Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The Causal Link Between Savings and Economic Growth in Bosnia and Herzegovina

The Causal Link Between Savings and Economic Growth in Bosnia and Herzegovina The causal link between savings and economic growth has been extensively discussed in the economic growth and development literature, but the question of the direction of this link has not yet been clearly defined. The aim of this paper is to determine the direction and intensity of savings causality (components of private savings) and economic growth in Bosnia and Herzegovina. Granger’s causality test, the Toda- Yamamoto procedure, was applied to test for causality between savings and economic growth. The results of Granger’s causality test indicated that there is no causal link between components of private savings with economic growth in Bosnia and Herzegovina. It was found that among the variables there is cointegration, but not causality, which means that the variables have a common stochastic trend. This fully corresponds to the characteristics of the Bosnia and Herzegovina’s economy. Keywords: Saving, economic growth, macroeconomics, econometric analysis JEL Classification: C32, E21 1. Introduction One of the most important issues in any economy, direction and intensity of the causality of savings and regardless of the degree of its growth and develop- economic growth, neither in the short nor in the long ment, is to ensure sufficient and quality funds to run. finance the economic growth. Given that foreign capi- Therefore, this paper aims to examine the prob- tal is quite expensive and risky to finance domestic lem of endogenous savings and economic growth investment and economic growth, domestic capital is in the small, underdeveloped economy of Bosnia considered a much more favorable and secure source and Herzegovina. In this case, the focus is on private of funds. Savings are most often mentioned in the lit- erature as a domestic source of funds. The relationship between savings and economic growth is defined Irma Đidelija, PhD in the economic literature through the three most Assistant Professor important theories of economic growth, which are: Faculty of Economics Keynesian theory of growth, Harrod-Domar’s theory University „Džemal Bijedić“ in Mostar and Solow’s theory. E-mail: irma.djidelija@unmo.ba The relationship between savings and economic Address: Vrapčići bb, 88000 Mostar, growth, given its economic significance, has been the Bosnia and Herzegovina subject of numerous empirical studies. These works ORCID: 0000-0003-1666-7363 did not give unambiguous results in terms of the Copyright © 2021 by the School of Economics and Business Sarajevo 114 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA savings, looking at its individual components sepa- principles of Granger’s causality test. The results of rately: household savings and savings of private the research, the fifth part of the paper, is the most for-profit companies. The reason for this is the fact important part, where the results of the research are that private savings, and especially its component of presented with an appropriate discussion. At the end, household savings, is recording continuous growth in concluding remarks were given with summarized re- Bosnia and Herzegovina, despite the great financial sults of the conducted research. crisis in 2008. Also, private savings, according to the Central Bank of Bosnia and Herzegovina (quarterly bulletins from 2000-2016), is the dominant part of na- 2. Theoretical framework tional savings, which is another reason to include only private savings in the model of this survey. When it comes to theories dealing with economic Private savings in Bosnia and Herzegovina are growth, meaning the the relationship between sav- mostly realized in the form of deposits given the ings and economic growth, two/three significant the - dominance of the banking sector in the financial mar - ories stand out in the literature: Keyenes’ hypothesis, kets. Thus, private savings in Bosnia and Herzegovina, neoclassical (Harrod-Domar’s and Solow’s) theory of viewed as the sum of long-term deposits, are growing growth, and endogenous theory of economic growth. despite the recent global financial crisis, but also the One of the most significant works in this field slowdown in economic growth and political instabil- was offered by Keyens, in 1936, and that is the ity in the country. It can be seen from the bulletin “General Theory of Employment, Interest and Money”. of the Central Bank of Bosnia and Herzegovina that According to Keyens, only in conditions of full em- at the end of September 2016 (in this analysis, eight ployment can the classical theory be considered years before and eight years since the beginning of valid. However, full employment conditions are the the crisis were observed), total household deposits expectation rather than the rule. The real conditions amounted to BAM 10.25 billion. This is also the larg- are different, and they are based on a series of causal est growth in deposits so far. The share of household relations as follows: entrepreneurs earn income by deposits in total deposits of commercial banks in selling goods and services; as income grows, so does Bosnia and Herzegovina is 59.6%. Compared to the consumption, but to a lesser extent than income; end of September of the previous year, 2015, house- differences in the percentages of income and con- hold deposits increased by 6.6% or BAM 633.5 million. sumption growth are reflected in savings; there is an Compared to December 2008, when the economic imbalance between supply and demand, if savings are crisis began, deposits are constantly growing in the not put into operation through investment activities; range of 8-15% annually. Thus, a cumulative growth of entrepreneurs’ expectations regarding future incomes deposits of 97% or BAM 5.05 billion was achieved. become pessimistic, and therefore they do not create Based on the above, the main hypothesis of the new jobs, but also abolish existing ones; eventually, research is that household savings in Bosnia and unemployment gradually rises and demand for goods Herzegovina initiate economic growth. For the pur- and services declines. pose of comprehensive analysis, the research also The theory assumes that the amount of savings includes another component of private savings, sav- and the amount of investment are necessarily equal ings of private for-profit companies, but given the since both amounts represent the value of income negligible share in total private savings in Bosnia and that exceeds consumption. The amount of savings is Herzegovina, no specific hypothesis is defined for the the result of the collective behavior of individual con- relationship between private profit savings and eco - sumers, and the amount of investment is the result of nomic growth. the collective behavior of individual entrepreneurs. In order to test the set hypothesis, Granger’s cau- This conclusion, as stated, is not the result of any per- sality test, the Toda-Yamamoto procedure, will be sonality or subtlety of the set definition of income. applied. Namely, the equality of savings and investment neces- The paper is divided into five parts. After introduc - sarily appears in traditional theory, if income is equal tory remarks, a theoretical framework is given with a to the value of current output, current investment brief overview of the postulates of basic theories of equal to the value of the part of current output that economic growth. The third part of the paper pro- is not consumed, and savings equal to the value of in- vides an overview of the empirical literature dealing come that exceeds consumption. with the problem of endogenous savings and eco- The Harrod-Domar model is the first and simplest nomic growth. The fourth part of the paper presents growth model, which is a long-term projection of the research methodology, more precisely the basic Keynes’ short-term analysis. Harrod-Domar’s theory SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 115 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA originated in the late 1930s and mid-1940s (Harrod In addition to the previously mentioned dominant 1939; Harrod 1948; Domar 1946). Capital factors are theories of economic growth, it is certainly impor- cited as crucial factors of economic growth in the tant to mention the endogenous theory of economic Harrod-Domar model. Based on the adjustment of growth. This theory was officially published in his supply and demand, the model tries to explain the article, “A contribution to the Empirics of Economic possibility of stable growth. Growth”, by David Romer, Gregory Mankiw and David According to the model, in order to ensure a state Weil in 1992, and is an upgrade of the neoclassical of equilibrium, it is necessary that the expenditures theory of growth. generated by investments be large enough to ab- The theory of endogenous growth is an economic sorb the growth of output, as a result of investments. theory that claims that economic growth is generated Investments not only cause consumption, but also within a system as a direct result of internal processes. lead to the expansion of the productive capacities of According to this theory, increasing the human capital the economy. The theory argues that there must be a of a nation will lead to economic growth through the balance between the volume of consumption gener- development of new forms of technology and efficient ated by investments and the productive capacities and effective means of production. The theory argues created by investments. The model points out that that knowledge-based industries play a particularly excessive capital accumulation unaccompanied by important role - especially telecommunications, soft- adequate income growth causes hyperproduction, ware, and other high-tech industries -as they become unemployment, and economic depression. That is increasingly influential in developed and developing why net investment must grow continuously in order economies. to maintain full employment in the long run. If net Some other important postulates of the endog- investments grow, then real income must grow at a enous theory of economic growth are: government rate that will ensure the use of the full capacity of the policies can increase a country’s growth rate if they increased capital stocks. The required income growth lead to stronger competition in markets and help rate is called the “full capacity growth rate” or guaran- encourage product and process innovation; returns teed growth rate. on capital investment, especially in infrastructure and Solow’s theory of growth is the result of the work investment in education, health and telecommunica- of authors Robert Solow and Trevor Swan from 1956, tions; private sector investment in research and de- and is sometimes called the neoclassical model of velopment is a key source of technological progress; growth. According to Solow’s model, the drivers the protection of property rights and patents is key to of economic growth are three factors: technology, providing incentives for companies and entrepreneurs capital accumulation and labor. This theory extends to engage in research and development; investing in Harrod-Domar’s model to the following elements: la- human capital is a vital component of growth; govern- bor factors are included in the factors of production; ment policy should encourage entrepreneurship as a declining returns on labor and capital are required means to start new businesses and ultimately as an when variables are viewed separately, and constant important source of new jobs, investment and further returns when variables are viewed in combination; in- innovation. troduction of variable technology, which depends on time and differs from labor and capital. The Solow model points out that an increase in 3. Literature review the accumulation of capital and labor leads to a high- er rate of economic growth, but this is of a short-term Domestic savings can be a significant catalyst for nature given the declining returns. If the economy had growth, as higher savings rates allow investment to only one worker, adding another would result in a sig- grow at higher rates and further economic growth. nificant increase in output. However, if thousands of The causal link between savings and economic growth workers are employed in the economy, one new work- has been much discussed in the literature on econom- er will not significantly contribute to output growth. ic growth and development, however the question of In this case, there can be a growth of the economy at the direction of this link has not yet been unambigu- a stable rate, where the gross domestic product grows ously determined. Proponents of Solow’s theory point at the same rate as productivity and labor. The rate of out that saving alone produces economic growth. economic growth, according to the model, can only Jagadeesh (2015), Nwanne (2016), Kim (2017), Patra et be increased by innovations and improvements in al. (2017) for long-term savings, also confirm that sav - technology once a stable rate of growth is achieved ings initiate economic growth. with resource utilization. On the other hand, using regression and Granger’s 116 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA causality test, the authors found that economic continuation of its research, Tang (2009a) conducts growth is a positive function of savings, e.g. Bayar an analysis of the sensitivity of the proven causality of (2014). To model an effective growth policy, it is im- savings and economic growth to the tests used. Five portant to know the direction of the causality of sav- different causality tests were used to test sensitivity: ings and economic growth. Thus, if savings really drive Granger’s, Geweke’s, Meese’s and Dent’s, Hsiao’s, Toda’s growth then a development policy to encourage sav- and Yamamoto’s, and Holmes ’and Hutton’s causal- ings will be applied. If savings do not cause economic ity test. The bidirectional causality of savings and growth, a policy of discouraging savings and strength- economic growth was determined regardless of the ening consumption will be applied. test used. Thus, the method of determining causality, Also, AbuAl-Foul (2010), in econometric analysis in this case, does not affect the results. Tang (2009b) of the long-run relationship of savings and growth showed in his further studies that savings and eco- shows that the link between savings and economic nomic growth cause each other to cause each other in growth exists and is two-way in Morocco, but not in the economies of East Asia as well. Tunisia (the link in this case was one-way, from growth Tang and Tan (2014) conducted an empirical anal- to savings). Shah (2015), applying the unit root test, ysis of the relationship between savings and econom- cointegration, Granger’s causality test and vector er- ic growth, now in Pakistan, using Solow’s neoclassical ror correction model, finds that savings contribute to growth model. It was determined that the variables economic growth in Nepal and India, i.e. growth con- are cointegrated, and that they are bidirectionally tributes to savings in Bhutan and Sri Lanka, while cau- causal. sality does not exist at all in Pakistan. Adam, Musah Guma and Bonga-Bonga (2016) observe for the and Ibrahim (2017) conduct an econometric analysis first time the separate impact of individual savings of the causality of savings and economic growth in 10 components on economic growth in South Africa. sub-Saharan African countries. Granger’s causality test The authors look separately at the impact of savings: and Toda-Yamamoto approach give consistent results households, private companies and government on for only three countries: Benin, Mali, and South Africa, economic growth. According to the results of the where savings causes economic growth. For other econometric analysis, the greatest impact on eco- countries, no consistent results were obtained using nomic growth is achieved by the savings of private the same methodology. Wyk and Kapingura (2021) for-profit companies. also conduct an analysis of the relationship between A certain number of authors have tried to explain savings and economic growth in South Africa and find the