Determinants of saving in Rwanda

DETERMINANTS OF SAVINGS IN RWANDA, 1978 – 2012.

AN EMPIRICAL ANALYSIS

By IRAGENA JOYEUSE

A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE

REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN ECONOMICS

OF THE OPEN UNIVERSITY OF TANZANIA

2015

i

CERTIFICATION

The undersigned certifies that has read and hereby recommends for acceptance by the Open

University of Tanzania a dissertation titled: “Determinants of savings in Rwanda from 1978 to

2012: an empirical analysis.” in partial fulfilment of the requirements for the degree of Master

of Science in Economics of the Open University of Tanzania.

___________________________________________

Dr. Felician Lugemalila Mutasa

(Supervisor)

____________________________

Date

ii

COPYRIGHT

No part of this dissertation may be reproduced, stored in any retrieval system, or transmitted in

any form by any means, electronic, mechanical, photocopying, recording or otherwise without

prior written permission of the author or the Open University of Tanzania in that behalf.

iii

DECLARATION

I, JOYEUSE IRAGENA, do hereby declare that this Dissertation is my own original work, and

that it has not been submitted to any other University for a similar degree or any other degree

award.

___________________________________________

Signature

1/15/2015

____________________________

Date

iv

DEDICATION

To my beloved spouse KABERUKA Casimir,

To my beloved daughter, KARIZA IDA Christa

To my late family

v

ACKNOWLEDGEMENTS

It was not that much easy to prepare this research to its finality and get a shape it looks as of

now. Further, I can’t say I could manage the entire process to make it on my own without great

assistance from other persons. In the circumstances, that is the reason I heartedly feel obliged to

acknowledge and appreciate the efforts of all those who played part in one way or the other for

me to undertake the research and prepare this Dissertation to completion.

The list is long, but I could manage to mention only a few persons, but not in the order of their

respective contributions. In the row is my research Supervisor, Dr. Felician Lugemalila Mutasa.

He took all his tireless efforts and support to ensure that I complete this dissertation not only

within time but also at the quality and standard it deserve.

I would also convey my gratitude all Course Lectures for their well organized and dedicated

efforts during our course work session. In addition, many thanks should be directed to my fellow

students of the first batch of M.Sc Economics for their cooperation during the study. Further, my

special gratitude should go to my entire family especially my uncle MWEBAZE R. Emmanuel.

I would like to thank my beloved spouse Casimir, my daughter Christa for their tolerance and

affording to sacrifice their opportune time that would otherwise be spent with me socially if not

in attendance of the M.Sc. Economics programme. It is sincerely hard to leave aside an

acknowledgement the good work of all previous researchers and academicians in the similar

areas, whom I have consulted and in one way or the other used their work as reference materials

in building up my research. I salute them all for their wonderful jobs well done, which had set

the research ball roll forever.

vi

ABSTRACT

The purpose of this study is to examine the determinants of Rwandan's national savings (1978-

2012).

The specific objectives are to determine the major macroeconomic determinants of savings in

Rwanda (1978-2012), to assess the existence of co-integration and causal direction among

determinants of savings in Rwanda (1978-2012) and to find out the relationship between savings

and economic growth in Rwanda (1978-2012)

We utilized Johansen co-integration and Granger causality testing approaches to check the

robustness for long run relationship and Error Correction Model (ECM) for short run dynamics

during the 1978-2012. It is found that the per capita income inversely related with national

saving, in short run and positively related in long span of time significantly. The capital

formation has a positive impact on national saving both in short and long run. The consumption

and interest rate have an inverse relation with savings in short and long run, however, the

inflation has a positive influence on savings both in long and short run. Keynesian and

permanent income hypotheses of income and savings are not valid for Rwanda in short run

because per capita income inverse function of savings at national level. However, Keynesian and

permanent income hypotheses of income and savings are valid in long run in Rwanda.

vii

TABLE OF CONTENTS

CERTIFICATION ........................................................................................................................... i

DECLARATION ........................................................................................................................... iii

DEDICATION ............................................................................................................................... iv

ACKNOWLEDGEMENTS ............................................................................................................ v

ABSTRACT ................................................................................................................................... vi

TABLE OF CONTENTS .............................................................................................................. vii

LIST OF TABLES .......................................................................................................................... x

LIST OF FIGURES ....................................................................................................................... xi

LIST OF ABBREVIATIONS AND ACRONYMS ..................................................................... xii

CHAPTER 1: INTRODUCTION .................................................................................................. 1

1.1. Background of the Study ...................................................................................................... 1

1.2 Statement of the problem ...................................................................................................... 5

1.3 Objectives of the study .......................................................................................................... 6

1.3.1 The general objective ...................................................................................................... 6

1.3.2 Specific objectives .......................................................................................................... 6

1.4 Research Hypotheses............................................................................................................. 6

1.6 Scope of the study ................................................................................................................. 7

1.7 Significance of the study ....................................................................................................... 7

1.8 Organization of the Study ..................................................................................................... 8

CHAPTER 2: LITERATURE REVIEW ........................................................................................ 8

2.1. Introduction .......................................................................................................................... 8

2.2. Definition of Key Concepts.................................................................................................. 8

2.2.1. Savings........................................................................................................................... 8

2.2.2. A Determinant ............................................................................................................... 9

2.3. Theoretical framework ......................................................................................................... 9

2.3.1. Keynes’ Absolute Income Hypothesis (AIH) ................................................................ 9

2.3.2. Modigliani’s Life Cycle Hypothesis of saving behavior (LCH) ................................. 12

2.3.3. Friedman’s Permanent Income Hypothesis (PIH) ....................................................... 14

2.3.4. Deaton’s Theory and Review of saving in developing countries ................................ 15

viii

2.3.5. Saving and Income ...................................................................................................... 16

2.3.6. Saving and Interest Rate .............................................................................................. 17

2.3.7. Saving and Exchange Rate .......................................................................................... 18

2.3.8. Saving and Inflation..................................................................................................... 18

2.3.9. Saving and Fiscal Policy.............................................................................................. 19

2.3.10 Saving and consumption ......................................................................................... 19

2.4. Empirical Studies ............................................................................................................... 21

2.5. Research Gap Analysis ...................................................................................................... 32

CHAPTER 3: AN OVERVIEW OF SAVING BEHAVIOUR IN RWANDAError! Bookmark

not defined.

3.1 Savings Historical Trends .................................................... Error! Bookmark not defined.

3.2 Rwanda’s goals ................................................................................................................... 30

3.3 Rwanda Financial inclusion and saving culture .................................................................. 27

CHAPTER 4: RESEARCH METHODS ..................................................................................... 35

4.1 Conceptual framework ........................................................................................................ 35

4.2 Research Design and Model Specifications ........................................................................ 36

4.2.1 Model Specification ...................................................................................................... 36

4.3. Data Analysis ..................................................................................................................... 40

4.3.1 Unit root test ................................................................................................................. 40

4.3.2 Co-integration test ........................................................................................................ 41

4.3.3 Error correction model (ECM) ..................................................................................... 41

CHAPTER 5: DISCUSSION OF FINDINGS ........................................................................... 42

5.1 Descriptive statistics of variables used in the econometric model ...................................... 42

5.2 Trend of variables in model ................................................................................................ 43

5.3. Empirical results ................................................................................................................. 44

5.3.1 Long-run relationship ................................................................................................... 44

5.3.2 Short run relationship ................................................................................................... 52

5.3.3 Regression Analysis ..................................................................................................... 54

5.3.4 Economic Growth and Savings relationship ................................................................ 56

CHAPTER 6: CONCLUSION ..................................................................................................... 58

6.1. Summary ................................................................................................................................ 58

ix

6.2. General Conclusion ............................................................................................................ 58

6.3. Suggestions......................................................................................................................... 59

6.3.1. Encouraging Economic Growth. ................................................................................. 60

6.3.2. To curb inflation .......................................................................................................... 60

6.3.3. Macroeconomic Stability ............................................................................................. 60

6.3.4. Institutionalized Savings.............................................................................................. 60

6.3.5. Expansion of the Financial Infrastructure and Intermediation .................................... 60

6.3.6. Secure and Diversified Means of Savings ................................................................... 61

6.3.7. Building Capacity and Efficiency of Intermediation ................................................... 61

6.3.8. Increased Awareness and Positive Perception of Tangible Benefit of Savings .......... 61

REFERENCES ............................................................................................................................. 62

APPENDICES .............................................................................................................................. 67

x

LIST OF TABLES

Table 1: Keynesian motives for saving ......................................................................................... 11

Table 2: Current trends of national saving and gross investment in Rwanda ............................... 26

Table 3: GDP Growth, Savings and Investment (as percentage of GDP) ...................................... 2

Table 4: Key EDPRS results indicators ........................................................................................ 31

Table 5: Variables in the model .................................................................................................... 42

Table 6: Unit Root test .................................................................................................................. 46

Table 7: Co-integration analysis ................................................................................................... 48

Table 8: Granger Wald causality tests .......................................................................................... 49

Table 9: Johansen normalization coefficients ............................................................................... 52

xi

Table 10: Vector error correction ................................................................................................. 53

Table 11: OLS Test for gross national savings (a) ....................................................................... 55

Table 12: OLS Test for GDP per capita (b) .................................................................................. 56

LIST OF FIGURES

Figure 1: Income, Consumption and Life-Cycle Saving .............................................................. 13

Figure 2: Rwanda Gross Domestic Savings and Gross National Savings

(%) of GDP (1995-2010) ................................................................................................. 3

Figure 3: Comparison of saving situation of Rwanda and other countries ..................................... 4

Figure 4: Trend of RR, CPI and GCF ........................................................................................... 43

Figure 5: Trend of Gross savings and GDP per capita ................................................................. 43

xii

LIST OF ABBREVIATIONS AND ACRONYMS

∇ : Differencing operator

µ : Residuals

ACET : African Centre for Economic Transformation

ADF : Augmented Dickey Fuller

AIC : Akaike Information Criteria

APS : Average Propensity for Saving

BD : Budget Deficit.

CA : Current Account

CPI Consumer Price Index

DNR : Do Not Reject

DS : Domestic Savings

e.g : Example

xiii

ECM : Error Correction mechanism

EDPRS : Economic Development and Poverty Reduction Strategy

EICV : Integrated Household Living Conditions Survey

FDI : Foreign Direct Investment

FSDP : Financial Sector Development Program

GCF : Gross Capital Formation

GNEP : Gross National Expenditure

GNP : Gross National Product

GNS : Gross National Savings

GoR : Government of Rwanda

GPCI : Growth rate of fixed per Capita Income.

GPD : Gross Domestic Product

H0 : Null Hypothesis

HQIC : Hannan-Quinn Information Criterion

ICOR : Incremental Capital Output Ratio

IMF : International Monetary Fund

INF : Inflation rate

IPAR : Institute of Policy Analysis in Rwanda

LCH : Life Cycle Hypothesis

M2 : Money supply

MDGs : Millennium Development Goals

MFI : Microfinance Institution

MINECOFIN: Ministry of Economy and Finance

MPC : Marginal Propensity for Consumption

NBR : National Bank of Rwanda

NGOs : Non-Governmental Organizations

PCY : GDP per capita growth

PIH : Permanent Income Hypothesis

PP : Phillips–Perron

RR : Real Interest Rate

SACCO : Saving and Credit Cooperative

xiv

SBC : Schwarz Bayesian Criteria

SNA : Systems of National Accounts

Std. Error : Standard Error

UBPR : Union des Banques Populaires du Rwanda

UN : United Nations

UNCTAD : United Nations Conference on Trade and Development

UNDP : United Nations for Development

VECM : Vector Error Correction Model

WDI : World Development Indicators

1

CHAPTER 1: INTRODUCTION

This chapter aims at providing the background, the statement of the problem, the general and

specific objectives, and hypotheses, the scope, and the significance of the study

1.1. Background of the Study

Rwanda’s savings rate is very low meaning that both domestic and foreign savings are less

mobilized. This indicates that saving instruments are underdeveloped. Due to low levels of

savings, Rwanda depends a lot on foreign assistance and this creates a deficit problem where by

the needed investment in the country cannot be covered by the available savings, MINECOFIN

(2010)

The table below outlines the five year trends in savings and investment in Rwanda. The economy

has been growing strongly, averaging 6.5 percent per annum during this period. Investment

levels have been rising from 13 to 18 percent of GDP. Although private sector investment has

been growing incredibly fast (especially in 2007/08) the majority of investment is still sourced

from the public sector; averaging 12 percent of GDP whilst the private sector averages 4.5

percent.

Over the past five years Foreign Direct Investment (FDI) has been growing sharply from 2 to 26

percent; on average accounting for 10 percent of private sector investment. This reflects the

improvements in macroeconomic stability and the government’s policy to attracting FDI. On the

savings side domestic savings are growing but remain low; averaging around -1 percent of GDP

over the five years. National savings have performed better averaging 13 percent of GDP. The

resulting savings-investment gap has increased over the last five years from -1 to -6 percent of

GDP. This large rise in 2007/08 marks the beginning of the EDPRS where large infrastructure

projects are implemented. This has lead to a rise in domestically and foreign funded investments.

The private net savings gap has been growing smaller over the last five years.

The private sector have benefited from the introduction of financial sector reforms and economic

growth. Data from the National Bank of Rwanda (NBR) 5 shows private sector credit growth of

22 percent in 2007 and around 30 percent in 2008 and bank deposits rising on by 27 percent per

annum on average over the past three years. However, despite strong performance and growth

2

rates savings and investment indicators are still below international and regional averages.

(MINECOFIN, 2010)

Table 1: GDP Growth, Savings and Investment (as percentage of GDP)

Indicators 2003/04 2004/05 2005/06 2006/07 2007/08 5Yr Av

Real GDP Growth rate (%) 2.6 6.5 7.6 7.2 8.5 6.5

Investment 13.1 13.2 16.2 17.7 18.2 15.7

Public 12.0 11.4 14.9 13.2 8.4 12.0

Private 2.6 4.1 1.3 4.6 9.8 4.5

Domestic Savings -3.2 -2.9 0.7 2.1 -1.0 -0.9

National Savings 11.9 11.2 14.6 15.6 11.9 13.0

Savings-Investment Gap -1.2 -2.0 -1.7 -2.1 -6.3 -2.6

Government Net Savings(-

deficit)

3.0 3.3 2.6 1.6 -4.6 1.2

Private Net Savings (deficit) -4.2 -5.3 -4.3 -3.7 -1.7 -3.8

ICOR 5.3 2.2 2.0 2.3 2.0 2.8

Source: MINECOFIN

According to ACET Report (2012), Rwanda’s gross domestic savings as a proportion of GDP

have shown a positive growth trend from negative levels prior to 1998 to a positive level of

about 5% of GDP in 2009. Although domestic savings have grown steadily over time, they are

still less than 10%, which is low when compared to the benchmark countries like Vietnam and

Malaysia whose current savings rates are above 10%. The low rates of domestic resource

mobilisation constitute a major bottleneck to sustaining productivity-driven economic

transformation in the largely informal private sector.

3

Figure 1: Rwanda Gross Domestic Savings and Gross National Savings (%) of GDP (1995-

2010)

Source: MINECOFIN (2010)

The low domestic saving rates in Rwanda have been partly due to a low saving culture, limited

access to banking facilities especially in the rural areas and low incomes which translates into

low savings for a significant portion of the unbanked population. In order to increase domestic

savings and access to finance for the unbanked population the government has rolled out the

SACCO Umerenge program, which is a savings cooperative at grass roots level. Here, local

people put their savings in a cooperative with government supplementing the funds once they

reach a certain threshold. Other members of the community can then borrow from the SACCOs

and pay back in order to sustain the revolving funds under the SACCO Umurenge program

(IPAR, 2012).

