with moderate confidence level. The Pearson test returned an 
inverse correlation between CPI inflation and investment spending with a 
moderate value of - 0.543. This result is in line with the findings in the literature 
review. Huang, Lin, Kim and Yeh (2010) stated that private expenditures result in 
improved output and hence reduced inflation. However, Galiet al. , (2001) stated 
that government should create an environment for private spending through 
investment-friendly monetary policies and institutional support. Given that central 
government follows the political mandate of a nation (Surico, 2007), fuelling 
private expenditure is the accountability of government. In South Africa, private 
players have contributed significantly but the moderate inverse correlation 
indicates that it is not sufficient to generate the desired output and reduce 
reliance on importing of essential commodities. An influence of exchange rates 
on inflation can be reduced if the local production is sufficient to meet the 
demand, and on the other hand the demand is sufficient to reduce the economic 
dependence of the nation on exports. In other words, the dollarization effects in 
South Africa need to be reduced. 
Page 89 of 112 
(c) The hypothesis no. Hco stating that government spending affects inflation in 
South Africa is accepted with a moderate confidence level. The Pearson test 
returned an inverse correlation between CPI inflation and government spending 
with a moderate value of - 0.635. The effects of government expenditure are 
almost the same as the effects of private expenditure. It is important not only to 
increase output but also to encourage private expenditure. A very good sign in 
South Africa is that government and private expenditures are significantly 
correlated (Pearson coefficient is 0.978). This indicates a consistent contribution 
by government and private players in economic development of South Africa. 
However, the inflation is only moderately reduced due to government spending 
because the contribution to output is still not optimum and the effects of 
exchange rate fluctuation on CPI inflation are prevalent. 
(d) The hypothesis no. H0 1 stating that taxation does not affect inflation in South 
Africa is accepted. A Pearson test was not possible because the direct tax rates 
have not changed in South Africa in the sampled period, rather the taxable slabs 
have been increased gradually. However, it is observed that the inflation has 
varied stochastically in the sampled five years although the direct taxable slabs 
have been increased gradually by the government. Hence, it is evident that there 
is very low correlation between direct tax variations and CPI inflation variations . 
This reveals that the government's support to direct tax payers in the form of 
reduced burden has not influenced inflation. 
Page 90 of 112 
The above findings indicate that inflation in South Africa is moderately influenced by 
changes external to the nation (dollarization effects) that is preventing the nation in 
deriving full benefits from the local expenditures made by private parties and 
government. This finding is analysed in detail in the next Chapter because the first 
research question is an inquisition of the factors causing high inflation in South Africa. 
Based on the above findings, the research questions have been answered in the next 
Chapter. In addition, a discussion has been presented on justification of how the 
objectives have been met and to what extent. 
5.5 SUMMARY OF CHAPTER 5 
In this Chapter, the five year data collected comprising quarterly values is tabulated and 
the Pearson correlation analysis has been conducted. Based on the analysis , the 
transitional probabilities for constructing Markov chains have been estimated and the 
chains prepared for four scenarios. The revelations about hidden transitions in Markov 
chains, by complying with the stochastic matrix rules have been highlighted. In addition, 
a discussion on the accepted hypotheses is presented along with an analysis of its 
interpretations indicating what is applicable in South Africa, and what changes are 
needed. In the next Chapter, the findings have been concluded in the context of South 
Africa . 
Page 91 of 112 
CHAPTER 6: 
CONCLUSIONS AND RECOMMENDATIONS 
6.1 INTRODUCTION 
In this Chapter, the conclusions are presented based on an analysis of how the 
research questions have been addressed, in line with the research design formulated at 
the beginning of this study. 
6.2 CONCLUSIONS 
The primary aim of this study was to establish the probability that inflation in South 
Africa, together with the variables causing it, would increase or decrease. The primary 
aim has been fulfilled by collecting data of five years pertaining to CPI inflation and the 
variables causing it , conducting a Pearson correlation test, constructing Markov 
transitional matrices for four identified scenarios, and constructing the corresponding 
Markov chains. In addition, the study targeted the determination is strategies good 
enough to produce optimum outputs in South Africa and be able to reduce inflation. 