problem of endogenous savings and economic that the relationship exists only in the short run, while growth by introducing additional variables into the tests do not show a link between the two variables model. Mohan (2006) also includes the size of income in the long run. The Granger test found that when a among different countries in the analysis and finds relationship is positive it goes in the direction that that income plays a significant role in the relationship economic growth causes savings. between growth and savings, and that growth affects An alternative approach to exploring the link be- savings in a number of countries considered. Olajide tween savings and economic growth in Mexico, for (2009), emphasizes the importance of foreign direct the period 1970-2000, was given by Alguacil et al. investment on the ratio of savings and growth, where (2004). Namely, the authors use Toda and Yamamoto’s, the growth of savings and investment produces eco- and Dolado’s and Lutkepohl’s (non) causality test, nomic growth. Joshi, Pardhan and Bist (2019) analyz- stating that Granger’s test is more sensitive to the in- ing the relationship between savings, investment and cluded variables in the model. The analysis shows that economic growth determine that the investment have savings are causing economic growth in Mexico. a positive impact on economic growth, and savings Tang (2008) in his study uses a relatively new have a negative impact on economic growth in the cointegration test, the “border test” in the process of long run. investigating the existence of a long-term link be- It is especially interesting that the analyzes show tween savings and economic growth. The analysis was different results for the same countries, e.g. for the conducted for Malaysia. The intensity and direction of United States. Dhakal et al. (1992). apply Granger’s the link between savings and economic growth was causality finding that, in the United States, there is a conducted using Toda and Yamamoto, Dolado and two-way link between savings and economic growth, Lutkepohl approaches within the augmented-vector but also that growth exclusively implies savings. Or, autoregressive (VAR) methodology. Using this pro- Cullison (1993), also applying Granger’s test of causal- cedure, the author found that there is a bidirectional ity, shows that at the same time, in the US, there is no link between savings and economic growth. In the causality between savings and economic growth, but SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 117 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA also savings exclusively imply economic growth. enough funds for more demanding and larger finan- In the countries of Southeast Europe, banks play a cial projects, or to save in order to raise funds for such central role in the financial markets, so bank deposits projects. appear as the most common form of savings, instead In Bosnia and Herzegovina, it is evident that the of, for example, investments in the capital market. population does not deposit all funds separated from If the capital market was more developed in these spending on savings in the form of deposits with com- countries, in the sense that the state issues securities mercial banks or similar savings and loan institutions. for infrastructure, holding bonds of this type would A large part of the funds, especially the anchor cur- be more profitable than demand deposits because it rency (euro) and foreign currency in general, is held would bring higher interest yields. Thus, short money, outside the financial and banking sector in Bosnia which includes foreign bank assets and central bank and Herzegovina. Proof of the above is the fact that reserves above required reserves, would be trans- in 2001, and in the first quarter of 2002, the Central formed into investments through the capital market, Bank purchased and transposed German marks (DEM) i.e. short money would actually be a source of savings to Germany in the amount of 4.87 billion DEM. During through the capital market. The condition for the for- this period, about 20 million DEM was bought daily, mer is market liquidity. and that speaks volumes about the foreign currency However, the presence of the dominant form of stocks that the population kept outside commercial savings in the countries of Southeast Europe (deposit banks. savings) does not give significantly different results It can be seen from the bulletin of the Central in terms of the problem of uniformity of savings and Bank of Bosnia and Herzegovina (2000-2016) that at economic growth. Turan and Gjergij (2014) analyz- the end of September 2016, total household deposits ing the relationship between savings and economic amounted to BAM 10.25 billion. This is also the largest growth in Albania found that there is cointegration growth in deposits so far. The share of household de- among the variables involved, but not causality. On posits in total deposits of commercial banks in Bosnia the other hand, Ribaj and Mexhauni (2021), analyz- and Herzegovina is 59.6%. ing the relationship between savings and economic Compared to the end of September of the previ- growth in Kosovo, find that savings in the form of de - ous year, 2015, household deposits increased by 6.6% posits have a significant positive impact on the coun- or BAM 633.5 million. Compared to December 2008, try’s economic growth. when the economic crisis began, deposits are con- Based on the above, it can be concluded that the stantly growing in the range of 8-15% annually. Thus, direction and intensity of the relationship between a cumulative growth of deposits of 97% or BAM 5.05 savings and economic growth is significantly deter - billion was achieved. mined by the applied methodology and by character- The growth rate of deposits changes differently if istics of the considered economy. This fact provides different maturity structures of deposits are observed. additional confirmation of the correctness of the The fastest growth is achieved by short-term depos- observations of individual countries in the analysis of its, more precisely transaction accounts and demand this problem of endogeneity. deposits. Compared to September 2015, transaction accounts increased by 329 million, or 16.3%, while demand deposits increased by BAM 300 million or 13.7%. 4. Research methodology The largest percentage of household deposits, in 4.1. Overview of the state of savings in September 2016, were savings and time deposits of Bosnia and Herzegovina 52.8%. Transaction accounts and demand deposits ac- count for 47.2% of total household deposits. Private savings in Bosnia and Herzegovina recorded a When it comes to the currency structure of house- continuous growth in the period 2000-2016, accord- hold deposits, in September 2016, the largest share ing to the Central Bank of Bosnia and Herzegovina. was occupied by deposits in euros 51.1% or 5.25 bil- The dominant part of private savings is household lion BAM, then in BAM currency 43.6% share or 4.47 savings. The share of household savings in total pri- billion BAM, and deposits in other foreign currencies vate savings ranges from 40-90%. 5.3% or BAM 539 million. Since savings are the most liquid part of assets, Savings and time deposits of private for-profit then households try to invest such funds in a way companies achieved the highest growth from 2005 to that achieves certain returns. There are opinions that the last quarter of 2008 and the beginning of the glob- households opt for savings because they do not have al financial crisis. The cumulative growth of deposits in 118 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA the observed period amounted to 78.72%. The decline Bosnia and Herzegovina’s commercial banks have in savings and time deposits of private for-profit com- more assets than liabilities abroad, net foreign assets panies was recorded by the end of 2010. is positive, which means that this method of matu- Unlike the savings of private companies, house- rity adjustment of funds is extremely expensive and hold savings have been growing continuously since results in a high imbalance of interest rates between the beginning of the global economic crisis, December foreign assets and obligation. The study also points to 2008. In the period from the beginning of the eco- the importance of developing the domestic financial nomic crisis to the third quarter of 2016, household market because it would enable more efficient use savings grew continuously at a rate of 8-15%, which is of available funds and reduce the costs of adjusting a cumulative as much as 97%, as noted earlier. funds because it would avoid the impact of political It can be concluded that private corporate depos- risk on interest rates and increase competitiveness. its were more affected by the global crisis than house - The development of financial markets and a deep - hold deposits, which is to be expected, given that it is er financial structure would also mean more effective a for-profit sector that is much more sensitive to eco - work of the Central Bank and the overall monetary nomic fluctuations than the household sector, where system, where the effect of expansion would expand precautionary savings occur. in times of crisis. from financial markets to the real economy. Greater Savings and time deposits of private for-profit efficiency of the monetary multiplier effect would companies recorded a gradual, slight increase from improve the effectiveness of available resources in the 2010-2016, but with constant oscillations to various financial system. changes. As previously pointed out, the financial market of Bosnia and Herzegovina is dominated by the bank- 4.2. Defining model variables ing sector, with banks that are subsidiary companies of foreign “mother” banks. These institutions have In this research, two models were developed, as men- contributed to increasing market competitiveness tioned earlier. One model is related to the relationship and consequently reducing interest rates. This fact between household savings and economic growth, becomes questionable if we take into account that the and the other model is related to the relationship be- financial market is not sufficiently developed to en- tween private profit enterprises savings and economic sure the competitiveness of various financial institu- growth. Therefore, the following variables are includ- tions. Also, the interest rate is above those in the euro ed: household savings, savings of private for-profit area. The influence of political risk on the interest rate companies and GDP. should not be neglected. Household savings were measured as the sum of The overall situation in the Bosnia and household savings and time deposits and this was Herzegovina’s financial market shows a mismatch be - used as a variable expressed in absolute terms in BAM tween short-term sources and long-term investment and in relative terms, as a percentage of GDP. needs, as is generally the case in underdeveloped and Savings of private companies represent the sum of low-income developing countries. The functioning of savings and time deposits of profitable private com- the financial market is significantly influenced by the panies. In this case, too, savings in absolute amount CBR (currency board regime - this system implies a and as a measure expressed as a percentage of GDP fixed exchange rate of BAM. Each issued BAM has cov - are used. erage in foreign currency or monetary liabilities must Data on household savings and profits of for- never be higher than the foreign exchange assets of profit companies were taken from the database of the Central Bank of Bosnia and Herzegovina). The cur- the Central Bank of Bosnia and Herzegovina (bulletins rency BAM is linked to the currency “anchor” euro, so from 2000-2016). According to the methodology of the purchase of BAM is made at a fixed exchange rate the Central Bank of Bosnia and Herzegovina, a house- of 1.95583 BAM for one euro), which imposes very hold or a population means one or more persons who rigid and strict rules. The mismatch between short- earn income on the market through their work, spend term sources of funds and long-term investments, ac- the earned income, and if it is a community, share a companied by the underdevelopment of the domes- place of residence. Private for-profit/non-financial tic market for the necessary maturity transformation enterprises, according to the same methodology, are and the rigidity of the monetary system, is resolved enterprises in majority private ownership of legal or externally by financing through “parent” banks that natural persons engaged in the production and sale perform the necessary transformation. of goods and/or services for the purpose of gaining Analyzes show (e.g. Kreso and Begović 2012) that profit. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 119 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Time series of gross domestic product at current Therefore, a large variance, prognostic error, occurs and constant prices required additional prepara- in the case of neglecting the importance of seasonal tion in order to be used as variables in defined mod- influence, so it is necessary to conduct the analysis els. Namely, the relevant institutions in Bosnia and on seasonally adjusted, time-adjusted data. The ap- Herzegovina do not have quarterly data on gross plication of seasonal adjustment methods is aimed domestic product for the observed period 2000q1- only at assessing and eliminating the seasonal impact, 2016q4, so additional recalculations were needed. The without analyzing or explaining the cause of seasonal time series of the quarterly gross domestic product at phenomena. current prices was obtained from the Release of the During this research, X-13 ARIMA and TRAMO/ Agency for Statistics of Bosnia and Herzegovina (2013 SEATS methodologies were used for the purpose of and 2016). The estimate of the time series of the quar- seasonal adjustment of time series data. The aim was terly gross domestic product in current prices in the to check whether both methods give equal results. period from 2000q1 to 2005q4 is taken from the paper Both approaches to seasonal adjustment were found by Čolaković and Hlivnjak (2007). to give almost identical results, so it was decided to The time series of the quarterly gross domestic use only the latest version of the first X-13 ARIMA pro - product at constant prices (prices from 2010) was gram in this study. obtained from Release Agency for Statistics of Bosnia Two-unit root tests were used to examine the sta- and Herzegovina (2016, this is a time series of chain- tionarity of the included variables: the Augmented linked values in pr ices in 2010). In the Release of the Dickey-Fuller test (ADF) and the Kwiatkowski-Phillips- Agency for Statistics of Bosnia and Herzegovina 2013, Schmidt-Shin (KPSS) test. Unlike the ADF test, which the values of the quarterly gross domestic product takes the nonstationarity of a process as a null hy- of chain-linked values in 2005 prices are given. These pothesis, the KPSS test claims the stationarity of a values were converted into chain indices which were process with a null hypothesis (Bahovec and Erjavec then used to calculate the value of the quarterly gross 2009). Therefore, the KPSS test is mainly used for the domestic product of chain-linked values, but now at purpose of confirmatory analysis, verification of re - 2010 prices. This reconstructed the series of real GDP sults obtained using the ADF test or some other single in the period from 2006q1 to 2016q3 in 2010 prices. root test. The real GDP series in the period from 2004q1 to When assessing whether one time series can serve 2005q4 was reconstructed taking into account the in predicting another, the Granger test of causality estimates of the GDP series in current prices from the (Granger 1969) is most often used, as can be seen from paper Čolaković and Hlivnjak (2007). The ratio of the a review of the empirical literature. Unlike regression, real GDP series to the GDP series at current prices in which only indicates a mere relationship between 2006 varied from quarter to quarter, taking values of variables, Granger’s causality test tests whether the 1.34, 1.21, 1.25, and 1.02 from the first to the fourth future value of one