4

Figure 2: Comparison of saving level of Rwanda and other countries

Source: MINECOFIN (2010)

The study relied on econometric methodology to investigate the determinants of savings. Indeed,

it is thought that a certain degree of precision and accuracy might be brought about when using

such a technique.

Because of the Occam’s razor principle, the research will not try to identify all determinants of

savings in Rwanda. Only the main ones will be found. Indeed, according to Gujarati (2006: 46-

47), following Occam‘s razor, we would like to, keep our regression model as simple as possible.

Occam’s razor principle of econometrics implies that an econometric model should not be a

perfect representation of reality.

Furthermore, Rwanda takes a developmental state approach with the key objective being

sustainable economic growth and social development. The main aim of EDPRS was to overcome

the key constraints to economic growth identified through a growth diagnostic and investment

climate analysis by: systematically reducing the operating costs of business; investing in the

private sector’s capacity to innovate; and widening and strengthening the public sector.

5

Government policy is to promote private sector investment through good governance, a legal

framework, promoting savings and the banking sector and investment in infrastructure, health

and education including vocational training. IPAR report (2012)

Rwanda’s gross domestic savings as a proportion of GDP have shown a positive growth trend

from negative levels prior to 1998 to a positive level of about 5% of GDP in 2009 .Although

domestic savings have grown steadily over time, they are still less than 10%, which is low when

compared to the benchmark countries like Vietnam and Malaysia whose current savings rates are

above 10%. The low rates of domestic resource mobilisation constitute a major bottleneck to

sustaining productivity-driven economic transformation in the largely informal private sector.

IPAR report (2012)

In addition, there are many economic indicators that may help predict savings, this study will

limit the list to those that are more closely linked to savings by economic theory and can be

easily measured at least in the context of the availability of data in Rwanda. Thus, in the study

we have been restricted by the availability of data on determinants that are likely to influence

savings in Rwanda between 1978 and 2012 period.

1.2 Statement of the problem

The current reports on Rwanda and official statistics (EICV1, 2, 3), UNDP Reports, World Bank

Data 2012) reveal that, constant Economic Growth (DGP) that the country experienced over

last12 years.

Vision 2020 calls for the achievement of the MDGs and for moving Rwanda out of

underdevelopment and poverty. Its targets for 2020 include: a GDP per capita of $900, an

increase from $230 in 2000; a reduction in the incidence of poverty from 60 per cent to

25percent; an increase in life expectancy from 49 years to 65 years; and an increase in adult

literacy from 48 per cent to 90 per cent, MINECOFIN (2010:3)

Moreover, vision 2020 assumes that the population will grown at an average annual rate of 2.7%,

that the will be 4.3 and Rwanda hopes to be a middle income country with per capita GDP of

USD 900. But analysis on the current EDPRS projections and finds that savings had been

underperforming for several reasons and unless the savings-investment gaps can be reduced, it

may be difficult to reach the goals, MINECOFIN (2010)

6

In addition, according to MINECOFIN(2009:6) saving are still low, for example gross national

saving as a percentage of GDP in 2007 was 12.7%, 15.7% in 2008, 13.6% in 2009 and 10.5% in

2010, and yet local savings are very important for any country’s economic development.

This is against African average savings rate of approximately 18%. Economic theory holds that,

for a country to accumulate savings sufficient to induce growth and thus development, these

have to be above 23% mark of the GDP. This indicates the monumental task we have to

overcome as a nation, if we are latch into sustainable development cycle and break the poverty

trap. In this framework, the study will examine the main determinants of savings in Rwanda

(1978-2012).

1.3 Objectives of the study

The objectives of the study are divided into general and specific objectives.

1.3.1 The general objective

This research aims to examine determinants that influence savings in Rwanda (1978-2012)

1.3.2 Specific objectives

� To determine the major macroeconomic determinants of savings in Rwanda (1978-2012),

� To assess the existence of co-integration and causal direction among determinants of

savings in Rwanda (1978-2012),

� To find out the relationship between savings and economic growth in Rwanda (1978-2012)

1.4 Research Hypotheses

A hypothesis is an anticipated answer to the research question. This provisional answer is

conformed or rejected through observations or experimentations. To guide the study, the

following hypotheses are tested:

H0: The factors that influence savings have no significant determinant in Rwanda (1978-2012).

H0: There is no relationship between saving and economic growth in Rwanda (1978-2012).

7

1.6 Scope of the study

The research as any other scientific work is limited in time, space and in the domain. In time, the

researchers focused the analysis on the period from 1978 to 2012 because the period allowed the

researcher to get data related to the subject. Regarding the delimitation of space, the study

focused on Rwandan territory. In the field, the study focused on Macroeconomics.

1.7 Significance of the study

The fundamental basis of carrying out this research is for academic achievements, being a

prerequisite condition for fulfilment of the Masters of Science Degrees in Economics.

Since savings is one of the channels that spur the economic growth and an important indicator of

economic development where it is viewed as an element which finances domestic investment to

achieve economic growth, we are interested to analyze its determinants.

In the previous researches, some researchers have attempted to undertake relatively similar

studies to assess the impact of savings on the economy. However, it has been learnt that most of

the earlier studies conducted focused on economic developments, or other sectors of the

economy without much touch on the determinants of savings which is a crucial factor for a

country to achieve sustained development especially developing countries like Rwanda. This

Study therefore strives to cover the observed gaps.

Accordingly the research tries to fill in the research gaps left by previous scholars as it has been

revealed that, not much have been studied about the determinants of savings in Rwanda with

such long period. At the end of the Study, it is expected that the results gave light on the

empirical determinants of savings in Rwanda during the period under the study.

Furthermore, the research will also help the policy makers to understand the determinants of the

aggregate savings in order to design a number of policy interventions; from the design of the tax

and social security system to the layout of financial markets regulations in order to achieve

economic development.

8

1.8 Organization of the Study

This study is organized in six chapters. The first chapter is an introductory part. It is composed of

the introduction, background, statement of the problem, objectives of the study, research

hypotheses, the scope of the study, significance of the study. The second chapter provided the

economic theories, empirical literature and research gap behind saving determinants, while the

third chapter focuses methodology. The fourth chapter is on discussion of findings. The fifth

chapter presents conclusion

CHAPTER 2: LITERATURE REVIEW

2.1. Introduction

In this chapter, the researcher sets out to define the key concepts used in the study and to

examine the existing literature relevant to the subject matter in order to ascertain what has been

discovered on the topic under study, which helped the researcher to draw conclusions from the

research findings. Therefore different literatures have been consulted and this helped the

researcher to understand the determinants of savings in Rwanda between 1978 and 2012.

2.2. Definition of Key Concepts

2.2.1. Savings

Before examining the savings theories and savings determinants, it is firstly necessary to have a

closer look at the definition of the term savings. As defined by Wärneryd (1989:516), Savings

means “refraining from consumption during one period in favor of later possibilities for

consumption”. Therefore, saving in general can be seen as an instrument of financial precaution.

According to Richard et al (1991:68), economists define savings as “the part of after-tax income

which is not consumed”. Basing on the above features, saving is the proportion of disposal

income that is directed to the acquisition of financial assets (monetary savings) and to the

9

purchase of durable goods (real saving). Therefore saving corresponds to the systematic

accumulation of capital.

‘Savings’ can mean a variety of different things; it could mean insurance, simple savings

accounts, capital such as livestock or other tangible assets, or could mean putting cash under the

mattress. This research will focus solely on the formal sector of savings, this means the financial

institutions and markets, additionally it also covers the policies and legislation, monetary policy

bodies that control bank and insurance etc. This later part was neglected by many developing

countries for a long time, thus limiting the success of the financial sector. (Hans: 2003).

2.2.2. A Determinant

Hornby (1995:317) defines a determinant as “a thing that decides whether or how something

happens” therefore we say that, the determinants of saving rate are things that decide whether or

how saving happens. There are a number of factors that are termed as the determinants of saving

rate such as the Gross Domestic Product (at constant prices), the inflation rate, the real interest

rate, the real exchange rate, fiscal policy, etc…

2.3. Theoretical framework

In the literature on economic development much of interest in saving has been focused on the

relation between saving and growth. But saving is not only about accumulation. It is about

smoothing consumption in the face of volatile and unpredictable income and helping to ensure

the living standards of poor people, whose lives are difficult and uncertain, Deaton (1989:61).

Keynes (1936) developed early thinking on the subject in the form of the absolute income

hypothesis, which was the followed by Modigliani’s life-cycle hypothesis of saving (1954) and

later by the permanent income hypothesis devised by Friedman (1957).

2.3.1. Keynes’ Absolute Income Hypothesis (AIH)

Keynes (1936 in Modigliani, 1986, 298) views savings in the context of the theory of demand

and the consumption function, and regards income as the main systematic determinant of both

individual and national saving, asserting that the average household’s propensity to save increase

as the household reaches a higher income level.

10

Saving is therefore regarded as one of the many ‘goods’ on which a consumer can spend his

income. In contrast to other goods, the ‘expenditure’ on saving could however be negative, with

the result that dis-saving is seen as typical of people or countries below a certain ‘break-even’

level of income. The most common form of the Keynesian savings function is linear with a

constant marginal propensity to save (MPS), as follows:

S = a0 + a1Yg (2.1)

Where:

S = Gross Domestic Saving

a0= the intercept (with a0<0)

a1= Constant Marginal Propensity to Save (with 0 <a1< 1)

Yg= Gross National Product.

With the result that, as the level of income rises, the average propensity to save (APS) will also

increase. However, if the intercept a0 is positive or a1 is negative, then the APS will decrease

with increase in income (Mikesell and Zinser, 2001).

Furthermore, Keynes psychological law expressed that “(…) men are disposed, as a rule and on

the average, to increase their consumption as their income increases, but not by as much as the

increase in their income” (Keynes, 1936 in: Wärneryd, 1989, p. 523). Concerning saving

behavior, this means that an increase in income will also lead to an increase – although not to the

same extent as the income increase – of savings.

Keynes (as quoted and re-classified in Browning and Lusardi, 1996, 1797) identifies eight

motives for saving:

11

Table 1: Keynesian motives for saving

Keynesian motives ( 1936) Browning and Lusardi (1996)

‘To build up a reserve against unforeseen

circumstances’

‘ The pre-cautionary motive’

‘To provide for an anticipated future

relationship between the income and the needs

of the individual…’

‘The life-cycle motive’

‘ To receive interest and capital appreciation’ ‘The intertemporal substitution motive’

‘To enjoy a gradually increasing level of

expenditure’

‘The improvement motive’

‘To enjoy a sense of independence and the

power to do things, though without a clear idea

or definite intention of specific action’

‘The independence motive’

‘To secure a masse de manoeuvre to carry out

speculative or business projects’

‘The enterprise motive’

‘To bequeath a fortune’ ‘The bequest motive’

‘To satisfy pure greed, i.e. unreasonable but

insistent inhibitions against acts of expenditure

as such( avarice)’

‘The avarice motive’

Source: Browning and Lusardi (1996:1797)

It is interesting to note that Browning and Lusardi, in research dated almost half a century after

Keynes’ work, add only a single further motive in the form of ‘the down-payment motive (to

accumulate deposits to buy houses, cars and other durables), thereby underscoring the continued

validity of Keynes’ saving motives as an explanation of economic agents savings.

12

2.3.2. Modigliani’s Life Cycle Hypothesis of saving behavior (LCH)

This hypothesis has been the centerpiece of the modern, mainstream theory of saving behavior

and is based on the core idea of individual utility maximization over the life time. Modigliani

and Brumberg (1954:65) propose the consumption theory known as the life-cycle hypothesis of

saving, which assumes that individuals make rational choices and plan their consumption and

savings behavior by considering their needs at different ages. The basic underlying assumption is

therefore that individuals spread their lifetime consumption evenly over their lives; by

accumulating savings during earning years and maintaining consumption levels during

retirement (Muradoglu and Taskin, 1996).

The LCH therefore emphasizes the importance of the long-term income over the life cycle of the

individual and the savings behavior of individuals who ‘spread their lifetime consumption over

their lives by accumulating savings during early years and maintaining consumption levels

during retirement’(Nga.2007:11). In comparison to Keynes’ absolute income hypothesis, the key

savings motive is to acquire wealth and monies for retirement. A fundamental assumption of the

LCH is therefore that households are forward-looking and will try to maximize total utility as a

function of current and future consumption.

The LCH predicts that the household saving rate will be a function of both ‘the growth rate of

per capita income and the age structure of the population’ (Horioka and Wan. 2007:2084), and

that an increase in the young dependency ratios (namely the ration of young economically

inactive individuals to older, economically active individuals) will have a negative impact on the

saving rate.

Therefore, this model implies that in a society with a stationary population and income there

would be no aggregate net personal saving, as the dis-saving of the retired would offset the

saving of employed. However, in a society with a growing population and/or growing per capita

income, aggregate saving should be positive (Mikisell and Zinzer, 1973)

According to Jappelli & Modigliani (2003:25), the LCH predicts that individuals accumulate

savings while they are younger and dis-save these assets after retirement to maintain and finance

more or less their previous living standard.

13

The positive saving rates – and thus wealth accumulation – during the working span and the

negative saving rates – and thus wealth dissolution – during the retirement phase lead to the

classic hump shape of wealth and asset holdings over the life span.

Figure 1: Income, Consumption and Life-Cycle Saving

Source: Börsch-supan et al (2008), p. 12

Thereby, the total lifespan can broadly be divided into three phases: Education phase, earning

phase and retirement phase, whereby the consumption level is assumed to stay constant over the

life cycle. During the first stage where the income is low, people take up a loan and go in debt

because they assume to be able to pay back these debts later. In their income phase, people pay

back these debts and additionally build up further assets by e.g. saving on bank accounts,

investing in capital life insurance or stock markets. These previous built assets are then

consumed in the third phase of retirement.

The basic LCH underlies the assumption that savings are only made because of the financial

precaution motive. However, Arnds and Bonin (2003) show that the precaution motive

considered in an isolated way is not strong enough to explain the observable saving behavior and

asset building in America, because it is too simplistic and neglects other important determinants

of consumption and saving behavior. Also Börsch-supan (2000a) states that the “elegance of the

basic LCH model stays in a large contrast to its predictive success”.

14

2.3.3. Friedman’s Permanent Income Hypothesis (PIH)

FRIEDMAN (1957:45) proposes a further theory based upon the consumption function, in the

form of the permanent income hypothesis. Similarly to Modigliani’s LCH theory, this hypothesis

stresses the importance of long-term income as the main determinant of household consumption

(Strydom, 2007), but differs from the LHC primarily in that it ‘models rational consumption and

savings decisions under the ‘simplifying’ assumption that life is ‘indefinitely long’ (Modigliani,

1986: 229)

In the Friedman view income consists of two components: “permanent” income and “transitory

income. The permanent income hypothesis may be expressed as:

S= b0+b1YPt+b2YTt (2.2)

Where:

S = saving

b = a constant

YPt is permanent income in year t

YTt is transitory income in year t

According to the equation (2.2), we note that in most extreme b1=0 and b2=1, so all savings

arises from transitory component of income and this entire component is saved. However, there

is a modified version of PIH, which contends that only saving out of permanent income is

constant over a person’s lifetime, but can be positive, and that while the propensity to save out of

transitory income is high; all transitory income may not be saved. The equation to represent this

version is similar to the above but with 0<b1<b2<1.