In Chapter 3, an organised study of inflation and the macro-economic variables causing 
it was studied by reviewing a large number of relevant literature. The studies by Harriott 
(2000), Salmanpour and Bahloli (2011 ), HM Treasury (n . d.), Guerrero (2006), Mollick, 
Cabral and Carneiro (2011 ), Ireland (1996), Tillmann (2008) and Crowder and Hoffman 
(1996) emphasised the primary reason for inflation as a gap in the national output that 
Page 92 of 112 
applies significant pressure on the supplies to meet the demand. The output is 
influenced by the cost of inputs (materials and labour). Hence, the macro-economic 
variables that influence the output affect inflation directly are the ones causing variations 
of inflation. Scholars like Reynard (2007), Bernasconi and Kirchkamp (2000), Brumm 
(2006), Cukierman, Miller and Neyapti (2002), Brimmer (2002), and Tillmann (2008) 
identified the monetary policy of the government as the key macro-economic factor 
affecting inflation. Wu and Ni (2011) identified oil price shocks as one of the greatest 
factor influencing the output. Dwyer and Fisher (2009) identified 20% or more excess 
money in the market as highly correlated with inflation. 
There are other factors identified as well, such as: institutional support (availability of 
loans), trade liberalisation, and globalisation (Baltensperger, Fischer and Jordan, 2007; 
Hung, 2003; Mallick, Cabral and Carneiro, 2011 ). However, these do not lie at the core 
of the gamut of macro-economic variables affecting inflation. It is important that the 
government of a country establishes growth targets to reduce the output gap. Hence, all 
the independent macro-economic variables influencing output gap need to be 
considered. In this study, the following macro-economic variables have been chosen as 
a result of an extensive literature review on the macro-economic variables affecting 
inflation, and a separate study of the state of inflation and the variables affecting it in 
South Africa: 
(a) Government expenditures 
(b) Private expenditures 
( c) Exchange rates 
Page 93 of 112 
(d) Taxation 
All other variables in some form are dependent upon these variables. For example: 
• Output growth is dependent upon government and private expenditures. 
• Increase in employment is dependent upon the output growth caused by 
government and private expenditures. 
• Increase in wages is dependent upon increase in employment and output growth. 
• Balance of payments is a result of inadequate output growth such that essential 
commodities are imported from external sources. 
• Increase in supply costs depend upon high reliance on external sources and 
increase of exchange rates. 
• Surplus money in market is dependent upon taxation policies. 
There are many such factors that are directly or indirectly related to these fundamental 
variables. A quantitative study of the influence of above variables on the consumer price 
index inflation (the primary indicator of inflation in South Africa) has been conducted in 
this study using two sophisticated instruments: Pearson's correlation tests and Markov's 
chain analysis (includes formulation of Markov's matrix of transitional probabilities). 
Fundamentally, the outcomes of this study are the following: 
(a) An in-depth literature review 
(b) A Table of data pertaining to fundamental macro-economic variables and inflation 
figures of five years - 2007 to 2011 . 
Page 94 of 112 
( c) A time series analysis 
(d) A Pearson's correlation test 
(e) A list of four scenarios of change in inflation in South Africa 
(f) Markov's matrices of transitional probabilities pertaining to the four scenarios 
(g) Markov's chains in the four scenarios 
The following is a description of the outcome of the study outcomes with respect to the 
research questions: 
(a) Why is South Africa experiencing a sudden increase in inflation? 