variable can be predicted based quarter, respectively. Precisely these quarterly values on the past values of another v ariable. From an econo- were used to multiply the values of the GDP ser ies in metric point of view, the Granger test is considered to current prices in the period 2004q1 to 2005q4, thus give “predictive causality” (Granger 1988). The basic obtaining a rough estimate of real GDP in 2010 prices. definition of the test does not include latent confu- As the variable GDP growth rate variable was used sion effects or current nonlinear causal relations. in the analysis instead of the level of that series, this In the following, explanations and the basic formu- rough estimate of the value of real GDP satisfies the las of the Granger test will be given in accordance with needs of the analysis in which the dynamics of the se- the article published by the author Granger in 1988. ries change is observed, and not its level in absolute A variable X can be said to cause a variable Y ac- amount. cording to the Granger test, if by t-test and F-test lagged values of X it can be proved that the values of X give enough statistically significant values about future values of Y. Thus, the variable X causes accord- 4.3. Research methods ing to Granger’s test, the variable Y if the predictions When using economic indicators in the form of time of the Y series are better based on past data and X and series, it is necessary to do their seasonal adjustment Y, than only based on Y. in order to exclude the influence of the season, i.e., to The Granger test is based on two principles. First, exclude short-term fluctuations in the phenomenon. the causality of variables arises before their effects. The total variance of the time series is significantly Here the term cause corresponds to the term pre- determined by the seasonal variation of the series. cede. The assumption is that the past and the present 120 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA cannot be caused by the future. Thus, if event A is The regression equation can retain all individual preceded by event A then A cannot cause A . Also, statistically significant X lags according to t-statis- 2 1 2 t if A occurs before A it does not necessarily mean tics. The condition for this setting is that in the total 1 2 that A causes A . Another assumption is that causal- amount, all lags, according to F-statistics, contribute to 1 2 P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x ity contains unique information about future effects. the increase of the regression power. When determining the causal effects of X on Y , the A multivariate Granger causality test (Granger t t Y  a  a Y  a Y  ... a Y  . following hypothesis is tested taking into account the 1988) is used to define vector autoregressive models. t 0 1 t1 2 t2 m tm t dxi given assumptions: Here it is set that X(t) ∈ R for t=1,...,T is a d-dimen- sional multivariate series. The VAR model uses the fol- Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x lowing form of the Granger causality test: Y  a  a Y  a Y  ... a Y  . In this expression, the notations are used: P for X (t )  A X (t  ) . t 0 1 t1 2 t2 m tm t   t 1 probability, A for an arbitrary non-empty set, and with I (t) and I (t) available information at time t. It is con- In the preceding expression, X is called the -x i Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t cluded that X causes Y according to the Granger test if Granger causality of some other variable X provided this hypothesis can be confirmed. that at least one element, for, in the absolute value is T Xhe gener (t )  al rule of the Gr A X (t  ) . anger test (Granger significantly greater than zero.   t 1 1988) is that in the case of a time series analysis that Linear testing in the Granger test is suitable for is stationary, the absolute values of the variables are determining causality in the mean values of v ariables. used, while for nonstationary series, the correspond- In this way, Granger’s causality for “larger moments” ing differences for which the variables are stationary such as variance cannot be determined. Therefore, a are used. Similar to previous tests, Granger’s meth- nonparametric Granger causality test was developed. odology uses Akaike or Schwarz information criteria The definition of causality in this test is general so that to select the lags length. The regression equation re- it does not include some modeling assumptions, as tains the individual lagged value of a variable if two is the case with linear autoregressive models. Better conditions are met: the value is statistically significant parametric models, including those with more mo- according to the t-test and if the observed value and mentum and/or nonlinearity, can be set up based on other lagged values of the variable together increase Granger’s nonparametric causality test. the exploitative power of the model according to Granger’s causality test, however, has certain F-statistics. If the lagged values of the explanatory limitations and shortcomings during its application variable are retained in the regression, then the null (Maziarz 2015). The first drawback is that Granger’s hypothesis of the absence of Granger causality cannot causality does not necessarily mean “true causality.” be rejected. The only thing that Granger’s test satisfies in terms of The results of the Granger test may show that nei- causality is Humaean’s definition of causality. Another ther variable causes the other, or, if two variables are drawback is that the alternative Granger hypothesis observed, that the variables cause each other. cannot be rejected if both variables, X and Y , are de- t t The simplest form of the Granger test (Granger rived from a common third process with different lags. 1988) is for models that include two variables, Y and Also, it is possible that there are no changes in the X . The starting point is the assumption that these are other variable if only one variable is manipulated. The stationary series. In order to test the null hypothesis, it Granger test is generally developed for testing pairs is necessary to first determine the appropriate num- of variables, so deficiencies may occur if the “right” ber of Y lags that will be included in the one-variant relation includes three or more variables. Incorrect P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x autoregression Y : test results can also be caused by: a nonlinear causal relationship between variables; non-stationarity and Y  a  a Y  a Y  ... a Y  . nonlinearity of the observed series; insufficient or t 0 1 t1 2 t2 m tm t excessive sampling; and the existence of rational ex- In the next step, the autoregressive equation is ex- pectations. By vector autoregression, a similar test can Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x tended with lags of the variable X : be applied to a number of variables included in the model. Y  a  a Y  a Y  ... a Y  . t 0 1 t1 2 t2 m tm t X (t )  A X (t  ) . Granger’s causality test continued to evolve de-   t 1 spite the aforementioned limitations. Thus, a form of where p and q denote the shortest and longest dis- the test was developed that is not sensitive to violat- Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t placements, respectively, for which the lagged values ing the assumption about the normality of the model of the variable X are significant. error distribution. Since most financial variables are X (t )  A X (t  ) .   t 1 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 121 not normally distributed then the extended version the hypothesis that the coefficients at the first p lags of the Granger test finds special application in ana- of the first variable are equal to zero. Then the same is lyzes of causality in finance. Asymmetric causality is done with the coefficients on the first p lags of the sec - especially emphasized in recent works in the field of ond variable. Rejecting the null hypothesis that these causality, with the aim of separating the impact of coefficients are equal to zero, is equivalent to rejecting positive and negative changes on causality. the hypothesis of the absence of causality in Granger’s sense between the two observed variables. In other words, the rejection of the hypothesis indicates the existence of Granger causality among the variables. 5. Research results The results of Granger’s test of causality among Two software programs were used in the study in- the variables of savings and economic growth are cluding: EViews 9.5 and R with a number of packages given in Table 1. Granger’s test was applied to both created for time series analysis. the savings of households and private for-profit com- In order to verify the existence of causality be- panies, as components of private savings. tween household savings and economic growth, as The results of Granger’s test of causality indicated investigated earlier, Granger’s causality test was ap- that there is no causal link between household sav- plied, namely the Toda-Yamamoto procedure. ings and economic growth, thus rejecting hypothesis The Toda-Yamamoto procedure first required to we tested, which claims that household savings in determine the order of integration of the two vari- Bosnia and Herzegovina initiate economic growth. ables between which it was sought to check whether The same results were obtained by: Cullison (1993) there was a causal relationship in Granger’s sense. This for the USA; Sinha (1996) for India; Saltz (1999) for: was done using ADF and KPSS unit root tests (results Colombia, Jamaica, Peru and the Philippines; Agrawal of unit root tests, cointegration and VAR models are (2000) for Nepal; Agrawal (2001) for Thailand and attached to the paper). Korea; Konya (2005) for a large number of countries, When the results of unit root tests are obtained, Shah (2015) for Pakistan. it is necessary to determine the maximum order of The results only confirm the non-existence of a integration m of the observed two variables. Thus, it strict rule on the presence and direction of the link be- was obtained that household savings are I (1) process tween savings and economic growth, even when ap- while GDP growth rate I (0) is a process. This means plying the same methodology in individual countries. that the maximum order of integration is m = 1. Savings and economic growth show cointegra- The next step defines the Vector autoregres- tion, which means that the variables have a common sion (VAR) model based on the level of the variables stochastic trend. However, the underdevelopment regardless of the results of the unit root tests. Here, of the financial and overall economic market, along household savings and GDP growth rates enter di- with the rigidity of the political system, prevents the rectly into the VAR model. The maximum lag length necessary flow of sufficient capital for faster economic for the variables in the vector autoregressive model growth, which is an imperative in the Bosnia and is then determined. The obtained optimal lag length Herzegovina’s economy. As in the case of other poor is denoted by p. It is checked whether the VAR model economies, the Bosnia and Herzegovina’s economy is correctly specified by determining whether there is finances its economic growth with expensive foreign autocorrelation of VAR model residuals. capital. If the variables are of the same order of integra- A possible cause of such results is perhaps the more tion, then it can be checked whether they are coin- significant effect of demographic and cultural fac - tegrated using Johansen’s methodology, in order to tors on economic growth in Bosnia and Herzegovina. further verify the results of Granger’s causality test. Numerous authors emphasize the importance of in- Namely, if two or more variables are cointegrated, cluding these variables in the research: Prskawetz et then there must be a Granger causality between them al. (2007), Crenshaaw and Robinson (2010), Basu et al. in either one or both directions. However, the oppo- (2013), Diep (2015), Yanagizawa-Drott and Compante site does not apply, i.e., if there is a Granger causality (2015), Gorodnichenko and Roland (2016). among the variables it does not mean that the series The inclusion of demographic variables in the must also be cointegrated. model in this study was not possible due to the short- Once the appropriate VAR model has been deter- ness of time series data. mined, m lags of each variable are added to each of Turan and Gjergij (2014) analyzing the relation- the equations of the VAR model. Granger causality ship between savings and economic growth in testing is then performed by applying the Wald test to Albania, an economy quite similar to that in Bosnia 122 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Table 1. Granger causality test (Toda-Yamamoto procedure) Null hypothesis Hi-square p-value Granger causality Households Savings do not cause GDP growth rate 0.872 0.351 Absence of G causality Savings do not cause GDP level 1.827 0.177 Absence of G causality Savings in % of GDP do not cause GDP growth rate 3.487 0.175 Absence of G causality Savings in % of GDP do not cause the level of GDP 1.036 0.309 Absence of G causality GDP growth rate does not cause savings 0.256 0.613 Absence of G causality GDP level does not cause savings 0.613 0.434 Absence of G causality GDP growth rate does not cause savings in % of GDP 0.483 0.786 Absence of G causality GDP level does not cause savings in % of GDP 0.002 0.964 Absence of G causality Private for-profit companies Savings do not cause GDP growth rate 2.760 0.097 Absence of G causality Savings do not cause GDP level 5.946 0.114 Absence of G causality Savings in % of GDP do not cause GDP growth rate 0.849 0.357 Absence of G causality Savings in % of GDP do not cause GDP level 4.525 0.033 Granger causality GDP growth rate does not cause savings 0.850 0.357 Absence of G causality GDP level does not cause savings 4.152 0.246 Absence of G causality GDP growth rate does not cause savings in % of GDP 2.761 0.097 Absence of G causality GDP level does not cause savings in % of GDP 0.102 0.750 Absence of G causality Source: Author and Herzegovina, also found that there is cointegra- 6. Conclusion tion among the variables involved, and stressed that the government should pay special attention to this The importance of the relationship between sav- important issue for faster economic growth. ings and economic growth, in addition to basic eco- In this research the null hypothesis of the absence nomic theories, has resulted in numerous empirical of causality was rejected only in the case of the vari- papers that have tried to at least determine the direc- ables of the share of private for-profit companies sav - tion of this relationship. However, no uniform results ings in the gross domestic product and the level of have been obtained. the gross domestic product. But strictly speaking, this This research also examined the direction and result does not change the general conclusion made intensity of the relationship between savings and above about the absence of a causal link between sav- economic growth in the underdeveloped economy ings and economic growth, because economic growth of Bosnia and Herzegovina. To verify the existence is measured by GDP growth rates, not its level. of causality between household savings and eco- If the level of significance of the test is reduced to nomic growth, the Granger causality test was applied, 10% then the hypothesis of the absence of Granger namely the Toda-Yamamoto procedure. The results of causality can also be rejected in two cases when the Granger’s test of causality indicated that there is no following variables are observed: private for-profit causal link between household savings and the pri- companies savings and GDP growth rate and GDP vate for-profit companies and economic growth. growth rate and private for-profit companies savings According to the results of this research, savings in% of GDP. But given the significance level of 10% this and economic growth in Bosnia and Herzegovina can be considered as a very weak confirmation of the show cointegration, which means that the variables existence of Granger causality among the observed have the same stochastic trend. However, the under- variables. development of the financial and overall economic market, along with the rigidity of the political system, SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 123 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Literature prevents the necessary flow of sufficient capital for faster economic growth, which is an imperative in the Bosnia and Herzegovina’s economy. As in the case of AbuAl-Foul, B. 2010. The causal relation between savings other poor economies, the Bosnia and Herzegovina’s and economic growth: Some evidence from MENA coun- economy finances its economic growth with expen- tries. Topic in Middle Eastern and African Economies 12: sive foreign capital. 