' J 

I 

I 

. I 

, I 

.I 

.I 

~ I 
I 

I 

I J 

. I 

I 

j 

' j 

J 
u. 
: I 

~strathrnore 

U NI V E RSITY 

EFFECT OF EXTERNAL DEBT ON INFLATION: THE CASE OF KENYA, TANZANIA 

AND UGANDA 

GATHENDU JANE JOY WAIRIMA 

101372 

Submitted in partial fulfilment of the requirements for the Degree of 

Bachelor of Business Science in Financial Economics at Strathmore University 

Strathmore Institute of Mathematical Sciences 

Strathmore University 

Nairobi, Kenya 

February 2021 



'I 
' ] 

.I 
' ] 

~ l 

I 
. j 

.I 
I 

- j 

.I 

J 
_I 

' J 

I declare that this work has not been previously submitted and approved for the award of a 

degree by this or any other University. To the best of my knowledge and belief, the 

Research Project contains no material previously published or written by another person 

except where due reference is made in the Research Project itself. 

©No part of this Research Project may be reproduced without the permission of the author and 

Strathmore University. This Research Project is available for Library use on the understanding that it 

is copyright material and that no quotation from the Research Project may be published without 

proper acknowledgement. 

J.~.n~)QY.W~!rjm.~ .. G.!:!~h~nQ~. [Name of Candidate] 

·······---~- ~ ---············· [Signature] 

JQ.t_~.f.~!:?r.l,J.~XY..fR:?.l ..................... [Date] 

This Research Project has been submitted for examination with my approval as the 

Supervisor. 

. Steve Anyona Makambi, PhD 
J 

-~------
[Name a/Supervisor] 

[Signature] 

[Date] 

Strathmore Institute of Mathematical Sciences 

Strathmore University 



.l 

I 
.I 
. I 

. I 

.I 

I 

.I 

.I 
J 
.I 

Abstract 

High inflation is considered to be a threat to growth of emerging countries. However, most 

of these countries need to borrow in order to cater for budget deficits and sustain economic 

growth. The aim of this study is to empirically investigate the effect of external debt on 

inflation in Kenya, Uganda and Tanzania for the period 1988-2018. Other independent 

variable is included, broad money growth rate, and analysed if it helps in explaining the 

relationship between external debt and inflation. The study uses a balanced annual panel 

data obtained from World Bank International Financial Statistics and Data Files. The 

variables are tested for stationarity to establish their order of integration and select a 

suitable model. Two tests are done for robustness, Engle and Granger, and Johansen's 

Cointegration to test for availability of co-integration relationship among the variables. A 

Vector Error Correction Model {VECM) is employed to estimate long run dynamics and 

Granger Causality to test if the co-integrated variables can help in predicting each other. The 

results show that external debt has a positive long-term effect on inflation and that money 

growth helped in explaining this relationship. Moreover, it was found that there is a 

unidirectional causality between external debt and inflation. This causality was found to be 

homogeneous. It is therefore necessary for governments to effectively manage external debt 

so as to avoid negative economic circumstances which may affects other countries in the 

region . 

Key words: external debt, inflation . 

ii 



Table of Contents 
CHAPTER ONE: INTRODUCTION ..................................................................... ! 

1.1. Background ofthe problem .............................................................................................. 1 

1.2. The statement of the problem ........................................................................................ 2 

1.3. Research objective ......................................................................................................... 3 

1.4. Research questions ........................................................................................................ 3 

1.5. Significance of the study ................................................................................................. 3 

1.6. Scope and organisation of the study ............................................................................... 4 

CHAPTER TWO: LITERATURE REVIEW ............................................................. S 

2.1. Theoretical literature ....................................................................................................... 5 

2.1.1. Demand-pull inflation theory ........................................................................................ 5 

2.1.2. Cost-push inflation theory ............................................................................................ 5 

2.1.3. Monetary theory of inflation ........................................................................................ 5 

2.2. Empirical literature ............................................................................................................. 6 

2.2.1. Cross country analysis .................................................................................................. 6 

2.2.2. Country specific analysis ............................................................................................... 8 

2.3. Summary of literature ..................................................................................................... 10 

CHAPTER THREE: METHODOLOGY ............................................................... 11 

3.1. Research design ................................................................................................................ 11 

3.2. Population and sampling ................................................................................................. 11 

3.3. Data analysis ..................................................................................................................... 12 

3.3.1. Stationarity Test ......................................................................................................... 12 

3.3.2. Engle and Granger(1987) Cointegration Test... ............................................................. 12 

3.3.3. Johansen's Cointegration Test ....................... .............................................................. 12 

3.3.4. Error Correction Model ............................................................................................... 13 

3.3.5. Granger Causality Test ................................................................................................ 13 

CHAPTER FOUR: DATAANALYSISAND FINDINGS ......................................... l4 

4.1. Descriptive statistics ........................................................................................................ 14 

4.2. correlation tests .............................................................................................................. 15 

iii 



l 
l 

.I 

I 

I 

j 

j 

J 
.I 

4.3. Stationarity Test ............................................................................................................... 15 

4.4. Cointegration Tests .......................................................................................................... 17 

4.4.1. Engle-Granger based Test ....... ................................................................................. .... l7 

4.4.2. Johansen Fisher Panel Cointegration Test ...... ... ........... ...... ...... ... .. ... ........ .. .............. .. . 18 

4.5. Vector Error Correction Model ....................................................................................... 19 

4.6. Granger causality test ...................................................................................................... 21 

4.6.1. Block Exogeneity Wald Tests .. .. .. .. ....... .... ...... .. .. .. ... ....... .............. ... ... ... ..... .. ; ................ 21 

4.6.2. Dumitrescu-Hurlin Panel Granger Causality Test ............. .. .. ..... ... ..... .... ................ .... .... 22 

CHAPTER FIVE: CONCLUSION ....................................................................... 23 

4.1. Summary of findings ........................................................................................................ 23 

4.2. Recommendations .......................................................................................................... 23 

4.3. limitation of study ........................................................................................................... 23 

REFERENCES ................................................................................................ 24 

~F»J»I:~[)I>C ............••..........•.......•..•......•..••...•....••.•...............•.••.••.....•••.........•• 2E; 

LIST OF TABLES 

Table 1: summary of data ..... .. ... .. ..... .. .................. ..... .. .... ... .. ................ .... .. .... ..................... ... 11 

Table 2: Descriptive statistics ........... ............... .. ...................................................................... 14 

Table 3: Correlation matrix ... .. ........... .. ......................... .. ........... .... ....................... .. ... ....... .. .... . 15 

Table 4: unit root test ........................ ........................................ .. ............................................ 16 

Table 5: Pedroni Residual Cointegration test ... .......... .... ... ......... .... ............ ............................. 17 

Table 6: Kao Residual Cointegration test.. ............................................................................... 17 

Table 7: Johansen Fisher Panel Cointegration Test .. .. .. .. ........................... .. ..... .. .... .. .............. 18 

.Table 8: Short-run adjustment ...... ..... .. ..................... ... .............. .. ........................................... 20 

Table 9: Block Exogeneity Wald Tests .. .... .. ............................................................... ..... .. ... .. .. 21 

Table 10: Dumitrescu-Hurlin Panel Granger Causality Test .. ...................................... ............ 22 

iv 



l 

.l 

J 
_ I 

LIST OF SYMBOLS 

CPI -Consumer Price Index 

PPI- Producer Price Index 

GMM -Generalized Methods of Moment 

CLRM - Classical Linear Regression Model 

ADF- Augmented Dickey Fuller 

VAR- Vector Auto Regression 

VECM -Vector Error Correction Model 

v 



. I 

, I 

~ I 

.I 
,_ I 

. I 

: J 

.I 
I 

.I 

. J 

J 
_l 

J 

CHAPTER ONE: INTRODUCTION 

1.1. Background of the problem 

Sustainable economic growth and economic development is a major concern to all 

economies. There is increase in public expenditure in many countries which are unable to 

meet their target revenues {Aiiko & Arowolo, 2010}. This has led to widening of the 

budgetary deficits in which the total expenditure (both on revenue and total accounts) 

exceeds the total receipts of the government. To fill the gap, majority are forced to pursue 

debt either internally or/ and externally. 

Debt plays a vital role in supplementing government budget for development . 