Outcomes: The findings to this question have been discussed in Section 5.3. The 
analysis in Section 5.3 revealed that South Africa is moderately affected by both 
internal expenditures (government and private) and external factors (dollarization of 
South African economy). However, a very good sign in South Africa is that 
government and private expenditures are highly correlated (0 .978). In this sample, 
the beginning quarter (Q1 2007) and the ending quarter (Q4 2011) have witnessed a 
CPI inflation rate of 6.1. This may reveal that inflation has not changed in this 
sample. However, 6.1 % cannot be treated as the benchmark because South Africa 
has witnessed inflation rates of the order of 3.2% (Q3 2010), 3.5% (Q4 2010) and 
4.1 % (Q1 2011 ). The CPI inflation has been at its peak at 9.0, 10.6, 12.2, 13.1 and 
9.5 (in the period Q4 2007 to Q4 2008) after which, the inflation subsided for a few 
quarters. However, it is increasing again. Given the study of the correlation test of 
the chosen macro-economic variables, it is evident that government expenditure 
hardly increased during the period of peak inflation albeit the private expenditure 
Page 95 of 112 
gradually increased at a slower pace. The USD to ZAR exchange rate increased 
significantly during this period (with the peak value at 9.5160). A combined effect of 
the high impact of dollarization during this period (recall that this was the peak of the 
economic crisis) and slowed pace of government and private expenditures caused a 
steep rise in CPI inflation. The exchange rate has again increased significantly (it 
was 8.0763 in 04 2011 and is 8.3451 on September 20th 2012). The latest recorded 
rate of CPI inflation is 5.0% (end of 02 2012; source: Liberta.co.za). This is a good 
sign because the trend of 04 2007 to 04 2008 is not getting repeated . This means 
that the government and private expenditures have improved output to such an 
extent that the effect of dollarization has reduced on South Africa. By the end of 02 
2012, the total government expenditure is R 696.213 billion and the total private 
expenditure is 1882.204 billion (Source: RBSA). These values in 01 2012 are R 683 
billion and R 1844.643 billion (Source: RBSA). All the three factors (government 
expenditure, private expenditure and exchange rate) have increased and hence, as 
per the Markov chain analysis in this dissertation, it is scenario A. Therefore, there 
should be a probability of 0.51 that inflation will reduce and a probability of 0.39 that 
the inflation will increase. As correctly predicted by our Markov chain , the inflation 
has reduced from 01 2012 (6 .1%: source: Liberta.co.za) to 02 2012 (5%: source: 
Liberta.co.za). This has happened due to significant private and government 
expenditures reducing the impact of external factors (reflected in the exchange rate 
rise). The nation has significant local potential that needs to be tapped and made 
independent of external disturbances. 
Page 96 of 112 
(b) Why is managing GDP, inflation, unemployment and the balance of payment 
important? 
Findings: This question has been addressed in the Section 1.7.2 and is further 
clarified here, taking into account the findings of Pearson's correlation test and 
construction of Markov's chain. GDP is a measure of output of a nation. Our study 
showed that both private and government expenditures, which contribute 
significantly to output, are highly correlated in South Africa. Hence, both are 
collective indicators of output optimisation. However, if output is optimised but 
inflation still grows, there are external factors affecting South African economy 
(dollarization), or else the expenditures are not returning desired results. If the 
desired returns are not achieved , and there is high level of dollarization, the 
unemployment cannot be reduced , the balance of payment will increase and the 
taxation may be increased to reduce fiscal deficit. The good sign in South Africa is 
that inflation is moderately correlated in the reverse direction to government and 
private expenditures (in the sample collected from 2007 to 2011 ). Hence, there is 
scope for improvement. The national efficiency will be indicated if inflation is 
significantly inversely correlated with government and private expenditures. This will 
nullify the effects of dollarization on the South African economy. Increased desired 
output will definitely reduce unemployment, and balance of payment. 
(c) How are the various macro-economic variables affecting inflation and output in 
South Africa? 
Page 97 of 112 
Findings: As a result of Pearson's test, the following facts are evident in the sample 
of five years collected from RBSA and OECD pertaining to the chosen macro-
economic variables of South Africa: 
• CPI inflation is inversely proportional to government expenditure, and the 
correlation is moderate (- 0.635). 
• CPI inflation is inversely proportional to private expenditure, and the 
correlation is moderate (- 0.543). 
• CPI inflation is directly proportional to exchange rates , and the correlation is 
moderate(+ 0.526). 
An interpretation of the gradually increasing taxable slabs for direct taxing in South 
Africa and the CPI inflation values indicate that the latter is poorly correlated with 
direct taxing. Keeping in view these correlations, four scenarios have been created 
in Section 5.1 that may influence the CPI inflation. These scenarios have been used 
to construct the transitional probability matrices and the Markov chains, as 
presented in Section 5.2. 