1-12. A possible cause of such results is perhaps a more Adam, I. O., Musah, A., Ibrahim, M. 2017. Putting the cart be- significant effect of demographic and cultural factors fore the horse? Re-examining the relatonship between on economic growth in Bosnia and Herzegovina. domestic savings and economic growth in selected Sub- The contribution of the conducted research is Saharan African countries. Journal of African Business, reflected in the examination of the direction and ma- 18 (1): 102-123. turity intensity of causality between savings and eco- Agrawal, P. 2000. Savings, investment and growth in South nomic growth, in small, open economy of Bosnia and Asia. Mumbai: Indira Gandhi Institute of Development Herzegovina. It was found that there is cointegration Research. among the variables, but not causality. Agrawal, P. 2001. The relation between savings and growth: The contribution is reflected in the fact that by Cointegration and causality evidence from Asia. Applied examining the direction and intensity of the causal- Economics, 33(4): 499-513. ity of savings and economic growth, it contributed Agency for Statistics of Bosnia and Herzegovina. 2016. to the identification of potential sources of financing Announcement, Gross Domestic Product-Quarterly economic growth through the optimal distribution of Data, No. 4. accumulated capital. The results show that in Bosnia Agency for Statistics of Bosnia and Herzegovina. 2013. and Herzegovina there are enough funds in the form Announcement, Gross Domestic Product-Quarterly of household savings to finance economic growth, Data, No. 1. the variables show cointegration, but inadequate Alguacil, M., Cuadros, A., Orts, V. 2004. Does saving re- and rigid distribution prevents causality among these ally matter for growth? Mexico (1970–2000).  Journal of variables. International Development. 16(2): 281-290. The biggest limitation in writing this paper was Bahovec, V. and Erjavec, N. 2009). Uvod u ekonometrijsku the availability of data. So, time series of gross domes- analizu, Element. tic product at current and constant prices required ad- Basu, T., Barik, D., Arokiasamy, P. 2013. Demographic ditional preparation in order to be used as variables in Determinants of Economic Growth in BRICS and se- defined models. lected Developed Countries, Presented in XXVII IUSSP It would be useful to conduct such research after International Population Conference (IUSSP-2013). a certain period of time, in order to verify the obtained August 2013 at BEXCO. Busan, Republic of Korea. results with longer observed time spans. Also, as Bayar, Y. 2014. Financial development and domestic savings noted earlier, it would be important to include de- in emerging Asian countries,  Theoretical and applied mographic and cultural determinants in the form of economics. 21(7 – 596): 55-66. mediation or moderation variables in models of the Central Bank of Bosnia and Herzegovina, Quarterly Bulletins: relationship between savings and economic growth, from Bulletin 2 in 2000 to Bulletin 3 in 2016. although it is a huge challenge with quartery data. In Crenshaw, E., Robison, K. 2010. Socio-demographic the continuation of the research, it would be useful to Determinants of Economic Growth: Age Structure, conduct such analyzes for the Western Balkans region Preindustrial Heritage and Sociolingustic Integration. and make a comparative analysis. Social Forces. 88 (5): 2217-2240. Čokalović, B., Hlivnjak, S. 2007. Estimate of quarterly GDP for the period 2000-2006. ISSN 1840-2593. Central Bank of Bosnia and Herzegovina. Department of Economic Research, Statistics and Publications. Cullison,W. E. 1993. Saving measures as economic growth indicators. Contemporary Policy Issues. 11(1): 1-8. Dhakal, D., Grabowski, R., Shields, M. P. 1992. The role of sav- ings in economic development: The U.S. and Japanese experiences. Japan and the World Economy. 3(4): 331-340. Diep, V. T. T. 2015. Demographic Factors and Economic Growth: The Bi-directional causality in South East Asia. 124 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA The ninth Vietnam Economists Annual Meeting. Da Olajide S. O. 2009. Does Saving Really Matter for Growth Nang City. August 2015. in Developing Countries? The Case of a Small Open Domar, E. D. 1946. Capital expansion, rate of growth, and Economy. http:// economics.ca /2009/ papers/ 0619 .pdf. employment. Econometrica, Journal of the Econometric (accessed November 10, 2016) Society: 137-147. Patra, S. K., Murthy, D. S., Kuruva, M. B., Mohanty, A. Granger, C. W. J. 1969. Investigating Causal Relations by 2017. Revisiting the causal nexus between savings Econometric Models and Cross-Spectral Methods. and economic growth in India: An empirical analy- Econometrica. 37: 424 - 438. sis. Economia. 18 (3): 380-391. Granger, C. W. 1988. Causality, cointegration, and control. Prskawetz, A., Fent, T., Barthel, W. 2007. The Relationship Journal of Economic Dynamics and Control.  12(2-3): Between Demographic Change and economic Growth in 551-559. EU. Institut fur Demographie. Osterreichische Akademie Gorodnichenko, Y., Roland, G. 2016. Culture, institutions and der Wissenschaften, Forschungsbericht. No. 32. https:// the wealth of nations. Review of Economics and Statistic. www.oeaw.ac.at/fileadmin/subsites/Institute/VID/PDF/ 99 (3): 402-416. Publications/Forschungsberichte/FB32.pdf. (accessed Guma, N., Bonga-Bonga, L. (2016). The relationship between November 15, 2016). savings and economic growth at the disaggregated lev- Ribaj, A., Mexhuani, F. 2021. The impact of savings on eco- el. Economia Internazionale/ International Economics, nomic growth in a developing country (the case of Camera di Commercio Industria Artigianato Agricoltura Kosovo). Journal of Innovation and Entrepreneurship. 10 di Genova. 70(1): 1-24. (1): 3-13.  Harrod, R. F. 1939. An essay in dynamic theory. The Economic Saltz, I. S. 1999. An examination of the causal relationship Journal. 49(193): 14-33. between savings and growth in the Third World. Journal Harrod, R. F. 1948. Towards a dynamic economics, London, of Economics and Finance. 23(1): 90-98. Macmillan. Shah, R. K., (2015), Saving and Economic Growth in South Jagadeesh, D. (2015). The impact of savings in eco- Asia: Evidence from Time Series Analysis.  Economic nomic growth: an empirical study based on Journal of Nepal. https://www.econbiz.de/Record/sav- Botswana. International Journal. 2: 10 - 21. ing-and-economic. (accessed November 10, 2016) Joshi, A., Pardhan, S., Bist, J. P. 2019. Savings, investment Sinha, D. 1996. Saving and economic growth in India. and growzh in Nepal: an empirical analysis. Financial Economia Internazionale. 49 (4): 637-647. Innovation. 5 (1): 1-13. Solow, R. M. 1956). A contribution to the theory of economic Keynes, J. M. 1936. The general theory of interest, employ- growth. The quarterly journal of economics. 70 (1): 65-94. ment and money. Macmillan. Tang, C. F. 2008. An empirical modelling on savings be- Kim, N. T. and Nguyen, H. H. 2017. Impacts of domestic sav- haviour in Malaysia. Labuan Bulletin of International ings on economic growth of Vietnam.  Asian Journal of Business and Finance. 6: 57-76. Economic Modelling. 5(3): 245-252. Tang, C. F. 2009a. Does causality technique matter to sav- Konya, L. 2005. Saving and growth: Granger causality analy- ings-growth nexus in Malaysia?. Malaysian Management sis with bootstrapping on panels of countries. Journal of Journal. 13(1-2): 1-10. Economic Research. 10: 231-260. Tang, C. F. 2009b. The savings-growth nexus for Asian econ- Kreso, S. and Begović, S. 2012. Maturity (mis) matching in omies: New evidence from the timevarying economet- currency board (like) regimes. Interdisciplinary manage- rics methods. Paper presented at the Symposium of ment research. 8: 423 – 428. USM Fellowship Holders 2009. Universiti Sains Malaysia. Mankiw, N.G, Romer D., Weil, D. N. 1992. A Contribution to Penang. Malaysia. 14-15 November. the Empirics of Economic Growth. The Quarterly Journal Tang, C. F. and Tan, B. W. 2014. A revalidation of the savings- of Economics. Oxford University Press. 107(2): 407 - 437. growth nexus in Pakistan. Economic Modelling. 36: Maziarz, M. 2015. A review of the Granger-causality falla- 370-377. cy. The journal of philosophical economics: Reflections Turan, G. and Gjergji, O. 2014. What is the Impact of Savings on economic and social issues 8(2): 86-105. on Growth? The Case of Small Open Economy (Albania). Mohan, R. 2006. Causal Relationship Between Savings and Mediterranean Journal of Social Sciences. 5 (13): 360-368. Economic Growth in Countries with Different Income Yanagizawa-Drott, D. and Compante, F. 2015. Does religion Levels. Economics Bulletin. 5: 1 - 12. Aec ff t Economic Growth and Happiness? Evidence from Nwanne, T. F. I. 2016. Implication of savings and investment Ramadan. The Quarterly Journal of Economics. 130 (2): on economic growth in Nigeria. International Journal 615-658. of Small Business and Entrepreneurship Research.  2(4): Wyk, B. F. and Kapingura, F., M. 2021. Unerstanding the nexus 74-86. between savings and economic growth: A South African context. Development Southern Africa. 38 (5): 828-844. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 125 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA APPENDIX 1: Granger causality test Household savings vs. GDP growth rate Step 1: Unit root tests Null Hypothesis: STEDOM_D11 has a unit root Exogenous: Constant Lag Length: 0 (Automatic - based on SIC, maxlag=10) t-Statistic Prob.* Augmented Dickey-Fuller test statistic  0.745  0.992 Test critical values: 1% level -3.544 5% level -2.911 10% level -2.593 *MacKinnon (1996) one-sided p-values. Augmented Dickey-Fuller Test Equation Dependent Variable: D(STEDOM_D11) Method: Least Squares Date: 05/11/17 Time: 02:18 Sample (adjusted): 2001Q1 2015Q4 Included observations: 60 after adjustments Variable Coefficient Std. Error t-Statistic Prob. STEDOM_D11(-1) 0.009 0.012 0.745 0.459 C 79.751 45.196 1.765 0.083 R-squared 0.009 Mean dependent var 108.185 Adjusted R-squared -0.008 S.D. dependent var 186.709 S.E. of regression 187.417 Akaike info criterion 13.337 Sum squared resid 2037261. Schwarz criterion 13.407 Log likelihood -398.120 Hannan-Quinn criter. 13.365 F-statistic 0.555 Durbin-Watson stat 2.624 Prob(F-statistic) 0.459 126 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Null Hypothesis: STEDOM_D11 is stationary Exogenous: Constant Bandwidth: 6 (Newey-West automatic) using Bartlett kernel LM-Stat. Kwiatkowski-Phillips-Schmidt-Shin test statistic  0.960 Asymptotic critical values*: 1% level  0.739 5% level  0.463 10% level  0.347 *Kwiatkowski-Phillips-Schmidt-Shin (1992, Table 1)  Residual variance (no correction)  4271797.  26754360 HAC corrected variance (Bartlett kernel) 0 KPSS Test Equation Dependent Variable: STEDOM_D11 Method: Least Squares Date: 05/11/17 Time: 02:20 Sample (adjusted): 2000Q4 2015Q4 Included observations: 61 after adjustments Variable Coefficient Std. Error t-Statistic Prob. C 3259.592 266.827 12.216 0.000 R-squared 0.000000 Mean dependent var 3259.592 Adjusted R-squared 0.000000 S.D. dependent var 2083.985 S.E. of regression 2083.985 Akaike info criterion 18.138 Sum squared resid 2.61E+08 Schwarz criterion 18.173 Log likelihood -552.215 Hannan-Quinn criter. 18.152 Durbin-Watson stat 0.011 (Other appendices of this step are available from the author on any request.) Conclusion: Household savings are I (1), GDP growth rate is I (0). SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 127 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 2: Order the VAR model VAR Lag Order Selection Criteria Endogenous variables: STEDOM_D11 RBDP_D11  Exogenous variables: C  Date: 05/11/17 Time: 02:31 Sample: 2000Q1 2016Q4 Included observations: 39  Lag LogL LR FPE AIC SC HQ 0 -237.791 NA   750.883  12.297  12.382  12.328 1 -140.014 180.511* 6.127* 7.488* 7.744* 7.580* 2 -138.860  2.012  7.106  7.634  8.060  7.787 3 -137.614  2.046  8.225  7.775  8.372  7.989 4 -137.154  0.707  9.952  7.957  8.724  8.232 5 -132.120  7.229  9.576  7.904  8.842  8.240 6 -131.834  0.381  11.837  8.094  9.203  8.492 7 -130.342  1.836  13.878  8.223  9.502  8.682 8 -127.907  2.747  15.677  8.303  9.753  8.823  * indicates lag order selected by the criterion  LR: sequential modified LR test statistic (each test at 5% level)  FPE: Final prediction error  AIC: Akaike information criterion  SC: Schwarz information criterion  HQ: Hannan-Quinn information criterion Conclusion: A lag of one quarter 128 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 3: Check the residual VAR (1) of the model VAR Residual Portmanteau Tests for Autocorrelations Null Hypothesis: no residual autocorrelations up to lag Date: 05/11/17 Time: 02:34 Sample: 2000Q1 2016Q4 Included observations: 46 Lags Q-Stat Prob. Adj Q-Stat Prob. df 1  1.511 NA*  1.545 NA* NA* 2  2.057  0.7252  2.116  0.715 4 3  3.739  0.8799  3.915  0.864 8 4  4.815  0.9639  5.093  0.955 12 *The test is valid only for lags larger than the VAR lag order. df is degrees of freedom for (approximate) chi-square distribution VAR Residual Serial Correlation LM Tests Null Hypothesis: no serial correlation at lag order h Date: 05/11/17 Time: 02:35 Sample: 2000Q1 2016Q4 Included observations: 46 Lags LM-Stat Prob 1  2.086  0.720 2  0.490  0.975 3  1.521  0.823 4  0.977  0.913 Probs from chi-square with 4 df. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 129 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 4: Cointegration Date: 05/11/17 Time: 02:37 Sample (adjusted): 2004Q4 2015Q4 Included observations: 45 after adjustments Trend assumption: Linear deterministic trend Series: STEDOM_D11 RBDP_D11  Lags interval (in first differences): 1 to 1 Unrestricted Cointegration Rank Test (Trace) Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None *  0.373  21.045  15.495  0.007 At most 1  3.09E-05  0.001  3.841  0.969  Trace test indicates 1 cointegratingeqn(s) at the 0.05 level  * denotes rejection of the hypothesis at the 0.05 level  **MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue) Hypothesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None *  0.374  21.043  14.265  0.004 At most 1  3.09E-05  0.001  3.841  0.969  Max-eigenvalue test indicates 1 cointegratingeqn(s) at the 0.05 level  * denotes rejection of the hypothesis at the 0.05 level  **MacKinnon-Haug-Michelis (1999) p-values  Unrestricted Cointegrating Coefficients (normalized by b’*S11*b=I):  STEDOM_D11 RBDP_D11  0.000  66.958  0.001 -8.993  Unrestricted Adjustment Coefficients (alpha):  D(STEDOM_D11)  12.200  0.655 D(RBDP_D11) -0.014  1.20E-05 1 Cointegrating Equation(s):  Log likelihood -163.913 Normalized cointegrating coefficients (standard error in parentheses) STEDOM_D11 RBDP_D11  1.000  288124.2  (56761.5) Adjustment coefficients (standard error in parentheses) D(STEDOM_D11)  0.003  (0.004) D(RBDP_D11) -3.35E-06  (6.8E-07) Conclusion: There is cointegration. 130 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 5: Add one lag to VAR (1) and rate VAR (2) VAR Granger Causality/Block Exogeneity Wald Tests Date: 05/11/17 Time: 02:43 Sample: 2000Q1 2016Q4 Included observations: 45 Dependent variable: STEDOM_D11 Excluded Chi-sq df Prob. RBDP_D11  0.256 1  0.613 All  0.256 1  0.613 Dependent variable: RBDP_D11 Excluded Chi-sq df Prob. STEDOM_D11  0.872 1  0.351 All  0.872 1  0.351 Other appendices of this step are available from the author on any request. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 131 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png South East European Journal of Economics and Business de Gruyter

The Causal Link Between Savings and Economic Growth in Bosnia and Herzegovina

Loading next page...