Governments in developing countries are increasingly investing in education, health and 

infrastructure. According to Nelasco {2012}, to increase capital formation and 

industrialization, heavy investments like road construction is required. Furthermore, rapid 

growth in industrialization necessitates heavy import of capital goods such as machinery 

{Nelasco, 2012}. This prompts not only domestic borrowing but also, external borrowing to 

alleviate the deficit in the balance of payments. 

Total public debt has two main divisions, domestic debt and foreign debt. Domestic debt are 

funds that are raised internally through taxation or internal debt by issuing of treasury bills 

and government bonds. However, a high domestic debt may lead to crowding-out of private 

sector thus the need for external debt. According to World Bank (2019}, external debt is 

foreign monetary obligations, that requires payment(s) of interest and/or principal by 

debtor in the future and that are owed to non-residents in other currencies. It is inclusion of 

public debt, public guaranteed debt and private nonguaranteed debt. External borrowing is 

indispensable during situations such as epidemics, pandemics and natural calamities such as 

the 2008 financial crisis, which resulted in mounted borrowing by economies.(Nelasco, 

2012). Therefore, while external debt may foster development in both public and private 

sectors of a country, there is need for caution as it may be counterproductive . 

A situation in which external debt cannot be sustained poses a risk to the economy of a 

country as it may cause some countries to default especially during pandemics or its 

1 



.I 

J 
. I 

servicing may lead to debt overhang. This may stunt or decline economic growth as less 

investment is made in education, health and infrastructure sectors of developing countries. 

This becomes a source of weakness when debt is not being utilized productively. The 

aggregate demand by consumers increases which causes inflation. 

Inflation being a key macroeconomic indicator provides insights on the state of the 

economy. The continuous and persistent increase in general prices in an economy decreases 

the purchasing power of money. This can reduce the standards of living of people over time. 

Therefore, it poses undesirable effects hence a global concern and threat to all economies 

(Asmamaw, 2011). 

The focus of this study is the impact of external debt on inflation on three East Africa 

countries, Kenya, Tanzania and Uganda, that are considered to be among the top fifty 

countries in the world which are highly indebted to China, according to Brookings Research 

Firm based in US. According to East Africa Economic Outlook (2019}, current account deficits 

and external debt posed a downside risk to economic growth. Moreover, the report 

indicated that the debt owed to China has expensive terms which may cause some countries 

to default payment when it fell due. There are two opposing views in this study. One view 

suggested that external borrowing has inflationary effect while the other suggested that it is 

deflationary in short run dynamics and long run dynamics. This study seeks to establish 

where the three countries lie in these views. 

1.2. The statement of the problem 

External debt is expected to bring a positive impact to the economy of developing countries. 

However, this may not be the case as Reinhart & Rogoff (2010) in their evaluation of debt, 

economic growth and inflation, showed that economic growth detoriates faster with 

increase in external debt levels than with increase in total debt. 

However, borrowing is inevitable in both developed and developing countries. The domestic 

debt markets in some states are less developed hence, the need to rely on foreign debt to 

bridge the domestic gap(Halima, 2013). This is because their domestic debt market is 

underdeveloped and does not match the government financial needs. Consequently, a large 

proportion of their total debt is composed of external debt(Halima, 2013) . 

2 



.I 

I 

·. I 

.. J 

J 
J 

In the context of East Africa, countries in this region are low-income, developing countries 

hence a large portion of their debt is foreign. Through the East African Community (EAC), 

some of the countries in this region can trade with less restrictions. However, this may be 

risky as changes in economic circumstances in one of these countries may affect other 

countries may affect countries in EAC and extend to countries in the East African region. The 

Mexican Peso crisis which occurred in 1994 is an example of how one country's economic 

circumstances affects or spreads to other countries in the region. 

Several studies have analysed the effect of external debt on inflation with regards to their 

relationship. This study, therefore, sought to investigate the long-term effects of external 

debt on inflation in Kenya, Uganda and Tanzania. 

1.3. Research objective 

• The objective of this study is to investigate the effect of external debt on inflation in 

Kenya, Tanzania and Uganda. 

1.4. Research questions 

1. What is the long-term effect of external debt on inflation in Kenya, Tanzania and 

Uganda? 

1.5. Significance of the study 

Kenya, Uganda and Tanzania have increasingly accumulated debt over the years with the 

amount of outstanding foreign debt being a cause for concern about the future 

sustain ability of these developing countries. Moreover, the weak arrangements by 

institutions and the regulatory framework governing debt management, makes it important 

to establish if there is a causal relationship between external debt and inflation in the short 

run and long run(Halima, 2013). This will help provide a prudent approach to managing 

debt . 

Furthermore, few studies have concentrated on external borrowing, if its inflationary or 

deflationary in East African countries. This study is of great importance to policy makers and 

macroeconomists to understand how governments should implement monetary and fiscal 

3 



--, 
• 

policies and efficiently manage external borrowing especially to foreign countries or 

institutions with expensive terms. 

1.6. Scope and organisation of the study 

-j The study is divided into five sections. Chapter two reviews both theoretical and empirical 

literature done on previous studies . Chapter three will entail the type of data and its 

source, and the methodology the study aims to use. Chapter four will discuss the empirical 

findings while chapter five will encompass the conclusions of the study's results. 

'l 

! 
. .l 

. i 

-I 

~-' 

-j 

' ·' 

I 

I __ J 4 



_j 

I 

) 

CHAPTER TWO: LITERATURE REVIEW 

Introduction 

In this chapter, the study delves into the theoretical issues surrounding external debt and 

inflation and concludes by highlighting empirical literature on the relationship between 

external debt and inflation in developing countries. 

2.1. Theoretical literature 

2.1.1. Demand-pull inflation theory 

The Demand-pull inflation theory states that price levels rise due to aggregate demand in an 

economy being greater than aggregate supply. Several factors cause the demand driven 

inflation. A growing economy encourages consumers to spend more money and be 

confident that their wages will increase. Consumers become inclined to borrow money and 

save less today. This increases general demand, and many companies may have difficulty 

keeping pace with the growing demand. Moreover, overexpansion of the money supply 

brings out the idea of" too much money chasing too few goods". Printing of too much 

money by the government to meet its financial requirements may cause demand pull 

inflation. 

2.1.2. Cost-push inflation theory 

The Cost push inflation theory states that price levels rise due to increase in cost of 

production caused by shortage or scarcity of capital or monopolistic segments of the 

economy driving wages above average levels. Keynesian suggests the same idea of costs of 

production causing inflation but notes that future price levels would be constant if it were 

determined by past and expected price levels. 

2.1.3. Monetary theory of inflation 

The monetary view is based on the quantity theory of money borrowed from 

Friedman(1968}, that "inflation is always and everywhere a monetary phenomenon". 

However, this does not exclude the role of non-monetary variables in determining inflation. 

5 



'l 

l 
' 1 

, I 

. I 

I 

, I 

" ] 

.I 

.I 

. I 

. I 

. j 

I 

. j 

j 

. j 

. j 

I 

The theory assumes that inflation is affected by changes in money density due to changes in 

aggregate demand and aggregate supply. Furthermore, it is believed that the additional 

demand for money generated by expectations by consumers on increased inflation or 

increase in utility prices by government adds to inflation(Aii, 1996). This generated demand 

is met by either increase in velocity of money stock or accommodating increase in money 

supply(Aii, 1996} . 

From the above theory reviews, ingredients that cause price escalation are demand and 

supply disturbances, monetary shocks and inflation expectations. Moreover, theories of 

demand-pull and cost-push inflation have similar results, increasing price levels, but have 

different inherent sources. It may therefore be difficult to distinguish both cases in reality. 

2.2. Empirical literature 

The following section provides a review of some of the empirical literature that has been 

done in the areas of external debt and inflation for cross country reviews and country 

specific analysis reviews . 

2.2.1. Cross country analysis 

Reinhart & Rogoff (2010) explored the relationship of total debt, economic growth and 

inflation in 20 developed economies and 24 developing economies. The study grouped data 

into four categories according to ratio of public debt to GDP during a particular year for 

each economy: low debt, medium debt, high debt and very high debt. The ratios showed 

that the link between debt and growth is relatively weak for countries with "normal" debt 

levels while those with very high debt levels have growth rates about one percent less than 

those with "normal" debt levels in both advanced and emerging economies. However, the 

similarity in both economies ends there as the study indicated that there is no significant 

correlation between high debt levels and inflation in advanced economies contrast to 

emerging countries where there is a casual relationship between high debt levels and 

inflation. The authors further evaluated external debt in emerging economies since most 

countries have a high dependence on external borrowing. The growth level detoriated faster 

6 



.I 

I 

J 
.l 

with increase in external debt levels than with increase in total debt at external debt levels 

above 60%. When external debt exceeds 90% emerging countries showed a declining 

growth rate and a significantly high inflation rate. 