6.3 RECOMMENDATIONS 
In this study, it is recommended that the Pearson correlation (and other forms of 
correlation tests) should be conducted on a moving five year sample (like, the new 
window at the end of 2012 should be 2008 to 2012). All correlations should be 
revalidated , the transitional probability matrices should be reconstructed keeping in 
mind the four scenarios analysed, and the Markov chains should be reconstructed. The 
reconstructed transitional probability matrices and the new Markov chain can then be 
Page 98 of 112 
used for a quarterly forecast of CPI inflation. In this study, the researcher has taken into 
account one-way of studying correlations. However, there can be more sophisticated 
methods to evaluate correlations more accurately. For example, the effects of multiple 
correlation coefficients can be combined to finalise a net value. This process may be 
useful in making a more accurate estimation of transitional probabilities. Moreover, the 
moving sample window may be increased from five years to seven years or ten years to 
determine more accurate estimation of correlations. 
6.4 SUGGESTIONS FOR FUTURE STUDIES 
Essentially, it is suggested that this study should be repeated every year by moving the 
sample window gradually by six months to one year. This will be a useful method to 
continuously monitor the national efficiency and the reduction of the dollarization of the 
South African economy. However, this study has missed one key aspect - a 
benchmark. A benchmark can be created by conducting this analysis for all countries 
having economic structures comparable to that of South Africa. An immediate idea that 
comes to mind is that the South African economy may be compared with the other 
nations in the BRICS group (Brazil, Russia, India and China). By repeating this analysis 
for these nations, a benchmark may be created , such that the performance of the South 
African economy can be compared with the benchmark. The future researchers may 
like to study how such a benchmark can be created. The researcher of this study is also 
keen to contribute in this direction in her future academic studies. 
Page 99 of 112 
6.5 SUMMARY OF CHAPTER 6 
In this Chapter, the conclusion has been presented and the research questions have 
been addressed. The results of a literature review, Pearson's correlation testing and 
Markov's chain analysis have been used to present the findings against the research 
questions. In addition, the recommendations from this study and the suggestions for 
future studies have been made. 
References 
Aizenman, J., Hutchison, M. and Noy, I. (2011 ). "Inflation Targeting and Real 
Exchange Rates in Emerging Markets". World Development, Vol. 39 (5): p. 712-
724. Elsevier. 
1. Alfaro, L. (2005). "Inflation, openness, and exchange-rate regimes: The quest for 
short-term commitment". Journal of Development Economics, Vol. 77: p. 229-
249. Elsevier. 
2. Arize, A. C., Malindretos, J. and Nippani , S. (2004). "Variations in exchange rates 
and inflation in 82 countries: an empirical investigation". North American Journal 
of Economics and Finance, Vol. 15: p. 227-247. Elsevier. 
3. Aron , J. and Muelbauer, J. (2012). "Improving forecasting in an emerging 
economy, South Africa: Changing trends, long run restrictions and 
Page 100 of 112 
disaggregation". International Journal of Forecasting, Vol. 28: p. 456-476. 
Elsevier. 
4. Aron, J. and Muellbauer, J. (2001 ). "Interest rate effects on output: evidence from 
a GDP forecasting model for South Africa". Centre for the Study of African 
Economies, Department of Economics, University of Oxford, p. 3-43. 
5. Baltensperger, E. , Fischer, A. M. and Jordan, T. J. (2007). "Strong goal 
independence and inflation targets". European Journal of Political Economy, Vol. 
23: p. 88-105. Elsevier. 
6. Bernasconi , M. and Kirchkamp, 0 . (2000). "Why do monetary policies matter? An 
experimental study of saving and inflation in an overlapping generations model" 
Journal of Monetary Economics, Vol. 46: p. 315-343. Elsevier. 
7. Brimmer, A. F. (2002). "Central banks and inflation targeting in perspective". 
North American Journal of Economics and Finance, Vol. 13: p. 93-97. Elsevier. 