 
/lp/de-gruyter/the-causal-link-between-savings-and-economic-growth-in-bosnia-and-M7xr15aFwb
Publisher
de Gruyter
Copyright
© 2021 Irma Đidelija, published by Sciendo
ISSN
2233-1999
eISSN
2233-1999
DOI
10.2478/jeb-2021-0018
Publisher site
See Article on Publisher Site

Abstract

The causal link between savings and economic growth has been extensively discussed in the economic growth and development literature, but the question of the direction of this link has not yet been clearly defined. The aim of this paper is to determine the direction and intensity of savings causality (components of private savings) and economic growth in Bosnia and Herzegovina. Granger’s causality test, the Toda- Yamamoto procedure, was applied to test for causality between savings and economic growth. The results of Granger’s causality test indicated that there is no causal link between components of private savings with economic growth in Bosnia and Herzegovina. It was found that among the variables there is cointegration, but not causality, which means that the variables have a common stochastic trend. This fully corresponds to the characteristics of the Bosnia and Herzegovina’s economy. Keywords: Saving, economic growth, macroeconomics, econometric analysis JEL Classification: C32, E21 1. Introduction One of the most important issues in any economy, direction and intensity of the causality of savings and regardless of the degree of its growth and develop- economic growth, neither in the short nor in the long ment, is to ensure sufficient and quality funds to run. finance the economic growth. Given that foreign capi- Therefore, this paper aims to examine the prob- tal is quite expensive and risky to finance domestic lem of endogenous savings and economic growth investment and economic growth, domestic capital is in the small, underdeveloped economy of Bosnia considered a much more favorable and secure source and Herzegovina. In this case, the focus is on private of funds. Savings are most often mentioned in the lit- erature as a domestic source of funds. The relationship between savings and economic growth is defined Irma Đidelija, PhD in the economic literature through the three most Assistant Professor important theories of economic growth, which are: Faculty of Economics Keynesian theory of growth, Harrod-Domar’s theory University „Džemal Bijedić“ in Mostar and Solow’s theory. E-mail: irma.djidelija@unmo.ba The relationship between savings and economic Address: Vrapčići bb, 88000 Mostar, growth, given its economic significance, has been the Bosnia and Herzegovina subject of numerous empirical studies. These works ORCID: 0000-0003-1666-7363 did not give unambiguous results in terms of the Copyright © 2021 by the School of Economics and Business Sarajevo 114 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA savings, looking at its individual components sepa- principles of Granger’s causality test. The results of rately: household savings and savings of private the research, the fifth part of the paper, is the most for-profit companies. The reason for this is the fact important part, where the results of the research are that private savings, and especially its component of presented with an appropriate discussion. At the end, household savings, is recording continuous growth in concluding remarks were given with summarized re- Bosnia and Herzegovina, despite the great financial sults of the conducted research. crisis in 2008. Also, private savings, according to the Central Bank of Bosnia and Herzegovina (quarterly bulletins from 2000-2016), is the dominant part of na- 2. Theoretical framework tional savings, which is another reason to include only private savings in the model of this survey. When it comes to theories dealing with economic Private savings in Bosnia and Herzegovina are growth, meaning the the relationship between sav- mostly realized in the form of deposits given the ings and economic growth, two/three significant the - dominance of the banking sector in the financial mar - ories stand out in the literature: Keyenes’ hypothesis, kets. Thus, private savings in Bosnia and Herzegovina, neoclassical (Harrod-Domar’s and Solow’s) theory of viewed as the sum of long-term deposits, are growing growth, and endogenous theory of economic growth. despite the recent global financial crisis, but also the One of the most significant works in this field slowdown in economic growth and political instabil- was offered by Keyens, in 1936, and that is the ity in the country. It can be seen from the bulletin “General Theory of Employment, Interest and Money”. of the Central Bank of Bosnia and Herzegovina that According to Keyens, only in conditions of full em- at the end of September 2016 (in this analysis, eight ployment can the classical theory be considered years before and eight years since the beginning of valid. However, full employment conditions are the the crisis were observed), total household deposits expectation rather than the rule. The real conditions amounted to BAM 10.25 billion. This is also the larg- are different, and they are based on a series of causal est growth in deposits so far. The share of household relations as follows: entrepreneurs earn income by deposits in total deposits of commercial banks in selling goods and services; as income grows, so does Bosnia and Herzegovina is 59.6%. Compared to the consumption, but to a lesser extent than income; end of September of the previous year, 2015, house- differences in the percentages of income and con- hold deposits increased by 6.6% or BAM 633.5 million. sumption growth are reflected in savings; there is an Compared to December 2008, when the economic imbalance between supply and demand, if savings are crisis began, deposits are constantly growing in the not put into operation through investment activities; range of 8-15% annually. Thus, a cumulative growth of entrepreneurs’ expectations regarding future incomes deposits of 97% or BAM 5.05 billion was achieved. become pessimistic, and therefore they do not create Based on the above, the main hypothesis of the new jobs, but also abolish existing ones; eventually, research is that household savings in Bosnia and unemployment gradually rises and demand for goods Herzegovina initiate economic growth. For the pur- and services declines. pose of comprehensive analysis, the research also The theory assumes that the amount of savings includes another component of private savings, sav- and the amount of investment are necessarily equal ings of private for-profit companies, but given the since both amounts represent the value of income negligible share in total private savings in Bosnia and that exceeds consumption. The amount of savings is Herzegovina, no specific hypothesis is defined for the the result of the collective behavior of individual con- relationship between private profit savings and eco - sumers, and the amount of investment is the result of nomic growth. the collective behavior of individual entrepreneurs. In order to test the set hypothesis, Granger’s cau- This conclusion, as stated, is not the result of any per- sality test, the Toda-Yamamoto procedure, will be sonality or subtlety of the set definition of income. applied. Namely, the equality of savings and investment neces- The paper is divided into five parts. After introduc - sarily appears in traditional theory, if income is equal tory remarks, a theoretical framework is given with a to the value of current output, current investment brief overview of the postulates of basic theories of equal to the value of the part of current output that economic growth. The third part of the paper pro- is not consumed, and savings equal to the value of in- vides an overview of the empirical literature dealing come that exceeds consumption. with the problem of endogenous savings and eco- The Harrod-Domar model is the first and simplest nomic growth. The fourth part of the paper presents growth model, which is a long-term projection of the research methodology, more precisely the basic Keynes’ short-term analysis. Harrod-Domar’s theory SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 115 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA originated in the late 1930s and mid-1940s (Harrod In addition to the previously mentioned dominant 1939; Harrod 1948; Domar 1946). Capital factors are theories of economic growth, it is certainly impor- cited as crucial factors of economic growth in the tant to mention the endogenous theory of economic Harrod-Domar model. Based on the adjustment of growth. This theory was officially published in his supply and demand, the model tries to explain the article, “A contribution to the Empirics of Economic possibility of stable growth. Growth”, by David Romer, Gregory Mankiw and David According to the model, in order to ensure a state Weil in 1992, and is an upgrade of the neoclassical of equilibrium, it is necessary that the expenditures theory of growth. generated by investments be large enough to ab- The theory of endogenous growth is an economic sorb the growth of output, as a result of investments. theory that claims that economic growth is generated Investments not only cause consumption, but also within a system as a direct result of internal processes. lead to the expansion of the productive capacities of According to this theory, increasing the human capital the economy. The theory argues that there must be a of a nation will lead to economic growth through the balance between the volume of consumption gener- development of new forms of technology and efficient ated by investments and the productive capacities and effective means of production. The theory argues created by investments. The model points out that that knowledge-based industries play a particularly excessive capital accumulation unaccompanied by important role - especially telecommunications, soft- adequate income growth causes hyperproduction, ware, and other high-tech industries -as they become unemployment, and economic depression. That is increasingly influential in developed and developing why net investment must grow continuously in order economies. to maintain full employment in the long run. If net Some other important postulates of the endog- investments grow, then real income must grow at a enous theory of economic growth are: government rate that will ensure the use of the full capacity of the policies can increase a country’s growth rate if they increased capital stocks. The required income growth lead to stronger competition in markets and help rate is called the “full capacity growth rate” or guaran- encourage product and process innovation; returns teed growth rate. on capital investment, especially in infrastructure and Solow’s theory of growth is the result of the work investment in education, health and telecommunica- of authors Robert Solow and Trevor Swan from 1956, tions; private sector investment in research and de- and is sometimes called the neoclassical model of velopment is a key source of technological progress; growth. According to Solow’s model, the drivers the protection of property rights and patents is key to of economic growth are three factors: technology, providing incentives for companies and entrepreneurs capital accumulation and labor. This theory extends to engage in research and development; investing in Harrod-Domar’s model to the following elements: la- human capital is a vital component of growth; govern- bor factors are included in the factors of production; ment policy should encourage entrepreneurship as a declining returns on labor and capital are required means to start new businesses and ultimately as an when variables are viewed separately, and constant important source of new jobs, investment and further returns when variables are viewed in combination; in- innovation. troduction of variable technology, which depends on time and differs from labor and capital. The Solow model points out that an increase in 3. Literature review the accumulation of capital and labor leads to a high- er rate of economic growth, but this is of a short-term Domestic savings can be a significant catalyst for nature given the declining returns. If the economy had growth, as higher savings rates allow investment to only one worker, adding another would result in a sig- grow at higher rates and further economic growth. nificant increase in output. However, if thousands of The causal link between savings and economic growth workers are employed in the economy, one new work- has been much discussed in the literature on econom- er will not significantly contribute to output growth. ic growth and development, however the question of In this case, there can be a growth of the economy at the direction of this link has not yet been unambigu- a stable rate, where the gross domestic product grows ously determined. Proponents of Solow’s theory point at the same rate as productivity and labor. The rate of out that saving alone produces economic growth. economic growth, according to the model, can only Jagadeesh (2015), Nwanne (2016), Kim (2017), Patra et be increased by innovations and improvements in al. (2017) for long-term savings, also confirm that sav - technology once a stable rate of growth is achieved ings initiate economic growth. with resource utilization. On the other hand, using regression and Granger’s 116 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA causality test, the authors found that economic continuation of its research, Tang (2009a) conducts growth is a positive function of savings, e.g. Bayar an analysis of the sensitivity of the proven causality of (2014). To model an effective growth policy, it is im- savings and economic growth to the tests used. Five portant to know the direction of the causality of sav- different causality tests were used to test sensitivity: ings and economic growth. Thus, if savings really drive Granger’s, Geweke’s, Meese’s and Dent’s, Hsiao’s, Toda’s growth then a development policy to encourage sav- and Yamamoto’s, and Holmes ’and Hutton’s causal- ings will be applied. If savings do not cause economic ity test. The bidirectional causality of savings and growth, a policy of discouraging savings and strength- economic growth was determined regardless of the ening consumption will be applied. test used. Thus, the method of determining causality, Also, AbuAl-Foul (2010), in econometric analysis in this case, does not affect the results. Tang (2009b) of the long-run relationship of savings and growth showed in his further studies that savings and eco- shows that the link between savings and economic nomic growth cause each other to cause each other in growth exists and is two-way in Morocco, but not in the economies of East Asia as well. Tunisia (the link in this case was one-way, from growth Tang and Tan (2014) conducted an empirical anal- to savings). Shah (2015), applying the unit root test, ysis of the relationship between savings and econom- cointegration, Granger’s causality test and vector er- ic growth, now in Pakistan, using Solow’s neoclassical ror correction model, finds that savings contribute to growth model. It was determined that the variables economic growth in Nepal and India, i.e. growth con- are cointegrated, and that they are bidirectionally tributes to savings in Bhutan and Sri Lanka, while cau- causal. sality does not exist at all in Pakistan. Adam, Musah Guma and Bonga-Bonga (2016) observe for the and Ibrahim (2017) conduct an econometric analysis first time the separate impact of individual savings of the causality of savings and economic growth in 10 components on economic growth in South Africa. sub-Saharan African countries. Granger’s causality test The authors look separately at the impact of savings: and Toda-Yamamoto approach give consistent results households, private companies and government on for only three countries: Benin, Mali, and South Africa, economic growth. According to the results of the where savings causes economic growth. For other econometric analysis, the greatest impact on eco- countries, no consistent results were obtained using nomic growth is achieved by the savings of private the same methodology. Wyk and Kapingura (2021) for-profit companies. also conduct an analysis of the relationship between A certain number of authors have tried to explain savings and economic growth in South