RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN

Transcription

1 Iteratioal Joural of Iovatio ad Scietific Research ISSN Vol. 12 No. 1 Nov. 2014, pp Iovative Space of Scietific Research Jourals RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN Irfa Hussai Kha 1, Prof. Dr. Bashir Ahmad Khilji 1, ad Ghulam Abbas 2 1 Departmet of Ecoomics, Presto Uiversity Kohat, Islamabad Campus, Pakista 2 Departmet of Sociology, Govermet College Uiversity Faisalabad, Pakista Copyright 2014 ISSR Jourals. This is a ope access article distributed uder the Creative Commos Attributio Licese, which permits urestricted use, distributio, ad reproductio i ay medium, provided the origial work is properly cited. ABSTRACT: The preset study is udertake to examie the impact of iterest rate, iflatio rate ad per capita icome o household s savig i Pakista for the period of 1981 to The study uses simple descriptive statistics to aalyze the tred i household s savig. It also uses advace techiques of time series ecoometrics for modelig the household s savig fuctio. It icludes Grager causality test (Block erogeeity test), Johaso Co-itegratio, ad geeral to specific methodology for modelig the appropriate Vector Error correctio model. The applicatios of the techiques are based o the order of itegratio of time series data. It is, therefore, the reaso that Augmeted Dickey Fuller test has bee carried out for testig the order of itegratio of the time series variables, ad thereby, selectig the estimatio techiques for estimatig parameters of the model. The careful ispectio of the data reveals that all the variables are o-statioary at level, statioary at their first differece. Based o this characteristic of the data Johso co-itegratio techique has bee used for estimatio, rather tha simple OLS which is biased at the preset case. The empirical results show that there exist equilibrium i log ru relatioship amog per capita icome, iflatio, iterest rate ad household s savig. There is also sigificat error correctio compoet which reveals that system reverts to its log ru equilibrium i respose to exteral shocks. These results are as per ecoomic theory. O the basis of empirical evidece, it is suggested that iflatio rate may be kept low so that real iterest rate may be stimulated for acceleratig rate of household s savig rate. KEYWORDS: Ecoomy, Iterest rate, per capita, icome, household, savig, Pakista. 1 INTRODUCTION Savig is the differece betwee icome ad cosumptio. It is the postpoemet of curret cosumptio for future cosumptio istead of or it is the substitutio of curret cosumptio for havig more budles of goods i future. Savig ad cosumptio are the two sides of the same coi. The decisio to save ad cosume are iter-related, ad there are a umber of factors which may ifluece idividual s preferece for savig ad spedig. These factors iclude iterest rate or rate of retur o savig, iflatio rate, spedig habits of a society etc. Natioal savig is the amout of moey which is ot speds by a atio or coutry. It is the differece betwee atioal icome of a coutry mius its cosumptio o various goods ad services. Natioal icome of a couty ca be calculated by differet ways such as atioal icome at curret market prices, costat market prices or base year prices ad factor cost. The cocept of atioal savig ca also vary accordigly, for istace atioal savig at market prices would be obtaied or omial atioal savig is resulted. Similarly, real atioal savig or atioal savig at costat prices is the amout obtaied after deductig aggregate cosumptio from atioal icome at base year prices, (Bishop & Cassidy, 2012). Natioal savig has three mai compoets such as household savig, corporate savig ad govermet savig. Household savig is the aggregate amout which is saved doe by all households i a couty or it is the summatio of savigs of all households i a coutry. It is obtaied by deductig household cosumptio from after tax household icome or it is the differece of household disposable icome ad it s spedig o various durable ad o-durable goods ad services. The household disposable icome is the differece of persoal icome from all sources ad direct taxes or icome taxes. Household s spedig is the expediture of a household for the purpose of obtaiig direct satisfactio or utility. Household savig is the aggregate amout of icome which is ot spet by all Correspodig Author: Ghulam Abbas 295

2 RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN households i a coutry, (Bishop & Cassidy, 2012). Corporate savig is the secod importat compoet of a atioal savig. It is the amout of after tax profit which is ot distributed amog shareholders. The cocept of after tax profit ca be defied as the amout of gross profit of a corporatio mius corporate icome tax. Like idividual icome tax, there are corporate icome taxes which are imposed o the icome or profit of corporatio as its ame idicates. It is very importat for a corporatio to keep aside amout of profit as savig so that it ca be used whe it is highly eeded. It reduces corporatio depedece o market or bak for fiacig her ivestmet eeds. It, therefore, reduces its elasticity of demad for loa i various busiess eviromets, (Bishop & Cassidy, 2012). Pakista s savig rate is ot very impressive. It has bee classified amog those coutries which are characterized by low savig rate. The highest savig rate of 21.3 percet i 1972 is lower tha the regioal coutries with similar icome ad social structure, (State Bak of Pakista, ). There are a umber of factors which are probably cosidered to be the mai determiats of low savig rate i Pakista. O the basis of ecoomic theories, these factors iclude ecoomic growth, demographic structure, fiscal stace of the govermet, real rate of retur, expected iflatio rate, state of fiacial sector, cultural ad psychological issues etc, (State Bak of Pakista, 2006). Savig rate is also depedet o real rate of retur o deposits which depeds o omial iterest rate, iflatio rate ad expected iflatio rate. If omial iterest is higher tha the iflatio rate, ad if it is expected that iflatio rate would be cosistetly lower tha omial iterest rate, the savig would icreases because real rate of retur o deposit would be positive ad et household s wealth would grow i future. Similarly, there are cultural ad psychological factors