The model for the determinants of saving

The coefficient on the current account balance is positive and is statistically significant. The results show that the foreign saving displaces domestic saving since the coefficients on foreign saving have a negative sign for both developed and developing countries. He finds that political instability, economic growth rate, and current account balance are statistically significant determinants of government saving. Demographic variables, social security expenditure and the development of financial sector are not statistically significant variables for government saving.

E denotes the individual’s expectation and U is the utility function of the individual. The

utility function is assumed to be concave. p is the rate of time preference and r is the interest rate. Private and public goods consumptions in period t are denoted by c and g, respectively. For simplification, supplies of c and g are assumed to be exogenous. Total wealth and the tax rate in the period are denoted by W andτ_t, respectively. y_t(1−τ)is the net income or the disposable income of the individual. stands for saving. Saving is defined as net income minus private goods consumption based on the rate of time preference. If saving is negative in a given period, the individual has to borrow. Thus, the third and final restriction shows that in any given period, saving ought to be less than the net income of the individual and borrowing cannot exceed k.

St

As income tends to fluctuate over the course of an individual’s lifetime, he or she will use saving and borrowing to smooth consumption through time. Individuals have smooth consumption over their lifetimes, and as a result, they are net savers during years which thy have high income to secure funds for hard times and dis-savers during years which they have low income.

Despite its simplicity, the model can be used to study several empirical and theoretical issues of saving behavior that have not been resolved conclusively. The main unresolved issues of saving behavior that can be explained by using the model are as follows.

4.1. Possible impact of fiscal policy

Empirical results on private saving crowding out government saving show that the Ricardian equivalence hypothesis is ambiguous. The model presented above implies that changes in either g orτ can have different effects on the household saving behavior. The

model suggests that both present and future tax rates have effects on household saving.

Also, the model is able to capture important implication of social security programs. The life-cycle hypothesis implies that the presence of government-run social security systems leads to a decrease in individuals’ saving because social security programs reduce the uncertainty of individuals in their retirement.

4.2. Does income growth raise saving?

Modigliani (1966) argues that a higher economic growth rate would, with unchanged saving rates by age groups, increase aggregate saving because it would increase the aggregate income of workers. In other words, when the economy is growing, workers’

saving will increase more than that of retired and old people. Hence, aggregate saving will increase. However, there are arguments in the opposite direction as well. Tobin (1967) argues that if workers are certain that their income will grow in the future, workers will consume more today and save less. Therefore, aggregate saving will decrease.

4.3. The role of inflation in determining saving

The effects of inflation on saving behavior have been studied extensively. However, results

in the earlier studies are mixed. In the above model, the impact of inflation on saving is through its role in determining the real interest rate. In other words, inflation could have an effect on saving through its impact on real wealth. In the saving literature, there are two strands of arguments. First, higher inflation tends to lead to a higher nominal interest rate and hence, leads to higher disposable income and household saving. Also, Athukorala and Sen (2003) argue that saving rises with inflation. They argue that if households seek to maintain a target level of wealth or liquid assets, inflation will lead to an increase in aggregate saving. Households increase their saving because if inflation increases, nominal value of liquid assets or wealth will decrease. This leads to an increase in saving because the households try to maintain a target level of real wealth. Second, higher inflation decreases saving by increasing the uncertainty about future price levels and thus, consumers tend to consume more.

4.4. Demographic variables as determinants of saving

The life-cycle hypothesis highlights the importance of age structure of the population. The impact of demographic variable for the above model is mainly based on aggregation across households in a country. The life-cycle hypothesis argues that economic growth makes the middle-age workers richer than young and retired people who do not have a steady income source. Furthermore, the life-cycle hypothesis argues that when the economy is flourishing, middle-aged workers save more than young workers and old people dis-save. Therefore,

economic growth with unchanged saving rate of workers leads to an increase in the aggregate saving. Empirical results on the relationship between dependency ratio and saving rate are often ambiguous. Having more children may increase the household’s desire to leave a larger bequest for the children. Therefore, a higher dependency ratio may lead to a higher saving rate. Conversely, having more children may increase the living expenses of the household such as expenditure on education and health, and may decrease the household’s potential to save. In that case, the higher dependency ratio would decrease aggregate saving.

4.5. Does a higher real interest rate lead to higher saving?

The relationship between the interest rate and the saving rate depends on substitution and income effects. A higher interest rate increases the present price of consumption relative to the future price of consumption i.e. the substitution effect, and therefore, leads to an increase in saving. If a person is a net lender, his or her lifetime income will increase and thus he or she will consume more than what he or she will save i.e. the income effect. At the aggregate level, a higher interest rate leads to an increase in saving only if the substitution effect is stronger than the income effect. If the income effect is stronger than the substitution effect, saving will decrease.

4.6. Borrowing constraint

The borrowing constraint is a very crucial variable in developing countries. The model

assumes that the individual can borrow without any difficulty which is unlikely in many developing countries or for low income workers. Also, the relationship between the borrowing constraint and saving is affected by the depth of the financial market. Though one can assume that the tendency to dis-save depends on the ability to borrow, empirical studies on the relationship between the borrowing constraint and saving show mixed results.