Africa and find the problem of endogenous savings and economic that the relationship exists only in the short run, while growth by introducing additional variables into the tests do not show a link between the two variables model. Mohan (2006) also includes the size of income in the long run. The Granger test found that when a among different countries in the analysis and finds relationship is positive it goes in the direction that that income plays a significant role in the relationship economic growth causes savings. between growth and savings, and that growth affects An alternative approach to exploring the link be- savings in a number of countries considered. Olajide tween savings and economic growth in Mexico, for (2009), emphasizes the importance of foreign direct the period 1970-2000, was given by Alguacil et al. investment on the ratio of savings and growth, where (2004). Namely, the authors use Toda and Yamamoto’s, the growth of savings and investment produces eco- and Dolado’s and Lutkepohl’s (non) causality test, nomic growth. Joshi, Pardhan and Bist (2019) analyz- stating that Granger’s test is more sensitive to the in- ing the relationship between savings, investment and cluded variables in the model. The analysis shows that economic growth determine that the investment have savings are causing economic growth in Mexico. a positive impact on economic growth, and savings Tang (2008) in his study uses a relatively new have a negative impact on economic growth in the cointegration test, the “border test” in the process of long run. investigating the existence of a long-term link be- It is especially interesting that the analyzes show tween savings and economic growth. The analysis was different results for the same countries, e.g. for the conducted for Malaysia. The intensity and direction of United States. Dhakal et al. (1992). apply Granger’s the link between savings and economic growth was causality finding that, in the United States, there is a conducted using Toda and Yamamoto, Dolado and two-way link between savings and economic growth, Lutkepohl approaches within the augmented-vector but also that growth exclusively implies savings. Or, autoregressive (VAR) methodology. Using this pro- Cullison (1993), also applying Granger’s test of causal- cedure, the author found that there is a bidirectional ity, shows that at the same time, in the US, there is no link between savings and economic growth. In the causality between savings and economic growth, but SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 117 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA also savings exclusively imply economic growth. enough funds for more demanding and larger finan- In the countries of Southeast Europe, banks play a cial projects, or to save in order to raise funds for such central role in the financial markets, so bank deposits projects. appear as the most common form of savings, instead In Bosnia and Herzegovina, it is evident that the of, for example, investments in the capital market. population does not deposit all funds separated from If the capital market was more developed in these spending on savings in the form of deposits with com- countries, in the sense that the state issues securities mercial banks or similar savings and loan institutions. for infrastructure, holding bonds of this type would A large part of the funds, especially the anchor cur- be more profitable than demand deposits because it rency (euro) and foreign currency in general, is held would bring higher interest yields. Thus, short money, outside the financial and banking sector in Bosnia which includes foreign bank assets and central bank and Herzegovina. Proof of the above is the fact that reserves above required reserves, would be trans- in 2001, and in the first quarter of 2002, the Central formed into investments through the capital market, Bank purchased and transposed German marks (DEM) i.e. short money would actually be a source of savings to Germany in the amount of 4.87 billion DEM. During through the capital market. The condition for the for- this period, about 20 million DEM was bought daily, mer is market liquidity. and that speaks volumes about the foreign currency However, the presence of the dominant form of stocks that the population kept outside commercial savings in the countries of Southeast Europe (deposit banks. savings) does not give significantly different results It can be seen from the bulletin of the Central in terms of the problem of uniformity of savings and Bank of Bosnia and Herzegovina (2000-2016) that at economic growth. Turan and Gjergij (2014) analyz- the end of September 2016, total household deposits ing the relationship between savings and economic amounted to BAM 10.25 billion. This is also the largest growth in Albania found that there is cointegration growth in deposits so far. The share of household de- among the variables involved, but not causality. On posits in total deposits of commercial banks in Bosnia the other hand, Ribaj and Mexhauni (2021), analyz- and Herzegovina is 59.6%. ing the relationship between savings and economic Compared to the end of September of the previ- growth in Kosovo, find that savings in the form of de - ous year, 2015, household deposits increased by 6.6% posits have a significant positive impact on the coun- or BAM 633.5 million. Compared to December 2008, try’s economic growth. when the economic crisis began, deposits are con- Based on the above, it can be concluded that the stantly growing in the range of 8-15% annually. Thus, direction and intensity of the relationship between a cumulative growth of deposits of 97% or BAM 5.05 savings and economic growth is significantly deter - billion was achieved. mined by the applied methodology and by character- The growth rate of deposits changes differently if istics of the considered economy. This fact provides different maturity structures of deposits are observed. additional confirmation of the correctness of the The fastest growth is achieved by short-term depos- observations of individual countries in the analysis of its, more precisely transaction accounts and demand this problem of endogeneity. deposits. Compared to September 2015, transaction accounts increased by 329 million, or 16.3%, while demand deposits increased by BAM 300 million or 13.7%. 4. Research methodology The largest percentage of household deposits, in 4.1. Overview of the state of savings in September 2016, were savings and time deposits of Bosnia and Herzegovina 52.8%. Transaction accounts and demand deposits ac- count for 47.2% of total household deposits. Private savings in Bosnia and Herzegovina recorded a When it comes to the currency structure of house- continuous growth in the period 2000-2016, accord- hold deposits, in September 2016, the largest share ing to the Central Bank of Bosnia and Herzegovina. was occupied by deposits in euros 51.1% or 5.25 bil- The dominant part of private savings is household lion BAM, then in BAM currency 43.6% share or 4.47 savings. The share of household savings in total pri- billion BAM, and deposits in other foreign currencies vate savings ranges from 40-90%. 5.3% or BAM 539 million. Since savings are the most liquid part of assets, Savings and time deposits of private for-profit then households try to invest such funds in a way companies achieved the highest growth from 2005 to that achieves certain returns. There are opinions that the last quarter of 2008 and the beginning of the glob- households opt for savings because they do not have al financial crisis. The cumulative growth of deposits in 118 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA the observed period amounted to 78.72%. The decline Bosnia and Herzegovina’s commercial banks have in savings and time deposits of private for-profit com- more assets than liabilities abroad, net foreign assets panies was recorded by the end of 2010. is positive, which means that this method of matu- Unlike the savings of private companies, house- rity adjustment of funds is extremely expensive and hold savings have been growing continuously since results in a high imbalance of interest rates between the beginning of the global economic crisis, December foreign assets and obligation. The study also points to 2008. In the period from the beginning of the eco- the importance of developing the domestic financial nomic crisis to the third quarter of 2016, household market because it would enable more efficient use savings grew continuously at a rate of 8-15%, which is of available funds and reduce the costs of adjusting a cumulative as much as 97%, as noted earlier. funds because it would avoid the impact of political It can be concluded that private corporate depos- risk on interest rates and increase competitiveness. its were more affected by the global crisis than house - The development of financial markets and a deep - hold deposits, which is to be expected, given that it is er financial structure would also mean more effective a for-profit sector that is much more sensitive to eco - work of the Central Bank and the overall monetary nomic fluctuations than the household sector, where system, where the effect of expansion would expand precautionary savings occur. in times of crisis. from financial markets to the real economy. Greater Savings and time deposits of private for-profit efficiency of the monetary multiplier effect would companies recorded a gradual, slight increase from improve the effectiveness of available resources in the 2010-2016, but with constant oscillations to various financial system. changes. As previously pointed out, the financial market of Bosnia and Herzegovina is dominated by the bank- 4.2. Defining model variables ing sector, with banks that are subsidiary companies of foreign “mother” banks. These institutions have In this research, two models were developed, as men- contributed to increasing market competitiveness tioned earlier. One model is related to the relationship and consequently reducing interest rates. This fact between household savings and economic growth, becomes questionable if we take into account that the and the other model is related to the relationship be- financial market is not sufficiently developed to en- tween private profit enterprises savings and economic sure the competitiveness of various financial institu- growth. Therefore, the following variables are includ- tions. Also, the interest rate is above those in the euro ed: household savings, savings of private for-profit area. The influence of political risk on the interest rate companies and GDP. should not be neglected. Household savings were measured as the sum of The overall situation in the Bosnia and household savings and time deposits and this was Herzegovina’s financial market shows a mismatch be - used as a variable expressed in absolute terms in BAM tween short-term sources and long-term investment and in relative terms, as a percentage of GDP. needs, as is generally the case in underdeveloped and Savings of private companies represent the sum of low-income developing countries. The functioning of savings and time deposits of profitable private com- the financial market is significantly influenced by the panies. In this case, too, savings in absolute amount CBR (currency board regime - this system implies a and as a measure expressed as a percentage of GDP fixed exchange rate of BAM. Each issued BAM has cov - are used. erage in foreign currency or monetary liabilities must Data on household savings and profits of for- never be higher than the foreign exchange assets of profit companies were taken from the database of the Central Bank of Bosnia and Herzegovina). The cur- the Central Bank of Bosnia and Herzegovina (bulletins rency BAM is linked to the currency “anchor” euro, so from 2000-2016). According to the methodology of the purchase of BAM is made at a fixed exchange rate the Central Bank of Bosnia and Herzegovina, a house- of 1.95583 BAM for one euro), which imposes very hold or a population means one or more persons who rigid and strict rules. The mismatch between short- earn income on the market through their work, spend term sources of funds and long-term investments, ac- the earned income, and if it is a community, share a companied by the underdevelopment of the domes- place of residence. Private for-profit/non-financial tic market for the necessary maturity transformation enterprises, according to the same methodology, are and the rigidity of the monetary system, is resolved enterprises in majority private ownership of legal or externally by financing through “parent” banks that natural persons engaged in the production and sale perform the necessary transformation. of goods and/or services for the purpose of gaining Analyzes show (e.g. Kreso and Begović 2012) that profit. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 119 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Time series of gross domestic product at current Therefore, a large variance, prognostic error, occurs and constant prices required additional prepara- in the case of neglecting the importance of seasonal tion in order to be used as variables in defined mod- influence, so it is necessary to conduct the analysis els. Namely, the relevant institutions in Bosnia and on seasonally adjusted, time-adjusted data. The ap- Herzegovina do not have quarterly data on gross plication of seasonal adjustment methods is aimed domestic product for the observed period 2000q1- only at assessing and eliminating the seasonal impact, 2016q4, so additional recalculations were needed. The without analyzing or explaining the cause of seasonal time series of the quarterly gross domestic product at phenomena. current prices was obtained from the Release of the During this research, X-13 ARIMA and TRAMO/ Agency for Statistics of Bosnia and Herzegovina (2013 SEATS methodologies were used for the purpose of and 2016). The estimate of the time series of the quar- seasonal adjustment of time series data. The aim was terly gross domestic product in current prices in the to check whether both methods give equal results. period from 2000q1 to 2005q4 is taken from the paper Both approaches to seasonal adjustment were found by Čolaković and Hlivnjak (2007). to give almost identical results, so it was decided to The time series of the quarterly gross domestic use only the latest version of the first X-13 ARIMA pro - product at constant prices (prices from 2010) was gram in this study. obtained from Release Agency for Statistics of Bosnia Two-unit root tests were used to examine the sta- and Herzegovina (2016, this is a time series of chain- tionarity of the included variables: the Augmented linked values in pr ices in 2010). In the Release of the Dickey-Fuller test (ADF) and the Kwiatkowski-Phillips- Agency for Statistics of Bosnia and Herzegovina 2013, Schmidt-Shin (KPSS) test. Unlike the ADF test, which the values of the quarterly gross domestic product takes the nonstationarity of a process as a null hy- of chain-linked values in 2005 prices are given. These pothesis, the KPSS test claims the stationarity of a values were converted into chain indices which were process with a null hypothesis (Bahovec and Erjavec then used to calculate the value of the quarterly gross 2009). Therefore, the KPSS test is mainly used for the domestic product of chain-linked values, but now at purpose of confirmatory analysis, verification of re - 2010 prices. This reconstructed the series of real GDP sults obtained using the ADF test or some other single in the period from 2006q1 to 2016q3 in 2010 prices. root test. The real GDP series in the period from 2004q1 to When assessing whether one time series can serve 2005q4 was reconstructed taking into account the in predicting another, the Granger test of causality estimates of the GDP series in current prices from the (Granger 1969) is most often used, as can be seen from paper Čolaković and Hlivnjak (2007). The ratio of the a review of the empirical literature. Unlike regression, real GDP series to the GDP series at current prices in which only indicates a mere relationship between 2006 varied from quarter to quarter, taking values of variables, Granger’s causality test tests whether the 1.34, 1.21, 1.25, and 1.02 from the first to the fourth future value of one variable can be predicted based