Friedman draws a distinction between two sources of income, namely:

Permanent income, as defined in terms of long-run expectation over a planning period (Miksell

and Zinser. 1973) and a steady rate of consumption maintained over a lifetime given the present

level of wealth ( Muradoglu and Taskin, 1996); and

15

Transitory income, which is constituted of the difference between actual and permanent income,

with differences arising as a result of temporary influences’ such as ‘a windfall gain or loss’

(Samuelson and Nordhaus. 1995:430)

Friedman assumes that individuals do not consume from transitory income and that transitory

income will therefore immediately be channeled to saving, with the result that the marginal

propensity to save on transitory income will approach unity (Muradoglu and Taskin, 1996).

Pertinently, for developing economies, Friedman (1957, as quoted in Nga, 2007) stated with

reference to income inequality and saving that reduction in the permanent income status is

neutral with reference to the saving ratio ( ceteris paribus).

In this regard, Mikesell and Zinzer (1973:9) confirm that ‘empirical studies for developing

countries show quite divergent marginal propensities to save out of permanent and transitory

income’ and find that ‘most of the studies support the permanent income hypothesis’, namely

that saving out of transitory income is greater than saving from permanent. It can therefore be

argued that an inequality in transitory income increases the need to save, as a result of

uncertainty about income prospects and resultant need to save for emergencies (Nga.2007)

In a retrospective of Friedman’s permanent income theory, Meghir (2004:35) concludes that the

PIH “(…) originates from the basic intuition that individuals would wish to smooth consumption

and not let it fluctuate with short-run fluctuations in income”.

Interestingly, both the LCH and the PIH predict that a temporary increase in the current income

will mostly result in additional saving, whilst a permanent increase in current income will mostly

result in additional consumption (Harjes and Ricci, 2005).

2.3.4. Deaton’s Theory and Review of saving in developing countries

Deaton developed a model of households which cannot borrow but which accumulate assets as a

buffer stock to protect consumption when incomes are low. Such households dis-save as often

they save, do not accumulate assets over the long term and have on the average very small asset

holdings. But their consumption is markedly smoother than their income.

16

Much of the evidence is as consistent with this view of saving as it is inconsistent with the

standard views of smoothing over life cycle and with explanations of the link between saving

and growth in terms of life cycle saving behavior.

Deaton (1989:62) argued that at macro level, both developing and developed countries are

concerned with saving and growth, with the possible distortion of aggregate saving and with

saving as a measure of economic performance. But few developing countries possess the sort of

fiscal system that permits deliberate manipulation of personal disposable income to help stabilize

output and employment.

The determinants of savings have been examined extensively in the literature on savings

behaviour. The application of the theoretical determinants of savings to specific economies,

however, has been the subject of debate as a result of the dynamic nature of savings behaviour

and the fact that the interaction between theoretical savings determinants (and therefore the

anticipated savings behaviour outcome) is influenced by country-specific social, demographic

and economic conditions.

Various scholars have highlighted the relationship between saving rate and various economic

variables. It has been shown that the relationship has been direct in most variables except for

some certain few variables.

2.3.5. Saving and Income

Change in income directly affects more than one economic phenomena; savings in both real and

monetary savings, and consumption in all its forms. Increment in households or enterprises’

income increases households or enterprises’ holdings in durable goods, and financial assets in all

its forms. Andrew et al (1998:115) argued that one benefit from an increase in current income is

that, in addition to enjoying more consumption today, individuals can also plan for greater future

consumption by saving a part of their current income. Therefore, they viewed income as the

main determinant of savings, where an increase in income leads to an increase in savings for the

individuals that cater for future consumption.

In a standard Keynesian model, saving is a positive function of income because people’s ability

to save begins to rise after their income exceeds subsistence level of consumption. Lower-

income people thus tend to consume a larger share of their income than higher-income people.

17

This is easily seen if we derive the Keynesian saving equation from a typical Keynesian

consumption equation below:

C= Co+αY (2.3)

Where C is consumption, Co is the subsistence consumption, Y is income, and α is the

propensity to consume.

The corresponding saving equation would be like the equation below:

S=-Co+ (1-α) Y (2.4)

Implying that

S/Y = (1-α)-Co/Y (2.5)

Thus, d(S/Y)/dY> 0: the saving/income ratio is a positive function of income.

An important insight is that in making economic decisions, people should think about the future.

They should not care on consumption to enjoy today but also hope to enjoy next week, next year

and when they will retire.

2.3.6. Saving and Interest Rate

According to Balassa (1989:25), the effect of a higher real interest rate depends on the relative

strength of substitution and income effects. The substitution effect is positive: an increase in the

real interest rate will increase saving by increasing the rate of return on saving in the current

period relative to that in the next period.

The income effect is negative: an increase in the real interest rate will lower saving because it

increases income (an increase in wealth), and thus consumption, in the current period. In many

developing countries, governments are known to have kept their real interest rates low as a

means of financial repression to force the national saving rate to rise – providing cheap credit to

industries serve to promote production while the low interest rate suppresses consumption

through lower interest income.

According to Thaler (1994:24), the direction and size of impact of the real interest rate on saving

rate in the short and long run depends upon the offsetting influences of substitution, income and

human wealth effects.

18

A rise in the real interest rate raises the cost of current consumption relative to future

consumption, providing an incentive to rise saving. However, the effect of interest rate on saving

rate behavior is not clear in theoretical terms. If interest rates high, households will limit their

present consumption and save more in order to consume more in the future.

2.3.7. Saving and Exchange Rate

According to Montiel, (2007:65), the mercantilist view that a country can boost its net exports –

and thus its national income and national saving – by undervaluing its currency is a long-held

one. Its presumption has been that a country, to the extent it succeeds in devaluing its currency

and keep it undervalued, can boost and preserve the price competitiveness of its tradable. Of

course, it is not a consensus view among economists. Nevertheless, the mercantilist view is

popular among contemporary commentators in reaction to China’s large trade surplus and high

saving rate.

There is also a growing literature that shows an undervalued currency has a positive effect on

saving through non-mercantilist channels. For example, Levy-yeyati and Sturzzenegger (2007)

also claim that a more depreciated real exchange rate results in higher saving, but through a

different channel: a more depreciated exchange rate is associated with lower real wages,

inducing firms to invest more and to increase their saving to finance the additional investment,

thereby raising overall saving and show that an undervalued currency boosts output growth by

increasing savings and capital accumulation.

2.3.8. Saving and Inflation

Both the positive and negative impacts of inflation on saving behavior have been defended in

empirical literature on savings behavior.

Schmidt (1999:62) pointed out that, “in many developing countries, inflation means the amount

of consumer goods that wage-earners can afford will fall. Inflation thus may increase saving by

redistributing wealth from workers (who tend to have a lower saving rate) toward capital-owners

(who tend to have a higher saving rate)”.

19

Many researchers have also included inflation in a saving equation as a proxy for

macroeconomic uncertainty, an increase of which is expected to have a positive effects on

precautionary saving. In such case a rise in inflation would be expected to raise savings for

precautionary motives. Also, an unexpected rise in inflation may lead households to raise savings

to compensate for the capital loss on a fixed-income. However, Clements (1984:45) put forward

two arguments that the inflation rate could have a negative effect on saving.

2.3.9. Saving and Fiscal Policy

Mankiw (2002:212-213), stressed that, “higher national saving means higher public saving, or

some combination of the two”. Much of the debate over policies to increase growth centers on

which of these options are likely to be most effective. The most direct way in which the

government affects national saving is through public saving – the difference between what the

governments receive in tax revenue and what it spends. When the government’s spending

exceeds its revenue, the government is said to run a budget deficit, which represents negative

public saving.

A budget deficit raises interest rates and crowds out investment; the resulting reduction in the

capital stock is part of the burden of the national debt on future generations. Conversely, if the

government spends less than it rises in revenue, it is said to run a budget surplus. It can then

retire some of the national debt and stimulate investment.

The government also affects national saving by influencing private saving— the saving done by

households and firms. In particular, how much people decide to save depends on the incentives

they face, and these incentives are altered by a variety of public policies. Many economists argue

that high tax rates on capital— including the corporate income tax, the federal income tax, the

estate tax, and many state income and estate taxes—discourage private saving by reducing the

rate of return that savers earn. However, tax-exempt retirement accounts are designed to

encourage private saving by giving preferential treatment to income saved in these accounts.

2.3.10 Saving and consumption

Consumption means expenditure during a particular period on goods and services used in

satisfaction of needs and wants. John Maynard Keynes developed a theory of consumption that

focused primarily on the importance of people’s disposable income in determining their

20

spending. A rise in real income gives people greater financial resources to spend or save. The

rate at which consumers increase demand as income rises is called the marginal propensity to

consume (MPC).

Savings, investment and consumption are closely related. There will be no investment without

savings. Investment, in turn, creates employment and income for people. Without it and,

therefore, without income, we shall have nothing to save and nothing to spend on consumer

goods and services. What we do not spend is what is saved. Consumption, therefore, is affected

by decisions to spend. If we spend all our income, there will be no capital accumulation for

saving investment (Funom Makama, 2009).

According to Rose and Kolari (1995), savings is referred to the postponement of current

consumptions. The volume of savings by individuals’ consumers is a function of number of

factors, including the amount of current and expected income, the stock wealth held by the

individual, the level of interest rate, expectations concerning the future rate of inflations and

other variables. Regina Chang (1994) noted that, the idea of consumption smoothing gives a

positive correlation between current consumption and saving because transitory income shocks

lead to a higher current income and increased saving.

As was pointed out by Caroll (1992), this may create a significant correlation between

consumption growth, lagged income and saving ratio. Russian households saved much the same

in the 1990s, which was higher than savings in 1976 (Gregory et al. 1999). This finding is

contrary to expectations since there is widespread belief that savings were actually relatively

high in the Soviet era because of shortages of consumption goods (a form of forced saving) and

that saving was lower with price liberalization.

According to the William and Micheal (2006) in their Economics book state that both

consumption function and saving function have positive slope. As disposable income rises,

consumption and saving rise. Consumption and saving then are positive functions of disposable

income. Loayza and Shankar (2000), advocate the use of measures of savings that correct for

consumer durables. In their study stated that, saving and consumer spending have positive

relationship.

W.S.Woytinsky (1948) remarks on the relationship between consumers’ expenditures, savings

and disposable income. He said that these three indicators have positive relationship.

21

There are several reasons why one may be interested to study the saving and consumption

behavior of households in developing countries. Saving is related to growth and economic

development. There is a close link between household consumption and national saving rates

over time (Deaton, 1997).

Athukorala (1998) and Sen (2002) noted that higher interest rate increases the present price of

consumption relative to the future price (the substitution effect), and thus provides an incentive

to increase saving. However, if the household is a net lender, the interest rate rise also raises

lifetime income, and thus tends to increase consumption and decrease saving (the income effect)

2.4. Empirical Studies

Almost all research done on various development paradigms pursued by various countries at

different times are all conclusive on the significant correlation between savings, investments, and

growth. There is no substitute for such a relationship now, or in near future, with regard to

growth and development pursuits of nations.

Along with the recent revival of interest in and the consequent expansion of the literature on

economic growth, the behavior of saving rates also underwent an upsurge in attention. Among

other things, the long-debated relationship between savings and the level and growth rate of

income has provided a strong stimulus for analyzing the determinants of saving more thoroughly.

This relation has become even more solid with the studies confirming that despite the occasional

importance of international flows of capital, the most important factor for a country’s investment

is indeed its own savings. There are many factors that determine the saving performance of a

country. The most important factors are those related to income, demographic structures, the

interest rate, macroeconomic stability, the extent of financial sector development, and external

variables.

Different empirical studies investigated these factors such as Mahmoud(2008), analyzed

determinants of Domestic Saving Performance in Egypt during the period 1975-2006, the unit

root test was used to test the stationary of all time series. He concludes that the growth of per

capita income is found to have positive influence on domestic savings, especially on the long-

run, budget deficit ratio appears to have a negative effect on domestic saving ratio, the

22

development of financial market as proxied by the increase in the M2/GNP ratio shows a

positive and significant effect on domestic savings, the real interest rate, and inflation rate prove

to have positive and significant impact on the level of domestic savings in Egypt and finally,

current account deficit recorded a negative and statistically significant effect on both the short

run and the long run, which imply that external saving may tend to act as a substitute to domestic

private saving.

Ozcan et al (2002) examines the empirical determinants of private saving for Turkey (1968-

1994), the findings support the hypothesis that the private saving rates have strong inertia. The

evidences indicate that the government saving does not tend to crowd out private savings and the

Ricardian equivalence does not hold strictly. Income level has a positive impact on the private

saving rate and growth rate of income is not statistically significant. From a policy point of view,

financial depth and development measure of Turkey suggests that countries with deeper financial

systems will tend to have higher private saving rates. Private credit and real interest rates try to

capture the severity of the borrowing constraints and the degree of financial repression for

Turkey. Moreover, negative impact of life expectancy rate lends support to the life-cycle

hypothesis. The precautionary motive for saving is supported by the findings that inflation

captures the degree of macroeconomic volatility and has a positive impact on private saving in

Turkey.

Along with the recent revival of interest and the consequent expansion of the literature on

economic growth, the behavior of saving rates also underwent an upsurge in attention. Among

other things, the long-debated relationship between savings and the level and growth rate of

income has provided a strong stimulus for analyzing the determinants of saving more thoroughly.

This relation has become even more solid with the studies confirming that despite the occasional

importance of international flows of capital, the most important factor for a country’s investment

is indeed its own savings .Ozcan et al (2009:1)

23

Obwana and Ssentamu(1995), examined nature and determinants savings in Uganda and the

results show that income, wealth and dependency ratio variables affect saving strongly and in

ways consistent with standards theories. Khan and Abdullah (2010), examine the saving

determinants in Malaysia by using Life Cycle Model setting, the study employs a saving function

which includes per capita income, rate of returns on savings deposit, government fiscal balance,

young age dependency rate, and old age dependency rate and also inflation rate as the potential

determinants of saving.

According to Kudaisi (2010) understanding the nature of national savings behavior is critical in

designing policies to promote savings and investment. It is therefore not surprising that the

analysis of saving behaviour has become one of the central issues in empirical macroeconomics

(Jappelli and Pagano, 1998). Along with the recent revival and the consequent expansion of the

literature on macroeconomic growth, interest in the saving’s determinants underwent an upsurge

attention (Ozcan et al, 1998).

Deaton, 1989; Jappelli and Pagano, 1998; Ozcan, 2000;Schmidt-Hebbel et al, 2000 and

Elbadawi and Mwega, 2000 further stressed the important of saving behavior in developing

countries.

Among other things, the long-debated relationship between saving and the level of growth rate of

income has provided a strong stimulus for analyzing the determinants of saving more thoroughly

in most countries of the world.

According to Musonera and Karuranga (2012). After the genocide against the Tutsis in 1994,

Rwanda embarked on a continuous, aggressive, and ambitious agenda of political, financial and

economic reforms to establish an attractive environment for both domestic and foreign

investments. Rwanda has continuously improved its policy and institutional reforms towards

poverty reduction. As a result, Rwanda is one of the fastest growing economies in Central Africa

and has average a GDP Growth Rate of 8.23 Percent from 2000 to 2012. The government of

Rwanda is committed to achieving sustainable economic growth, and its overall economy is

growing at a significant rate: Its GDP per capita has increased from less than $200 in 1994 to

$540 in 2010. It is evident that Rwanda is an example of success stories in post-conflict

reconstruction (Bigsten and Isaksson, 2008).

24

Saving is a key macroeconomic variable, as it is a potential source of investment and thus

economic growth. The economic and social situation in sub-Saharan Africa remains fragile and

vulnerable to domestic and external shocks (Ulku, 2004). Investment remains subdued, limiting

efforts to diversify economic structures and boost growth (Nkurunziza and Bates, 2004).