(Nguyen, 2015) This study used PMG (Pooled Mean Group) estimation and panel 

differenced GMM Arellano-Bond regression to examine the impact of public debt on 

inflation with control variables in 15 Asian developing countries for the period 1990-2012. 

The control variables are money supply, private investment, GDP per capita, budget 

revenue, government investment, government current expenditure and trade openness. 

The results from the GMM regression indicate public debt, economic growth and trade 

openness have a positive significant impact on inflation, therefore, increase in the above 

independent variables increases the general price levels while increase in first lag of 

inflation and private investments causes general prices to reduce(deflation) . However, the 

GMM regression did not consider stationarity of variables hence PMG-error correction 

model is employed to check robustness of the GMM estimation. Results from wasterlund 

panel co-integration test indicates that all the independent variables are co-integrated with 

inflation. PMG estimation indicated that in the short run, public debt is deflationary. It also 

confirmed that the GMM regression is consistent and robust as in the long run, the impact 

of public debt, economic growth and private investments are similar. However, the study 

did not consider if there is a unidirectional or bidirectional causality between external debt 

and inflation. 

(Karakaplan, 2009) The author conducted a research based on two hypotheses: external 

debt has less effect on inflation when financial markets are well developed and the effects 

of the determinants of inflation are heterogeneous across countries. The study employed a 

GMM technique to prevent estimation problems of the unbalanced data of 121 countries 

for the period 1960-2004. The results support the first hypothesis of external debt having 

less impact on inflation in countries with well-developed financial markets if, the effects of 

the determinants are assumed to be homogeneous across countries. However, the results 

from the first hypothesis cannot be generalized as the second hypothesis that suggests the 

relationship are heterogeneous is supported by the presence of different coefficients of 

variables across the countries. Despite this study not looking at long run inflationary effects, 

it indicates that impact of determinants of inflation differ across countries which is in line 

7 



1 

:I 

. j 

j 

. J 

.l 
l 

with the case of this study: what is the long run relationship between external debt and 

inflation in Kenya, Tanzania and Uganda. 

2.2.2. Country specific analysis 

Ahmad, Sheikh & Tariq(2012) explored how domestic debt and servicing of this type of debt 

affected inflation in Pakistan for the period 1972 to 2009. This study used two econometric 

models to study the ingredients of price escalation . These ingredients are demand and 

supply disturbances, monetary shocks and inflation expectations which are based on 

theoretical stances on inflation. The first model has five independent variables, total 

domestic debt, money supply to show consistency with Expectation of quantity theory of 

money and demand-pull inflation effects are shown by variables of private investments, 

exports and government expenditures. The variables were found to have a significant 

impact on inflation. Domestic debt and money supply had a positive effect on inflation with 

the latter being consistent with the quantity theory of money. The variables export and 

government purchases are consistent with the demand-pull inflation theory. However, 

private investments are negative hence, does not contribute to aggregate demand thus 

contradicting the demand-pull inflation theory. The second model has four variables, 

domestic debt servicing, indirect tax in light of cost push inflation, exchange rate under 

theory of devaluation and budget deficit according to neoclassical and Keynesian 

framework. All variables are positive and statistically significant except for exchange rates 

which is statistically insignificant. Also, the effect of debt servicing on inflation is very small 

.J but it has a high statistical significance. This study covers a vast arena on the ingredients of 

inflation but fails to tell us how these ingredients respond in the short run and long run. 
I 
I 

(Mweni, Njuguna & Oketch, 2016) The study employed an ordinary least square regression 

) to establish the relationship between external debt on inflation in Kenya for the period 
. l 

.I 

I 

. .J 

.l 

1972-2012. The authors carried out various tests on the data to ensure it is consistent with 

classical linear regression model (CLRM) assumptions. Using the Augmented Dicky 

Fuller(ADF), external debt is stationary at first difference while inflation is stationary at level. 

The study regresses balance of trade, exchange rate, foreign reserves, total investment, GDP 

growth rate and inflation on external debt at their stationary levels. The Breusch­

Pagan/Cook-Weisberg test indicated that the variance of the residual term of the model is 

8 



' j 

I 

. I 

.I 
' l 

J 
. I 

constant hence homoscedastic. Moreover, the Lagrangian multiplier test indicated that the 

autoregressive conditional heteroscedasticity(ARCH} effect on data is not significant at lag 

one. Furthermore, the data does not suffer from serial correlation as the first and second 

order autocorrelations are insignificant. Having met the CLRM assumptions, the authors 

establish that increase on inflation rate increases the level of external debt in Kenya. 

However, the authors fail to establish if inflation helps in prediction of level of external debt 

or vice versa. 

Ekinci (2016} used a simple linear regression to analyse how both consumers and producers 

are affected by external borrowing in terms of inflation in Turkey from 2003 to 2015. The 

study has two regression models each with different dependent variables, CPI and PPI, and 

the same explanatory variable, external debt. The author carried out various tests to 

ascertain that the data is consistent with the Classical Linear Regression Model(CLRM) 

assumptions. The Augmented Dickey Fuller (ADF} test showed that CPI, PPI and external 

debt are stationary at first difference. The Lagrangian multiplier test for autocorrelation 

indicated that the variables do not suffer from serial correlation when a lag length of 2 is 

used. Granger causality test done through VAR analysis shows that only CPI can help predict 

PPI and no relation between other variables. The white heteroskedasticity test indicated 

that the error variance term is constant for all observations. Having met the CLRM 

assumptions, the authors establish that the strong positive correlation between CPI and 

external debt is similar to the case of PPI as the R-squared of both models are quite high, 

96.6%(CPI} and 97%(PPI}. Despite the study conducting a VAR analysis to determine 

causality, it does not examine in the short-run and long-run relationships. 

2.3. Summary of literature 

The theoretical literature looked at some theories, the demand-pull inflation and cost push 

inflation and how some factors relate to these theories. This study includes two other 

independent variables, growth of gross domestic product and growth of broad money and 

analysed if they helped in explaining the relationship between external debt and inflation. 

These two variables are factors identified to cause a demand driven inflation. Furthermore, 

the monetarist view suggested that changes in demand and supply changes money supply 

9 



I 
. I 

. J 

. ) 

. ) 

J 
. ) 

that affects inflation. Therefore, it suggests a positive relationship between money supply 

and inflation. 

The empirical analysis covered six published works of which all involved analysis of 

developing countries except for Growth in a time of Debt (Reingart & Rogoff, 2010) and the 

conditional effects of external debt on inflation (Karakaplan, 2009) which analysed both 

developed countries and developing countries. It is clear that empirical analysis for both 

cross-country analysis and country specific analysis have consistent results for relationship 

between total public debt and inflation, domestic debt and inflation, and external debt and 

inflation. The results showed positive correlations of the three types of debt with inflation. 

This study adopted a model, vector error correction model, contrast with those presented in 

the empirical review so as to answer the research question on the long-term relationship 

and the direction of causality between external debt and inflation in Kenya, Uganda and 

Tanzania. 

10 



. ' 

:.1 
l 
I 

. ) 

I 

CHAPTER THREE: METHODOLOGY 

Introduction 

This section included the approaches that were undertaken to prepare and study the data, 

including comprehensive description of the variables, the tests and the models used in the 

analysis of the data. 

3.1. Research design 

The study took a causal research design, to investigate the effect of external debt on 

inflation. This study included other independent variables, broad money growth rate and 

analysed if they help in explaining the relationship between external debt and inflation, 

which involved quantitative analysis. 

3.2. Population and sampling 

This study used secondary data. All the data was obtained from the World Bank 

International Financial Statistics and Data Files. This study analysed annual panel data 

spanning 1988 to 2018 on the following macroeconomic variables: inflation rate, external 

debt stocks, and broad money growth rate. This study picked a sample of data from 1988 to 

2015 for Kenya, Uganda and Tanzania. The table below summarizes the variables: 

Variable 

external debt stocks 

inflation rate 

broad money growth rate 

Description 

Total external debt stocks as a percentage 

of Gross Nationallncome(GNI) 

Changes in the consumer price index 

It is the sum of currency outside banks and 

proxy for money supply 

. ) Table 1: summary of data variables 

I 
. I 

J 
.I 

11 



3.3. Data analysis 

3.3.1. Stationarity Test 

This is the first step in analysis of the panel data. A variety of tests are used in this to check 

for unit root such as the Levin-Lin-Chu, Harris-Tzavalis, Fisher-type and lm-Pesaran-Shin. 