8. Brumm, H. J. (2006). "The effect of central bank independence on inflation in 
developing countries". Economics Letters, Vol. 90: p. 189- 193. Elsevier. 
9. Bryman, A. and Bell , E. (2007). "Business Research ·Methods". Second Edition. 
OUP, London. 
Page 101 of 112 
10. Bryman, A. (2004). "Social Research Methods". Second Edition. London: Oxford 
University Press. 
11. Budina, N. and van Wijnbergen, S. (2000). "Fiscal Deficits, Monetary Reform, 
and Inflation Stabilization in Romania". Policy Research Working Paper WPS-
2298, Macroeconomics and Growth, The World Bank Research Group, p. 3-32. 
12. Campillo, M. and Miron, J. A. (1997). "Why does inflation differ across 
countries?". p. 335-362. Book Chapter: Reducing Inflation: Motivation and 
Strategy. Romer, C. D. and Romer, D. H. (Eds). University of Chicago Press. 
13.Carranza, L., Galdon-Sanchez, J . E. and Gomez-Biscarri , J. (2009). "Exchange 
rate and inflation dynamics in dollarized economies". Journal of Development 
Economics, Vol. 89: p. 98-108. Elsevier. 
14. Catao, L.A. V. and Terrones, M. E. (2005). "Fiscal deficits and inflation". Journal 
of Monetary Economics, Vol. 52: p. 529-554. Elsevier. 
15. Collis, J. and Hussey, R. (2009). ""Business Research: A Practical Guide for 
Under Graduate and Post Graduate Students". Third Edition. US: Palgrave 
McMillan 
Page 102 of 112 
16. Coolen, A. C. C. (2009). "Markov Chains: Compact Lecture Notes and 
Exercises". Department of Mathematics, King's College London, p. 2-40. 
17. Crowder, W. J. and Hoffman, D. L. (1996). "The Long-Run Relationship between 
Nominal Interest Rates and Inflation: The Fisher Equation Revisited". Journal of 
Money, Credit and Banking, Vol. 28 (1 ): p. 102-118. Ohio State University Press. 
JSTOR. 
18. Cukierman, A. , Miller, G. P. and Neyapti , B. (2002). "Central bank reform , 
liberalization and inflation in transition economies-an international perspective". 
Journal of Monetary Economics, Vol. 49: p. 237-264. Elsevier. 
19. David Begg, Stanley Fischer and Rudiger Dornbusch (2005), economics 8th 
edition by publisher: mcgraw-hill higher education. 
20. Dwyer, G. P. and Fisher, M. (2009). "Inflation and monetary regimes". Journal of 
International Money and Finance, Vol. 28: p. 1221- 1241. Elsevier. 
21 . Fallahi , F. and Rodriguez, G. (2007). "Using Markov-Switching Models to Identify 
the Link between Unemployment and Criminality". Department of Economics, 
Faculty of Social Sciences, University of Ottawa, p. 2-53. 
Page 103 of 112 
, 
r= - -----· 
22. Fielding, D. and Shields, K. (2006). "Regional asymmetries in monetary 
transmission: The case of South Africa". Journal of Policy Modelling, Vol. 28: p. 
965-979. Elsevier. 
23. Gali , J. , Gertler, M. and Lopez-Salido, J. D. (2001 ). "European inflation 
dynamics". European Economic Review, Vol. 45: p. 1237-1270. Elsevier. 
24.Gandolfi, A. E. (1982). "Inflation, Taxation, and Interest Rates". The Journal of 
Finance, Vol. 37 (3): p. 797-807. American Finance Association. JSTOR. 
25. Garcia, C. J. , Restrepo, J. E. and Roger, S. (2011 ). "How much should inflation 
targeters care about the exchange rate?". Journal of International Money and 
Finance, Vol. 30: p. 1590-1617. Elsevier. 
26. Guerrero, F. (2006). "Does inflation cause poor long-term growth performance?". 
Japan and the World Economy, Vol. 18: p. 72- 89. Elsevier. 
27. Gupta, R. and Kabundi , A. (2011 ). "A large factor model for forecasting 
macroeconomic variables in South Africa". International Journal of Forecasting, 
Vol. 27: p. 1076- 1088. Elsevier. 