which may also affect savig of a coutry, such as, cosumptio habits. If society is cosumptio orieted i.e. if society prefers to sped lavishly may be either due to lack of foresight or lack of visio, the savig rate is expected to be lower. STATEMENT OF THE PROBLEM Natioal savig plays vital role i ecoomic growth ad developmet of a couty. It is the supply of fuds for ivestmet projects of a coutry. If there is regular supply of loa available for ivestors, there would be smooth busiess operatios ad hece would result i expasio of productive capacity of a ecoomy. If there is expasio of productivity at higher sustaied rate. With opeig of ew ad ew employmet opportuities of a coutry, uemployed labor would be absorbed i productio process ad thereby icrease the overall productivity of iputs. As stadard of livig directly depeds o rate of expasio of atioal ecoomy, therefore, atioal ecoomy would share the aggregate prosperity amog the eglected ad margialized segmet of the society. The log ru impact of high sustaied savig rate ca be viewed as lower poverty, abudat facilities of health ad educatio which result i eve more productive huma resources i future, high growth with low uemploymet ad massive icetive for research ad developmet, techical spillover ad higher stadard of livig. Household s savig costitute a sigle larger compoet of atioal savig of Pakista. Fluctuatios i household savig ca greatly affect the performace of atioal ecoomy through ups ad dows i availability of atioal savig for ivestmet i the coutry. I other words, household savig is the largest compoet of atioal savig which ca cause busiess cycle with sigificat fluctuatios i plaed ivestmet i Pakista. I this cotext, the determiats of household s savig are vital to ivestigate i the Pakista or i other words, it is very urget to kow the leadig factors which ca fluctuate the household savig at macroecoomic level. The respose of household s savig to various macroecoomic eviromets i both short ru ad log ru is crucial to be kow to policy makers i order to prepare policy packages for smooth uplift of atioal ecoomy. There are a umber of studies which are related to the determiats of atioal savigs ad its lik to ivestmet ad ecoomic growth i various settigs; however, the studies related to aalyses of factors which ca cause cosiderable fluctuatio i household s savig are ot properly aalyzed usig loger time series data with stadard macro-ecoometric models. This study is a attempt to ivestigate the short ad log ru respose of household savig to the leadig factors i Pakista ad a costitute useful ivestigatio i the light of its importace for atioal ecoomy. OBJECTIVES OF THE STUDY To aalyze patter of household s savigs i Pakista, over a log period of time. To assess the role of iterest rate i determiatio of level of household s savigs i Pakista 2 REVIEW OF LITERATURE Summer (1982) estimated the iterest elasticity of savig that depeds o the life-cycle hypothesis from a cosumptio fuctio icorporatig data o the after-tax real rate of iterest. It shows a positive relatioship betwee savig ad iterest rate. Ali (1985) exteds his earlier study to estimate icome ad price elasticities of household s cosumptio ad savig for Pakista usig Exteded liear Expediture system approach. He exteds sample from 1979 household s icome ad expediture survey. He quatifies the cosumer resposiveess to the chage i icome ad relative commodity prices usig ISSN : Vol. 12 No. 1, Nov

3 Irfa Hussai Kha, Prof. Dr. Bashir Ahmad Khilji, ad Ghulam Abbas sigle equatio model. He estimates oe demad equatio at a time. This approach has a umbers of advatages, but its major shortcomig lies i the difficulty ad of drivig cosistet estimates of parameters because he igores the problem of iterdepedece amog differet goods. The empirical results ca be used for studyig polices effects at micro level. Khatkhate (1988) divides developig coutries i three groups, depedig o whether they had positive real iterest rate, moderately egative real iterest rate ad strogly egative real iterest rate. He estimates the respose of savig rate to these types of iterest rates ad cocludes that savig rate is high for positive real iterest coutries vice versa. Kazmi (1990) examies the causes of low savig rate i Pakista compared with the persistetly risig rates i the regioal coutries. The factors of low savig rate are the outcomes of may ecoomic, demographic, political, social, cultural ad religious factors. I which some factors are quatitative ad other are qualitative. However socio-political ad religious factors are basically the leadig factors resposible for low savig rate i Pakista. Burey & kha (1992) test the hypothesis of low savig i Pakista for last two decades. They argue that there are may factors behid it. These iclude socioecoomic ad admiistrative factors, which are resposible for it. High propesity to cosume due to itroductio of ew products i the market that is also a major factor for low savig i the ecoomy. These factors slow dow the savig rate i ecoomy. Empirical results show that savig deped o it depedecy ratio, educatio, employmet, opportuities, earig status of the household, professio, residetial factor ad secodary earig sources. Lavoto (1992) aalyses the impact of the iterest rate o the savig rate i the developig coutries. He also compares the iterest elasticity of savig i the Uited States with the developed coutries. Wealth effect outweigh i importace with the elasticity of substitutio betwee preset ad future cosumptio. Study reveals that iterest elasticity of savigs is high ad the value of substitutio s elasticity is slow. Siddiqui & Siddiqui (1993) argue that household s savig is vital for ecoomic for progress ad prosperity. I his study, they discuses that i rate of the household s savig i the total atioal savig is about 83 percet ad the Gross domestic product is about 6.6 percet to 10 percet. This is the highest rate of household s savig that emerged i the history of Pakista. Empirical results idicate that there are may factors, which are resposible for it. Firstly, capital formatio helps i icreasig labor productivity ad future s icome. Secodly, chages i the ecoomic ad demographic factors also icrease the household s savig more rapidly tha cosumptio. Smyth (1993) implies that there are