These tests are applied to the variables at level to test for stationarity, in order to avoid a 

spurious regression . The importance of this test is to examine whether the variables are 

stationary and determine the order of integration as this is important when deciding what 

model to use. The null hypothesis is defined as; all panels contain a unit root against the 

alternative hypothesis; all panels are stationary. Failure to reject the null implies that the 

data needs to be differenced and tested for stationarity at each differenced level. 

3.3.2. Engle and Granger {1987} Cointegration Test 

After conducting stationarity tests for each of the variables, finding the order of integration 

and specifying optimal lag length, the next step was to determine whether the variables are 

co-integrated or not, using Engle and Granger and Johansen's Cointegration test. 

Cointegration shows existence of long-run relationship between the dependent variable, 

which is inflation in this study, and the independent variables. This test required that the 

variables be not stationary at level but become stationary when they are differenced . This 

test involves testing the null hypothesis that there is no cointegration, against the 

alternative hypothesis that cointegration exists. 

3.3.3. Johansen's Cointegration Test 

For robustness, this study also conducted the Johansen's test for cointegration so as to 

provide for all the weaknesses of the Engle and Granger test which are the Engle and 

Granger test is weak for finite sample data and permits more than one co-integration 

relationship. 

3.3.4. Error Correction Model 

12 



I 

I 

. j 

. j 

J 

After conducting the cointegration tests and finding the existence of cointegrating 

relationships between the dependent and independent variables, a vector error correction 

model is estimated. The error correction model is advantageous as it captures not only the 

long-run equilibrium to but also the short run dynamics and the rate of adjustment to 

equilibrium after a disequilibrium; and it is of the form: 

~inflation t = ao + Ll=o {3i ~X t- i + LJ=o Yj ~inflation t- j + P E t-1 + llt 

Where~ shows the first difference aspect of the model, Et is the estimated residual from the 

selected Engle and Granger cointegration equation while Xit is the vector of exogenous 

variables, including external debt stocks, and broad money growth rate. For stability of the 

system in the model, the P coefficient, which measures the speed of adjustment of inflation 

rate to the value implied by the long run equilibrium relationship should be statistically 

significant. 

3.3.5. Granger Causality Test 

Granger (1969) suggested an approach to measure causality between co-integrated 

variables. This study tested whether there is unidirectional or bidirectional causality 

between external debt and inflation. The test is carried out on in order to determine if 

historic values of one random variable can help predict the future values of the other 

random variable, whether there is unidirectional or bidirectional causality between external 

debt and inflation. When carrying out the test, the implied null hypothesis is that external 

debt does not Granger cause inflation and inflation does not Granger cause external debt. 

13 



~l 

CHAPTER FOUR: DATA ANALYSIS AND FINDINGS 

4.1. Descriptive statistics 

Variables INFLATION RATE EXTERNAL DEBT MONEY_GROWTH 

Mean 14.33156 56.85042 21.75008 

'I 
Median 

I 
8.169021 46.79714 17.71556 

. I Maximum 196.1000 165.8710 117.6640 

.I 
Minimum -0.287509 13.47453 2.931252 

.I 
Std. Dev. 22.16245 37.61375 16.72408 

: l Skewness 6.224646 1.299648 2.742660 

: I Kurtosis 50.14035 4.017277 ~4.01860 

I 
.l Jarque-Bera 9211.640 30.19088 587.0562 

.I Probability 0.000000 0.000000 0.000000 

.I Observations 93 93 93 

.. l Table 2: Descriptive statistics 

I As shown in table 2 above, inflation was quite high at an average of 14.33 percent 
• I 

J over the sample period. The standard deviation was also high at 22.16 percent 

__I 14 



I 

' l 
: I 
: I 

' l 
. I 

J 

J 

. I 

'• J 

. I 

' J 

__j 

J 

which means that there were price levels fluctuated during the sample period. Over 

the sample period the maximum value of inflation has been surprisingly high 

at 196 percent in Uganda. 

External debt was quite high with an average of 56.85 percent with a standard 

deviation of 37.61 percent which is high. 

Skewness is a measure of the symmetry of the data. If the skewness value is 

between -0.5 and 0.5 the data is fairly symmetrical, and the results showed that 

the data on the three variables were highly skewed 

4.2. Correlation test 

Variables INFLATION RATE EXTERNAL_DEBT MONEY_GROWTH 

INFLATION_RATE 1.0000 

EXTERNAL_DEBT 0.2224 1.0000 

MONEY GROWTH 0.7779 0.3137 1.0000 

Table 3: Correlation matrix 

Table 3 above shows results on the relationship among the variables. Correlation between 

external debt and inflation, and external debt and money growth is low while inflation and 

money growth have a high correlation . 

4.3. Stationarity Test 

The data is first tested for stationarity using the Levin-Lin-Chu and lm, Pesaran and Shin unit 

root tests. The null hypothesis is that panels contain unit root, while the alternative is that 

panels are stationary. The results of the tests are shown in the table below: 

15 



i Series Levin, Lin & Chu lm, Pesaran and Shin 

Level First difference Level First difference 

constant Constant constant Constant Constant Constant Constant Constant 

., and and and and 

trend trend trend trend 

.! INFLATION RATE 0.2729 0.6915 o.oooo··· o.oooo··· o.ooo5*** 0.0034*** o.oooo··· o.oooo*** -

• J EXTERNAL DEBT 0.4053 0.8652 o.oooo··· o.oooo··· 0.6154 0.5372 o.oooo··· o.oooo··· 

I 

'j 

. l 

: l 
I 

. I 

MONEY_GROWTH o.oooo··· o.oooo··· o.oooo··· o.oooo*** o.oooo··· o.oooo··· o.oooo··· o.oooo*** 

Table 4: unit root test p values 

Note: ***, **and *show sigmficance at 1%, 5% and 10% levels, respectively. 

Lags were included in the test so as to eliminate autocorrelation. The Schwarz Bayesian 

Criterion was used as an automatic lag length selection given two as the maximum number 

of lags to avoid losing the degrees of freedom . 

Individual effects and Individual effects, individual linear trends were included in the test 

equations when conducting the unit root test for all the series at level. Both constant and 

constant trend values under Levin, Lin & Chu test fail to reject the null hypothesis of panels 

containing unit root at level except for MONEY _GROWTH. Under the lm, Pesaran and Shin 

. J test, we reject the null hypothesis of panels containing unit root at level in favour of the 

alternative of at least one panel is stationary. For both unit root tests, the null hypothesis of 

panels containing unit root is rejected at first difference. 

16 



I 

I 

. I 

. J 

I 

__I 

.l 

4.4. Cointegration Tests 

All the variables are integrated at order 1, 1(1). These variables were tested for cointegration 

properties to show if they have compatible long run properties. Three panel cointegration 

test were used, Pedroni Residual Cointegration test and Kao Residual Cointegration Test 

which are Engle-Granger based tests and the Johansen Fisher Panel Cointegration Test, a 

system-based test and a superior test of the three as it permits more than one co integration 

relationship. 

4.4.1. Engle-Granger based Test 

Trend assumption 

coi ntegration 

statistics 

probabilities 

No deterministic 

trend 

9/11 

Table 5: Pedroni Residual Cointegration test 

Deterministic 

intercept and trend 

8/11 

No deterministic 

intercept and trend 

10/11 

An automatic lag length selection, Schwarz Bayesian Criterion (SIC), was used with a 

maximum lag length of two. From table above, null hypothesis of no cointegration was 

rejected under all trend assumptions. 9 out of the 11 cointegration statistics are below the 5 

percent significance level when individual fixed effects are included. 8 out of the 11 

cointegration statistics are below the 5 percent significance level when both individual fixed 

effects and trends are included while 10 out of the 11 cointegration statistics are below the 

5 percent significance level when no regressors are included. The variables, 

INFLATION_RATE, EXTERNAL_DEBT and MONEY_GROWTH, were cointegrated . 

17 



. j 

.I 

.I 

.I 

.J 

.I 

t-Statistic Prob. 

ADF -9.823736 0.0000 

Table 6: Kao Residual Cointegration test 

An automatic lag length selection, Schwarz Bayesian Criterion (SIC}, was used with a 

maximum lag length of two. From table above, null hypothesis of no cointegration was 

rejected given that the statistic is below the 5 percent significance. The variables, 

INFLATION_RATE, EXTERNAL_DEBT and MONEY_GROWTH, were cointegrated. 