28. Gylfason, T. and Herbertsson, T. T. (2001 ). "Does inflation matter for growth?". 
Japan and the World Economy, Vol. 13: p. 405-428. Elsevier. 
Page 104 of 112 
29. Harriott, K. (2000). "Essential facts about inflation". Research & Economic 
Programming Division, Bank of Jamaica, p. 3-20. 
30. Henkel, S. J., Martin, and Nardari, F. (2008). "Time-varying Short-Horizon 
Predictability". Published thesis , Carnegie Mellon University, p. 1-49. 
31. Huang, H., Lin, S., Kim, D. and Yeh, C. (2010). "Inflation and the finance-growth 
nexus". Economic Modelling, Vol. 27: p. 229-236. Elsevier. 
32. lmmervoll , H. (2000). "The Impact of Inflation on Income Tax and Social 
Insurance Contributions in Europe". Department of Applied Economics, 
University of Cambridge, p. 1-30. 
33. "Influences on inflation". HM Treasury article. n. d. 
http://archive.treasury.gov.uk/pub/html/top/top6/inf.html [Accessed : March 22, 
2012] 
34. Ireland , P. N. (1996). "Long-Term Interest Rates and Inflation: A Fisherian 
Approach". Federal Reserve Bank of Richmond Economic Quarterly, Vol. 82 (1 ): 
p. 21-36. 
Page 105 of 112 
35. Joubert, J. W. and Steyn, D. (2003). "Operationalising Operations Research". 
World Transactions on Engineering and Technology Education. Vol. 21 (11 ): p. 
91-114. 
36. Lawrence L lapin (1976), 10th Edition Quantitative Method for Business Decisions 
37. Lin, S. and Ye, H. (2009). "Does inflation targeting make a difference in 
developing countries?". Journal of Development Economics, Vol. 89: p. 118-123. 
Elsevier. 
38. "Markov Chain Models". P. 2-20. 
URL:<http://faculty.oxy.edu/angela/prior%20to%200708/392/CH5-
MarkovChains.pdf> [Retrieved on: 28 August 2012] 
39. Meyer, A. and Nagpal , R. (2002). "Introduction to Probability". Course Notes 10, 
Mathematics for Computer Science, Massachusetts Institute of Technology: p. 1-
45. URL: <http://www.cs.brown.edu/courses/csci1510/reference/probability.pdf> 
[Retrieved on 30 August 2012] 
40. Mallick, A. V. , Cabral , R. and Carneiro , F. G. (2011 ). "Does inflation targeting 
matter for output growth? Evidence from industrial and emerging economies". 
Journal of Policy Modelling, Vol. 33: p. 537- 551 . Elsevier. 
Page 106 of 112 
41. Naraidoo, R. and Paya, I. (2012). "Forecasting monetary policy rules in South 
Africa". International Journal of Forecasting, Vol. 28: p. 446-455. Elsevier. 
42. "Pearson Correlation Coefficient r". University of West England , 2007, URL: 
<http://hsc.uwe.ac.uk/dataanalysis/quantinfasspear.asp> [Retrieved on 28 
September 2012] 
43. Poterba, J. M. and Rotemberg, J. J. (1990). "Inflation and Taxation with 
Optimizing Governments". Journal of Money, Credit and Banking, Vol. 22 (1 ): p. 
1-18. Ohio State University Press. JSTOR. 
44. Rattso, J. and Stokke, H. E. (2012). "Regional convergence of income and 
education: Investigation of distribution dynamics". Department of Economics, 
Norwegian University of Science and Technology, p. 2-23. 
45. Reynard , S. (2007). "Maintaining low inflation: Money, interest rates, and policy 
stance". Journal of Monetary Economics, Vol. 54: p. 1441-1 471. Elsevier. 
46. Salmanpour, A. and Bahloli , P. (2011 ). "Inflation, Inflation Uncertainty and 
Factors Affecting Inflation in Iran". World Applied Sciences Journal, Vol. 14 (8): p. 
1225-1239. 
