We classify West African economies broadly into three—the mineral (including oil) producing, the agricultural product exporters and the aid dependent. The mineral producing group is made up of those for which oil and solid minerals comprise at least 50 % of exports. The agricultural group consists of countries for which such commodities as cotton, coffee, animals, cocoa, peanuts tea, tobacco, seafood, and sugar (canes) carry large weight in their export basket (again comprising 50 % or more in the export basket). The aid dependent group comprises of countries for which official development assistance form up to 50 % of government budget. The study will be based on a panel analysis. The research shall adopt an eclectic growth model that combines structuralist variables with monetary variables to generate estimates for the West African region as a whole and for each group of countries. In addition to the product grouping, the work will also obtain estimates for the two major regional blocs, the UEMOA group and 'the rest'. The approach is to specify a highly over-parameterized model incorporating both structural and monetary variables and then let the data generating process establish what matters and what does not. In effect, we limit the imposition of arbitrary restrictions.

The model for analysis is a modification of the Bassanini and Scarpetta (2001) model for growth in the OECD. The approach used pooled cross-country timeseries data. In addition, the econometric technique allows short-term adjustments and convergence speeds to vary across countries while imposing (and testing) restrictions only on long-run coefficients (i.e. those related to the production function). The researchers tried to shed light on these issues by presenting evidence on the long-term links between policy settings, institutions and economic growth in OECD countries while controlling for underlying differences in technological progress. In particular, the focus is twofold: first, on the possible influences of human capital, research and development activity, macroeconomic and structural policy settings, trade policy and financial market conditions on economic efficiency; second, on the effects of many of the same factors on the accumulation of physical capital.

Formally, the policy-augmented growth equation can be derived from a growth model built around a constant-returns-to-scale technology (Bassanini and Scarpetta 2001). Output is a function of capital, employment, the efficiency with which they act together, and the level of technology. “Given straightforward assumptions on how the factors of production evolve over time, the steady-state level of output per capita can be expressed as a function of the propensity to accumulate physical capital, the population growth rate, the level and growth rates of technological and economic efficiency, and the rate of depreciation of capital”. Moreover, if the concept of capital is widened to include human capital, then the propensity to accumulate the latter is also a factor shaping the steady-state path of output per capita.

The Bassanini and Scarpetta (2001) approach started with a parsimonious specification of the growth equation and then analysis of extended models. The initial specification is consistent with the standard neo-classical growth model and includes only a convergence factor and the basic determinants of the steady state, namely the accumulation of physical capital and population growth. The first extension involves the introduction of human capital while further extensions consider R&D and a set of policy and institutional factors potentially affecting economic efficiency. The growth equation, in its more general form, can be written as follows:

(1)

where y is GDP per capita, sk is the propensity to accumulate physical capital; h is human capital; n is population growth the Vj is a vector of variables affecting economic efficiency; t is a time trend; the b-regressors capture short-term dynamics and ε is the error term. It should be stressed that Eq. (1) is a fairly general specification, and different growth models are nested in it. This is important for the interpretation of the policy variables which could be taken to represent either growth effects or level effects.

The current study retains the influence of physical capital formation through gross fixed capital formation and domestic savings. Given the paucity of data on employment and that population would be a very biased proxy for human, capital, the study drops the influence of human capital and restricts itself to inflows from indigenes resident abroad through remittances. We make additional modifications in the treatment of prices (including relative prices) as well as the incorporation of structural variables regularly taken for granted in most developed countries, but which are still rudimentary and may affect growth in West Africa. Given that the effects of prices have been removed from most real variables (including output), they could enter the modeling process as either policy-induced or exogenous variables. Diverse measures of domestic and relative prices that affect output will therefore be included in the models. Some of the prices to be considered include interest rate (and its spread), the real exchange rate, export and import value indices and interest on external debt. Structural factors to be considered include age dependency, electric power transmission and distribution, domestic credit (to the private sector) and tax revenue. We also incorporate effects of exogenous variables from the external sector including remittances (which also is able to reflect effects of human capital formation, albeit externally), foreign direct investment and reserves to debt ratio. We leave off considerations of short term trends and specifically define the factors that affect economic efficiency to be those relating to policy, including but not limited to monetary and fiscal policy stance of government.

Where the P, S and E vectors are prices, structural variables and external factors (all as listed earlier). As in the Bassanini and Scarpetta (2001) model, the study will adopt the pooled mean group (PMG) estimator. It is an intermediate choice between imposing homogeneity on all slope coefficients (DFE) and imposing no restrictions (MG). The PMG allows intercepts, the convergence parameter, short-run coefficients and error variances to differ freely across countries, but imposes homogeneity on long-run coefficients. Under the assumption of long-run slope homogeneity, the PMG estimator increases the efficiency of the estimates with respect to mean group estimators. The main advantage of pooled cross-country time-series data for the analysis of growth equations is that the country-specific effects can be controlled for, e.g. by using a dynamic fixed-effect estimator (DFE). However, this estimator generally imposes homogeneity of all slope coefficients, allowing only the intercepts to vary across countries. The validity of this approach depends critically on the assumption of a common growth rate of technology and a common convergence parameter. The first assumption is difficult to reconcile with evidence on multifactor productivity patterns across countries. The second assumption is not consistent with the underlying growth model, where the speed of convergence depends, amongst other factors, upon the rate of population growth. An alternative approach is to use the mean-group approach (MG) that consists of estimating separate regressions for each country and calculating averages of the countryspecific coefficients. While consistent, this estimator is likely to be inefficient in small country samples, where any country outlier could severely influence the averages of the country coefficients, thus the choice of PMG estimator.

Found a mistake? Please highlight the word and press Shift + Enter