4.4.2. Johansen Fisher Panel Cointegration Test 

Further evaluation of the existence of long-run equilibrium relations is estimated using the 

Fisher (combined Johansen) test which incorporate both the trace test and maximum eigen 

test. 

Unrestricted Cointegration Rank Test (Trace and Maximum Eigenvalue) 

Hypothesized 

No. of CE(s) 

None 

At most 1 

At most 2 

Fisher Stat.* 

(from trace 

test) 

31.81 

8.388 

14.83 

Individual cross section results 

Pro b. 

0.0000 

0.2111 

0.0216 

18 

Fisher Stat.* 

(from max­

eigen test) 

30.51 

4.932 

14.83 

Pro b. 

0.0000 

0.5526 

0.0216 



'l 

.l 

.I 
: l 
: l 

.I 

. j 

. I 

.I 

! 

.J 
I 

. I 

Cross Section Trace Test Prob.** 

Statistics 

Hypothesis of no cointegration 

Kenya 39.3332 0.0030 

Tanzania 24.9617 0.1628 

Uganda 46.6544 0.0003 

Table 7: Johansen Fisher Panel Cointegration Test 

Max-Eigen 

Test 

Statistics 

30.0984 

15.3068 

34.3349 

Pro b.** 

0.0021 

0.2680 

0.0004 

The result displayed in table above were obtained when the linear deterministic trend 

(unrestricted) was included in the cointegration test. Both Johansen Fisher trace and 

maximum eigen test reject the null hypothesis of no cointegration indicating that there is 

long run panel relationship between the variables. These results are consistent when the 

trend assumptions; no intercept or trend, intercept no trend, intercept and trend are used 

as shown in APPENDIX 2. The result showed that there is one cointegrating relationship with 

a 5 percent critical value among the series. The individual cross sections reveal the same 

result for the rank tests except for Tanzania . 

4.5. Vector Error Correction Model 

The vector error correction model requires that the variables be integrated at order 1 and 

co integrated. 

D(INFLATION_RATE) = C(1)*( INFLATION_RATE(-1) + 0.1511 *EXTERNAL_DEBT(-1) + 

0.1288*MONEY_GROWTH(-1)- 19.3799) + C(2)*D(INFLATION_RATE(-1)) + 

C(3)*D(INFLATION_RATE(-2)) + C(4)*D(INFLATION_RATE(-3)) + C(5)*D(INFLATION_RATE(-4)) 

19 



. ) 

. l 

1 

' t 

I 

1 

' 

'l 
\ 

. l 

I 
l 

, I 
J 

+ C(6)*D(INFLATION_RATE(-5)) + C(7)*D(EXTERNAL_DEBT(-1)) + C(8)*D(EXTERNAL_DEBT(-2)) 

+ C(9)*D(EXTERNAL_DEBT(-3)) + C(lO) * D(EXTERNAL_DEBT(-4)) + C(ll)*D(EXTERNAL_DEBT(-

5)) + C(12)*D(MONEY_GROWTH(-1)) + C(13)*D(MONEY_GROWTH(-2)) + 

C(14)*D(MONEY_GROWTH(-3)) + C(lS)*D(MONEY_GROWTH(-4)) + 

C(16)*D(MONEY_GROWTH(-5)) + C(17) 

This equation above shows the error correction model with the coefficient of the first 

difference sum showing the short run dynamics. Thus, in the short run, C (1) is the speed of 

adjustment of the model to equilibrium in the short run which is at a rate of -0.193342 as 

shown in the table below. C(l) is significant as its coefficient is negative and statistically 

significant. Furthermore, all coefficients in the table are insignificant except for first lag of 

INFLATION_RATE, second lag of INFLATION_RATE, third lag of INFLATION_RATE, first lag of 

MONEY_GROWTH and second lag of MONEY_GROWTH. 

Coefficient Std. Error t-Statistic Pro b. 

C(1) -0.1933* 0.05664 -3.4129 0.0008 

C(2) -0.3117* 0.1156 -2.6965 0.0077 

C(3) -0.2861* 0.09694 -2.9515 0.0036 

C(4) -0.1764*** 0.09427 -1.8713 0.0630 

C(5) -0.08628 0.07910 -1 .09078 0.2769 

C(6) -0.01195 0.04630 -0.2581 0.7966 

C(7) 0.09193 0.06310 1.4568 0.1470 

C(8) -0.06172 0.05856 -1 .0539 0.2934 

C(9) -0.01165 0.05543 -0.2102 0.8337 

C(10) 0.08069 0.05277 1.5290 0.1281 

C(11) -0.01045 0.05192 -0.2013 0.8407 

C(12) 0.2443* 0.08036 3.0411 0.0027 

C(13) 0.1521 *** 0.07940 1.9163 0.0570 

C(14) 0.04706 0.07756 0.6067 0.5448 

C(15) 0.1020 0.07796 1.3089 0.1923 

C(16) 0.02698 0.07196 0.3749 0.7082 

C(17) -0.72500 0.6819 -1.0630 0.2892 

. j Table 8: Short-run adjustment 

Note: *, **, ***represent 1%, 5% and 10% significance level respectively 

) 

__j 

20 

.l 



_j 

I 

Long Run Dynamics 

The long run relationship between EXTERNAL_DEBT, INFLATION_RATE and 

MONEY _GROWTH for one co integrating vector for Kenya, Uganda and Tanzania in the 

period 1988-2018 is displayed below. This is the first part of the VECM resultant equation . 

INFLATION_RATE(-1) = 0.1511 *EXTERNAL_DEBT(-1) + 0.1288*MONEY_GROWTH(-1)-
19.3799 

The equation above shows long run elasticities when there is one cointegration equation. 

Increase in level of external debt and money supply is related to increase in inflation, thus, the 

estimated model was able to produce expected results that is consistent. The results displayed 

in equation above show that a one percent increase in EXTERNAL_DEBT, ceteris paribus, is likely 

to increases INFLATION_RATE by 0.1511 percent while a one percent increase in 

MONEY_GROWTH, ceteris paribus, is likely to increases INFLATION_RATE by 0.1288 percent. 

The result above is consistent with the monetary theory of inflation; increase in money supply 

leads to higher prices and inflation. 

4.6. Granger causality test 

4.6.1. Block Exogeneity Wald Tests 

Dependent variable: D(INFLATION_RATE) 

Excluded Chi-sq df Prob. 

D(EXTERNAL_DEBT) 2.697755 2 0.2595 

D(MONEY_GROWTH) 2.484214 2 0.2888 

All 5.044300 4 0.2828 

Table 9: Block Exogeneity Wald Tests 

21 



.I 
'] 

~ I 

: I 

.I 

.I 

. J 

. I 

. I 

.J 

.I 

j 

Also, known as the VEC granger causality test; test on lagged explanatory variables where 

the null hypothesis is that the lagged coefficient(s) are equal to zero. The table below shows 

that the null hypothesis fails to be rejected given the prob-value of the chi-sq statistic is 

greater than the 5 percent significance level. There is no short run causality among the 

variables. 

4.6.2. Dumitrescu-Hurlin Panel Granger Causality Test 

Null Hypothesis: Vl does not homogeneously cause V2 

Vl V2 Pro b. 

EXTERNAL DEBT INFLATION RATE 0.0084 

INFLATION RATE EXTERNAL DEBT 0.3445 

Table 10: Dumitrescu-Hurlin Panel Granger Causality Test 

The Granger Causality test shows that there is unidirectional causality between external 

debt and inflation for the sample data of the three countries. The test null hypothesis that 

external debt does not Granger cause inflation is rejected, and the test fails to reject that 

inflation does not Granger cause inflation. The study therefore concluded that external debt 

helped in the prediction of inflation . 

22 



1 
1 

I 

. I 

. I 

. I 

' ! 

1 

.l 

: I 

: I 

J 
I 

. ' 

I 
... I 

' 

·j 

j 

CHAPTER FIVE: CONCLUSION 

4.1. Summary of findings 

The study conducted an empirical analysis into the relationship between external debt and 

inflation. liquidity measures and top twenty performing stocks of Nairobi Stock Exchange: 

stocks in NSE 20. The study adopted annual panel time-series data under the time spanning 

1988 to 2018. The study utilized cointegration tests as well as granger causality test. The 

results from Engle-Granger and Johansen cointegration approach revealed the existence of 

long-run relationship among the variables. This results were further established under the 

long-run scenario of the VECM estimation output. The VECM result revealed positive 

relationships among inflation rate, external debt and the control variable, money growth. 