Furthermore, a number of countries have only recently emerged from civil wars that have

severely set back their development efforts while in other parts of the continent, new armed

conflicts have erupted (Basu et al., 2005). These conflicts and other adverse factors, notably poor

weather conditions and deterioration in the terms of trade, have led to loss in economic

momentum in the region over the last two decades (Ulku, 2004; Nkurunziza and Bates, 2004;

Basu et al., 2005).

Sub-Saharan African countries therefore face major challenges like raising economic growth,

reducing poverty and their economic integration into the world economy. Economic growth rates

are still not high enough (Nkurunziza and Bates, 2004) to make a real dent in the pervasive

poverty and enable these countries to catch up with other developing nations. An empirical study

by Basu et al (2005) suggest that what is needed is a sustained and a substantial increase in real

per capita GDP growth rates in these countries, coupled with significant improvements in social

conditions, Ndambiri et al (2012:1,18-24)

The growth of any economy depends on capital accumulation, which in turn depends on

investment and an equivalent amount of savings to match it. Two of the most important issues in

development economics and for developing countries are how to stimulate investment and

increase the level of saving to fund increased investment. Understanding the determinants of the

aggregate savings is a crucial prerequisite in designing a number of policy interventions; from

the design of the tax and social security system to the layout of financial markets regulations.

Due to the absence of efficient credit and insurance markets in developing countries, domestic

savings are a crucial determinant of welfare. It is therefore not surprising that the analysis of

saving behaviour has become one of the central issues in empirical macroeconomics.

According to Carroll and Weil (1994:2), Economic theory suggests that productivity and

population growth are the main candidates to explain variation in savings, both across countries

and over time. This prediction is validated by much empirical work showing that a positive

25

correlation between savings and growth is one of the most robust stylized facts in

macroeconomics.

According to Humbert et al (1998:1), concluded that, the world’s average savings has declined

for the last two decades but countries’ savings exhibit a large dispersion, especially in

developing regions. While in a small number of developing countries’ savings rate has risen

substantially (together with growth), it has stagnated or even declined in most other developing

countries.

According to Khaled and A.P(1999:1), important variables that determine the ability to save

include the level of per capita income; the growth of income (which comprises the growth of per

capita income and population growth); the age structure of the population (or dependency ratio)

if population is not in balanced growth, and the distribution of income. Key variables that

determine the willingness to save include the rate of interest, the degree of financial deepening,

and the rate of inflation. The overall domestic savings ratio will also be affected by the

government’s fiscal stance.

Rwanda’s financial system is small, unsophisticated and dominated by commercial banking. A

mere 7 per cent of the population has a bank account, mostly with the Union des Banques

Populaires du Rwanda (UBPR), a union of cooperative banks. The client base of the six

commercial banks is limited to around 10 000 commercial and 100 000 individual clients.89

Among these, commercial banks compete for a restricted core of about 50 large clients, which

include the largest corporations, embassies, international organizations and NGOs. The

requirements of these main clients consist mainly of simple foreign exchange transactions and

short- to medium-term borrowing in local and foreign currency (UNCTAD, 2006)

According to IPAR Report (2012), Rwanda’s gross domestic savings as a proportion of GDP

have shown a positive growth trend from negative levels prior to 1998 to a positive level of

about 5% of GDP in 2009. Although domestic savings have grown steadily over time, they are

still less than 10%, which is low when compared to the benchmark countries like Vietnam and

Malaysia whose current savings rates are above 10%. The low rates of domestic resource

mobilisation constitute a major bottleneck to sustaining productivity-driven economic

transformation in the largely informal private sector.

26

MINECOFIN (2010) Rwanda’s savings rate is very low meaning that both domestic and foreign

savings are less mobilized. This indicates that saving instruments are underdeveloped. Due to

low levels of savings, Rwanda depends a lot on foreign assistance and this creates a deficit

problem where by the needed investment in the country cannot be covered by the available

savings. On average, Rwanda’s gross national savings is still low as compared to many other

countries on Africa; however a number of policies are being put in place in order to boost it.

According to Kanimba (2005:10), the very low level of domestic saving in private sector coupled

with a structural current fiscal deficit cannot support the Rwanda development agenda” Rwanda

National savings as a % of GDP stood at 10.5% in 2010 which still very low. Kanimba added

that, Rwanda should target a saving rate of 20% as minimum and mobilize external saving as a

supplement to national saving.

He further went on to state that, Rwanda’s saving performance is below the sub-Saharan

benchmark. While the poverty profile of the country is part of the explanation, the low levels of

savings in Rwanda reflect a lack of saving culture for Rwandan. Table 2 below is a comparison

between Gross Investment and National Savings of Rwanda

Table 2: Current trends of national saving and gross investment in Rwanda

Indications 2005 2006 2007 2008 2009 2010

Gross investment as a % of GDP 15.8% 16.0% 18.0% 22.8% 22.0% 21.9%

National savings as a % of GDP 13.3% 8.9% 12.7% 15.7% 11.5% 10.5%

Source: MINECOFIN (2010)

As shown by the Table 2, in 2010, Rwanda’s gross national savings rate was low, around 10.5 of

GDP. The average African savings rate is approximately 18%, and economic theory holds that

for a country to accumulate savings sufficient to induce growth and development, savings should

be above 23% of its GDP. “The persistent low growth and savings rate in Africa is due to the

vicious cycle of poverty, a trap in which low incomes, and thus low savings, reinforce each

other, to the extent that, growth can only be anticipated if, and only if, this trap is broken.”

27

What is immediately seen in the table 3.4 is that the difference between investment and national

savings has been associated with huge balance of payments deficits. These deficits pose

problems for countries unless in flows are non-commercial on very generous terms or consist of

foreign direct investments with no repayment obligations. Saving is an important indicator of

economic development where it is viewed as an element which finances domestic investment to

achieve economic growth.

Indeed, the task of poverty alleviation and meeting the Rwandan 2020 vision requires targeting

and achieving around 8% annual growth over next 12 years. This will need substantial

investment both in public and private sectors. Meeting and funding this increased investment

requirements through mobilization of savings nationally is the focus of Rwanda savings

mobilization strategy. The EDPRS has set a target of achieving a gross national savings of 18%

of GDP to attain a gross national investment target of 30% of GDP, MINECOFIN (2010)

Encouraging growth in an economy depends to a large extent, upon capital accumulation. This

requires investment, which according to economic theory, must be matched by saving.

However, domestic savings and investment are not always equal as countries with low savings

can attract investment from overseas and foreign savers lacking opportunities at home can invest

abroad. This can lead to a saving-investment gap (also known as a resource gap). In Rwanda this

gap is negative, i.e. investments are greater than savings. To close this gap it is not possible to

reduce investment – due to the large infrastructure requirements for development – and so the

solution is to increase savings (MINECOFIN, 2010).

2.5 Rwanda Financial inclusion and saving culture

Finscope report (2012) indicates that 68% (3 million) of Rwandan adults save or put money

away, Rwandan adults do not perceive saving as a means of wealth accumulation, 40% of adults

regard saving as “putting money in a special place to keep it safe”, 29% of adults regard saving

as “putting money aside to stop you from spending it immediately so that you have it later when

you need it”. Savings behavior does not have a long-term or investment orientation; most adults

(63%) who save, put money away to enable them to cover living expenses during times of

financial difficulty. A secondary driver of savings behavior is school fees – 8% of savers save for

this purpose, 5% of savers save for the purpose of investing in a house or land.

28

Table 3.3: Percentage of Rwandans who trust most with their savings

Financial group/institution

Kigali

City

Other urban

area Rural Total

Bank 60.3 33.8 27.5 33

Umurenge SACCO 29.1 45.9 53.9 49.4

Non-umurenge SACCO or a MFI 2.8 3.3 2.3 2.5

Saving groups like tontine, ikibina 3.1 9.5 9.5 8.6

Source: Finscope report (2012)

FinScope 2012 findings indicate that financial activity has increased amongst Rwandan adults

since 2008. In 2008, 52% of Rwandan adults saved, in 2012, 71% are saving.

The significant uptake of SACCO usage for savings has lead to decreased usage of “home

savings” and the use of “own” mechanisms such as buying assets or livestock as a means of

saving.In 2008, 27% reported that they borrowed money or accessed credit, in 2012 59% are

borrowing or accessing credit .It levealed that 88.2% of banked population, only 20.1% have

saving account.83% of MFI clients, 3% have savings facility. 90% of umurenge SACCO clients

only 19% have savings facility.

The Rwanda’s aim is to turn into a middle income country with per capita GDP of US$900,

currently Rwanda is engaged to reach the following targets: eliminating extreme poverty,

reducing poverty to 30%, achieving an average GDP growth rate of 8%, achieving gross national

savings of 18% (of GDP), attaining gross national investment of 30% (of GDP) (MINECOFIN,

2010)

Given the strong correlation between all those key elements, especially the direct correlation in

“savings-investment-economic growth”, the GoR has embarked on various programs aimed at

putting in place, conditions for the development of private investment: creating a legal and

administrative environment which is conducive for business, developing human resource

capacity, developing the financial sector, especially for savings mobilisation and access to credit

and financial services.

A strategy to improve Rwanda’s savings rate needs to benefit from the recent insights in the

savings, investment and growth literature.

29

Tax and interest rate instruments are important policy tools but need to be a part of an overall

policy package aimed at fostering growth through improved productivity and financial

liberalization. The financial sectors requiring particular attention are a well functioning banking

sector, social security, pensions, insurance and a low-cost remittance schemes for Diasporas

(MINECOFIN, 2010)

There are six Pillars that have been identified as Rwanda National Savings Mobilization strategy.

When these key issues are addressed properly, these will strengthen the financial infrastructure,

mobilize savings and help to create a culture of savings. These are macroeconomic Stability,

institutionalized savings, expansion of the financial infrastructure and intermediation, secure and

diversified means of savings, building capacity and efficiency of intermediation, and increased

awareness of tangible benefit of savings. (MINECOFIN, 2010)

In addition, the report offers a major piece of policy recommendation and examines potential of

SACCOs (savings and credit co-operatives) and how they can be utilized to mobilize savings for

low income individuals. There are six main keys to the success of the SACCO program. These

are security, low minimum balance, liquidity, government support, tailored products to meet the

needs, public education and the capacity building through training (MINECOFIN, 2010)

Rwanda’s economic growth over the last decade has been remarkable. With a government that is

committed to achieving sustainable economic growth coupled with growth in employment

opportunities for its people, Rwanda has made impressive progress in rehabilitating and

stabilizing its economy to exceed pre-1994 levels. The overall economy is growing at a

significant rate. The average annual growth rated in GDP was 8.8 per cent between 2005 and

2009. Rwanda’s GDP per capita has increased from less than 200 USD in 1994 to 540 USD in

2010.

Although still at an early stage, the GoR has set a set path towards economic transformation in

Rwanda .There is evidence of a significant increase in private sector investment following the

introduction of a revised tax code and implementation of the doing business reforms since 2005

although there was a downturn due to the World economic crisis in 2009.

Both foreign and domestic investment have increased with FDI exceeding local investment and

new jobs have been created (Dickson and Serge, 2012:5)

30

Rwanda generally has a good track record in recent years of maintaining a stable macroeconomic

environment. However, recent developments in international financial and commodity markets

have increased the risks to the macroeconomic stability, particularly to savings and investments.

Due to increased inflation, interest rates have also gone up in order to pursue tightened monetary

policies. However, there is a need to create an environment with moderate inflation accompanied

with positive interest regimes to help mobilization of savings (MINECOFIN, 2010)

Vision 2020 set the stage for the financial sector reform process in Rwanda. The Rwandan

Financial Sector Development Programme (FSDP) was launched in 2006 with the vision to

“develop a stable and sound financial sector that is sufficiently deep and broad, capable of

efficiently mobilizing and allocating resources to address the development needs of the economy

and reduce poverty”.

The FSDP is one of the key components in the implementation strategy for Vision 2020 (the

Economic Development and Poverty Reduction Strategy of Rwanda, EDPRS) and has four core

objectives:

• To enhance access and affordability of financial services by developing a strong, efficient

and competitive banking sector offering a diversified array of financial products and

services. This includes support for the development and broad outreach of a healthy,

well-regulated and professionally managed microfinance sector as a tool to extend

financial services to the unbanked and to contribute to poverty reduction;

• To enhance savings mobilization by creating an appropriate environment, developing

institutions and fostering market incentives for the development of long-term financial

instruments and an efficient capital market;

• To develop an appropriate policy, legal and regulatory framework for non-bank financial

institutions;

To develop an efficient, secure and technology-based modernized national payment system

(FinScope, 2012)

3.2 Rwanda’s goals

31

Like many developing countries, the primary focus of policies in RWANDA is to have high and

sustainable growth. The EDPRS and Vision 2020 set out targets for the economy of which

savings are a crucial linchpin to success.

According to Finscope report (2012), Rwanda’s development policy is guided by Vision 2020

(MINECOFIN, 2000). This statement articulates the Rwandan Government’s aim to “transform

Rwanda into a middle income country, as well as an economic trade and communications hub by

the year 2020”. Illustrating its commitment to this vision, Rwanda has made significant progress

over the last decade which has been characterized by sustained economic growth, significant

poverty reduction and the beginning of economic transformation (2010/11 Integrated Household

Survey (EICV3))

Rwanda has set two different policy objectives to be achieved in 2012. Firstly, there are certain

goals which are milestones on a longer journey. These include the Millennium Development

Goals (MDGs) which have targets set for 2015, and the objectives of Rwanda Vision 2020 which

have targets set for 2020.

Given a time path for achieving the MDGs, there will be an implicit set of targets for 2012.

However, since the EDPRS is a mechanism for implementing Rwanda Vision 2020 in the

medium term, there is no separate set of targets for Rwanda Vision 2020 in 2012. Secondly,

there are the EDPRS goals themselves which constitute a destination in 2012.

These goals include targets which differ from those of Rwanda Vision 2020 and the MDGs.

(MINECOFIN, 2007)

Table 2: Key EDPRS results indicators

Priority area Indicator Baseline

(2006)

Target

(2012/13)

Actual (2010/11)

Growth and

Real GDP growth (%

annual)

6.5

8.1

8.2

National investment (%

of GDP)

16.3

24.4

21.9

32

poverty

reduction

Share of population living

in poverty (%)

57 46 44.9

Share of population living

in extreme poverty (%)

37 24 24.1

Widen and

strengthen the

Financial

Sector

Private Sector credit (%

of GDP)

10 15 13.9

Financial depth (broad

money/GDP)

20 22.5 20.4

Source: MINECOFIN (2007)

In the macro and economic sectors, Rwanda has maintained a high economic growth rate in the

face of both external and internal shocks. During EDPRS, GDP growth rate has increased at an

average rate of 8.5% exceeding the anticipated requirement of 8.1%.

This was achieved against the backdrop of the world economic crises that affected fuel and food

prices as well as the financial sector. The percentage of investment to GDP increased from 16%

in 2006 to 21.9% in 2010/11 exceeding the target of 19% for 2010/11, the EDPRS target for

2012/13 is 23%.

2.6. Research Gap Analysis

The savings is a key performance indicator for development policy. Despite the crucial role that

savings plays in the development process, economists have not reached conclusive answers

about the role of various economic variables in determining the savings rate of any country.

In addition, an extensive literature on saving behaviour, several empirical issues have not been

resolved conclusively, including the effects of real interest rates, demographic factors, and per

capita income on private saving; the relationship between growth and saving; and the extent to

which private saving offsets movements in public (dis)saving.

