ABSTRACT
This paper examines the effectiveness of monetary policy and its implications for financially included and excluded households in Sub-Saharan African (SSA) economies, using an estimated New-Keynesian dynamic stochastic general equilibrium (DSGE) model. The model has financially included households coexisting with financially excluded households. We exploit time series data on four SSA economies, spanning 1985–2016, to estimate the model’s parameters through Bayesian inference methods. Our estimation results show that the share of financially excluded households in these economies is relatively small, usually between 35% and 42%. Further, our Bayesian impulse response analysis shows that a positive monetary policy shock significantly reduced inflation and output, despite a sizeable fraction of the population is financially excluded. Additionally, we find that contractionary monetary policy tends to have differentiated impacts; it decreases the consumption of financially excluded households more than that of financially included ones. The results reveal that financially included households are able to absorb shocks, and thus can smooth consumption more effectively than financially excluded households. Generally, although an increase in the number of financially excluded households reduces the effects of interest rate policies, we find an opposite result: the effectiveness of monetary policy improves as the fraction of financially included households falls.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Even, in response to the prospects of such persons falling on non-cash savings during income shocks, Mehrotra and Nadhanael (Citation2016) were unequivocal when they suggested that, if there are greater negative shocks to an economy, savings in the form of assets other than cash may not even be helpful to smooth consumption much.
2 To the best of our knowledge, this paper is the first to estimate that model for Ghana, Gabon, Mauritius, and Lesotho.
3 See the discussion paper version of this paper for the full sets of log-linearized equations.
4 Although there is a negative relationship between the interpolated variables and the factor variable, the Chow-Lin technique allows for both positive and negative relationships between the far variables.
5 We tried using large standard deviation such 0.1, 0.075, and 0.05. However, this often resulted in numerical problems for the algorithm, such as non-convergence. Here, we show a 95% confidence interval of the prior distributions we set for the parameter characterizing the fraction of financially excluded households: [0.4509 0.5489].
Additional information
Notes on contributors
Paul Owusu Takyi
Paul Owusu Takyi is a Ph.D. candidate and a Graduate Teaching Assistant at the National Graduate Institute for Policy Studies (GRIPS), Tokyo, Japan. His research interests are in the areas of Monetary Economics, Public Economics, Development Economics, Applied Econometrics and Time Series Analysis.
Roberto Leon-Gonzalez
Roberto Leon-Gonzalez is a Professor at the National Graduate Institute for Policy Studies (GRIPS) in Tokyo (Japan). He is also a senior fellow at the Rimini Centre for Economic Analysis (RCEA). He is specialized in Bayesian Econometrics.