ecoomic factor that motivate household s savig. O the oe had, future demads for credit ca motivate household s savigs but o the other had, supply of savig also creates demad for credit. The study is carry out i the rural area of Nakuru District The choice of a rural regio as area of study is based o the fact that a majority of the household s live i the rural areas, ad hece the ivestmet demad for loa will cotribute to the icrease household savig as well as growth ad developmet agricultural sector. The research study aalyzed savigs behavior amog is moetary savigs amog farmers, etrepreeurs ad teachers whose members live i the rural areas of Nakuru District over the period of November 2005 to May Azam et al (2010) examie relatioship betwee savig ad ivestmet which is importat for the socioecoomic developmet of the coutry via capital accumulatio. The study shows that durig there has bee dowward tred i household savig for may developig coutries. The empirical fidig of study reflects that there is a positive impact of the per-capita icome o the atioal savig ad there is egative impact of iflatio. The coclusio of the study is simple that there may be icrease i the level of the per-capita which ca icrease household savig ad ca accelerate ecoomic. They also discuses that uemploymet, icrease i the populatio growth, decrease i the ecoomic growth ad other supply shocks especially i food ad eergy supply also slow dow the savig rate i Pakista. PROFILE OF PAKISTAN S ECONOMY Pakista came ito existece at 14 th August of It is situated i South Asia. Its eighborig coutries iclude Idia, Ira, Tajikista, Afghaista, ad Chia. Pakista is a developig coutry ad oe of the 27 th largest ecoomy i the world 1 (Ecoomy Watch, 2010). The value of Pakista s GDP based o purchasig power parity (PPP), is $488.4 billio i fiscal year 2011 as compared to $474 billio ad $459.9 billio i fiscal year of 2010 ad 2009 respectively 2. Real growth rate of GDP is 3 percet i fiscal year 2011, whereas, it recorded as 3.1 percet ad 1.7 percet i the fiscal year of 2010 ad 2009 respectively (Cetral Itelligece Agecy - CIA, 2012). Pakista is depedet o agriculture sector 3 which accouts for more tha oe fifth of output ad two fifths of employmet level (Cetral Itelligece Agecy - CIA, 2012). It cotributes 21 1 It is o the basis of Purchasig power parity (PPP). 2 Estimates are i US dollars 3 Maufacturig is the third largest sector i Pakista s ecoomy. It accouts 18.5 percet to GDP ad 13 percet to total employmet. Large Scale Maufacturig domiates the whole idustrial sector. It cotributes to GDP by 12.2 percet ad cotais 66 percet of total sectoral share (Miistry of Fiace, ). ISSN : Vol. 12 No. 1, Nov

4 RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN percet to GDP ad geerates productive employmet opportuities for 45 percet of coutry s labor force. Almost, 60 percet of the rural populatio depeds for their livelihood o agriculture sector, (Miistry of Fiace, ; p-17). Table No.1 Shows a brief sketch of behaviors of leadig macro-ecoomic variables i Pakista, amely GDP, agriculture, maufacturig, commodity producig sector, services sector, atioal savig ad percetage chage i cosumer price idex (CPI). Average aual growth rates i 1960s to 2000s are reported. The behavior of Gross Domestic Product (GDP) is volatile. GDP growth rate i 1960s is 6.8 percet, ad falls to 4.8 percet i 1970s. It makes it mius two percet fall durig 1970s. GDP growth recovers to 6.5 percet i 1980s; however, it falls back to 4.6 percet i 1990s. There has bee oly 0.7 percet rise i growth durig 2000s as compared to 1990s. There are several factors which may be cosidered as pullig factors of GDP growth such as war with Idia durig 1960s, agai there had bee tesios i East Pakista durig 1970s, ad eve East Pakista got a status of a idepedet state as a Bagladesh. There are policies of atioalizatio ad privatizatio i 1980s ad 1990s coupled with Afgha war i 1980s. Recet war o terror which started i 2001 ad cotiue till date also suffered Pakista ecoomy a lot. Table No. 3.1: Growth Rates of Leadig Macro-Ecoomic Idicators of Pakista (Aual average at costat factor cost i percet) Idicators/Time 1960s 1970s 1980s 1990s 2000s GDP Agriculture Maufacturig Commodity producig sector Services sector Uemploymet Rate Natioal savigs at Market prices CPI (percet) Source: Ecoomic Survey of Pakista ( ) Like GDP growth rates, growth of agriculture, maufacturig ad services sectors also show the similar patter over 1960s to 2000s. The growth of agriculture, maufacturig, ad services sectors is 5.1 percet, 9.9 percet ad 6.7 percet i 1960s respectively. Similarly, with the falls i GPD growth i 1970s, there had bee sigificat dowsizig i above three sectors with 2.4 percet for agriculture, 5.5 percet for maufacturig, ad 6.3 percet for services sector. Natioal savig has importat implicatio for growth ad developmet i Pakista (see Table No3.1). The growth rate of atioal savig has bee 11.2 percet i 1970s, icreases to 14.8 percet i 1980s. It falls to 13.8 percet i 1990s ad rises to 15.8 percet i 2000s. The growth rate of GDP has bee closely related with the growth rate of atioal savigs. GDP growth rate has bee lower whe growth of atioal savigs has bee decreased ad vice versa (See Figure No 3.1). The relatioship betwee atioal savig ad CPI iflatio is also show i the Figure No 3.1. It is evidet that iflatio rate iversely affect growth of atioal savigs. If iflatio rate is high, the growth rate of atioal savig has bee low. It may be due to the chael of real rate of retur o deposits. Ecoomic theory predicts that if iflatio rate is higher tha the omial market iterest rate, there would be egative rate of retur o savig, therefore, atioal savig falls. ISSN : Vol. 12 No. 1, Nov