33

Masson et al (1994:483), found that the empirical positive correlation of saving with income

growth is not, on the face of it, consistent with the life-cycle hypothesis, unless the higher

income growth is at least partly transitory.

Ozcan et al (2003) Determinants of private savings behavior in Turkey, conclude that private

saving rates have strong inertia and they are highly serially correlated, government savings to

GPDI ratio has a negative impact on the saving rate, Income level has a positive impact on the

private saving rate for Turkey. It is found that terms of trade shocks increase private saving in

Turkey. The empirical findings presented here indicate a number of variables that affect private

savings in Turkey.

The complexity of the relationship between saving and other variables are examined. These

variables clearly indicate the role of policies pursued by the country that affect saving.

According to the empirical findings, it can be said that financial market development,

macroeconomic stability, life expectancy, external factors and economic crisis may be the core

policy instruments in Turkey for the saving behaviour.

Mahmoud A. Touny(2008) determinants of domestic saving performance in Egypt, he found

that the growth of per capita income is found to have positive influence on domestic savings,

especially on the long-run, budget deficit ratio appears to have a negative effect on domestic

saving ratio, the development of financial market shows a positive and significant effect on

domestic savings, the real interest rate, and inflation rate prove to have positive and significant

impact on the level of domestic savings in Egypt ,current account deficit recorded a negative and

statistically significant effect on both the short run and the long run,

Khalil Ahmad and Haider Mahmood (2013) give a detailed explanation of different determinants

of national savings in the process of economic growth, in the glimpse of Pakistani experience.

Using Autoregressive Distributed Lag Model (ARDL) bound testing approach for co-integration

techniques to check the robustness for long run relationship and Error Correction Mechanism

(ECM) for short run dynamics during the 1974-2010.

It is found that the per capita income inversely related with national saving rate, both in long run

and as well in short run significantly. The exchange rate and inflation rate have a negative impact

on national saving but lagged exchange rate has significantly impact. Because of floating

34

exchange rates and the decrease in capital controls, the volume of international capital flows in a

country, has increased significantly. Trade openness is positive associated with national savings

in Pakistan because trade openness cause to increase the income and welfare of the society in

through market economy.

Money supply positive linked with national saving due to seigniorage effect. The growth of the

income level has negatively related with national savings. Keynesian and permanent income

hypothesis of income and savings is not valid for Pakistan because per capita income and income

growth inverse function of savings at national level.

There are many studies on Rwanda economic growth such, “Foreign Direct Investment and

Economic growth in Rwanda,2013” by Bruno et al, “Rwanda case study on economic

transformation,2012” by Dickson and Serge and several reports of different ministries, agencies

and UN institutions talking about economic growth and saving slightly.

However, there are few researches carried out on savings. Most of documents are reports from

MINECOFIN, UNCTAD, Finescope and IPAR. The common factor of all reports is the low

level of savings in Rwanda and lack of saving culture.

The recent research conducted by Uwamariya (2010) on the analysis of determinants of saving in

Rwanda, using regression analysis conclude that Real Gross Domestic Product influence

positively savings in Rwanda, inflation and Deposit Interest Rate influence negatively savings.

Having motivated from the previous literature, we try to analyze the determinants that are likely

to influence saving in Rwanda (1978-2012). In this context, the key variables that will be

analyzed are income, interest rate, inflation, investment and consumption as factors that

influence saving behavior in Rwanda.

In this context, the purpose of this research is to examine deeply the savings determinants in

Rwanda during 1978 to 2012.

35

CHAPTER 4: RESEARCH METHODS

This chapter deals with how the research is to be conducted in order to achieve the stated

objectives. It indicates the research design, the methods and techniques that will be used within

the study. It further will describe research instruments in carrying out the research and how data

will be processed.

According to Bailey (1987:26), the methodology is a philosophy of the research process. It

involves assumptions and values that serve as a rationale for the research and standard or criteria

the research uses for interpreting data in order to derive conclusions. In brief therefore,

methodology refers to a systematic or an orderly procedure followed in carrying out any

scientific activity. Therefore, this chapter gives a detailed account of how the research will be

carried out.

The study covers the period of 1978-2012 and employed an econometric methodology. The data

used in this study are secondary data from the World Development Indicators (WDI, 2012), as

well as reports published by the National Bank of Rwanda (NBR).

4.1 Conceptual framework

Independents variables in this study are determinants of savings (consumption, interest rate,

inflation rate, income and investment) have been tested to evaluate the level at which influence

savings in Rwanda (1978-2012). In this study government expenditures proxied consumption,

capital formation proxied investment and Gross Domestic Product as proxy for income. To

define all variables we referred to World Development Indicators (WDI, 2012)

Gross national expenditure (% of GDP); Gross national expenditure (formerly domestic

absorption) is the sum of household final consumption expenditure (formerly private

consumption), general government final consumption expenditure (formerly general government

consumption), and gross capital formation (formerly gross domestic investment).

Real interest rate (%); Real interest rate is the lending interest rate adjusted for inflation as

measured by the GDP deflator.

36

Inflation, consumer prices (annual %); Inflation as measured by the consumer price index

reflects the annual percentage change in the cost to the average consumer of acquiring a basket

of goods and services that may be fixed or changed at specified intervals, such as yearly. The

Laspeyres formula is generally used.

Aggregates are based on constant 2005 U.S. dollars. GDP per capita is gross domestic product

divided by midyear population. GDP at purchaser's prices is the sum of gross value added by all

resident producers in the economy plus any product taxes and minus any subsidies not included

in the value of the products. It is calculated without making deductions for depreciation of

fabricated assets or for depletion and degradation of natural resources.

Gross capital formation (% of GDP); Gross capital formation (formerly gross domestic

investment) consists of outlays on additions to the fixed assets of the economy plus net changes

in the level of inventories. Fixed assets include land improvements (fences, ditches, drains, and

so on); plant, machinery, and equipment purchases; and the construction of roads, railways, and

the like, including schools, offices, hospitals, private residential dwellings, and commercial and

industrial buildings. Inventories are stocks of goods held by firms to meet temporary or

unexpected fluctuations in production or sales, and "work in progress." According to the 1993

SNA, net acquisitions of valuables are also considered capital formation.

Dependent variable is Gross savings (% of GDP) .Gross savings are calculated as gross national

income less total consumption, plus net transfers.

4.2 Research Design and Model Specifications

Most studies about the determinants of savings used different econometric models, so there is no

single theory that would be used, only a set of assumptions that need to be refined and tested.

4.2.1 Model Specification

a) Theoretical model

Both theoretical and empirical work on savings, have consistently outlined the major potential

determinants of savings, which can be grouped loosely under the headings of government policy

variables, financial variables, income and growth variables, demographic variables, financial

variables, uncertainty measures, and external variables, K. Metin Ozcan et al.(2003)

37

The relationship between the savings and macroeconomic variables has been the subject of many

researches. The life-cycle approach (Modigliani, 1970), is the model most commonly referred to

in studies of saving, notwithstanding the fact that no single model actually has the ability to

account for every aspect of such a broad subject. The major argument of the model is that

individuals seek to smooth out consumption over time, saving in ‘good times’ to consume in

‘bad times’.

In order to define variables I referred to K. Metin Ozcan et al.(2003) who carried a research on

Determinants of private savings behaviour in Turkey(1968-1994). By examining potential

determinants of private savings in Turkey, they outlined a variety of policy and non-policy

variables relevant to saving.Inertia, it is an observed fact that saving rates generally contain

inertia; that is, they are serially correlated, even after controlling for other factors.

Government policies, various actions of government can have a bearing on saving. Among these,

the effect of fiscal policy has especially been the centre of debate. Theoretical views on this

relationship span a broad range. The neoclassical version of the life-cycle model asserts that a

decline in government saving will tend to raise consumption and discourage saving by shifting

the tax burden from present to future generations, and predicts that a decline in government

saving will cause a decline in national saving.

Income and growth variables, the relationship between savings and income, on the other hand,

the relationship between savings and growth have been a major subject of discussion in the

growth literature.

Subsistence-consumption theories suggest that countries with higher income levels tend to have a

higher saving rate, and the empirical evidence strongly supports this conclusion (Edwards,

1996;Dayal-Ghulati and Thimann,1997; Loayza et al., 1999; Metin-Ozcan and Ozcan, 2000).

Financial variables, the financial variables that have an impact on saving are usually ones that

capture the degree of development of the financial sector. The most ambiguous financial variable

that will be considered is the real interest rate. This is largely because of the fact that a change in

the interest rate entails opposing substitution and income effects.

The set of variables under the heading ‘demographic variables’ are usually the urbanization ratio,

the age distribution of the population, and life expectancy.

38

The age structure of the population is an important factor for savings because people, who seek

to smooth out consumption over their lifetime, save when they expect future income to be low

and dissave when they anticipate it to be high.

Uncertainty variables, the variables that capture the effects of uncertainty about the future bear

on saving rates primarily via their impact on precautionary savings. These variables can be

termed broadly as macroeconomic stability and political stability. Macroeconomic uncertainty,

usually proxied by the inflation rate, is expected to have a positive impact on saving, as people in

such an environment would try to hedge risk by saving.

External variables, the external variables that might be relevant to savings are the terms of trade

and the current account deficit. Moreover, to build a model, I adapted the model used by

Mahmoud A. Touny (2008) on Determinants of Domestic Saving Performance in Egypt (1985-

2006): An Empirical Study.

His model has been presented like this:

DSt= α0 +α1 GPCIt +α2 BDt +α3 RRt +α4Mt +α5 INFt +α6 CAt + ut

Where:

DS: is domestic savings = gross domestic product (GDP) – private consumption –

government consumption expenditure)

GPCI: is the growth rate of fixed per capita income.

BD: is the budget deficit ratio to GDP.

RR: is the real interest rate.

M: is the ratio of broad money supply (M2) to GDP,

INF: is the inflation rate, measured as the growth rate of consumer price index as a proxy of

macroeconomic stability.

CA: is the current account deficit ratio to GDP [- (exports – imports)/GDP].

µ: Residuals

39

b) Empirical model

Given the specific nature of the savings of Rwanda and by taking into account constraints

relating to data availability in Rwanda, Our models encompass the Keynesian, the Classical as

well as more recent and less conventional models. Particularly, we have adopted and modified

the theoretical model of Mahmoud A. Touny (2008) to come up with our models in explaining

the determinants of savings in Rwanda and to establish the relationship between savings an

economic growth. Accordingly, two empirical research models have been designed and are used

in the regression analysis.

The empirical equations are as follows:

Model I intends to examine the major determinants of savings in Rwanda during the period

under study. In this model, we use GNS =Gross National Savings. Model II emphasis on

establishing the relationship between saving and economic growth, where PCY is used as proxy

for GDP dependent variable representing economic growth.

In this research, based on the savings functions, we specify the econometric model as below:

GNS = ��+ ��PCYt + �� RRt+��CPIt+�� GNEPt+���t+ +εt ..................... (1)

PCYt = α0 + α1 GNS +α2 RRt +α3 CPIt + α4GNEPt +α5 GCF + µt ................. (2)

Where the dependent variable is the Gross Savings (% of GDP) “GNS” in equation (1),

PCY: GDP per capita growth (annual %). Almost all studies reviewed above as well as those

contained in the various literature surveys find this variable to be an important determinant of

savings. Higher GDP per capita is associated with a higher saving/GDP ratio in several studies.

Studies typically find a statistically positive and significant effect for real per capita GDP growth

on savings. The findings of a positive coefficient on growth are consistent with various theories

(e.g, Modigliani’s aggregation effect, habit formation). Accordingly, in this study we would like

to determine the relationship between savings and economic growth in Rwanda in equation (2),

thus PCY a proxy of economic growth. The economic growth is the dependent variable.

GNEP: is Gross National Expenditure (% of GDP) a proxy of consumption.

RR: Real interest rate (%), knowing the impact of real interest rates is crucial for informing

policy on stimulating saving

40

CPI: Inflation, consumer prices (annual %), measured as the growth rate of consumer price

index. Inflation is a measure of uncertainty, and this result is interpreted as being consistent with

a precautionary motive for saving.

GCF: is Gross capital formation (% of GDP) proxy of investment.

The expected coefficients are positive for α1, α2, and α6, β1, β2, and β6. The coefficient for α3, α4,

and α5, β3, β4 and β5 are expected to be negative, and t is a time series data and εt and µt are an

error term

4.3. Data Analysis

According to BERNARD S.PHILLIPS (1968: 217), data analysis techniques deal with the

manipulation of the information that has been gathered so as to present the evidence. The data

obtained is analysed using econometric methodology.

In order to examine the relationship between the potential explanatory variables and the savings

rates, we should first employ a unit root test before we can proceed with other econometric

estimation method.

4.3.1 Unit root test

Econometric theory requires all variables to be stationary if the regressions are to be realistic.

Therefore, all variables in the savings function should be tested to determine whether they are

influenced by economic factors of a relatively permanent nature or by self-correcting forces that

indicate temporary elements in their dynamics. In this analysis, we will employ the unit root test,

more specifically, using augmented Dicky-Fuller (ADF) tests to check the stationary of the

variables. It is an augmented version of the Dickey–Fuller test for a larger and more complicated

set of time series models. The augmented Dickey–Fuller (ADF) statistic, used in the test, is a

negative number. The more negative it is, the stronger the rejections of the hypothesis that there

is a unit root at some level of confidence. In order to test it, we consider the equation as follows:

∇yt= β1+ β2t + δyt-1+ αi∑∇yt-1+µt

Where yt is our variable of interest, ∇ is the differencing operator, t is the time trend and µ is the

white noise residual of zero mean and constant variance.

41

β1,β2, δ and αi are the set of parameters to be estimated.

Both the null and alternative hypothesis in unit roots tests are:

H0: δ = 0 (yt is non-stationary)

H1: δ ≠ 0 (yt is stationary)

The H0 hypothesis can be rejected if the t-test statistic from this test is negatively less than the

critical value tabulated. In other words, a unit root exists in the series yt (implies non-stationary)

if the null hypothesis of δ equals zero that is not rejected (Gujarati 1995).

4.3.2 Co-integration test

Given that the time series properties of the data are not stationary, one has to consider the long

run relationship between the different time series to see whether there is a co integration relation

among the variables of interest. A series is said to be integrated of order d if one can obtain a

stationary series by differencing the series for d times.

4.3.3 Error correction model (ECM)

According to the Granger Representation theorem, when variables are co integrated, there must

also be an error correction model (ECM) that describes the short run dynamics or adjustments of

the co integrated variables towards their equilibrium values. ECM consists of one period lagged

co integrating equation and the lagged first differences of the endogenous variables. In particular,

ECM can be constructed by expressing changes in the dependant variables as a function of the

level of disequilibrium in the co integrating relationship (captured by the error correction term)

as well as changes in other explanatory variables. The following error correction model is

developed:

∇GNSt = a0 +a1∇GNS+ a2∇ PCYt + a3∇RRt + a4∇CPIt + a5∇GNEPt + a6∇GCFt+ +a7ECM–1 + vt

Where ECM is the error correction component and is the lagged estimated error series from

Equation (1) while vt are the random error terms. From the regression analysis, we are able to

interpret the coefficient for the explanatory variables and detect the sign. This approach shows us

the speed of adjustment of our model in short run. This interpretation leads to the determination

of the factors that influence savings in Rwanda.

42

CHAPTER 5: DISCUSSION OF FINDINGS

After analyzing theoretical view of savings in general and in Rwanda particularly during the

selected period, this chapter intends therefore, to verify empirically the concordance between

theories developed in previous chapters with the economic practice. It is to show the main

determinants that influence the level of savings in Rwanda during the selected period. We used

econometric procedures of estimating econometric models to make conclusion.