However, the short run effects were not statistically significant enough to cause a desired 

change in the short run. Granger causality tests reveal that external debt causes inflation 

homogeneously this means that a crisis in one of these countries may spread the crisis to 

the other countries similar to the Mexican peso crisis that occurred in 1994. 

4.2. Recommendations 

Based on the results, the governments of Kenya, Tanzania and Uganda should effectively 

manage debt to foreign countries and institutions with expensive terms. This is because in 

the sample period of the study, external debt was found to have long term effects rather 

than short terms effects on inflation. This may be detrimental to the country(s) as it may 

limit other countries from establishing relationships with them. Also, external debt is likely 

to cause inflation as shown in chapter four therefore, policy makers should set effective 

debt limits to lower this probability. 

The study leaves room for further exploration on this topic using additional variables such as 

imports and exports. This is because regional integration involves a lot of trading amongst 

the member countries, thus imports and exports are expected to play a major role. 

4.3. Limitation of study 

The study was limited to the period 1988 to 2015 for lack of data, and also limited to Kenya, 

J Uganda and Tanzania for the same reason. 

'J 

~ 1 23 



l 

l 
J 

.l 
_ ] 

l 
.l 
J 
.l 

REFERENCES 

Ahking, F. W., and Miller, S. M. (1985). The Relationship Between Government Deficits, 

Money Growth and Inflation. Journal of Macroeconomics, Vol.l, pp.447-467. 

Ahmad,M,J., Sheikh,M,R., & Tariq,K. (2012). Domestic Debt and Inflationary Effects: An 

Evidence from Pakistan. International Journal of Humanities and Social Science, Vol. 2 No. 

18; October 2012. 

Ajayi, L. B., & Oke, M . 0. (2012). Effect of External Debt on Economic Growth and 

Development of Nigeria. International Journal of Business and Social Science, 3, 12. 

Ali,S, H. (1996). Analysing inflation: Monetary and Real theories. The Pakistan Development 

Review, 35:4 Part //(Winter 1996} pp.761-771. 

Aluko, F., & Arowolo, D. (2010). Foreign Aid, the Third World's Debt Crisis and the 

Implication for Economic Development: The Nigeria Experience. African Journal of Political 

Science and International Relations, Vol. 4(4}, pp. 120-127, April2010. 

Asmamaw,M. (2012). Relationship between Inflation and Economic Growth in Ethiopia. 

Budapest International Research and Critics Institute-Journal (BIRC/-Journal}, Volume I, No 3, 

October 2018, Page: 264-271. 

Cheong, C., Lau, E., Liew, V. K., & Puah, C. (2010). Does debts foster economic growth? The 

experience of Malaysia. African Journal of Business Management, 4(8}, 1564-1575. 

Ekinci,M,B. (2016). External Borrowing and Inflation in Turkey Between 2003 and 2015: A 

Simple Linear Regression Analysis. International Journal of Economics and Financial Issues, 

vol. 6{1}, pages 45-54. 

Granger, C.W.J., Newbold, P. (1974), Spurious Regressions in Econometrics. Journal of 

Econometrics, Volume 2, Issue 2, July 1974, Pages 111-120 

Halima,l. (2015). Effect of external public debt on economic growth: An empirical analysis of 

East African countries. 

24 



J 
I 

Kwon, G., McFarlane, L. and Robinson, W., 2006, Public debt, money supply, and inflation: A 

cross-country study and its application to Jamaica, International Monetary Fund. 

Metin, K. (1998).The Relationship between Inflation and the Budget Deficit in Turkey. Journal of 

Business and Economics Statistics, 16{4}, 412-422. 

Karakaplan,M,U. {2009).The conditional effects of external debt on inflation. 

Mweni,F,T., Njuguna,A., & Oketch,T. (2016). The Effect of External Debt on Inflation Rate in 

Kenya. International Journal of Financial Research, Vol. 7, No.4; 2016. 

Nguyen,B,V. (2015). The effects of public debt on inflation in developing economies of Asia: 

An empirical evidence based on panel differenced GMM regression and PMG estimation. 

The Empirical Economics Letters, 14{4): {April 2015). 

Reinhart, C., & K. Rogoff. (2009). This Time is Different: Eight Centuries of Financial Folly. Princeton, 

NJ: Princeton University Press. 

Reinhart, Carmen M., & Kenneth, S. Rogoff. (2010). Growth in a time of debt. American 

Economic Review, 100 {2}: 573-78. 

25 



l 
l 

.I 

_ I 

: I 

. j 

. J 

J 
I 

APPENDIX 

@ G< ' ,_ < f [~]l_§lj@] liD )0 U' .. '' . "- l ' ~[[][]] liD Grou1r. UNTillED Worl:file: PROJECTOA!I\.,Untitled\ ~[[]~ 
ViM[ Proc[oujmll Prnrt[ tll:no[ frruejj mpte[ ll ~.tl_lt.Jt< [ lpec [ ! [Vi.,,[Proc[rluj..:t[[ Prira[ tllme[ Fr..,.MIJ"ple [llre."'[ lt.lt<[ ll"'' I [ [Vi.,·r[ Pro<[<luj•rtll Print[ tlanr<[ Frrut JI Iample[llrtet[lt.lt<[ l pec[ I 
Pedrolli Re~i<IJ~ Coilltegm~o/1 Te~t 

!>enes: INFlAT!ON_RATE EXTERNAL_DEBT I.IONEY _GROWTH 
Dote: 01/17121 Time: 14:o5 
S;unple: 19ijij201ij 
lndudedoiJServaions: 93 
Cros:i ·SECtion.; indudect :3 
Null H)~otlresi.s : No Cllirrtegmlion 
Trend a> .su n J) J~on : No detenninis~c ~end 

AutOIIL1ticlag IJ!Jigtl i ~eii!C!ionlra.>ed on SIC lli UI a Ill:];( lag of2 
Newey '!lest 01rtolllalic ll8J~irltt r s ei i!C!i O<J and BarUett kemel 

Mtem;E11e l~·poUtesi:r conunou AA cot.lf.'i . (wmJin-dmetlsion) 
Weigllti!d 

stalistic PralL St1ti:;tic Prolr. 
P;umi•St1tis~c 1.596489 0.05.12 0.0096.16 0.4961 
Pallet rtJG{>t:Jtisllc -5.074847 0.0000 .. :J.168972 0.0008 
P;uretPP-!>'t:Jtistic -6.381926 0.0000 -4.304481 0.0000 
P;urel AOF,'lt1~s ti c -6.534872 0.0000 -4.349510 0.0000 

1\ttem:rtM! l~pon~e.sis : irHI>i~aJ i\tR coffi. (lw~vo8r<limffiroion) 

Si1~-~~c PralL 
Group rtm.:JI:J~s tic .. 2_f1122l 0.0034 
Group PP ·StJHs~c -4.910094 0.0000 
GrOll!! ADF~'tltis~c -6.122014 0.0000 

{ ) 

Johansen Fzsher Panel CcLnli!Cr3bon Test 
St~rH!S !tlFLATIOn_RATE EXlERrJ.=·L_OEBT I.!OtlEY_GRO'.'l TH 
Dati!: 01f2 112 1 Tz me: 05 0 ~ 
Sam,cle 1938 20 18 
lncl u<:ec: o~se r .. atlcns. 93 
Trend assurn pt!on Lznelr ae:crmLnLS!LC uenc ( r es!lld~Cl) 

Laos ln!Nvar (1n rzrst azlfNences} 1 1 

H 1 ootnes1:~d FzsMrStJt" FI Sher Slat • 

A 

v 

rJo. or CE(s} (fromU J c~ te sl) Pro:;, (from ma·H: Icen I ·~ sf} 

!!one 1888 O.GO•" 2l1" 
r.rmos t1 3.729 0.7132 371 5 
Armost2 2 570 0 8505 2 570 

• Pro:>atahtles are compute<l usmo Js)·mptotlc Cni·st::uJ.re a•stntutlon 

lnC:r.1CU JI cross section results 

Cross Sec:non 
Trace Tes t 
St~lls cs 

Ho~olhes L s or no co in 1 e 2rat.~o n 

Kent"J ~1 .8832 

Tan:JniJ 3•.70" 2 
U QJ M J 5 1 9863 

Pro~: · 

0 0632 
0 2575 
o oo•9 

H.-~othesls or a: mosl1 cointe(jratlon rei:Jt1onshiE: 
Kenya 10 8.:2 1 0 BS-¢ 1 

Tan.:Jnia 16 20 17 0 " 76> 
ucaMa 1765 13 o 3;78 

H :~o!hesis of at mosl 2 comte~rJtlon retJbonshiE 
Kenra 28297 0 895 1 

TJ.O:.lnl l 5 92 12 0 "70" 
ucanc:a " 5880 0 6502 

• ' l.lacKrnnon-Hauo-Urcnehs { 1 999) 0·'.'"31ues 

P.IJ (· Er<;n Tes t 
S!JtlSbCS 

31 0"11 
18 : 025 
3" 33oo 

8 012> 
10 2805 
13 053.! 