5 Irfa Hussai Kha, Prof. Dr. Bashir Ahmad Khilji, ad Ghulam Abbas Figure No 3.1: GDP growth at costat market prices, growth of Natioal savig at curret market prices ad CPI iflatio rate. Source: Table No 3.1 The preset dyamics of Pakista s ecoomy reveals that it is tripped ito low growth high iflatio ad uemploymet zoe. It is due to the reaso that efforts for poverty reductio ad ehacemet of livig of stadard have ot bee remaied fully profitable. Icreasig food prices coupled with overall double digit iflatio, the level of poverty icreased to 50 percet of the total populatio i FY11, (Cetral Itelligece Agecy - CIA, 2012). GDP growth is the key yardstick to judge expasio of productivity amog the coutries with the similar coditio of social ad ecoomic set-ups. GDP growth rate of Pakista is uder pressure from 2005 to 2011 comparig it with its similar couter parts, (See Table No.3.2). For istace, Pakista s GDP growth rate has bee cotiuously fallig from 9 percet i 2005 to 2.4 percet i 2011 with a little improvemet i O the other had, real GDP growth rate of Chia has bee cotiuously icreasig from 11.3 percet i 2005 to 14.2 percet i There has bee a little pressure o chia s real GDP growth from 2008 till 2011; however, overall performace is outstadig i the regio eve i spite of recet fiacial crises. Figure No 3.2 presets a sketch of growth rates of some of the regioal coutries o the basis of the data i Table No 3.2. The average growth rate of Chiese ecoomy is percet durig 2005 to 2011, followed by 8.48 for Idia ecoomy, ad 6.45 percet for Sri Laka. The last seve years average of growth rate for Bagladesh is 6.24 percet, while it is oly 4.74 percet for Pakista. Pakista stads at last amog chia, Idia, Sri Laka ad Bagladesh i the growth performace of the atioal ecoomy. It meas that Pakista s ecoomy is very sesitive to global ecoomic developmet tha the other coutries just metioed above. This high sesitivity of atioal ecoomy to global ecoomic performace is the key parameter of busiess cycles i Pakista. Table No 3.2: Regioal Compariso i Real GDP Growth Rates (Percet) Year Chia Idia Sri Laka Bagladesh Pakista Source: Aual Report (2011, P -1), State Bak of Pakista ISSN : Vol. 12 No. 1, Nov

6 RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN Source: Aual Report (2011, P -1), State Bak of Pakista Figure No 3.2: Real GDP Growth Rates (Percet) Apart from the sesitivity of Pakista ecoomy to global ecoomic performace, there are some other structural factors which also keep it depressed i terms of productivity growth ad developmet. There is a log list of such factors, however, some most importat factors which are iadequate tax system, trade barriers with eighborig coutries, icompetecy ad iefficiecy of public corporatios or state owed eterprises, lack of soud privatizatio well, ad iappropriate model for huma resource developmet ad social protectio. The objective of sustaiig high growth, low iflatio ad exteral paymet viability requires removig these structural barriers, (Ecoomic Survey of Pakista ) THEORETICAL FRAMEWORK Farha & Akram (2011) develops a theoretical model for testig effects of iterest rate o houshold savig. The model icludes three idepet variables such as Per capita icome, iterest rate, cosumer price idex with Household savigs as a depedet variable. Household savig costitutes a sigle larger compoet of atioal savig of Pakista. Fluctuatios i household savig ca greatly affect the performace of atioal ecoomy through ups ad dows i availability of atioal savig for ivestmet i the coutry. I other words, household savig is the largest compoet of atioal savig which ca cause busiess cycle with sigificat fluctuatio i plaed ivestmet i Pakista. I this cotext, the determiats of household savig are vital to ivestigate i the Pakista or i other words, it is very urget to kow the leadig factors which ca fluctuate the household savig at macroecoomic level. The respose of household savig to various macroecoomic eviromets i both short ru ad log ru is crucial to be kow to policy makers i order to prepare policy packages for smooth uplift of atioal ecoomy. The relatioship betwee idepedet variables to depedet variable has bee explaied, oe by oe, i sufficiet detail. (a) PER CAPITA INCOME AND HOUSEHOLD S SAVING Per capita icome ad household s savig have positive relatioship betwee them. If the per capita icome icreases household savigs also icrease ad vice versa. Farha & Akram (2011). This is kow as icome effect of iducig savig. Keyes argues that margial propesity to cosume (MPC) is betwee zero ad oe, which meas that additioal icome may ot be cosumed fully. (b) INTEREST RATE AND HOUSEHOLD S SAVING Iterest rate ad household s savig are also positively related with each other. If the iterest rate icreases more it iduce savers to save more. Farha & Akram (2011)There are differet defiitios of iterest rate, e.g. ledig rate; deposit rate has bee used as proxy for iterest i the preset study. Whe Iterest rate icreases, people are ecouraged to save more ad cosume less just to ear more retur o deposit. There is a positive relatioship betwee Natioal Savig ad rate of iterest. ISSN : Vol. 12 No. 1, Nov

7 Irfa Hussai Kha, Prof. Dr. Bashir Ahmad Khilji, ad Ghulam Abbas Uderstadig the respose of persoal savig to chages i iterest rates is cetral to may issues i ecoomic policy. For example, a reductio i the budget deficit would probably cause iterest rates to declie. If persoal savig declied as a result, the overall icrease i atioal savig would be less tha the reductio i the budget deficit. Alteratively, cotractioary moetary policy geerally causes iterest rates to rise. If persoal savig icreases as a result, the correspodig fall i cosumer spedig helps to slow dow ecoomy. As a fial example, chages i the tax rates ca raise or lower the et-of-tax retur to savig. The effect of these chages o the amout of savig may play a importat role i tax collectio. The iterest elasticity of savig is defied as the percet chage i savig resultig from a oe-percet chage i the iterest rate. There is disagreemet amog ecoomists about both the sig ad magitude of this elasticity, as existig theory ad empirical evidece do ot appear to offer ay clear coclusios. David & Mackio (1990). (C) CPI AND HOUSEHOLD S SAVING Cosumer price idex ad household s savigs have egatively related. If CPI icreases it decreases the savig rate ad vice versa. Farha & Akram (2011) Cosumer Price Idex umber measures the average rate of Prices i the ecoomy. If Price levels i the ecoomy icreases it has two implicatios: (i) Due to a icrease i Price Level, Producers charge high prices of their product to ear more profits which raise Natioal Savig. There is positive relatioship betwee Cosumer Price. Idex ad Natioal Savig from Producer poit of view. Hasai et al (2006) have also foud positive