Though the above model is appropriate for studying important possible determinants of saving, the model has some limitations. The model is not able to capture variables such as households’ heterogeneity and foreign saving. Hence, in addition to the above variables, few other explanatory variables are used for the empirical analysis.

4.7. The effects of the terms of trade on saving

The possible relationship between the terms of trade and saving is an important issue. The terms of trade can be defined as the ratio of the export price index to the import price index.

According to the Harberger-Laursen-Metzler hypothesis, a deterioration in the terms of trade decreases domestic saving and an improvement in terms of trade leads to an increase in saving. A deterioration in the terms of trade means a fall in prices of domestically produced goods relative to that of foreign goods. A fall in the prices of domestically produced goods leads to a decrease in real income of households and thus, saving.

According to Masson, Bayoumi, and Samiei (1998), changes in the terms of trade can

either increase or decrease saving depending on whether the changes are transitory or permanent. In view of the above, a separate variable for the terms of trade is included in our model.

4.8. Foreign saving

In an open economy, households can use foreign borrowing to smooth consumption. This implies that foreign saving affects domestic saving. Masson, Bayoumi, and Samiei (1998) point out that foreign saving is a potential exogenous variable for determining domestic saving in developing countries. Baharumshah et al. (2003) indicate that there are two broad views on how foreign saving affects domestic saving. One view is that foreign saving adds to the overall availability of funds without substituting for domestic saving. The second view is that foreign saving substitutes domestic saving and thus, reduces domestic saving.

Whether foreign saving reduces domestic saving or not is ultimately an empirical issue.

4.9. Income distribution

The relationship between income inequality and saving has been receiving attention in both theoretical and empirical studies. According to Schmidt-Hebbel et al. (2000), there are two broad views on how income inequality may affect domestic saving. First, the recent political-economy literature postulates a negative relationship between the two variables.

Second, most of the empirical literature on the relationship between income disparity and saving based on cross-section data finds a positive relationship between the two variables.

4.10. Level of per capita GDP

Athukorala and Sen (2003) argue that absence of a link between current income and current saving is a limitation of the life-cycle hypothesis. One can not ignore the link especially in developing countries such as India. Households in developing countries are not always forward looking and do not take their saving decision based on lifetime income or wealth. These households may take their saving decisions based on current income rather than on future income because it is impossible for them to set aside resources now in order to provide funds for consumption later. Majority of the people living in South Asia are low income earners. According to Leff (1969), the literature on determinants of saving shows that the level of per capita affects saving rate in developing countries. Hence, the level of per capita income is included in our empirical model.

4.11. The possible explanatory variables

The above discussion and the literature review suggest a number of factors that may influence saving rate. Hence, the saving function for the determinants of saving is specified as follows.

t t t

t t

t t

t t

t t

O t

BC LPY

IDP FSA

TOT RIR

DPR INF

GDP SPB

SR

ε β

β β

β

β β

β β

β β

β

+ +

+ +

+ +

+ +

+ +

+

=

10 9

8 7

6 5

4 3

2

1 （3.12）

SRis the saving rate of the country in period t. The saving rate is defined as the ratio of gross domestic saving to GDP. SPBis the government saving rate. The government saving

rate is defined as the ratio of government budget deficit or surplus to GDP. The coefficient is expected to be negative because we assume that government saving will crowd out private saving. The saving literature shows that if the coefficient is significantly different from -1, increases in government saving, will crowed out private saving. GDP is the gross domestic product. As suggested by the life-cycle model, the coefficient of GDP growth rate is assumed to be positive. is the inflation rate. Following Edwards (1995), this variable is used to capture the macroeconomic stability as well. Its coefficient is expected to be negative. is the dependency ratio. The dependency ratio is defined as the ratio of population younger than 15 years old plus population over 65 years as a percentage of working age population. According to the life-cycle model, the coefficient should be negative.

INF

DPR

RIRis the real interest rate. The sign of the coefficient is expected to be positive.

TOT is the terms of trade. Terms of trade is defined as the index of prices of export goods

divided by the index of prices of import goods in domestic currency. Its coefficient is expected to be positive. is the foreign saving. Foreign direct investment is a broad measure of foreign saving. If foreign saving displaces domestic private saving, the coefficient would be positive.

FSA

IDP measures income inequality. Income share held by the

highest ten percent of the total population is the proxy for income inequality and its coefficient is expected to be positive.LPY is the growth rate of per capita GDP. The coefficient is expected to be positive. BCis the borrowing constraint. The coefficient is

proxied by domestic credit of the banking sector as a percentage of GDP. The coefficient is used to study the depth of the financial system. The relationship between the borrowing constraint and saving is expected to be negative.ε is the error term or the disturbance

term.

We prefer to use ratios rather than absolute values. By using ratios, instead of levels, comparisons between countries can be made without using appropriate exchange rates. We use data for different time periods for the four countries based on the availability of data.