A reassessment of the relationship between services and economic growth in low-income and lower-middle income countries: a non-linear approach

ABSTRACT

We test whether the service sector can contribute to economic growth in low-income and lower-middle income countries despite the likely presence of Baumol’s ‘cost disease’. We employ the non-linear auto-regressive distributed lag method, which investigates whether growth responds differently to a positive and negative change in the service sector’s share. By employing a sample of 62 countries from 1970–2019, we find that an increase in the sector’s share reduces short-run growth implying cost disease. In the long run, however, the results show that an increase in the sector’s share positively contributes to economic growth. Thus, the results support policy reforms that may strengthen service sector in these countries.

JEL CLASSIFICATION:

• O47
• O14
• C33
• C34

KEYWORDS:

• Service Sector
• Economic Growth
• Low Income and Lower Middle-Income Countries
• Sectoral Change
• Baumol’s Cost Disease
• Non-linear auto-regressive distributed lag model

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed here.

Notes

¹ One notable exception to this is the work of Lee and McKibbin (2018). Using a DSGE framework, the authors model different scenarios for the short-run behaviour of aggregate growth and long-run steady state contingent on different growth rates of service sector productivity. Our analysis differs from theirs in terms of the methodology employed and that we attempt to model possible nonlinearity between the service sector and growth.

² Eichengreen and Gupta (2013), however, hypothesize that the service sector’s share evolves in a quartic fashion with respect to per capita income and provide evidence that there are two ‘waves’ of service sector growth: the first wave begins at low levels of income with growth levelling off at around per capita income of USD 1, 800 and the second growth spurt happening around per capita income of USD 4, 000. There has, however, been no study addressing the potential non-linear response of economic growth to the rising share of the service sector.

³ It is the ratio of the total enrolments in the primary classes irrespective of age (that’s why gross) to the total population of the age-group that corresponds to the primary level of education. Source: https://data.worldbank.org/indicator/SE.PRM.ENRR

⁴ We use GDP per capita (constant 2010 USD) for the above analysis. The year 1980–2007 is chosen for normalizing as it is an intermediate period and, in addition, denotes the pre-crisis years.

⁵ Depending on data availability.

⁶ We also ran cointegration tests proposed by Kao (1999) and Pedroni (1999, 2004) which demonstrated that a long-run relationship exists among the dependent variable ($gdppc$) and the explanatory variables.

⁷ The application of the (second-generation) Pesaran (2007) CIPS unit-root test on the residuals confirmed the residuals’ stationarity in levels, thus, confirming the validity of our PMG estimates.

⁸ The application of Pesaran (2007) CIPS test to the residual estimates confirms their stationarity.

⁹ As suggested by one of the reviewers, to reduce the impact of outliers, we winsorize the observations for the GDP per capita growth variable ($gdppc$) at the 1^st and 99^th percentiles. On re-estimation, our results are found to be robust to the winsorization procedure.

¹⁰ For relevant literature, please see (Rodrik, Subramanian, and Trebbi 2004; Alajekwu and Ezabasili 2013; Olayungbo and Quadri 2019).

¹¹ We use the dynardl command in STATA to generate the graphs (Sarkodie and Owusu 2020).

¹² On the suggestion of one of the reviewers, we also divided the data into different groups namely LMICs, Sub-Saharan and non-Sub-Saharan African nations. The revised regression estimates (symmetric and asymmetric) for all the groups were similar to the baseline results in terms of sign and significance.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.