relatioship betwee Natioal Savigs ad iflatio rate. (ii) Due to a icrease i Price Level, Cosumers have to pay higher prices ad these results i low domestic savig. Hece atioal savig may fall. From Cosumers poit of view, there is egative relatioship betwee Cosumer Price Idex ad Natioal Savig. Kazmi (1993) has foud egative relatioship betwee Natioal savig ad iflatio rate. So the Cosumer Price Idex umber iflueces positively ad egatively the Natioal. Figure 4.1 liks the iterest rate, iflatio Rate, per capita icome to household s savig i Pakista. I below figure shows positive relatioship betwee per-capita icome ad Iterest rate with household savigs. Ad egative relatioship of CPI with household savigs. Figure 4.1 Household savigs relatioship with Per-capita icome, iterest rate Ad CPI Source: Author Costructio usig MS office HYPOTHESIS OF THE STUDY Based o the review of literature ad theoretical framework, the followig relatioship has bee hypothized. 1) Iterest rate positively affects household savig. 2) There is positive relatioship betwee per capita icome ad household savig. 3) There is egative relatioship betwee iflatio rate ad household savig. ISSN : Vol. 12 No. 1, Nov

8 RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN 3 RESEARCH METHODOLOGY This study use Farha & Akram (2011) empirical model to assess short ad log ru determiats of household s savigs. I the preset case, substatial modificatios have bee made to their model for makig it more appropriate i the curret cotext. These modificatios are: Firstly, depedecy ratio has bee excluded from the model because per capita icome has bee used istead of icome level. As per capita icome is the icome level divided by populatio, therefore, depedecy ratio may exhibit sigificat mutlicolliearity, if use combied with per capita icome, which would hamper cosistecy of the estimates. This is oe of the reasos that preset study cotributes to the modelig of household s savig i Pakista. Secodly, sice preset study uses time series data, therefore estimatio techiques eed to be cosistet with the stadard properties of time series data. Oe of them is the existece of uit root i the time series data. If time series data cotais uit root, the Ordiary Least Square (OLS) caot be used to report cosistet ad efficiet estimates. Therefore, Augmeted Dickey Fuller (ADF) test has bee used to check the existece of uit root i the data. However, i case of evidece of uit roots i the data, Johaso (1998) co-itegratio techique ca be used to estimate short ad log ru respose of household s savigs to various factors icludig iterest rate. These two modificatios cotribute to the validity of the estimates. The maximum possible umber of years of time series data has bee collected, that is, from 1981 to 2011 which make it of 30 years data. I this cotext, the study is restricted to the total of four variables for esurig sufficiet degree of freedom for macro ecoometric modelig. Out of these four variables, iflatio rate, iterest rate ad per capita icome are idepedet variables whereas; household s savig is the depedet variable. Ecoometric properties of the estimates are checked usig various tests. ECONOMETRIC MODEL The preset study uses Farha & Akram (2011) empirical model to assess the short ad log ru determiats of household s savig. I the preset case, substatial modificatios have bee made to their model to make it more appropriate i the curret cotext of Pakista. These modificatios are: First, depedecy ratio has bee excluded from the model because per capita icome has bee used istead of icome level. As per capita icome is the icome level divided by total populatio, therefore, depedecy ratio may exhibit sigificat mutlicolliearity, if used combie with per capita icome, which would hamper cosistecy of the estimates. This is oe of the reasos that preset study cotributes to the modelig of household savigs i Pakista. Secodly, the preset study deals with time series data, so it is most likely that there may be evidece of Uit Root i the data. If Uit root exist i the data, the Ordiary Least Square caot be used to estimate the co-efficiet. It is oe of the reaso that Co-itegratio based modelig techiques has bee used to report more efficiet ad cosistet estimates. The macro ecoometric modelig starts with testig for directio of causality, statioarity ad co-itegratio i.e. testig for existece of log ru relatioship ad costructig appropriate robust error correctio model for household savig as a depedet variable. First, directio of causality amog variables would be tested usig Grager Causality test (Grager, 1969) which is outlied by Karlsso et al., (2002) as give i equatio (1) to equatio (4): LHS t a 10 a1i LPCi t i a i LCPI t i a3i LIR t i 2 1 t i1 i1 i1 (1) LPCi t a a LHS 10 1i t i a ilcpi t i a ilir 2 3 t i 2 t i1 i1 i1 (2) LCPI t a 10 a1i LHS t i a i LPCi t i a i LIR 2 3 ti 3t i1 i1 i1 (3) LIR t a 10 a1i LHS t i a i LPCi t i a3i LCPI t i 2 4 t i1 i1 i1 (4) Where LHS is the log of household s savig at curret market prices, LPCi is the log of per capita icome at market prices, LCPI is the log of cosumer price idex ad LIR is the weighted average rate of retur o deposits. Equatio (1) to equatio (4) ca be estimated by ordiary least square (OLS) method to fid out causal relatioships amog the variables. For istace, if lag terms of LHS ad LPCi i equatio (1) ad equatio (2) are statistically sigificat as group, the there is bilateral causality betwee LHS ad LPCi. It meas that per capita icome would ISSN : Vol. 12 No. 1, Nov