2 8297 
5 92 12 
" 5330 

Pedroni Re~idual Cointegration Test 
Sene.<: INFlATION_ RATE EXTERNAL_DEBT IJONEY _GROWTH 
Dote: 01/17121 Tirne: 14:56 
Sample: 19ij8 2018 
lnduded oll.serv:~~ons : 93 
Cross -5edions indurled: 3 
Null 1-tiilOtlJesis: No coirrtegm~on 
Trend assu111p~on : Detemunis~c intercept and ~end 
Alrtollla~clag IJ!Jigtli ~ eii!C!ionlra.sed on SIC ~iUJalllll lag of2 
Newey·West autolll:JIJC lraod\',irlttr ~ el edionalld Bartlett kemel 

i\ltemalive IJ)~O tl Jesis : conunoJJ i\tR coef~ . (llittlill<lilllen.>ion) 
Weiglrted 

:Jta~s~c Prall. :Jtatls~c 

Panelv~t;Jtistic 0.017121 0.4932 -1.224052 
P;uJetrtr o.:Jt:Jtis~c -3.8:!3850 0.0001 ·2147094 
Pallet PP.:li1tiS~C -li.558053 0.0000 -4003090 
PO!ieiAOF.:lt1tis tic -6.737990 0.0000 -4.027007 

A/lemall'le hypoUresis: indilid1Jal i\tR coefo. (lret,,een<limen.sion) 

St1tis~c PIOil. 
Group rtro.:Jta tls~c ·1.493493 0.0677 
Group PP.:Jt:J~s~c -4.151686 0.0000 
Group ADF.St1ti:itiC -4.110294 0.0000 

Pr o:~ 

0.0007 
0.71 >1 
08605 

Pro:>." 

0.0091 
0 3399 
00030 

0820• 
0 5839 
0 3229 

0 890 1 
o .:7o.: 
0 6~52 

26 

A 
Pedroni Resi <IJ31 Cointegm~on Test 
!>enes: INFlAT!ON_RATE EXTERNAL_DEBT fJ<JNEY _ GRIYNTH 
D:Jte: 01117121 ·lime: 1457 
Sa111ple: 198ij 2018 
lnduded oll<erv:J~ons : 9:! 
Cross ~ediort>J indudel13 
NullftfpOUJesis:Nocoirrtegr:Jtiorr 
Trend assu1nption: No detemr in i s~cintercept or~elld 

Al rtollla~c lag IJ!JJgtiJ ~etedionlla>i!d on SIC ~iUJanJ:u lag of2 
N ~W~ey JNest autom:IIic IJ an ct,~idth selecto11 and B:unett kemel 

Altemalive tr,~oUJe~is : comnron i\tR coffi. (winlin<liiii i!J JSiO<l) 
V/eiglrtl!d 

Prall. Gtltl.>tiC Prall. st:Jti~~c Prall. 
0.8895 PauelvSt:ltistic 269203:l 0.0036 0.91ll04 0.180.1 
0.0159 Pauel rt ro~}tJti stic .. 5_4 70154 0.0000 -1631222 0.0001 
0.0000 Panel PP.:JI:JtisHc .. 5.704866 0.0000 -4.034886 0.0000 
0.0000 P;uret ADF.:ltJ!i.stic ·-5.826284 00000 -4.070749 00000 

Altemativa lrypottrosis: in<i'rid11al i\tR coefs. (lretween-<limon.sion) 

Stili .'i~C Prall. 
Group rt10.St1tistic ·l816400 0.0001 
Group PP.St11istic -5.74.1103 0.0000 
GrOll)! AOF.:JI.1ti:l~c -!i.!l'i4491 0.0000 

v - ~ 

JotlJnsen Flsner Panel CO-m!et;:rJbOn Test 
Serzes: tiJi:U .. TIOU R.;TE EXTERtF·L DEBT t.IOtJEY GRO'.'nH 
DJte· 011'2 11'2 1 T1rrl e· 05.03 - -
Samcle: 1938 20 18 
lncluC:ed o:>ser·r.,tJons 93 
Trend assumption· Lln eJr actermm•stl c trend 
LJQS 1nter1al (In rlrst CitTetentes} 1 1 

~ 

Unre-stncteC: Co:nt.:cratrcn Rano; Test(Ttace ana I,IJ•imum E1cem·atue) 

H1potnes1.:ea F1stlt:r Stat ' F1 Shfr Stat. • 
llo o!CE(S) (r:om trace tes t) Pro:l. (rrom ma:t-eu:;en tes t) 

r:one- 31.8 1 00000 3.0 S t 
.:..rmost1 8 388 0 2 11 1 " .932 
;..:mosl 2 1: 83 0 0215 1' 83 

• Pro:>J ~zhtle s Jre com~ utea usmo JS)mpto c Chi-square C1stn:~ubon 

lncr,,dua) cross sectiOn results 

Cross Section 
Trace Tes t 
StJ:tshcs 

H.fotnesis of no comteqr.at:on 
KenfJ 39 3332 

Pro:l •· 

0003~ 
Tan.:.aniJ 2..: 95 17 01 628 
UQJndJ ..: 5 65.!..: 0 0003 

H.poth eszs or Jl most1 com!ecration reiJbonsh1p 
Ken fa 9 n.:s o J.:~1 

Tan:an1l 9 6550 0 3082 
uoanaa 12 31 95 o '"23 

H.-Fotneszs or J! most 2 co:ntecrJtLon relationship 
Kenra 2 566 ..: 0.1092 

Tan: .an1J 2 2658 0.1322 
UgJOCa ..: t.:S1 0.0..:17 

··r.facKmnon-Hauc-1-hchehs (1999} p.va!ues 

l.la(-Eicn Test 
StJ!ISbCS 

30 098" 
15 3068 
J.~ 33.~9 

6 658> 
7.3882 
8 173< 

. 2 s;;" 
22658 
" ·'"51 

ProD 

0 0000 
0 . 526 
002 15 

Pro~ .. 

0 0021 
0 2580 
0 000< 

0 5191 
o.••"3 
0 JS13 

0.1092 
0.1322 
o o• 11 

A 



: 1 

' l 

J 
J 

Joh:'ln:;en Rslu~r Panel Cointegr;J.!jon Test 
Seri es: INFLATIQN_RATE EXTERNAL OEBf MONEY GROWTH 
Date: 0 1121121 Time: 06:06 - -
3arnpl~ : 19Slll '2:010 
lrrdrJr1ed Oll~itHV:Jtions: !H 
Treml Ol:isumptirm: No deterministic trend 

I L.l!}.'i lnterv:JI (Ill nr.il dirl'erence:;): 11 

Unrestrtded Colntegmtion R.lnk Te:;t (frace and U:uillllJitl BgermhJe) 

HypoUle:iil.P.d Asher Stat• Rsl!er StJt• 
No. ofCE(s) {from trace ta:it) Pro h. (from m:u.eiQen te:;t) 

None 16.20 0.0000 :34.01 
AttiiO:il1 9.7:!4 0.1:36:1 6.792 
/1Jmos t '2: 11.:11 o.orn 11 .31 

• Proll:lllililies :1re cornputed IJSiiiQ 3.'iymptolic Clil -squ:ne distrilmtion. 

ln c1i~oid!J :ll crO.'iS section res•Jil•i 

Grosll Sedlon 
Tra.ceTe!lt 
Stltistic; 

Hypoth esis or r1 o coilrtegration 

Prou.• • 

Krmy:J 34.7:119 0.0017 
T:Jn.t.mria 20.0216 0 . 1 ~68 
IJ!].JIId.J 44.2953 0.0001 

Hypothesis or at mosl1 coinlegration relationship 
Kenya 6.227"!j 0.40611 

T;J.nt.;vria 9.2017 0.15 71) 
IJg:Jnda 9.9046 0.1194 

Hmothesis of:llmos t 2 coinlegralion relationship 
Keny:J 2.07:j9 0.1764 

Tnrll..1tlla 2.1924 0.'11i:!U 
IJ!J;Jild;'} 2.6650 0.1212 

..... MacKinnon+t:Juq-l.lidlelis (1999) p--.':}!ue-s 

M:u-Bqn fell ! 
St:Jtistics 

2ll.5044 
10.0200 
34.3107 

4.1517 
7.0093 
7.3196 

2.0759 
2.1924 
2.6650 

Pro b. 