quarter, respectively. Precisely these quarterly values on the past values of another v ariable. From an econo- were used to multiply the values of the GDP ser ies in metric point of view, the Granger test is considered to current prices in the period 2004q1 to 2005q4, thus give “predictive causality” (Granger 1988). The basic obtaining a rough estimate of real GDP in 2010 prices. definition of the test does not include latent confu- As the variable GDP growth rate variable was used sion effects or current nonlinear causal relations. in the analysis instead of the level of that series, this In the following, explanations and the basic formu- rough estimate of the value of real GDP satisfies the las of the Granger test will be given in accordance with needs of the analysis in which the dynamics of the se- the article published by the author Granger in 1988. ries change is observed, and not its level in absolute A variable X can be said to cause a variable Y ac- amount. cording to the Granger test, if by t-test and F-test lagged values of X it can be proved that the values of X give enough statistically significant values about future values of Y. Thus, the variable X causes accord- 4.3. Research methods ing to Granger’s test, the variable Y if the predictions When using economic indicators in the form of time of the Y series are better based on past data and X and series, it is necessary to do their seasonal adjustment Y, than only based on Y. in order to exclude the influence of the season, i.e., to The Granger test is based on two principles. First, exclude short-term fluctuations in the phenomenon. the causality of variables arises before their effects. The total variance of the time series is significantly Here the term cause corresponds to the term pre- determined by the seasonal variation of the series. cede. The assumption is that the past and the present 120 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA cannot be caused by the future. Thus, if event A is The regression equation can retain all individual preceded by event A then A cannot cause A . Also, statistically significant X lags according to t-statis- 2 1 2 t if A occurs before A it does not necessarily mean tics. The condition for this setting is that in the total 1 2 that A causes A . Another assumption is that causal- amount, all lags, according to F-statistics, contribute to 1 2 P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x ity contains unique information about future effects. the increase of the regression power. When determining the causal effects of X on Y , the A multivariate Granger causality test (Granger t t Y  a  a Y  a Y  ... a Y  . following hypothesis is tested taking into account the 1988) is used to define vector autoregressive models. t 0 1 t1 2 t2 m tm t dxi given assumptions: Here it is set that X(t) ∈ R for t=1,...,T is a d-dimen- sional multivariate series. The VAR model uses the fol- Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x lowing form of the Granger causality test: Y  a  a Y  a Y  ... a Y  . In this expression, the notations are used: P for X (t )  A X (t  ) . t 0 1 t1 2 t2 m tm t   t 1 probability, A for an arbitrary non-empty set, and with I (t) and I (t) available information at time t. It is con- In the preceding expression, X is called the -x i Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t cluded that X causes Y according to the Granger test if Granger causality of some other variable X provided this hypothesis can be confirmed. that at least one element, for, in the absolute value is T Xhe gener (t )  al rule of the Gr A X (t  ) . anger test (Granger significantly greater than zero.   t 1 1988) is that in the case of a time series analysis that Linear testing in the Granger test is suitable for is stationary, the absolute values of the variables are determining causality in the mean values of v ariables. used, while for nonstationary series, the correspond- In this way, Granger’s causality for “larger moments” ing differences for which the variables are stationary such as variance cannot be determined. Therefore, a are used. Similar to previous tests, Granger’s meth- nonparametric Granger causality test was developed. odology uses Akaike or Schwarz information criteria The definition of causality in this test is general so that to select the lags length. The regression equation re- it does not include some modeling assumptions, as tains the individual lagged value of a variable if two is the case with linear autoregressive models. Better conditions are met: the value is statistically significant parametric models, including those with more mo- according to the t-test and if the observed value and mentum and/or nonlinearity, can be set up based on other lagged values of the variable together increase Granger’s nonparametric causality test. the exploitative power of the model according to Granger’s causality test, however, has certain F-statistics. If the lagged values of the explanatory limitations and shortcomings during its application variable are retained in the regression, then the null (Maziarz 2015). The first drawback is that Granger’s hypothesis of the absence of Granger causality cannot causality does not necessarily mean “true causality.” be rejected. The only thing that Granger’s test satisfies in terms of The results of the Granger test may show that nei- causality is Humaean’s definition of causality. Another ther variable causes the other, or, if two variables are drawback is that the alternative Granger hypothesis observed, that the variables cause each other. cannot be rejected if both variables, X and Y , are de- t t The simplest form of the Granger test (Granger rived from a common third process with different lags. 1988) is for models that include two variables, Y and Also, it is possible that there are no changes in the X . The starting point is the assumption that these are other variable if only one variable is manipulated. The stationary series. In order to test the null hypothesis, it Granger test is generally developed for testing pairs is necessary to first determine the appropriate num- of variables, so deficiencies may occur if the “right” ber of Y lags that will be included in the one-variant relation includes three or more variables. Incorrect P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x autoregression Y : test results can also be caused by: a nonlinear causal relationship between variables; non-stationarity and Y  a  a Y  a Y  ... a Y  . nonlinearity of the observed series; insufficient or t 0 1 t1 2 t2 m tm t excessive sampling; and the existence of rational ex- In the next step, the autoregressive equation is ex- pectations. By vector autoregression, a similar test can Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t P (Y (t+1) ∈ A | I (t)) � P (Y (t+1) ∈ A | I (t)). -x tended with lags of the variable X : be applied to a number of variables included in the model. Y  a  a Y  a Y  ... a Y  . t 0 1 t1 2 t2 m tm t X (t )  A X (t  ) . Granger’s causality test continued to evolve de-   t 1 spite the aforementioned limitations. Thus, a form of where p and q denote the shortest and longest dis- the test was developed that is not sensitive to violat- Y  a a Y a Y ...a Y b X b X  , t 0 1 t1 2 t2 m tm p tp q tq t placements, respectively, for which the lagged values ing the assumption about the normality of the model of the variable X are significant. error distribution. Since most financial variables are X (t )  A X (t  ) .   t 1 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 121 not normally distributed then the extended version the hypothesis that the coefficients at the first p lags of the Granger test finds special application in ana- of the first variable are equal to zero. Then the same is lyzes of causality in finance. Asymmetric causality is done with the coefficients on the first p lags of the sec - especially emphasized in recent works in the field of ond variable. Rejecting the null hypothesis that these causality, with the aim of separating the impact of coefficients are equal to zero, is equivalent to rejecting positive and negative changes on causality. the hypothesis of the absence of causality in Granger’s sense between the two observed variables. In other words, the rejection of the hypothesis indicates the existence of Granger causality among the variables. 5. Research results The results of Granger’s test of causality among Two software programs were used in the study in- the variables of savings and economic growth are cluding: EViews 9.5 and R with a number of packages given in Table 1. Granger’s test was applied to both created for time series analysis. the savings of households and private for-profit com- In order to verify the existence of causality be- panies, as components of private savings. tween household savings and economic growth, as The results of Granger’s test of causality indicated investigated earlier, Granger’s causality test was ap- that there is no causal link between household sav- plied, namely the Toda-Yamamoto procedure. ings and economic growth, thus rejecting hypothesis The Toda-Yamamoto procedure first required to we tested, which claims that household savings in determine the order of integration of the two vari- Bosnia and Herzegovina initiate economic growth. ables between which it was sought to check whether The same results were obtained by: Cullison (1993) there was a causal relationship in Granger’s sense. This for the USA; Sinha (1996) for India; Saltz (1999) for: was done using ADF and KPSS unit root tests (results Colombia, Jamaica, Peru and the Philippines; Agrawal of unit root tests, cointegration and VAR models are (2000) for Nepal; Agrawal (2001) for Thailand and attached to the paper). Korea; Konya (2005) for a large number of countries, When the results of unit root tests are obtained, Shah (2015) for Pakistan. it is necessary to determine the maximum order of The results only confirm the non-existence of a integration m of the observed two variables. Thus, it strict rule on the presence and direction of the link be- was obtained that household savings are I (1) process tween savings and economic growth, even when ap- while GDP growth rate I (0) is a process. This means plying the same methodology in individual countries. that the maximum order of integration is m = 1. Savings and economic growth show cointegra- The next step defines the Vector autoregres- tion, which means that the variables have a common sion (VAR) model based on the level of the variables stochastic trend. However, the underdevelopment regardless of the results of the unit root tests. Here, of the financial and overall economic market, along household savings and GDP growth rates enter di- with the rigidity of the political system, prevents the rectly into the VAR model. The maximum lag length necessary flow of sufficient capital for faster economic for the variables in the vector autoregressive model growth, which is an imperative in the Bosnia and is then determined. The obtained optimal lag length Herzegovina’s economy. As in the case of other poor is denoted by p. It is checked whether the VAR model economies, the Bosnia and Herzegovina’s economy is correctly specified by determining whether there is finances its economic growth with expensive foreign autocorrelation of VAR model residuals. capital. If the variables are of the same order of integra- A possible cause of such results is perhaps the more tion, then it can be checked whether they are coin- significant effect of demographic and cultural fac - tegrated using Johansen’s methodology, in order to tors on economic growth in Bosnia and Herzegovina. further verify the results of Granger’s causality test. Numerous authors emphasize the importance of in- Namely, if two or more variables are cointegrated, cluding these variables in the research: Prskawetz et then there must be a Granger causality between them al. (2007), Crenshaaw and Robinson (2010), Basu et al. in either one or both directions. However, the oppo- (2013), Diep (2015), Yanagizawa-Drott and Compante site does not apply, i.e., if there is a Granger causality (2015), Gorodnichenko and Roland (2016). among the variables it does not mean that the series The inclusion of demographic variables in the must also be cointegrated. model in this study was not possible due to the short- Once the appropriate VAR model has been deter- ness of time series data. mined, m lags of each variable are added to each of Turan and Gjergij (2014) analyzing the relation- the equations of the VAR model. Granger causality ship between savings and economic growth in testing is then performed by applying the Wald test to Albania, an economy quite similar to that in Bosnia 122 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Table 1. Granger causality test (Toda-Yamamoto procedure) Null hypothesis Hi-square p-value Granger causality Households Savings do not cause GDP growth rate 0.872 0.351 Absence of G causality Savings do not cause GDP level 1.827 0.177 Absence of G causality Savings in % of GDP do not cause GDP growth rate 3.487 0.175 Absence of G causality Savings in % of GDP do not cause the level of GDP 1.036 0.309 Absence of G causality GDP growth rate does not cause savings 0.256 0.613 Absence of G causality GDP level does not cause savings 0.613 0.434 Absence of G causality GDP growth rate does not cause savings in % of GDP 0.483 0.786 Absence of G causality GDP level does not cause savings in % of GDP 0.002 0.964 Absence of G causality Private for-profit companies Savings do not cause GDP growth rate 2.760 0.097 Absence of G causality Savings do not cause GDP level 5.946 0.114 Absence of G causality Savings in % of GDP do not cause GDP growth rate 0.849 0.357 Absence of G causality Savings in % of GDP do not cause GDP level 4.525 0.033 Granger causality GDP growth rate does not cause savings 0.850 0.357 Absence of G causality GDP level does not cause savings 4.152 0.246 Absence of G causality GDP growth rate does not cause savings in % of GDP 2.761 0.097 Absence of G causality GDP level does not cause savings in % of GDP 0.102 0.750 Absence of G causality Source: Author and Herzegovina, also found that there is cointegra- 6. Conclusion tion among the variables involved, and stressed that the government should pay special attention to this The importance of the relationship between sav- important issue for faster economic growth. ings and economic growth, in addition to basic eco- In this research the null hypothesis of the absence nomic theories, has resulted in numerous empirical of causality was rejected only in the case of the vari- papers that have tried to at least determine the direc- ables of the share of private for-profit companies sav - tion of this relationship. However, no uniform results ings in the gross domestic product and the level of have been obtained. the gross domestic product. But strictly speaking, this This research also examined the direction and result does not change the general conclusion made intensity of the relationship between savings and above about the absence of a causal link between sav- economic growth in the underdeveloped economy ings and economic growth, because economic growth of Bosnia and Herzegovina. To verify the existence is measured by GDP growth rates, not its level. of causality between household savings and eco- If the level of significance of the test is reduced to nomic growth, the Granger causality test was applied, 10% then the hypothesis of the absence of Granger namely the Toda-Yamamoto procedure. The results of causality can also be rejected in two cases when the Granger’s test of causality indicated that there is no following variables are observed: private for-profit causal link between household savings and the pri- companies savings and GDP growth rate and GDP vate for-profit companies and economic growth. growth rate and private for-profit companies savings According to the results of this research, savings in% of GDP. But given the significance level of 10% this and economic growth in Bosnia and Herzegovina can be considered as a very weak confirmation of the show cointegration, which means that the variables existence of Granger causality among the observed have the same stochastic trend. However, the under- variables. development of the financial and overall economic market, along with the rigidity of the political system, SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 123 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Literature prevents the necessary flow of sufficient capital for faster economic growth, which is an imperative in the Bosnia and Herzegovina’s economy. As in the case of AbuAl-Foul, B. 2010. The causal relation between savings other poor economies, the Bosnia and Herzegovina’s and economic growth: Some evidence from MENA coun- economy finances its economic growth with expen- tries. Topic in Middle Eastern and African Economies 12: sive foreign capital. 