9 Irfa Hussai Kha, Prof. Dr. Bashir Ahmad Khilji, ad Ghulam Abbas cause household s savig ad vice versa. O the basis of ecoomic theory, it is highly possible that household s savig may cause per capita icome icrease through ivestmet multiplier. Similarly, other equatios ca also be iterpreted i the similar ways. For istace, if LHS i equatio (1) ad LCPI are sigificat as a group, the there are evidece of bilateral causality betwee iflatio ad household s savig. The impact of iflatio rate o savig is well documeted i literature. If iflatio rate is higher tha omial iterest rate, the there is egative real rate of retur o savig deposit. The egative real rate of retur o deposits mea that the value of savig deposit falls over a time. I this case, iflatio rate has egative impact o household's savig or it discourage household s to save more. It meas that egative rate of retur o savig makes the curret cosumptio more attractive tha the future cosumptio. The dyamic effects of per capita icome, iterest rate ad iflatio rate o household savig have bee estimated usig Johase (1988) ad Johase et al., (1990) co-itegratio approach. The Johase (1988) ad Johase et al., (1990) multivariate co itegratio approach has bee based o the followig steps. First, the estimatio of urestricted vector autoregressive model (VAR) is required as explaied by Halda et al., (2010), as give below, y t p i y i1 ti t (5) The equatio (5) is the urestricted VAR model, where, y t represet all N variables of the model which is four i the preset case, that is, per capita icome, iterest rate, iflatio rate ad household savig. is the N*N matrix of coefficiets ad εt is a vector of radom shocks to the system. It is assumed that εt is ormally distributed with zero mea ad costat variace. The VAR model (5) ca be writte i its error correctio (ECM) form as give i equatio (6) y t y t1 p1 iy i1 ti t (6) p p Where = - (I - ) ad Ai ad I is a idetity matrix. If all the variables are itegrated of order i Aj i1 ji1 oe i.e. o statioary of same order, tha y t are statioary. The estimated coefficiets of equatio (6) are cosistet if the assumptio that variables are co-itegrated is ot violated, the xt1 is also statioary or itegrated of order zero. The rak of shows umber of co-itegratig vectors, for istace, if =0, the matrix is ull or there is o coitegratig vector at all, ad equatio (6) shrik to usual first differece VAR model. If =1, the it would meas that there is oe co itegratig vector, ad xt1 is the error correctio term which shows how system revert to its log ru equilibrium i respose to ay shock i the short ru. For multiple co itegratig vectors, rak of ca be i 0 < < rage, ad there exist represetatio of such that = α. β. Where α ad β are both (.r) dimesio with r beig rak of. β represets the matrix of co-itegratig vectors ad α shows speed of adjustmet to its log ru equilibrium of the whole system. Basically there are two tests, that is, trace statistic (λ trace ) ad maximal Eige value statistic (λ max ) which ca be used to coclude about the umber of co-itegratig vectors i the whole system. Oce the existece of log ru relatioship is idetified, the ext step is to estimate a represetative Vector Error Correctio model for household s savig as a depedet variable. This ca be doe by followig geeral to specific approach to VECM i.e. a geeral VECM may be estimate i the first step. I the secod step, the geeral VECM model may look for matchig it with the stadard theory. However, if there are some variables which are isigificat or havig theoretically icorrect sig, they may be dropped, ad yet aother versio of VECM model may be obtaied. Followig this geeral to specific approach, researcher ca obtai such model which is more compatible with the stadard theories of ecoomics. 4 DATA ANALYSIS AND DISCUSSION Household s savig rate ad iflatio are idirectly correlated, that is, at ay give omial iterest rate i the market; high iflatio rate reduces real iterest rate ad hece savig too. Similarly, at ay prevailig omial iterest rate i the market, if iflatio rate falls, it raises the real rate of retur o savig. The household s savig respose positively to this reductio i iflatio rate because low iflatio reduces cost of livig, ad hece icrease savig. ISSN : Vol. 12 No. 1, Nov

10 RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN Table No 5.1 shows N way tabulatio of cosumer price idex based iflatio rate ad log of household s savig. It is evidet that high double digit iflatio i the rage of 20 to 25 percet discourages household s savig, as there is oly oe frequecy out of 31 years which correspods to a rage of 20 to 25 percet iflatio rate i the data. Similarly, if iflatio rate is i rage of 10 to 15 percet, the household s savig i rage ad are two ad three times i the history respectively. Table No 5.1 also reveals that i past 31 years data, 20 correspodece of household s savig are observed i the rage of iflatio from 5 to 15 percet. It shows that moderate level of iflatio promote household s savig i Pakista. Table No 5.1: N Way tabulatio of CPI ad LHS CPI Cout [0, 5) [5, 10) [10, 15) [20, 25) Total [10, 11) [11, 12) LNHS [12, 13) [13, 14) [14, 15) Total Source: Author s calculatio usig Eviews 6.0 portable software Iterest rate is the rate of retur o household s savig. Ecoomic theory predicts that other thigs remai costat; household s savig is positively related to iterest rate. Higher iterest rate icreases the opportuity cost of curret cosumptio, ad hece motivate household s to substitute curret cosumptio over future cosumptio, ad vice versa. Table No 5.2 shows the relatioship betwee iterest rate ad household s savig. It is evidet from the table that 6 ad 18 correspodece of household s savig fall with the iterest rate rag of 4 to 6 ad 6 to 8 respectively. Similarly, there is 4 ad 2 correspodece i the low rage of iterest rate which is from 0 2 ad 2 4 respectively. This meas that table o 5.2 favor a hypothesis that low iterest rate discourage savers ad vice versa. Table No 5.2: N Way tabulatio of IR ad LHS Cout IR [0, 2) [2, 4) [4, 6) [6, 8) [8, 10) Total [10, 11) [11, 12) LNHS [12, 13) [13, 14) [14, 15) Total Source: Author s calculatio usig Eviews 6.0 portable software Table No 5.3 shows the relatioship amog per capita icome ad household s savig. Ecoomic theory predict positive effect of icome o household savig, that is, other thigs costat; if household s icome icrease, the, its savig too. Table No 5.3 favors such theoretical relatioship. There are total of four icome brackets give i the table, that is, 8 9, 9 10 ad There are five highest correspodeces i frot of 8 9 icome rage. It further rises to 6 ad 7 correspodeces i the icome groups of 9 to 10 ad 10 to 11. Similarly, there are three correspodeces out of total of four correspodeces i frot of highest icome group. This shows that icome ad household s savig are positively related. ISSN : Vol. 12 No. 1, Nov