0.0000 
0.:1405 
0.0792 

Proll ... 

0.0009 
0.4026 
0.0001 

27 

Jollan . .;en R.:iller P:Jrt el Cointegr.Jtion fe~ t 

Series: INFL../\OON RATEEXTERNAL DEBT MONEY GROWfH 
Dale: 01121121 Ti11ie: 06:05 - -
Sample: 1988 2013 
lnduded ohservations: 93 
Trend assumpDorl: No detenninistic trend ( restridnd corrst;JSl t) 
Lags intP.rvJI (in firs t dllferences): 1 1 

IJnrestrideiJ Coinleqratiou Rank Tes t(frac:e :Utd Uaximum Eigenvalue) 

H~'J] oth eslmd Fisher Stat.• RsllerSt."ll• 
No.oiCE(s) (f'romll:'lce test) PnHJ. (from m;u~gentes l) Prolt 

None 24.54 0.0004 28.57 0.0001 
AJ.most1 4.521 0.60 65 3.6'n 0.7162 
/IJ III O~i l 2 4.60 7 0})!)51 4.607 f)}j9~1 

• Pro iJaiJilities are computed using :Jsymptotic Chi ·.squ::~re disbiiJution. 

Cross Section 
Trace Test 
Statistics 

HmotJresi . .; of no coi rrtP.-Qr.ttioll 

Pro!l.-

Kenya 39.0841 0.01 45 
T:Ji1L111i:l 27.4458 0 .. 2lHi9 
Uganda 48.1660 0.0012 

Hypothellls of:U most1 cointegmtiOI I relations ltln 
Kenya 9. 7Sl:H 0.61> 11 

Tai!L1flia 11.192'J 0.52:37 
Ugand:l 13.8:308 0.:1012 

H;Jlothesis of :JI It1ost2 cointeqrntion relationsl1iu 
Keny:J :\.0413 0.573{) 

Tm1L111i:'l :l7Til0 0.4461 
Uganda 4.1.W2 0.:3900 

.. l.lad<innon-Ha•JQ·Midlali~ (1999) p~r.llue:J 

M:u-EignTest 
St:Jtistics 

30.1008 
16.2'il:J 
34.3352 

6.7420 
7.4137 
9.6846 

:1.041:! 
1.7708 
4.1462 

Pro!J.•• 

0.0013 
0.2806 
0.0007 

0.7009 
0.6188 
0.36:l9 

0.57:30 
0.4401 
0.1908 



..., 

J 

Vector Error Correction Estimates 

I Date : 02/09/21 Time : 09:25 
Sample (adjusted) : 1994 2018 
Included observations: 75 after adjustments 

I Standard errors in ( ) & t-statistics in [] 

' l 
Cointegrating Eq : CointEq1 I 

INFLATION _R.A.TE(-1) 1.000000 

EXTERNAL _DEBT(-1) 0.151050 
(0 . 05004) 

[ 3.01 852] 

MC>NEY_GRC>VVTH(-1) 0.128757 
(0 . 23129) I [ 0.55668] 

c -19 . 37991 l 
Error Correction : D(IN FL.A.TIC> . . . D(EXTERN.A.. .. C>(IVIC>N EY ··- I 

CointEq 1 -0 . 193342 -0.551438 -0.201021 I (0.05665) (0.08192) (0.08269) 
[-3 .41300] [-6 .73172] [-2 .431 02] 

D(IN FL.A.TIC>N_R.A.TE(-1 )) -0.311752 -0 . 063732 -0 .019660 
(0.11561) (0 . 16718) (0.16876) 
[-2.69653] [-0.38121] [-0.11649] 

D(IN FL.A.TIC>N_R.A.TE(-2)) -0 .286127 0.009547 0.362736 

I (0 .09694) (0 . 14018) (0 . 141 51) 
[-2 . 95152] [ 0 .0681 0] [ 2 .56340] 

D(IN FL.A.TIC>N _ R.A.TE(-3)) -0 . 176408 -0.233811 0.378804 I (0.09427) (0.13632) (0 . 13761) 
[-1 . 87130] [-1 .71518] [ 2 .75282] 

C>(IN FL.A.TIC>N_R.A.TE(-4)) -0.086281 -0 . 078290 0 .315082 
(0.0791 0) (0.11438) (0.11546) 
[-1 . 09078] [-0 .68446] [ 2.72889] 

D(IN FL.A.TIC>N_R.A.TE(-5)) -0.011953 0 . 046378 0.101645 
(0 . 04631) (0.06696) (0.06759) 
[-0.25812] [ 0.69260] [ 1 .50376] 

I D(EXTERN.A.L_DEBT(-1 )) 0.091935 0 .255104 0.033887 
(0 .0631 0) (0 .09125) (0 .09211) I 

[ 1 .45688] [ 2.79561] [ 0 .36789] 

D(EXTERN.A.L_DEBT(-2)) -0.061726 0.171569 0 .001347 
(0 . 05857) (0.08469) (0 .08549) 
[-1 .05396] [ 2.02587] [ 0 .01575] 

D(EXTERN.A.L_DEBT(-3)) -0.011654 -0.173726 -0.157008 
(0 . 05544) (0.08017) (0 .08092) 
[-0.21 022] [-2 . 16709] [-1 .94022] 

D(EXTERN.A.L_DEBT(-4)) 0.080692 -0 . 023502 -0.160603 
(0.05277) (0.07631) (0.07703) 

[ 1 . 52908] [-0.30798] [-2 . 08493] 

.. I D(EXTERN.A.L_DEBT(-5)) -0 .010453 0.022272 0 .035633 
(0.051 92) (0 . 07509) (0.07579) 
[-0.20131] [ 0 .29662] [0 .47012] 

D(MC>NEY_ GRC>VVTH(-1 )) 0 .244395 -0.124834 -0.562399 
(0.08036) (0 . 11621) (0 . 11731) 

[ 3.04115] [-1 . 07423] [-4.79432] 

D(MC>N EY _GRC>VVTH (-2)) 0 . 152164 0 . 098158 -0 . 139873 
(0.07940) (0 . 11482) (0 . 11591) 

_j 
[ 1.91634] [ 0.85488] [-1 . 20679] 

I D(MC>NEY_ GRC>VVTH(-3)) 0.047066 0.071130 -0 .241971 
(0.07757) (0 . 11217) (0 . 11323) 

[ 0 .60677] [0 .63415] [-2.13707] 

C>(MC>N EY _GRC>VVTH (-4)) 0 . 102050 -0 . 095375 -0.029627 
(0.07797) (0.11274) (0 . 11381) 

[ 1.30892] [-0 . 84596] [-0 .26033] 

D(MC>NEY_GRC>VVTH(-5)) 0 . 026983 0 . 098874 0.084854 
(0.071 97) (0 . 1 0407) (0.1 0505) 

[ 0.37494] [ 0 . 95009] [ 0 .80774] 

c -0.725008 -3.091740 -1.845158 
(0 .68199) (0.98619) (0.99550) 
[-1 . 06308] [-3 . 13504] [-1 .85350] 

R-squared 0.510826 0.595766 0 .469725 
Adj . R-squared 0 . 375882 0.484253 0.323443 
Sum sq_ resids 1483. 119 3101 .267 3160 . 107 
S.E . equation 5.056780 7 . 312327 7.381368 
F-statistic 3 . 785453 5.342582 3.211079 
Log likelihood -218 . 3359 -245 . 9983 -246.7031 
.A.kaike .A.IC 6.275625 7.013288 7.032083 

I 
SchwarzSC 6.800922 7.538586 7 .557381 
Mean dependent -0.821972 -3 .705832 -1.345304 
S . D. dependent 6.400893 10. 18210 8 . 973968