Does the quality of institutions enhance savings? The case of Sub-Saharan Africa

ABSTRACT

Over the last three decades, there has been increasing disparity in savings across regions and income groupings globally. In this paper, we investigate whether the quality of institutions explains the saving disparities in Sub-Saharan Africa (SSA). Utilizing comprehensive panel data and spanning the period 1980–2015, we estimate a savings model using the two-step instrumental variable generalized method of moment (2SIV-GMM) estimator. Our results show that the impact of institutions on savings behaviour differs across regions and income groupings, and in SSA, in aggregate. We find that the level and growth of per capita income and terms of trade enhance savings whereas government consumption expenditure, financial sector development and the elderly dependency rate are savings impeding. The findings are robust to alternative model specification and highlight the importance of institutions in influencing savings behaviour in SSA.

KEYWORDS:

• Savings
• institutions
• governance
• 2SIV-GMM
• Sub-Saharan Africa

JEL CLASSIFICATION:

• E02
• O16
• O43
• O55

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Good institutions in this context imply those political, economic and social institutions that provide secure private property, encourage trust, prevent the risk of expropriation and foster exchange by reducing transaction cost for broad cross-section of a society. Good governance entails the process of policy decision-making and implementation through legitimate intermediate institutions to promote sustained society. For a review, see Shirley (2003), Acemoglu, Johnson, and Robinson (2005) and North (1990).

² The subsistence-consumption theory posits that at lower levels of income, the level and growth of income has a greater impact on the saving rate.

³ The difference between the two is the time horizon – the PIH assumes infinitely lived consumers, whereas the LCH considers finitely lived consumers.

⁴ Note that in Equation (8) additional restrictions are imposed: there are no borrowing restrictions, bequest motives or buffer-stock saving behaviour.

⁵ Laibson (2005) stressed that economic actors display a high degree of impatience when making decisions and prefer to immediately choose than to wait.

⁶ Loayza, Schmidt-Hebbel, and Servén (2000) noted that closed-form equations derived from consumption optimization models are usually narrow, focusing on one or two variables and recommended a reduced-form equation for aggregate studies.

⁷ Given the variant measures of financial development (FD) and no selection criteria of which proxy is the most appropriate, the FD variable is an index constructed by means of principal component analysis using three measures of FD variables extensively used in the empirical literature, that is M2, Credit to private sector, and credit to private sector by banks. The highest eigenvalue was used to measure financial development.

⁸ Note that when income uncertainty persists, consumers prefer to save for precautionary reasons in the short term.

⁹ We combined the lower-middle and upper-middle-income countries into one sub-group and labelled this group ‘middle-income countries’ given the relatively small number of upper-middle-income countries.

¹⁰ Table A2 summarizes the variable definitions and descriptive statistics and Table A3 displays the correlation matrix.

¹¹ Our sample has been subject to various checks, and we exclude countries, such as South Sudan and Somalia, for which we find inconsistencies in the basic national account figures (see the Table A2).

¹² This may arise because of the presence of common shocks and unobserved components of the error term.

¹³ See Loayza, Schmidt-Hebbel, and Servén (2000), Ando and Modigliani (1963), Edwards (1996) and Barro (1974).

¹⁴ See Kaufmann, Kraay, and Mastruzzi (2011) and Knack and Langbeign (2010) for details.