11 Irfa Hussai Kha, Prof. Dr. Bashir Ahmad Khilji, ad Ghulam Abbas Table No 5.3 N Way of Tabulatio of LPCi ad LHS LPCi Cout [8, 9) [9, 10) [10, 11) [11, 12) Total [10, 11) [11, 12) [12, 13) LNHS [13, 14) [14, 15) Total Source: Author s calculatio usig Eviews 6.0 portable software Table No 5.4 highlights the descriptive statistics of the key variables used i the model. Descriptive statistics covey a geeral picture of the data. Mea value of log of household s savig is the approximately with a stadard deviatio of The maximum value of log of household s savig is the 14.24, whereas, miimum value of log of household s savig is the The value of Jarque Bera test, a test of ormality of distributio, is give by 2.14 with probability value of 0.34 which is eough high to accept alterative hypothesis of o-ormal distributio of household s savig. The value of skewess is positive, which shows that distributio is positively skewed. Similarly, the value Kurtosis is lower tha 3, which meas that distributio is platykurtic, that is, distributio is flatter tha a ormal distributio with a wider peak or values are wide spread aroud mea value of log of household s savig. This type of aalysis are very much relevat for selectio of estimatio techique for the estimatig the ecoometric model. The average value of log of per capita icome durig 1981 to 2011 is 9.66 approximately with maximum of ad miimum value is 8.08 with a stadard deviatio of Stadard deviatio of log of per capita icome is lower tha log of household s savig which shows that observed data is more closed scatter to mea of log of per capita icome tha log of household s savig. Alteratively, it ca be said that log of per capita icome is more cosistet series tha log of household s icome, or eve i other words, log of per capita icome shows less variatios aroud its mea value tha log of household s savig. However, it is positively skewed series which is reflected i the value of co-efficiet of skewess with value greater tha zero. The co-efficiet of kurtosis is also less tha three, which shows that distributio of log per capita icome is platykurtic or o ormal. The o-ormality coclusio ca also be draw from the calculated value of Jargue Bera test which is 2.05 with probability value of It meas that, like log of household s savig, log of per capita icome data also suffer from o ormality issue. Table No 5.4 also shows the descriptive statistics related to iterest rate which is measured by weighted average rate of retur o savig deposits. The average rate of retur o savig, the iterest rate, is 5.87 for the period of 1981 to 2011 i Pakista. The highest iterest rate durig this period is 8.79 whereas lowest iterest is 0.96 per aum. The stadard deviatio is 2.23, with value of skewess less tha zero. It meas that iterest rate is more egatively skewed, ad there is evidece that distributio of iterest rate is ormal distributio. It meas that the observed iterest rates, from 1981 to 2011, is closed scattered aroud mea values with reasoable peaks which is ofte kow as ormal distributio. However, this coclusio is draw at 10 percet level of probability, that is, the ormality of distributio is oly accepted at 10 percet level of probability. Table No 5.4 Descriptive statistics of selected variables LNHS LNPCI IR CPI Mea Media Maximum Miimum Std. Dev Skewess Kurtosis Jarque-Bera Probability Observatios Source: Author s calculatio usig Eviews 6.0 portable software ISSN : Vol. 12 No. 1, Nov

12 RELATIONSHIP BETWEEN INTEREST RATE AND HOUSEHOLD S SAVINGS IN PAKISTAN Fially, the behavior of iflatio rate (based o cosumer price idex) is give i the table o 5.4. The average iflatio i durig 1981 to 2011 is 8.58 which is a sigle digit rate. However, there are very high ad very low iflatio rates i Pakista as well. For istace, the highest iflatio rate is percet, which is really high eough too beloved that it might be harmful for growth, poverty ad household savig. Similarly, lowest iflatio, from 1981 to 2011, is 3.2 which is eough low. The value of skewess suggests that distributio of iflatio is positively skewed; with leptokurtic distributio as value of kurtosis is greater tha 3. It meas that distributio of iflatio is sharper tha ormal distributio, which is also reflected i the sigificat value of Jargue Bera test. 5 CONCLUSION Savig is the egie of growth ad prosperity. It is a source of fiacig ivestmet i the coutry, which reduces depedece o foreig aid ad loas. There are a umber of defiitios of savig i Pakista, amely, public, private, corporate ad household s savig. These defiitios of savig are importat for overall macro-ecoomic performace of a ecoomy. The aalysis of savig tred i Pakista shows that all these defiitios of savig are volatile up to cosiderable extet. It is also cocluded that, o the basis of tred aalysis that private sector savig is the mai driver of overall savig i Pakista. Iterestigly, out of private sector savig, household s savig is early 90 percet which highlight its importace i ecoomic developmet of Pakista. It is also evidet that Pakista s savig tred is low if compared with regioal coutries havig almost same icome level. It is due to low savig rate that savig ivestmet gap is high i Pakista, ad Pakista is highly depedet o exteral borrowig to fiace the icreasig savig ivestmet gap. There are several reasos which ca be explaied for such dismal performace of low savigs rate. However, tred aalysis shows that lower growth of per capita icome, high iflatio particularly food iflatio ad low real iterest rate are the mai drivers of low savig rate i Pakista. The quatitative aalysis of 30 years past data shows that there are strog evideces that these variables are coitegrated. They cotai log ru equilibrium relatioships. The Error Correctio compoet shows that there may be disequilibrium i the short ru, however, there is a sigificat Error correctio compoet which revert the system back ito equilibrium. It is also show that the effects of idepedet variables are as per ecoomic theory. 