1-12. A possible cause of such results is perhaps a more Adam, I. O., Musah, A., Ibrahim, M. 2017. Putting the cart be- significant effect of demographic and cultural factors fore the horse? Re-examining the relatonship between on economic growth in Bosnia and Herzegovina. domestic savings and economic growth in selected Sub- The contribution of the conducted research is Saharan African countries. Journal of African Business, reflected in the examination of the direction and ma- 18 (1): 102-123. turity intensity of causality between savings and eco- Agrawal, P. 2000. Savings, investment and growth in South nomic growth, in small, open economy of Bosnia and Asia. Mumbai: Indira Gandhi Institute of Development Herzegovina. It was found that there is cointegration Research. among the variables, but not causality. Agrawal, P. 2001. The relation between savings and growth: The contribution is reflected in the fact that by Cointegration and causality evidence from Asia. Applied examining the direction and intensity of the causal- Economics, 33(4): 499-513. ity of savings and economic growth, it contributed Agency for Statistics of Bosnia and Herzegovina. 2016. to the identification of potential sources of financing Announcement, Gross Domestic Product-Quarterly economic growth through the optimal distribution of Data, No. 4. accumulated capital. The results show that in Bosnia Agency for Statistics of Bosnia and Herzegovina. 2013. and Herzegovina there are enough funds in the form Announcement, Gross Domestic Product-Quarterly of household savings to finance economic growth, Data, No. 1. the variables show cointegration, but inadequate Alguacil, M., Cuadros, A., Orts, V. 2004. Does saving re- and rigid distribution prevents causality among these ally matter for growth? Mexico (1970–2000).  Journal of variables. International Development. 16(2): 281-290. The biggest limitation in writing this paper was Bahovec, V. and Erjavec, N. 2009). Uvod u ekonometrijsku the availability of data. So, time series of gross domes- analizu, Element. tic product at current and constant prices required ad- Basu, T., Barik, D., Arokiasamy, P. 2013. Demographic ditional preparation in order to be used as variables in Determinants of Economic Growth in BRICS and se- defined models. lected Developed Countries, Presented in XXVII IUSSP It would be useful to conduct such research after International Population Conference (IUSSP-2013). a certain period of time, in order to verify the obtained August 2013 at BEXCO. Busan, Republic of Korea. results with longer observed time spans. Also, as Bayar, Y. 2014. Financial development and domestic savings noted earlier, it would be important to include de- in emerging Asian countries,  Theoretical and applied mographic and cultural determinants in the form of economics. 21(7 – 596): 55-66. mediation or moderation variables in models of the Central Bank of Bosnia and Herzegovina, Quarterly Bulletins: relationship between savings and economic growth, from Bulletin 2 in 2000 to Bulletin 3 in 2016. although it is a huge challenge with quartery data. In Crenshaw, E., Robison, K. 2010. Socio-demographic the continuation of the research, it would be useful to Determinants of Economic Growth: Age Structure, conduct such analyzes for the Western Balkans region Preindustrial Heritage and Sociolingustic Integration. and make a comparative analysis. Social Forces. 88 (5): 2217-2240. Čokalović, B., Hlivnjak, S. 2007. Estimate of quarterly GDP for the period 2000-2006. ISSN 1840-2593. Central Bank of Bosnia and Herzegovina. Department of Economic Research, Statistics and Publications. Cullison,W. E. 1993. Saving measures as economic growth indicators. Contemporary Policy Issues. 11(1): 1-8. Dhakal, D., Grabowski, R., Shields, M. P. 1992. The role of sav- ings in economic development: The U.S. and Japanese experiences. Japan and the World Economy. 3(4): 331-340. Diep, V. T. T. 2015. Demographic Factors and Economic Growth: The Bi-directional causality in South East Asia. 124 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA The ninth Vietnam Economists Annual Meeting. Da Olajide S. O. 2009. Does Saving Really Matter for Growth Nang City. August 2015. in Developing Countries? The Case of a Small Open Domar, E. D. 1946. Capital expansion, rate of growth, and Economy. http:// economics.ca /2009/ papers/ 0619 .pdf. employment. Econometrica, Journal of the Econometric (accessed November 10, 2016) Society: 137-147. Patra, S. K., Murthy, D. S., Kuruva, M. B., Mohanty, A. Granger, C. W. J. 1969. Investigating Causal Relations by 2017. Revisiting the causal nexus between savings Econometric Models and Cross-Spectral Methods. and economic growth in India: An empirical analy- Econometrica. 37: 424 - 438. sis. Economia. 18 (3): 380-391. Granger, C. W. 1988. Causality, cointegration, and control. Prskawetz, A., Fent, T., Barthel, W. 2007. The Relationship Journal of Economic Dynamics and Control.  12(2-3): Between Demographic Change and economic Growth in 551-559. EU. Institut fur Demographie. Osterreichische Akademie Gorodnichenko, Y., Roland, G. 2016. Culture, institutions and der Wissenschaften, Forschungsbericht. No. 32. https:// the wealth of nations. Review of Economics and Statistic. www.oeaw.ac.at/fileadmin/subsites/Institute/VID/PDF/ 99 (3): 402-416. Publications/Forschungsberichte/FB32.pdf. (accessed Guma, N., Bonga-Bonga, L. (2016). The relationship between November 15, 2016). savings and economic growth at the disaggregated lev- Ribaj, A., Mexhuani, F. 2021. The impact of savings on eco- el. Economia Internazionale/ International Economics, nomic growth in a developing country (the case of Camera di Commercio Industria Artigianato Agricoltura Kosovo). Journal of Innovation and Entrepreneurship. 10 di Genova. 70(1): 1-24. (1): 3-13.  Harrod, R. F. 1939. An essay in dynamic theory. The Economic Saltz, I. S. 1999. An examination of the causal relationship Journal. 49(193): 14-33. between savings and growth in the Third World. Journal Harrod, R. F. 1948. Towards a dynamic economics, London, of Economics and Finance. 23(1): 90-98. Macmillan. Shah, R. K., (2015), Saving and Economic Growth in South Jagadeesh, D. (2015). The impact of savings in eco- Asia: Evidence from Time Series Analysis.  Economic nomic growth: an empirical study based on Journal of Nepal. https://www.econbiz.de/Record/sav- Botswana. International Journal. 2: 10 - 21. ing-and-economic. (accessed November 10, 2016) Joshi, A., Pardhan, S., Bist, J. P. 2019. Savings, investment Sinha, D. 1996. Saving and economic growth in India. and growzh in Nepal: an empirical analysis. Financial Economia Internazionale. 49 (4): 637-647. Innovation. 5 (1): 1-13. Solow, R. M. 1956). A contribution to the theory of economic Keynes, J. M. 1936. The general theory of interest, employ- growth. The quarterly journal of economics. 70 (1): 65-94. ment and money. Macmillan. Tang, C. F. 2008. An empirical modelling on savings be- Kim, N. T. and Nguyen, H. H. 2017. Impacts of domestic sav- haviour in Malaysia. Labuan Bulletin of International ings on economic growth of Vietnam.  Asian Journal of Business and Finance. 6: 57-76. Economic Modelling. 5(3): 245-252. Tang, C. F. 2009a. Does causality technique matter to sav- Konya, L. 2005. Saving and growth: Granger causality analy- ings-growth nexus in Malaysia?. Malaysian Management sis with bootstrapping on panels of countries. Journal of Journal. 13(1-2): 1-10. Economic Research. 10: 231-260. Tang, C. F. 2009b. The savings-growth nexus for Asian econ- Kreso, S. and Begović, S. 2012. Maturity (mis) matching in omies: New evidence from the timevarying economet- currency board (like) regimes. Interdisciplinary manage- rics methods. Paper presented at the Symposium of ment research. 8: 423 – 428. USM Fellowship Holders 2009. Universiti Sains Malaysia. Mankiw, N.G, Romer D., Weil, D. N. 1992. A Contribution to Penang. Malaysia. 14-15 November. the Empirics of Economic Growth. The Quarterly Journal Tang, C. F. and Tan, B. W. 2014. A revalidation of the savings- of Economics. Oxford University Press. 107(2): 407 - 437. growth nexus in Pakistan. Economic Modelling. 36: Maziarz, M. 2015. A review of the Granger-causality falla- 370-377. cy. The journal of philosophical economics: Reflections Turan, G. and Gjergji, O. 2014. What is the Impact of Savings on economic and social issues 8(2): 86-105. on Growth? The Case of Small Open Economy (Albania). Mohan, R. 2006. Causal Relationship Between Savings and Mediterranean Journal of Social Sciences. 5 (13): 360-368. Economic Growth in Countries with Different Income Yanagizawa-Drott, D. and Compante, F. 2015. Does religion Levels. Economics Bulletin. 5: 1 - 12. Aec ff t Economic Growth and Happiness? Evidence from Nwanne, T. F. I. 2016. Implication of savings and investment Ramadan. The Quarterly Journal of Economics. 130 (2): on economic growth in Nigeria. International Journal 615-658. of Small Business and Entrepreneurship Research.  2(4): Wyk, B. F. and Kapingura, F., M. 2021. Unerstanding the nexus 74-86. between savings and economic growth: A South African context. Development Southern Africa. 38 (5): 828-844. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 125 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA APPENDIX 1: Granger causality test Household savings vs. GDP growth rate Step 1: Unit root tests Null Hypothesis: STEDOM_D11 has a unit root Exogenous: Constant Lag Length: 0 (Automatic - based on SIC, maxlag=10) t-Statistic Prob.* Augmented Dickey-Fuller test statistic  0.745  0.992 Test critical values: 1% level -3.544 5% level -2.911 10% level -2.593 *MacKinnon (1996) one-sided p-values. Augmented Dickey-Fuller Test Equation Dependent Variable: D(STEDOM_D11) Method: Least Squares Date: 05/11/17 Time: 02:18 Sample (adjusted): 2001Q1 2015Q4 Included observations: 60 after adjustments Variable Coefficient Std. Error t-Statistic Prob. STEDOM_D11(-1) 0.009 0.012 0.745 0.459 C 79.751 45.196 1.765 0.083 R-squared 0.009 Mean dependent var 108.185 Adjusted R-squared -0.008 S.D. dependent var 186.709 S.E. of regression 187.417 Akaike info criterion 13.337 Sum squared resid 2037261. Schwarz criterion 13.407 Log likelihood -398.120 Hannan-Quinn criter. 13.365 F-statistic 0.555 Durbin-Watson stat 2.624 Prob(F-statistic) 0.459 126 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Null Hypothesis: STEDOM_D11 is stationary Exogenous: Constant Bandwidth: 6 (Newey-West automatic) using Bartlett kernel LM-Stat. Kwiatkowski-Phillips-Schmidt-Shin test statistic  0.960 Asymptotic critical values*: 1% level  0.739 5% level  0.463 10% level  0.347 *Kwiatkowski-Phillips-Schmidt-Shin (1992, Table 1)  Residual variance (no correction)  4271797.  26754360 HAC corrected variance (Bartlett kernel) 0 KPSS Test Equation Dependent Variable: STEDOM_D11 Method: Least Squares Date: 05/11/17 Time: 02:20 Sample (adjusted): 2000Q4 2015Q4 Included observations: 61 after adjustments Variable Coefficient Std. Error t-Statistic Prob. C 3259.592 266.827 12.216 0.000 R-squared 0.000000 Mean dependent var 3259.592 Adjusted R-squared 0.000000 S.D. dependent var 2083.985 S.E. of regression 2083.985 Akaike info criterion 18.138 Sum squared resid 2.61E+08 Schwarz criterion 18.173 Log likelihood -552.215 Hannan-Quinn criter. 18.152 Durbin-Watson stat 0.011 (Other appendices of this step are available from the author on any request.) Conclusion: Household savings are I (1), GDP growth rate is I (0). SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 127 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 2: Order the VAR model VAR Lag Order Selection Criteria Endogenous variables: STEDOM_D11 RBDP_D11  Exogenous variables: C  Date: 05/11/17 Time: 02:31 Sample: 2000Q1 2016Q4 Included observations: 39  Lag LogL LR FPE AIC SC HQ 0 -237.791 NA   750.883  12.297  12.382  12.328 1 -140.014 180.511* 6.127* 7.488* 7.744* 7.580* 2 -138.860  2.012  7.106  7.634  8.060  7.787 3 -137.614  2.046  8.225  7.775  8.372  7.989 4 -137.154  0.707  9.952  7.957  8.724  8.232 5 -132.120  7.229  9.576  7.904  8.842  8.240 6 -131.834  0.381  11.837  8.094  9.203  8.492 7 -130.342  1.836  13.878  8.223  9.502  8.682 8 -127.907  2.747  15.677  8.303  9.753  8.823  * indicates lag order selected by the criterion  LR: sequential modified LR test statistic (each test at 5% level)  FPE: Final prediction error  AIC: Akaike information criterion  SC: Schwarz information criterion  HQ: Hannan-Quinn information criterion Conclusion: A lag of one quarter 128 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 3: Check the residual VAR (1) of the model VAR Residual Portmanteau Tests for Autocorrelations Null Hypothesis: no residual autocorrelations up to lag Date: 05/11/17 Time: 02:34 Sample: 2000Q1 2016Q4 Included observations: 46 Lags Q-Stat Prob. Adj Q-Stat Prob. df 1  1.511 NA*  1.545 NA* NA* 2  2.057  0.7252  2.116  0.715 4 3  3.739  0.8799  3.915  0.864 8 4  4.815  0.9639  5.093  0.955 12 *The test is valid only for lags larger than the VAR lag order. df is degrees of freedom for (approximate) chi-square distribution VAR Residual Serial Correlation LM Tests Null Hypothesis: no serial correlation at lag order h Date: 05/11/17 Time: 02:35 Sample: 2000Q1 2016Q4 Included observations: 46 Lags LM-Stat Prob 1  2.086  0.720 2  0.490  0.975 3  1.521  0.823 4  0.977  0.913 Probs from chi-square with 4 df. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 129 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 4: Cointegration Date: 05/11/17 Time: 02:37 Sample (adjusted): 2004Q4 2015Q4 Included observations: 45 after adjustments Trend assumption: Linear deterministic trend Series: STEDOM_D11 RBDP_D11  Lags interval (in first differences): 1 to 1 Unrestricted Cointegration Rank Test (Trace) Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None *  0.373  21.045  15.495  0.007 At most 1  3.09E-05  0.001  3.841  0.969  Trace test indicates 1 cointegratingeqn(s) at the 0.05 level  * denotes rejection of the hypothesis at the 0.05 level  **MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue) Hypothesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None *  0.374  21.043  14.265  0.004 At most 1  3.09E-05  0.001  3.841  0.969  Max-eigenvalue test indicates 1 cointegratingeqn(s) at the 0.05 level  * denotes rejection of the hypothesis at the 0.05 level  **MacKinnon-Haug-Michelis (1999) p-values  Unrestricted Cointegrating Coefficients (normalized by b’*S11*b=I):  STEDOM_D11 RBDP_D11  0.000  66.958  0.001 -8.993  Unrestricted Adjustment Coefficients (alpha):  D(STEDOM_D11)  12.200  0.655 D(RBDP_D11) -0.014  1.20E-05 1 Cointegrating Equation(s):  Log likelihood -163.913 Normalized cointegrating coefficients (standard error in parentheses) STEDOM_D11 RBDP_D11  1.000  288124.2  (56761.5) Adjustment coefficients (standard error in parentheses) D(STEDOM_D11)  0.003  (0.004) D(RBDP_D11) -3.35E-06  (6.8E-07) Conclusion: There is cointegration. 130 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 THE CAUSAL LINK BETWEEN SAVINGS AND ECONOMIC GROWTH IN BOSNIA AND HERZEGOVINA Step 5: Add one lag to VAR (1) and rate VAR (2) VAR Granger Causality/Block Exogeneity Wald Tests Date: 05/11/17 Time: 02:43 Sample: 2000Q1 2016Q4 Included observations: 45 Dependent variable: STEDOM_D11 Excluded Chi-sq df Prob. RBDP_D11  0.256 1  0.613 All  0.256 1  0.613 Dependent variable: RBDP_D11 Excluded Chi-sq df Prob. STEDOM_D11  0.872 1  0.351 All  0.872 1  0.351 Other appendices of this step are available from the author on any request. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 131

Journal

South East European Journal of Economics and Businessde Gruyter

Published: Dec 1, 2021

Keywords: Saving; economic growth; macroeconomics; econometric analysis; C32; E21

There are no references for this article.