5.1 Descriptive statistics of variables used in the econometric model

The behavior of Gross savings as a percentage of GDP presented in Table 3 shows small

fluctuations in its value during the period 1978-2012.The economy has been growing, averaging

12.9 percent per annum during this period with standard deviation of 5.3 percent per annum.

Table 3 outlines the trend of savings and its determinants during the period 1978-2012 in

Rwanda. On the other hand, the real interest rate witnessed high fluctuations with standard

deviation of 7.2 and averaging 7.4 percent per annum during the period under study. In addition

the inflation fluctuated highly since standard deviation and average values are closer 8.2; 8.6

respectively.

Moreover, the consumption shows small fluctuations since 115.1; 8.8 are average and standard

deviation respectively. The investment indicates small fluctuations with standard deviation of 3

and averaging 15.8 during the period under the study. The exchange rate records 291.4 averages

and 208.4 standard deviation. While the GDP per capita records 2.3 average per annum and 11.2

standard deviation during the period (1978-2012).

Table 3: Variables in the model

Variables GNS PCY RR GCF GNEP CPI

Minimum 5.06 -47.31 -4.59 9.98 101.70 -2.405

Maximum 33.66 36.77 28.66 22.85 158.49 45.66

Mean 12.91 2.32 7.39 15.75 115.14 8.62

Standard deviation 5.3 11.24 7.2 3.033 8.77 8.19

Jarque-Bera 46.158 0.395 1.170 3.315 3.962 18.477

Skewness 1.6741 -.19182 .20265 .75113 -.81489 1.324

43

Kurtosis 7.7286 2.6255 3.8285 3.3925 3.4745 5.535

Observations 35 35 35 35 35 35

Source: developed by the author

5.2 Trend of variables in model

The Figure 2 shows the trend of real interest rate, inflation and investments from 1978 to 2012.

The real interest rate exhibits the lowest record in 2003 and the highest record in 1999.

Investment levels have been rising during the period under study except the low value in 1994.

The inflation touched ceiling in 1995 due mainly to effects of 1994 Tutsi genocide and the

lowest record in 1999.

Figure 2: Trend of RR, CPI and GCF

Source: developed by the author

The Figure 3 illustrates the trend of gross savings (% of GDP) and GDP per capita during the

period (1978-2012). The graph displays a clear relationship between gross savings and GDP per

capita. Indeed, the gross savings exhibits almost constant trend with a peak value in 1994 and

low value in 1999. Furthermore, the GDP per capita has also almost constant trend but with some

negative values. The highest record is observed in 1995 and the lowest in 1994.

Figure 3: Trend of Gross savings and GDP per capita

020

4060

Per

cent

age

chan

ge

1980 1990 2000 2010YEAR

R R CPIGCF

This indicates trend of RR, CPI & GCF

Time series of RR , CPI & GCF(1978-2012)

44

Source: developed by the author

5.3. Empirical results

This study has mainly used statistical and econometric tools for the verification of assumptions

related to the study.

5.3.1 Long-run relationship

In order to establish the long run relation between variables, we firstly test stationarity of each

variable, and then Johansen Test for Co integration and analysis Granger Causality.

Finally, we examine normalized integrating coefficients to check the level at which savings

respond to every independent variable percentage change.

A) The test of stationarity

The regression of a non-stationary unit‘s time series to another non- stationary time series may

produce a spurious regression. In order to produce a meaningful analysis, it is important to

conduct a unit root test. The augmented dickey- fuller (ADF) and Phillips –Perron (PP) test are

important tools for doing this. This test involves a regression of the change in an underling

variable on the lagged variable and on error term. Broadly speaking, a stochastic process is said

to be stationary if its mean and variance are constant over time and the value of the covariance

between the two time periods depends only on the distance or gap or lag between the two time

-40

-20

020

40Per

cent

age

chan

ge

1980 1990 2000 2010YEAR

GNS PCY

This graph indicates a trend of gross savings and GDP per capital

Time series graph of gross savings and GDP per capita (1978-2012)

45

periods and not the actual time at which the covariance is computed . In other words, a no

stationary time series will have a time varying mean or a time-varying variance or both.

(GUJARATI, 2004:797-815).

So, direct stochastic processes are stationary if and only if:

E (Yt) = E (Yt+m) =µ, ∀ t et ∀ m = mean is constant and independent of time.

Var (Yt) ≠∞ = variance is finite and independent of time.

Cov (Yt, Yt+k) = E [(Yt - µ) (Yt+k - µ) = ץk = covariance is Independent of time.

The test of stationarity by using Dickey-Fuller test consists to test the following hypotheses:

Ho: α-1= 0 null hypothesis (non-stationary) say (Yt), has a unit root. That is, a series is non-

stationary.

H1: α-1< 0 alternative hypothesis that a series has not a unit root. That is, a series is stationary.

The following models are used:

∆Υt = β0 + (α-1) Υt−1 + νt (1): Model with drift, without trend;

∆Υt = β0 + β1 t+ (α-1) Υt−1 + νt (2): Model with trend and intercept;

∆Υt = (α-1) Υt−1 + νt (3): Model without drift, without trend (none).

We have a unit root if α=1 and the model is not stationary. τ (tau) statistic replaces, in this case

the student test. Augmented Dickey-Fuller test (ADF) is used by adding more differentiated terms

of Yt (i.e. ) to correct autocorrelation in errors to Dickey-Fuller simple models above.

Where ∆Υt=Υt -Υt-1, ∆Υt-i = Yt-i - Yt-i-1 and i= 1, 2, 3…, n. While α, β and δ= (α-1) are the parameters

to be estimated and νt is white noise error term such that νt each ∼N (O,δ) that is, follows the

standard normal distribution.

If the value of the ADF statistic is less than the critical value at the conventional significance

level [usually the five per cent significant level (5%)] then the series (Yt) is stationary and vice

versa. That is:

∑=

−∆n

iitYai

1

46

Calculated ADF test statistic > critical value, we accept Ho that a series has a unit root at

significant level of 5%. But if calculated ADF statistic < critical value, we reject Ho that a series

has a unit root at significant level of 5%.

The Phillips–Perron test involves fitting (ADF), and the results are used to calculate the test

statistics. They estimate:

yt = πyt−1 + (constant, time trend) + ut

In (DF) ut is I(0) and may be heteroskedastic. The PP tests correct for any serial correlation and

heteroskedasticity in the errors ut non-parametrically by modifying the Dickey Fuller test

statistics.

Phillips and Perron’s test statistics can be viewed as Dickey–Fuller statistics that have been made

robust to serial correlation by using the Newey–West (1987) heteroskedasticity- and

autocorrelation-consistent covariance matrix estimator.

It is compulsory to test the economic time series for stationarity before proceeding for

cointegration test and establishing long-run relationships. The study used two different tests, i.e.

Augmented Dickey Fuller (ADF) test, Phillips-Perron (PP) test for finding unit roots in time

series. All these tests revealed that all the variables were stationary in levels which is not the

common phenomenon in most of the economic time series because most of the time economic

times series become stationary after first difference. However, all variables are stationary in a

two models (with constant only and constant & trend) for both tests. Hence, all two tests were

undisputedly declared that all the variables were integrated of the same order, i.e. I (0) as shown

in Table 4.

Table 4: Unit Root test

Augmented Dickey Fuller(ADF) Test

vari

able

s

Non constant Constant & Trend Constant

Leve

l

Fir

st

diff

eren

c

e Dec

isio

n

Leve

l

Fir

st

diff

eren

c

e Dec

isio

n

Leve

l

Fir

st

diff

eren

c

e Dec

isio

n

GNS -1.435* -1.383 I(0) -3.789* -3.593 I(0) 3.844* -3.719 I(0)

47

PCY -7.285* -5.419 I(0) -7.683* -6.293 I(0) -7.557* -5.837 I(0)

RR -2.647* -2.034 I(0) -4.056* -3.455 I(0) -4.076* -3.463 I(0)

GCF 0.112 1.065* I(1) -2.706* -0.961 I(0) -1.904* -0.379 I(0)

GNEP -0.068 0.244* I(1) -4.491* -3.173 I(0) -4.122* -3.026 I(0)

CPI -2.749* -1.933 I(0) -4.258* -3.002 I(0) -4.327* -3.046 I(0)

Phillips-Perron(PP) Test

Var

iab

les

Non constant

Constant & Trend

Constant

Leve

l

Firs

t

diff

eren

c

e Dec

isio

n

Leve

l

Firs

t

diff

eren

c

e Dec

isio

n

Lev

el

Firs

t

diff

eren

c

e Dec

isio

n

GNS -3.435* -2.830 I(0) -21.22* -21.136 I(0) -21.046* -20.855 I(0)

PCY -41.904* -40.975 I(0) -44.434* -42.647 I(0) -43.567* -42.292 I(0)

RR -11.575* -10.626 I(0) -23.790* -24.215 I(0) -23.563* -24.215 I(0)

GCF 0.100 0.290* I(1) -13.915* -11.594 I(0) -9.378* -6.441 I(0)

GNEP -0.036* 0.024 I(0) -26.797* -26.453 I(0) -23.518* -22.423 I(0)

CPI -12.168* -10.009 I(0) -24.971* -24.294 I(0) -24.973* -24.303 I(0)

* implies that the coefficient is significant at 0.05 percent probability

Source: developed by the author

B) Johansen Test for Co integration analysis

The results of stationarity analysis illustrated by the Table 5 show that all the modeled variables

were integrated at same order, so the study applied the Johansen tests for co-integration

technique to explore the long-run relationships among the variables as this technique is

appropriate, if all the model variables are integrated at same order.

The first step in multivariate co-integration analysis is the appropriate lag selection for the

variables. For selection of appropriate lag length, the study used four criteria i.e Akaike

Information Criteria (AIC), Schwarz Bayesian Criteria (SBC), and Hannan-Quinn information

48

criterion (HQIC). All the criteria selected lag length of two. In order to find out the number of

co-integrating vectors, Trace statistic and Maximal Eigen value tests were used.

Table 5: Co-integration analysis

Hypothesis Trace Max Eigen value

Co-integration

rank

Critical value

at 5%

Co-integration

rank

Critical value

at 5%

r=0 96.4984 94.15 43.7179 39.37

r≤1 52.7805* 68.52 19.9736* 33.46

r≤2 32.8069 47.21 17.2676 27.07

r≤3 15.5392 29.68 12.2396 20.97

r≤4 3.2996 15.41 3.2413 14.07

r≤5 0.0584 3.76 0.0584 3.76

*denotes rejection of the hypothesis at the 0.05 percent critical value

Trace and Max-Eigen value tests indicate 1cointegrating eqn(s) at the 0.05 percent critical value

Source: developed by the author

If two or more time series are co-integrated, there is a long run relationship between them .The

results for both Trace statistic and Maximal Eigen statistic were reported in Table7. Both tests,

i.e. the Trace statistic and the Maximal Eigen statistics recognized one co-integrating vectors;

therefore, the study used one co-integrating vectors in order to establish the long-run

relationships among the variables.

There is a theorem that implies if x and y are co-integrated then some form of granger causality

must occur: either x must granger cause y or y must granger cause x (or both).Therefore, let

analyse Granger causality between savings and its determinants.

C) Granger Causality

Steps involved in testing for Granger causality (Gujarati, 1995).

The steps involved in testing for the direction of causality between two economic series say, Yt

and Xt are as follows:

49

1. Regress current Yt on all past values Yt and other variables, but do not include the lagged Xt

variables in this regression. Hence, from this regression, obtain the residual sum of squares.

2. Now run the regression including the lagged Xt variable (unrestricted regression).From this

regression, obtain the unrestricted residual sum of squares (RSSUR)

3. Test the null hypothesis Ho: i.e. lagged Xt terms do not belong in the regression.

4. To test this hypothesis, we apply the F-test given by;

F = (RSSR-RSSUR)⁄M:RSSUR/(N-K)

This follows the F-distribution with M and N-K degrees of freedom. M is the number of lagged

XT terms and K is the number of parameters of parameters estimated in the restricted regression.

5. If the F-value exceeds the critical F-values at the chosen level of significance, or if the P-value

is less than the alpha level of significance, we reject the null hypothesis in which case the lagged

Xt values belong in the regression. This is another way of saying that Xt Granger causesYt.

Gujarati (1995)

6. Step 1-5 can be repeated to test model (**) i.e. to test whether Yt Granger causes Xt.

This methodology is highly sensitive to lag length selection when conducting a Granger causality

analysis.

Table 6: Granger Wald causality tests

Pairwise

hypothesis

obs Chi2

statistics

p-value Decision Type for causality

GNS↗PCY 31 .26364 0.876 DNR no causality

PCY↗GNS 31 5.8836 0.053 Reject uniderectional

GNS↗GCF 31 5.2227 0.265 DNR no causality

GCF↗GNS 31 18.482 0.001 Reject uniderectional

GNS↗GNEP 31 44.194 0.000 Reject bi-directional

GNEP↗GNS 31 34.035 0.000 Reject bi-directional

GNS↗CPI 31 30.033 0.000 Reject bi-directional

CPI↗GNS 31 69.146 0.000 Reject bi-directional

50

GNS↗RR 31 2.6607 0.616 DNR no causality

RR↗GNS 31 8.1056 0.088 DNR no causality

GNS↗ALL 31 148.7 0.000 Reject uniderectional

Alpha (α) = 0.05

Decision rule: reject H0 if P-value < 0.05.

Key: DNR = Do not reject;

↗ = does not Granger cause.

Source: developed by the author

A major implication of Granger causality is that if two variables say, x and y, are co-integrated,

then either x must Granger cause Y or vice-versa. Therefore, we tested for the absence of

Granger causality by estimating the following VAR model:

Yt = a0+ a1Y t-1+…+apYt-p+ b1X t-1+…+bpXt-p+ut…….(*)

Xt= co+c1Xt-1+…+cpXt-p+d1Yt-1+…+dpY t-p+vt……….(**)

Testing

H0:b1=b2=…bp=0

Against

H1: not H0

is a test that Xt does not Granger-cause Yt.

Similarly, testing H0: d1= d2=…= dp=0 against

H1: Not H0 is a test that Yt does not Granger cause Xt.

In each case, a rejection of the null hypothesis implies there is Granger causality between the

variables.

In testing for Granger causality, two variables are usually analyzed together, while testing for

their interaction. All the possible results of the analyses are four:

• Unidirectional Granger causality from variable Yt to variable Xt.

• Unidirectional Granger causality from variable Xt to Yt

• Bi-directional causality and

• No causality

51

The main results obtained from the Pairwise Granger-causality analysis done in the study are the

above table. The results revealed that Gross National Savings(GNS) does not Granger cause

income per capita(PCY) of Rwandans during the period under the study while PCY Granger

causes GNS that is the performance of income per capita have implication on Gross National

Savings in long run. In addition, investments (GCF) granger cause Gross National Savings but

GNS do not Granger cause investment which is different from economic theories.

The consumption (GNEP) and Gross National Savings have bidirectional Granger cause. The

inflation (CPI) and Gross National Savings have also bidirectional causality. The interest rate

does not cause Gross National Savings and vice-versa. Finally, all variables joint together such

as PCY,GNEP,GCF,CPI and RR Granger cause Gross National Savings that is they forecast

savings in Rwanda .

D) Normalized integrating coefficients

The first normalized equation was estimated as shown below. Given the presence of stationary

variables, we use the general model, which aims to minimize the possibility of estimating

spurious relations while retaining long-run information since in some cases, even though two

series have unit root and follow a random walk individually, they move together in long run.

This is the Long run model and the coefficients indicate the long run dynamics relationship.