6 SUGGESTIONS Household s savig is sesitive to price chages, particularly, high food iflatio cause household expediture to rise, which reduce their ability to save more. It is, therefore, recommeded that overall iflatio, particularly, food iflatio may be kept i desirable limits so that savig ca be ehaced i Pakista. Real iterest rate is low i Pakista. There are two basic reasos for this; first, iflatio rate is high, ad secodly, omial iterest o deposit is low. Ideed, the first argumet is more valid as real iterest meas omial iterest rate mius iflatio rate. Now if omial iterest is stable, there real iterest rate fluctuates as result of iflatio rate. Per capita icome is the mai driver of household savig i Pakista, however, per capita icome is ot oly low but also have slow growth rate as compared to the eighborig coutries. There may be several factors which may accout for low per capita icome growth there are. Oe of them is the low productivity growth. Pakista s real GDP is growig very slowly. Growth of real GDP is the mai determiat of livig stadard of a ecoomy. o There has bee high populatio growth i the ecoomy. If populatio growth is icreasig, ad GDP growth is fallig, the resultig Per capita icome will be low. o There has bee high uproductive savig rate i Pakista which may lead to Per capita icome low. Pakista is primarily cosumptio orieted society. Its ecoomic growth is cosumptio led. There cosumptios are made at the cost of low savig. There may be several factors which ca be used to explai such pheomeo; however, there is less umber of fiacial istitutes i the rural areas of Pakista which is also oe of the hurdles. People i rural areas are ot educated about the value of savig. There are early marriages ad expasive traditios which adds more to the cosumptios. Over 67 percet of populatio lives i rural areas of Pakista, which is directly or idirectly depedet o agriculture sector. However, agriculture is highly pro-cyclical i ature. These fluctuatios also cause variability i household icome. I geeral, i periods of productivity losses, household depeds o debt which is used to repaid i the time of boom i the agriculture sector, so overall savig remai depressed. ISSN : Vol. 12 No. 1, Nov

13 Irfa Hussai Kha, Prof. Dr. Bashir Ahmad Khilji, ad Ghulam Abbas REFERENCES [1] Abdelkhalek, T., Arestoff, F., de Freitas, N. E., & Mage, S. (2009). A Microecoometric Aalysis of Household ss Savig Determiats i Morocco. Istitut atioal de statistique et d écoomie appliquée (INSÉA), Maroc, Pp [2] Abid, S., & Afridi, G. S. (2010). Assessig the Household s Savig Patter of Urba ad Rural Household ss i District Muzaffarabad. Pakista Joural of Life ad Social Scieces, Pp [3] Agrawal, P., Sahoo, P., & Dash, R. K. (2007). Savigs behaviour i South Asia. Expadig Freedom: Towards Social ad Ecoomic Trasformatio i a Globalizig World, Pp [4] Ahmad, M., & Asghar, T. (2002). Estimatio of Savig behaviour i Pakista usig Micro Data. The Lahore Joural of ecoomics, vol 9, No.2, Pp [5] Ahmad, M., & Asghar, T. (2002). Estimatio of Savig Behaviour i Pakista Usig Micro Data. The Lahore Joural of Ecoomics, Vol.9, No.2, Pp [6] Ali, M. S. (1985). Household s Cosumptio ad Savig Behavior i Pakista: A Applicatio of the Exteded Liear Expediture System. The Pakista Developmet Review vol.xx1v, No.1, Pp [7] Alves, N., & Cardoso, F. (Witer 2010 ). Household Savig i Portugal: micro ad macroecoomic evidece*. Ecoomic Bulleti Baco de Portugal, Pp [8] Azam, M., Kha, M., Kha, Z., Shafiqullah, Ali, S. I., & Qaiyum, A. (2010). Sigificace of Natioal savig i the socioecoomic developmet of Pakista: Sarhad j. Agric vol.26, o.3, 2010, Pp [9] Bishop, J., & Cassidy, N. (2012). Treds i Natioal Savig ad Ivestmet. Bulleti, Reseve Bak of Australia [10] Burey, N. A., & Kha, A. H. (1992). Socio-ecoomic Characteristic ad Household s Savig: A aalysis of the Household s's Savig Behavior i Pakista. The Pakista Developmet Review, Pp [11] Cetral Itelligece Agecy - CIA. (2012). Pakista Ecoomy. Retrieved Jue 16, 2012, from The World Fact Book: [12] Chaudhry, I. S., Faridi, M. Z., Abbas, M., & Bashir, F. (2010). Short Ru ad Log Ru Savig Behavior i Pakista: A Empirical Ivestigatio. Joural of Moey, Ivestmet ad Bakig, Pp [13] Christese, L. (2012). How Are Iterest Rates Affectig Household s. All Graduate Reports ad Creative Projects, Pp [14] Deato, A., & Paxso, C. (1998). Savig ad growth: aother look at the cohort evidece. Research Program i Developmet Studies Priceto Uiversity, Pp.1-32.Ecoomic survey of Pakista ( ), P-2 [15] Ecoomy Watch. (2010, March 29). Pakista Ecoomy. Retrieved Jue 15, 2012, from Ecoomy Watch: ecoomy/pakista/ [16] Elmedorf, D. W. (1996). The effect of iterest-rate chages o household s savig ad cosumptio:a survey. Federal reserve board, Pp [17] Ersado, L., Alwag, J., & Alderma, H. (2000). Chages i Cosumptio ad Savig Behavior before ad after Ecoomic Shocks: Evidece from Zimbabwe. Iteratioal Food ad Agribusiess Maagemet Associatio Coferece, Chicago, Pp [18] Farha, M., & Akram, M. (2011). Does Icome Level affect Savig Behaviour i Pakista?A ARDL approach to coitegratio for empirical assessmet. Far East Joural of Psychology ad Busiess, Pp [19] Fry, M. J. (1998). Savig, ivestmet, growth, ad fiacial distortios i pacific asia ad other developig areas. Iteratioal ecoomic joural volume 12, umber 1, Pp [20] Gaire, N. H. (2010). Real iterest Rate ad Savig Behavior i Nepal. Bakig joural,volume 2 (issue 1), Pp [21] Hagiwara, A. T. (2010). Have Filipio Households Become Less Prudet? Asia Developmet Bak; 6 ADB Aveue, Madaluyog City 1550, Philippies, Pp [22] Hussei, K. A., & thirlwall, a. P. (1999). Explaiig differeces i the domestic savigs ratio across coutries: a pael data study. Jel classificatio: e21, Pp [23] Kapoor, M., & Ravi, S. (2007). Expect of Iterest Rate o Cosumptio. Idia School of Busiess, Pp [24] Kazmi, A. A. (1990). A Study o Savig Fuctios for Pakista: The Use ad Limitatios of Ecoometric Methods. The Lahore Joural of Ecoomics, Vol.6, No.2, Pp [25] Hussei, K. A., & thirlwall, A. P. (1999). Explaiig differeces i the domestic savigs ratio across coutries: a pael data study. Jel classificatio: e21, Pp [26] Kapoor, M., & Ravi, S. (2007). Effect of Iterest Rate o Cosumptio. Idia School of Busiess, Pp [27] Kha, R. E., & Hye, Q. M. (2010). Fiacial sector reforms ad household s savigs i Pakista: A ARDL approach. Africa Joural of Busiess Maagemet Vol.4 (16), Pp [28] Kibet, L. K., Mutai, B. K., Ouma, D. E., Ouma, S. A., & Owuor, G. (2009). Determiats of household savig: Case study of mallholder farmers, etrepreeurs ad teachers i rural areas of Keya. Joural of Developmet ad Agricultural Ecoomics Vol. 1(7), pp , October, 2009 ISSN Academic Jourals, Pp ISSN : Vol. 12 No. 1, Nov