Page 107 of 112 
47. Saunders, M.N.K., Lewis, P., and Thornhill, A. (2011 ). "Research Methods for 
Business Students". Fifth edition. Prentice Hall, London. 
48. Sekaran, U. (2010). "Research methods for business: A skill building approach". 
Fourth Edition, NY: Wiley. 
49. Serban, M. (2002). "Budget Deficit and Inflation". Published PH D thesis, 
Academy of Economic Studies. p. 4-28. Proquest Information and Learning. ABI 
Informs. 
50. Smal , M. M. and de Jager, S. (2001 ). "The monetary transmission mechanism in 
South Africa. Occasional Paper No. 16, South African Reserve Bank, p. 1-19. 
51. Songezo Zibi, City Press, (14 August 2011 ), An economic "CODESA" must have 
real value. 
52. Struwig, F.W. and Stead, G.B. (2001 ), Planning, designing and reporting 
research. CapeTown: Masker Miller Longman 
53. Surico, P. (2007). "The Fed's monetary policy rule and U.S. inflation: The case of 
asymmetric preferences". Journal of Economic Dynamics and Control, Vol. 31: p. 
305-324. Elsevier. 
Page 108 of 112 
54. Talvi , E. and Vegh, C. A. (2005). "Tax base variability and pro-cyclical fiscal 
policy in developing countries". Journal of Development Economics, Vol. 78: p. 
156- 190. Elsevier. 
55. Tillmann, P. (2008). "Do interest rates drive inflation dynamics? An analysis of 
the cost channel of monetary transmission". Journal of Economic Dynamics and 
Control, Vol. 32: p. 272'J-27 44. Elsevier. 
56. Van der Merwe, E. J. (2004). "Inflation targeting in South Africa". Occasional 
Paper No 19, South African Reserve Bank, p. 1-19. 
57. Walsh, B. (2004). "Markov Chain Monte Carlo and Gibbs Sampling". Lecture 
Notes for EEB 581 , version 26. URL: 
<web. mit. edu/~wingated/www/introductions/mcmc-gibbs-intro. pdf> . [Retrieved on 
28 August 2012). 
58. Welman, C., Kruger, F. and Mitchell, B. (2005), Research Methodology, (3rd Ed), 
Cape Town: Oxford University Press Southern Africa. 
59. Wu, M. and Ni , Y. (2011 ). "The effects of oil prices on inflation, interest rates and 
money". Energy, Vol. 36: p. 4158-4164. Elsevier. 
Page 109 of 112 
Pages 110 
Appendix A 
Private total Government Exchange CPI Inflation 
expenditure (R total rate (USO to rate on the last 
Billions) expenditure SA Rand) on date of each 
(R Billions) the last day quarter(%) 
of the quarter 
1 Q1 2007 310.36 94.62 7.223 6.1 
2 Q2 2007 321.71 95.53 7.0798 7.1 
3 Q3 2007 332.12 96.61 6.89 7.2 
4 Q4 2007 336.73 104.45 6.8062 9 
5 Q1 2008 344.24 106.39 7.9803 10.6 
6 Q2 2008 354.04 104.61 7.895 12.2 
7 Q3 2008 357.95 110.97 8.2732 13.1 
8 Q4 2008 357.89 120.99 9.516 9.5 