GNS = 1.0955 PCY + 0.9349 GCF -0.8627 GNEP + 0.9039 CPI +1.1004 RR

After normalization the cointegrating vector on GNS normalized cointegrating coefficients were

estimated as reported in Table 7. According to the first normalized equation, Gross National

Savings (GNS) showed significantly positive relation with income per capita (PCY) in long-run

which suggested that GNS increase of 109% due to increase of income per capita.

Normalized equation above showed that there was an insignificant positive relationship between

Gross National Savings and investment (GCF) which implied that the GCF influence GNS 93%

even if is not significant at 5% degree of significance but is significant at 10%. This study is not

different from economic theories.

52

Table 7: Johansen normalization coefficients

GNS PCY GCF GNEP CPI RR

1 1.0955 .9349646 -.8627762 .9039048 1.100466

SE .3160126 .5179844 .2191895 .322786 .2959319

Z 3.47 1.81 -3.94 2.80 3.72

P>│Z│ 0.001 0.071 0.000 0.005 0.000

Source: developed by the author

GNS was influenced significantly by the consumption (GNEP) negatively that is the

consumption discourages Gross National Savings of 86% in the long term. The inflation captured

by CPI has a positive and significant relationship on Gross National Savings in long run by 90%.

In addition, interest rate has also significant positive influence on Gross National Savings of

110% as it is illustrated in the Table 7.

5.3.2 Short run relationship

When the variables of a VAR are co-integrated, we use a vector error-correction (VEC) model.

In order to capture the short-run dynamics of the model, error correction mechanism was applied.

When we examine the long run cointegration of variables the we are able to estimate the short

run analysis with respect to savings.

The coefficient of ECM explains adjustment speed to long run equilibrium. The sign of _C1must

be negative and significant, proof that significant _C1 is further explanation of long run

relationship among the said variables. It is the more reliable way to examine the co-integration

among variables.

The results of vector error correction model were reported in Table 8. The coefficient of_C1

showed the speed of adjustment of disequilibrium in the period of study. As the error correction

term was significant with negative sign, hence the results of vector error correction model

(VECM) depicted that the adjustments in GNS were due to the first error correction term (_C1)

and the second Equation below showed that the coefficient of _C1 was significant which implied

that GNS adjusted by 12.2 percent in one year to the long-run equilibrium. The results showed

that it took more than approximately 8 years (1/0.122= 8.2) to eliminate the disequilibrium.

53

_DGNS=-0.1598959-0.5063755_DGNS-0.1603911_DPCY+0.3217319_DGCF-

0.2469003_DGNEP +0.1235546_DCPI -0.054979 _DRR -0.1225277_C1

Where _DGNS, _DPCY, _DGCF, _DGNEP, _DCPI and _DRR are first difference of Gross

National Savings, income per capita, consumption, inflation and interest rate respectively. The

error correction term is _C1. As it is reported in the above model the results of previous period of

savings influence negatively the results of next period of savings, this is clear indication of low

level of savings in Rwanda.

The income per capita influence also negatively savings because of low level of income of

Rwandans. The outcome of the study indicates that there is an inverse correlation among per

capita income and GDP growth with national savings.

The Keynesian approach and the permanent income hypothesis it is hypothesized that the

savings rate is positively related to the growth in the national income because more surplus

income means a higher savings rate in the economy. In case of Rwanda per capita income is

negatively related, both hypotheses are not valid in Rwanda for the short run.

The investment, consumption and inflation influence positively savings. Because savings come

from investments that why the results of previous period have positive influence next period. The

Government expenditures as proxy of consumption have positive influence on savings, as

Government expenses encourage people to invest and doing business. In a country, there must be

a certain level of inflation so that economic activities move, that why inflation encourage

savings. Finally, interest rate demotivates savings.

We used different tests to check the validity of the model such as Lagrange multiplier test which

test for residual autocorrelation (Ho: no autocorrelation at lag order), since p-value is 0.68779 we

fail to reject null hypothesis.

Table 8: Vector error correction

Variable Coefficient Std. error chi2 p>chi2

D_GNS -.5063755 .2672874 9.935826 0.2696

D_PCY -.1603911 .1583625 62.28776* 0.0000

54

D_GCF .3217319 .4717829 21.16574* 0.0067

D_GNEP -.2469003 .2418223 21.82866* 0.0052

D_CPI .1235546 .1764493 91.77213* 0.0000

D_RR -.054979 .1638659 13.19555 0.1053

Cons -.1598959 1.012903 -0.16 0.875

_C1 -.1225277 .1133843 60.21905* 0.0000

AIC = 35.82441

HQIC = 36.63311

SBIC = 38.2279

No. of obs = 33

* Significant at 5%

Source: developed by the author

5.3.3 Regression Analysis

Fitting of the model is statistically acceptable. The adjusted R2 is about 52% and p-value<0.05,

which implies that changes in the explanatory variables explain 52% of the variations in the

Gross National savings in Rwanda (1978-2012).

The results indicate that per capita income has a negative effect and statistically significant at 5

percent level. The coefficient having negative sign suggesting that 1 percent change in income

per capita leads to 30.2 percent decrease in savings on the average. The Keynesian approach and

permanent income hypothesis it is hypothesized that the savings rate is surely linked with growth

in the national income because more surplus income means a higher savings rate in the economy.

The GDP growth rate and income per capita income are used alternatively as variables in all the

savings functions. In case of Rwanda per capita income is negatively related both hypotheses are

not valid in Rwanda.

Regarding the effect of the real interest rate (RR), it is found that it has a positive and statistically

insignificant effect at 5 percent level. The coefficient having positive sign suggesting that 1

percent increase in real interest rate leads to 1.68 percent increase in savings on the average.

We also found that the inflation rate (CPI) has a positive and statistically significant effect on

Gross National Savings. The coefficient having positive sign suggesting that 1 percent change of

55

inflation leads to 50.7 percent increase in savings. This provides support of precautionary

motives for saving in the face of increased economic uncertainty in Rwanda. In addition, higher

inflation rates may increase savings ratio through its effect on the distribution of income in favor

to entrepreneurs where their marginal propensity to save is higher than the low-income class.

High inflation will also increase profits, which if it is reinvested will result in increasing of

domestic savings of course increase in gross savings.

Table 9: OLS Test for gross national savings (a)

Variable Coeff Std. error t P>|t|

PCY -.3022491 .097447 -3.10 0.004

GCF .3014164 .2431513 1.24 0.225

GNEP -.0057255 .105505 -0.05 0.957

CPI .5078291 .1257929 4.04 0.000

RR .0168223 .1155292 0.15 0.885

Cons 5.020006 11.55019 0.43 0.667

Number of obs = 35

F( 5, 29) = 8.10

Prob > F = 0.0001

R-squared = 0.5828

Adj R-squared = 0.5108

Root MSE = 3.706

Source: developed by the author

We found that the consumption (GNEP) has a negative and statistically significant effect on

Gross National savings at 5% in short run. The coefficient having negative sign suggesting that 1

percent change in consumption leads to 0.5 percent decrease in savings on the average. We

finally investment (GCF) have positive effect on savings with statistical insignificance influences

on savings, the coefficient having positive sign suggesting that 1 percent increase in investment

capita leads to 30.1 percent increase in savings on the average.

We used different tests to check the validity of the model such as Breusch-Pagan / Cook-

Weisberg test for heteroskedasticity (Ho: Constant variance), Breusch-Godfrey LM test for

56

autocorrelation (H0: no serial correlation), Durbin's alternative test for autocorrelation (H0: no

serial correlation) ho is accepted since p- value is 0.5527, p-value is 0.5416, p-value is 0.5831

respectively.

5.3.4 Economic Growth and Savings relationship

Table 10 reports the final results of the estimated model (2), together with a set of commonly

used diagnostic statistics. First, fitting of the model (2) seems to be statistically acceptable.

The adjusted R2 is about 70%, which implies that changes in the explanatory variables explain

70% of the variations of economic growth in Rwanda from 1978 until 2012. Furthermore, the

results revealed that savings have a negative and significant influence on economic growth

which is different from what economic theories expectations.

Table 10: OLS Test for GDP per capita (b)

Variable Coeff Std. error t P>|t|

GNS -.8241591 .2657141 -3.10 0.004

GCF 1.048908 .363066 2.89 0.007

GNEP -.5807019 .1368478 -4.24 0.000

CPI 1.100211 .1601748 6.87 0.000

RR .4175939 .174377 2.39 0.023

_cons 50.72895 16.65529 3.05 0.005

Number of obs = 35

F( 5, 29) = 17.15

Prob > F = 0.0000

R-squared = 0.7473

Adj R-squared = 0.7037

Root MSE = 6.1196

Source: developed by the author

The estimated parameters of the explanatory variables point out that saving has a negative effect

and statistically significant at 5 percent level. The coefficient having negative sign suggesting

that 1 percent change in savings leads to 82.4 percent decrease in income per capita on the

average. Regarding the effect of the real interest rate (RR), it is found that it has a positive and

57

statistically significant effect at 5 percent level. The coefficient having positive sign suggesting

that 1 percent increase in real interest rate leads to 41.7 percent increase in income per capita on

the average.

We also found that the inflation rate (CPI) has a positive and statistically significant effect on

income per capita. The coefficient having positive sign suggesting that 1 percent change of

inflation leads to 50.7 percent increase in savings.

We found that the consumption (GNEP) has a negative and statistically significant effect on

income per capita at 5% in short run. The coefficient having negative sign suggesting that 1

percent change in consumption leads to 58 percent decrease in income per capita on the average.

Lastly we found that investment (GCF) have positive effect on income per capita with statistical

significance influences on income per capita, the coefficient having positive sign suggesting that

1 percent increase in investment capita leads to 104 percent increase in income per capita on the

average.

We used different tests to check the validity of the model such as Breusch-Pagan / Cook-

Weisberg test for heteroskedasticity (Ho: Constant variance), Breusch-Godfrey LM test for

autocorrelation (H0: no serial correlation), Durbin's alternative test for autocorrelation (H0: no

serial correlation) ho is accepted since p- value is 0.2995, p-value is 0.4390, p-value is 0.4850

respectively.

58

CHAPTER 6: CONCLUSION

The purpose of this paper is to examine the determinants of Rwandan's national savings. We

utilized Johansen co-integration and Granger causality testing approaches to check the

robustness for long run relationship and Error Correction Model (ECM) for short run dynamics

during the 1978-2012. In this section, we summarized the main results, and then the general

conclusion and we end up by some suggestions.

6.1. Summary

The objective of this study has been analysis of the determinants of savings in Rwanda using the

appropriately available econometric methodologies of a 35-years period (1978-2012). The unit

roots test to test the stationarity of all time series, co integration analysis, Granger causality,

Error-Correction Model of the determinants of savings function as well as others tests were used

in this study in order to get better and reliable results.

The First chapter strives to outline different points which guide the researcher throughout the

study; it is composed of the introduction, background, statement of the problem, objectives of the

study, research hypotheses, the scope of the study, significance of the study, and organization of

the study. The second chapter talks about the Literature Review, the definitions of conceptual

terms on determinants of savings as well as ascertaining what has been discovered on the topic

under the study.

The third chapter talks about saving behaviour in Rwanda. The fourth chapter describes the

methodology used in this study. The fifth chapter focuses on discussion of findings, such as

indicating trend of variables in the model, unit root test, co integration analysis, and VECM and

regression analysis. Finally, the study has ended by drawing the summary, conclusion and

suggestions.

6.2. General Conclusion

This study has analyzed the determinants of Gross National Savings in Rwanda during the period

1978-2012. The unit root test was used to test the stationary of all time series, and after that the

first the co-integration analysis is carried out. And the pair-wise causality is used.

59

Finally, the VECM was use to check the short run relationship and regression model. The results

of the study provide evidence that Gross National Savings in Rwanda is determined by the

following factors.

First, the growth of per capita income is found to have positive influence on gross savings, in the

long-run but negative in the short term. Second, investments have a positive effect on gross

savings both in short and long run. This explains the importance of investments on savings.

Third, the consumption as proxied the government expenditures show a negative and significant

effect on Gross savings. This explains how much consumption affect savings as it has been

shown by John M. Keynes’s consumption theory.

Fourth, inflation proves to have positive impact on the level of gross savings in Rwanda, This

provides support of precautionary motives for saving in the face of increased economic

uncertainty in Rwanda. In addition, higher inflation rates may increase savings ratio through its

effect on the distribution of income in favor to entrepreneurs where their marginal propensity to

save is higher than the low-income class. High inflation will also increase profits, which if it is

reinvested will result in increasing of domestic savings, thus gross savings increase.

Finally, interest rate recorded a negative effect on the short run and positive effect on the long

run. At the end, we would like to emphasis that this research found the negative relationship

between gross savings and economic growth in short run, however the relation is positive in long

run. The error correction term was significant with negative sign, hence the results of vector

error correction model (VECM) depicted that the adjustments in GNS were due to the first error

correction term (_C1) and the coefficient of _C1 was significant which implied that GNS

adjusted by 12.2 percent in one year to the long-run equilibrium. The results showed that it took

more than approximately 8 years (1/0.122= 8.2) to eliminate the disequilibrium.

6.3. Suggestions

After examining the determinants of savings and review the current Rwandan savings situation,

there are key suggestions which are evident. These can make up the basis for the Savings

Mobilisation Strategy in Rwanda and when addressed properly strengthen the financial

infrastructure and increase savings and ultimately growth.

60

6.3.1. Encouraging Economic Growth.

It is necessary to stimulate growth by promoting good macroeconomic management, creating an

attractive environment for private investment (domestic and external) in order to to increase the

productive base of the economy in order to promote real income growth and reduce

unemployment. Therefore, a robust growth induces higher rates of national savings and capital

formation.

6.3.2. To curb inflation

Besides, inflation rate (CPI) is positively significant in impacting on volume of savings

mobilized in Rwanda, hence there is need to stabilize its bad effect via minimizing all

inflationary pressures on the economy.

6.3.3. Macroeconomic Stability

Macroeconomic stability is vital in order to allow other factors flourish; if the superstructure is

unstable the rest of the system suffers. There need to moderate to medium inflation and nominal

interests rates with a positive real interest rate of 2 to 4%+ will be conducive to savings,

investment and growth.

6.3.4. Institutionalized Savings

Unless institutionalized, savings will note help in investment and hence economic growth.

Savings must be instituted throughout both the public and private sector, they cannot be cycled

and resource will not and cannot be directed to the growth areas. It is imperative that savings

become second nature to government organizations, individuals and institutions alike. If the

savings do not exist then the funds for investment and growth do not exist and the system

becomes unstable.

6.3.5. Expansion of the Financial Infrastructure and Intermediation

The lack of access to the financial infrastructure hinders savings mobilisation; the FinScope

survey showed 86% of adults in Rwanda have no access to formal banking products. This

equates to 52% of the population being financially excluded.

61

There are numerous examples from around the world which have shown how increasing access

to the financial sector increases savings and quickly.

6.3.6. Secure and Diversified Means of Savings

A diversified and secured financial sector, increases competition among institutions, provides a

larger variety of savings products in which savers at all levels can choose from and take part in,

savers must feel that they have choice and security. A competition of products on offer such as

insurance, mutual funds, pension funds and other long term savings instruments, provide an

attractive vehicle for individual savers with the main role to improve the allocation of savings;

thus, heavy regulation and limited portfolio distribution results in comparatively low returns and

little flexibility to react to market developments.

6.3.7. Building Capacity and Efficiency of Intermediation

The strategies for savings mobilization and the development of the Rwandan financial sector

must go beyond policies and include building capacity and efficiency. Increased efficiency

allows for operation costs to be reduced which can be passed on through credit costs and can

increase the success rate of the fund usage in growth and profitable sectors. Additionally, there

must be continued training and education of those directly involved on the institutional side and

with the public.

6.3.8. Increased Awareness and Positive Perception of Tangible Benefit of Savings

The concept in this recommendation is simple, if the population is not aware of the opportunities

or the benefits the size of the bankable population won’t increase. The government and the

financial sector must work together to communicate the benefits of savings to individuals and to

the country. By increasing the services financial institutions can provide and their visibility allow

for increased awareness and confidence, which leads to a more positive perception. All these

efforts will strengthen the financial infrastructure, mobilise savings and help to create a culture

of savings in Rwanda.

Finally, with regard to future researches, future researchers may investigate the possible effects

of demographic factors on savings behavior in Rwanda, which we could not investigate in this

study due to the lack of information available to us about time series data of Rwandan

demographic factors.

62

REFERENCES

Aghevli, B. (1990). The Role of National Savings in the World Economy: Recent Trends and

Progress. IMF Occasional Papers, No. 67

Ando, A. and Modigliani F. (1963). The Life Cycle Hypothesis of Saving Aggregate Implications

and Tests. American Economic Review 53.

Andrew et al (1998). Macroeconomics. European edition Addison-Welsey Longman London.

Bailey Kenneth D. (1978). Methods of social Research. (2nd Ed). London: Free press

BERNARD S. Phillips (1996). Social Research Strategy and Tactic. Macmillan publishers, New

York

Björn Andersson( ). On the Causality Between Saving and Growth:

Browning, M. &Crossley, F. (2001). The Life-Cycle Model of Consumption and Saving. Journal

of Economic Perspectives, Volume 15, Nr. 32

C.R COTHARI(2004). Research methodology,methods and techniques. New age publisher

limited, 2nd edit.New delhiCambridge, MA 02138

Campbell and Shiller (2014). Interpreting Cointegrated Models. Working Paper No. 2568

Cavaliere et al (2012). Lag length selection for unit root tests in the presence of nonstationary

volatility. Cowles foundation discussion paper no. 1844

Chete, L.N. (1999). Macroeconomic Determinants of Private Savings in Nigeria. NISER

Monograph Series, No. 7.

Christoffersen and Diebold (1998). Cointegration and Long-Horizon Forecasting. Journal of

Business and Economic Statistics, 16, 450-458.

Damodar N. Gujarati(2006). Essentials of econometrics, Vol. 2, McGraw-Hill/Irwin

De Mello and Nell (2005). The forecasting ability of a cointegrated VAR system of the UK

tourism demand for France,Spain and Portugal.

Deaton, A. and Christina Paxson(2000). Growth and Saving Among Individuals and households.

The Review of Economics and Statistics, pp212-225

63

Deaton, Angus (1991). Saving and Liquidity Constraints, Econometrica, pp.1221-1248

Economic Research Forum / Istanbul

Engle and Granger(1987). Cointegration and error correction: representation,estimation and

testing. Econometrica,vol.55,no2,pp 251-276

Eric and Muhsin (2001). A System Test of McKinnon's Complementarity Hypothesis with an

Application to Turkey. LE11 3TU England

Etienne Musonera and Egide Karuranga(). Institution fitness, investment and economic growth

in Rwanda. Université Laval-Quebec,

Faculty of Commerce, Benha University, No. 1

Finescope(2012). Financial inclusion in Rwanda (2008-2012).Kigali

Francois Kanimba. (2005). Domestic Saving and Investment, current trends gaps, weakness and

the way forward. retrieved on 12/1/2014 from http:/www.bnr/investeds

Friedman, M. (1957). A Theory of the Consumption Function.Princeton. University Pressfor the

National Bureau of the Economic Research.

GrawitzMr. (1986). Methods of social science. Paris, Dalloz

Gujarati (2004). Basic Econometrics. (4th Ed). New York: McGraw- Hill.

Gujarati (2009). Basic Econometrics, 5th edition. McGraw Hill, ISBN 978-007-127625-2.

Gujarati (1992). Essentials of Econometrics, International Edition. McGraw Hill, ISBN 0-07-

112624-4.

IPAR and ACET (2012). Rwanda case study on economic transformation.Kigali

Jappelli et al (2003). The Age-Saving Profile and the Life-Cycle Hypothesis. Vol. 10 Working

Paper Nr. 9, April 2003, Retrieved March 12, 2009,

Karl-Erik Wärneryd (1989). Improving Psychological Theory through Studies of Economic

Behaviour: The Case of Saving. Article first published online: 22 JAN 2008

Keynes, J.M. (1936). The General Theory of Employment, Interest and Money. London:

MacMillan and Company Ltd

64

Khaled and Thirlwall ( 1999), Explaining differences in the domestic savings ratio. Across

countries: a panel data. Published in journal of development studies, pp.31-

52

Khan and Abdullah (2010). Saving Determinants in Malaysia. Jurnal Ekonomi Malaysia.

Konya (2005). Saving and Growth: Granger Causality Analysis with Bootstrapping On Panels

of Countries. Journal of Economic Research pp.231–260

Kudaisi(2013).Savings And Its Determinants In West Africa Countries. Journal of Economics

and Sustainable Development.

Loizides and Vamvoukas (2005). Government expenditure and economic growth: evidence from

trivariate causality testing.

Mahmoud A. Touny(2008). Determinants of Domestic Saving Performance in Egypt: An

Empirical Study.Journal of Commercial Studies and Researches,

Mansor et al (2001). Macroeconomic variables, exchange rate and stock price: a Malaysian

perspective . IIUM Journal of Economics and Management pp.141-63

Masson et al (1998.International Evidence on the Determinants of Private Saving. World Bank

Economic Review, 12(3), pp. 483-501

Meghir, C. (2004). A retrospective on Friedmans`s theory of permanent income. The Journal of

Development Economics. 144 (june)

Mikesell, R. and Zinser, J. (1973).The Nature of the Saving Function in Developing Countries. A

Survey of the Theoretical and Empirical Literature. Journal of Economic

Literature,11

MINECOFIN(2009). Rwanda Development Indicators. Kigali

MINECOFIN(2007).Economic Development and Poverty Reduction Strategy: Lessons Learned

2008 – 2011,Kigali

MINECOFIN, (2007). National Savings Mobilisation Strategy. Kigali

65

Modigliani, F. and R. Brumburg, (1954). Utility Analysis and the Consumption Function in post

Keynesian Economics. K.K. Kurihara (ed), New Brunswick, NJ:Rutgers

University Press.

Montiel, Peter. (2007).The real exchange rate and the saving rate in an intertemporal model of

the current account. Williams College, mimeo.

Muhammad et al (2008). Economic Growth and Its Determinants in Pakistan. The Pakistan

Development Review pp. 471–486

Muradoglu and Taskin (1996). Differences in Household Savings Behaviour: Evidence from

Industrial and Developing Economies. The Developing Economies.34

N. Gregory Mankiw. (2002). Macroeconomics.( 5thed). New York: Irving place

Ndambiri et al (2012) .Determinants of economic growth in sub-saharan africa: A Panel Data

Approach International journal of economics and management

scinces.pp18-24

Nga M. (2007). Investigative analysis into the saving behavior of poor households in

developing countries: with specific reference to South Africa.

NurkseRagnar .(1966). Problems of capital formation in underdeveloped countries. Journal of

Economic Literature,15

Obwona and Ssentamu(). Nature and determinants of domestic savings in Uganda.Research

series no2,Makerere

OZCAN et al (2003). Determinants of private savings behavior in Turkey. Applied Economics,

pp.1405–1416

Pilinkus and Boguslauskas(2009). The Short-Run Relationship between Stock Market Prices and

MacroeconomicVariables in Lithuania:pp. 23 – 34

Richard et al (1991). Precis of strategic management and economics. Strategic Management

Journal pp.5-29

Richard H, Patten and Joy K. Rosengard. (1991).Progress with profits, the development of rural

banking in Indonesian. (4th Volume), ICS Press Son Francisco,California

66

Samuelsom, P.A.and Nordhaus,W.D.(1995).Economics.New York: McGraw-Hill

Hebbel et al (1999). Mobilising Domestic Savings and Resources for Capital Formation in

Nigeria. First Bank Review

Sebastian OfumbiaUremadu (2007).Core Determinants of Financial Savings in Nigeria. An

Empirical Analysis for National Monetary Policy Formulation.

Sohail and Hussain (2009). Long-run and short-run relationship between macroeconomic

variables and stock prices in Pakistan. The Case of Lahore Stock Exchange.

Soren (2002). The interpretation of cointegrating coefficients in the cointegrated vector

autoregressive model. ISSN1600-2660 HCO Tryk

Üçer (2009). The Evolution and Determinants of the Turkish Private Saving Rate:What Lessons

for Policy? Publication No: EAF-RR/09-01).

Umoh, O.J. (2003). An Empirical Investigation of the Determinants of Aggregate National

Savings in Nigeria.

UWAMARIYA(2010). Analysis of the determinants of savings in Rwanda (1995-2010). Inatek

UNCTAD(2006). Investment Policy Review Rwanda. New York and Geneva ISBN-13: 978-92-1-

112709-6

67

APPENDICES

i

Appendix I: Data on Variables during the Period of 1978-2012

YEAR RR CPI GNEP GCF PCY GNS

1978 8.522918 13.2705338 109.0191 16.60703 5.600792 17.92342

1979 6.57744 15.6733294 101.7053 12.03078 8.171847 22.68183

1980 10.12492 7.24939299 111.9456 16.14038 5.302376 13.26608

1981 5.374859 6.45066447 111.9292 13.29909 2.006469 9.754561

1982 8.215635 12.5651365 112.5979 17.77822 -1.37591 13.47702

1983 10.7866 6.59300214 109.1772 13.53058 2.584528 12.44386

1984 -2.87248 5.36957039 107.0328 15.80657 -7.61361 15.2134

1985 8.904459 1.75933006 109.1049 17.30935 0.324715 14.50917

1986 22.6093 -1.11706657 107.5702 15.87481 0.640749 14.13373

1987 12.25321 4.13301491 111.6912 15.66461 -4.99076 9.147336

1988 9.561837 2.97864356 110.9837 14.49244 -0.0234 9.172479

1989 6.365716 1.01027825 111.1492 13.42602 -2.49989 7.535286

1990 -0.25983 4.18576373 108.4549 14.64998 -2.27616 11.32506

1991 3.497218 19.6371658 110.7388 14.02175 0.855033 12.17545

1992 8.746361 9.56041188 112.6946 15.6341 12.80512 11.42308

1993 0.991914 12.3543888 115.3285 16.74702 -0.85539 10.20764

1994 0.991914 15.312211 158.4905 9.982509 -47.3142 33.65875

1995 1.300591 45.6628994 120.6706 13.40913 36.76702 20.21856

1996 6.868452 7.41137174 120.1673 14.37043 7.692939 14.21047

1997 0.517635 12.0154225 117.8728 13.80978 4.330004 10.54039

1998 14.57699 6.21006709 117.6243 14.80766 -1.75487 8.010251

1999 28.66086 -2.4059321 118.3034 13.14875 -1.78044 5.068694

2000 13.76913 3.8995298 118.5585 13.37598 1.318637 6.355106

2001 16.15763 3.34285507 115.7979 13.73552 4.148203 8.055477

2002 22.89378 1.99258542 116.6792 13.48131 10.637 7.285299

2003 -4.59417 7.44970014 115.0932 13.8539 -0.08991 9.369542

2004 2.484144 12.2507103 113.6356 15.02818 5.464734 14.8076

2005 6.034051 9.01408918 113.7926 15.77887 7.001424 15.11446

2006 5.694459 8.88282655 114.2259 16.00155 5.950932 10.36037

2007 4.917078 9.08072206 114.395 18.04045 4.736554 14.24578

2008 2.772403 15.4449312 115.7366 22.68581 7.95549 17.18829

2009 6.464107 10.3648345 119.3923 21.57334 3.149346 12.97864

2010 5.847457 2.30914619 120.5417 20.96962 4.180599 11.35934

2011 2.230807 5.67068273 117.3743 21.44232 5.252137 17.11014

2012 1.594173 6.27090301 120.4344 22.84879 5.026731 11.54641

Source: World Development Indicators (2012), NBR

ii

Appendix II: Additional Results on the Determinants of savings in Rwanda

ALL 73.477 12 0.00000 D_rr 1.170 2 0.55717 D_cpi 18.477 2 0.00010 D_gnep 3.962 2 0.13794 D_gcf 3.315 2 0.19062 D_pcy 0.395 2 0.82069 D_gns 46.158 2 0.00000 Equation chi2 df Prob > chi2 Jarque-Bera test

. vecnorm, jbera skewness kurtosis

ALL 32.238 6 0.00001 D_rr .20265 0.226 1 0.63461 D_cpi 1.324 9.641 1 0.00190 D_gnep -.81489 3.652 1 0.05599 D_gcf .75113 3.103 1 0.07814 D_pcy -.19182 0.202 1 0.65281 D_gns 1.6741 15.414 1 0.00009 Equation Skewness chi2 df Prob > chi2 Skewness test

ALL 41.238 6 0.00000 D_rr 3.8285 0.944 1 0.33128 D_cpi 5.535 8.836 1 0.00295 D_gnep 3.4745 0.310 1 0.57795 D_gcf 3.3925 0.212 1 0.64532 D_pcy 2.6255 0.193 1 0.66056 D_gns 7.7286 30.744 1 0.00000 Equation Kurtosis chi2 df Prob > chi2 Kurtosis test

Exogenous: _cons Endogenous: gns pcy gcf gnep cpi rr 2 -511.713 58.565* 36 0.010 1.8e+08 35.7402 36.9303 39.2774 1 -540.995 141.29 36 0.000 9.3e+07* 35.333* 35.9739* 37.2377* 0 -611.641 7.3e+08 37.4328 37.5243 37.7049 lag LL LR df p FPE AIC HQIC SBIC Sample: 1980 - 2012 Number of obs = 33 Selection-order criteria

. varsoc gns pcy gcf gnep cpi rr, maxlag(2)

iii

Source: developed by author

gnep ALL 13.126 10 0.217 gnep rr .54706 2 0.761 gnep cpi 6.1566 2 0.046 gnep gcf 2.0746 2 0.354 gnep pcy 3.7548 2 0.153 gnep gns 2.4273 2 0.297 rr ALL 14.516 10 0.151 rr gnep 5.4143 2 0.067 rr cpi 2.3432 2 0.310 rr gcf 3.1537 2 0.207 rr pcy 7.0499 2 0.029 rr gns 1.0489 2 0.592 cpi ALL 101.81 10 0.000 cpi gnep 14.326 2 0.001 cpi rr 4.9911 2 0.082 cpi gcf 5.5109 2 0.064 cpi pcy 2.5986 2 0.273 cpi gns 7.12 2 0.028 gcf ALL 16.299 10 0.091 gcf gnep .7614 2 0.683 gcf rr 4.1418 2 0.126 gcf cpi 2.4554 2 0.293 gcf pcy 6.6457 2 0.036 gcf gns 5.4076 2 0.067 pcy ALL 15.219 10 0.124 pcy gnep 4.1243 2 0.127 pcy rr .03244 2 0.984 pcy cpi 1.1651 2 0.558 pcy gcf 1.2002 2 0.549 pcy gns 5.8836 2 0.053 gns ALL 11.025 10 0.356 gns gnep .15388 2 0.926 gns rr 2.8817 2 0.237 gns cpi 1.3713 2 0.504 gns gcf 1.5076 2 0.471 gns pcy .26364 2 0.876 Equation Excluded chi2 df Prob > chi2 Granger causality Wald tests