9 Q1 2009 360.48 118.86 9.4402 8.5 
10 Q2 2009 367.46 124.87 7.7005 6.9 
11 Q3 2009 375.06 137.75 7.5222 6.1 
12 Q4 2009 383.89 139.65 7.41 6.3 
13 Q1 2010 388.04 141.38 7.2855 5.1 
14 Q2 2010 396.32 142.7 7.6755 4.2 
15 Q3 2010 407.66 149.72 6.9711 3.2 
16 Q4 2010 418.15 150.62 6.6026 3.5 
17 Q1 2011 428.2 153.6 6.7801 4.1 
18 Q2 2011 437.56 162.96 6.7717 5 
19 Q3 2011 453.35 169.25 8.0975 5.7 
20 Q4 2011 461.16 170.75 8.0763 6.1 
Page 111 of 112 
Tax Gulde 2007 Tax Gulde 2008 
• 2"7Jaa:RatnfM81 1Wd\2007to2tfebnaiary200I • 2MS Tu:a.tesJOt"t11.&1rd!Mllto11F~ry200I 
- Sout.h. A. fnctn lndtvidull Ta apayen - Sou..... .th..A _ trtun  ~ h);J)iYer• HltNdnQ - Tox Ta txueding -
•• •• ~ 122.'® '"' Tu Culde 201 1 Q 112.!® "
q112!-C1 A 1&0000 R2
"C 2iC • 2S,_Of N .tlftOlo'°'a,e R t220Ct ~ ,;-;etc R21.9&0 •2S%ofh~tbo\e 
112500 122.000 ..... • 1111 Ta:btesfa.-t1 u.rct.2011 to29Feb~ry2012 .,,. ... A31.125 - ~ofN a!.!Wft(,t,c~, A.l~.OC1 rt Z7C.OCC rt,0210 - 30'4ofthe ~ Solm.. A. fncan lnCtvidultl T11q>1yers AW,OC:1 1&0000 195 000 Income 
RE-2.110 • ?S"ofrN a!"IO..l'1atXht Tu 
Q2!00C1 R~OOOO R54.125 • ~"0fNamouftllOC\ I R2'."l°'t .,.. ... exceecing _ ...... 
2'0000 2TOCOO 
,,. ... ... •• R IS-0000 13'\ i:1350.0Ct Rl3.125•3"oofltite"'DW1ttoo\"1 R"50.000 R3800CI q.:so.coo ,R. 1.0 1210 • !4%oflht a"'°""'1..,• 
... R 1S<I001 
R 27.000 .. 2S~ of tttt al'!'IOw"I abo.e 
R:135.000 
R~,000._,IOC_. .A,1.3 1 .12S•'O%ofti. 1•!'JOlilf'lhbo, 1 1!0.000 i:U900C I N1bo,a Ii 1'l010•"0%of~ •~MIO\t '-iC.OCO 
Tu0Utde200I A:135001 Rmooo 
R '3 2!<1 • JC% of tt-t 1ffl0.;"t abo, • 
Tu-2010 235000 
R 5.2!0 • 35% ot ttlt "101,1t'lt abo,·• 
• 2M 1'a ..n fort1u.d'l2Mlo21f.,_,. 201t ► 2"1 Tallateatorl1 U.dl211tto21februl,y2t1 1 RJ:25.001 R '-55000 325.000 
- Soul-h Atrbn ~ T~rs ---~-.. A. fnearl lndMdull Ta..,,_r, ·- R 120.750 • 38,. ofltte , ...,v,n ebo'\ e R•.55.001 R!a0000 ""' ,55000 '"' In •• ,.. . ~,~coc R580001 WIIDO\e R 1&!.250 • ' ~ of the &tTOll"1 Ibo R1J2~ " "" SS0.000 
q 132..001 Al1000-, R23.160 • 2!'4cl !N ~ aott. e R1~ct, IU21011C R2!.
 200 • 2!% cl ttie~ .oc-e 
... "'"" l 'OliOO A 2l0®1 .,,. R'-.? 2EO • ~CflM 1~eoc- • RZ%10C1 R'~ '50 • 3C'4cl lle a.~abc. e R~CQC 
210,000 22, ... 
A~001 ,R.6.7 ...2.6 0•!5-'4ollhttm01H'l abo'.1 R'10000 R30Sett A10E!O · ?S"ofttlt a"'IOolfl1abe.-t R.'31tt.C ,., ... 
q,1oc:o1 A490000 RI09.260•31%cfttt ~ato,t l:Ul1. I R!!2CCO Rll'7S0 • 34'4oflf!t..-ount~• ,10.000 <31 ... 
R!l:S.001r.w211bo-!e R 1!2.96-11 • ~cf~ai,ouM 1!1cr.e 
A 160.730 • '4'4oftht amowl'ltabc,e 
,l~Q,ettled,bc:4 
525000 ~2.000 
End of Document 
Page 112 of 112