Does asymmetric birch effect phenomenon matter for environmental sustainability of agriculture in Tunisia?

Abstract

This paper investigates the validity of the Environmental Kuznets Curve (EKC) hypotheses for agriculture. The impact of precipitations on soil CO₂ emissions is incorporated in the EKC hypothesis to assess the so-called “Birch effect phenomenon.” Through an autoregressive distributed lag (ARDL) bound test modeling and using annual data from 1975 to 2014, we examine the short- and long-term relationships between net agricultural values added per rural capita, energy consumption, precipitation and agricultural CO₂ emissions in Tunisian agriculture. We then used the nonlinear version of ARDL (NARDL) to examine the asymmetric effect of precipitations on CO₂ emissions. The EKC assumption is validated in favor of the agricultural sector in both short and long-run associations, suggesting the adoption of cleaner production practices. The results also indicate that precipitation increases soil CO₂ emissions in the short-term as well as in the long-term, confirming the Birch effect phenomenon and reflecting the specificity of the Tunisian climate-ecosystem. The asymmetric findings provide evidence of different CO₂ emissions response to negative and positive shocks of precipitations in terms of magnitude, whereas the energy use in agriculture is found not to affect CO₂ emissions when switching from linear to a nonlinear model. As recommendations, promoting the agricultural productivity and preserving farmland in Tunisia should be among the main actions for any agricultural policy to improve economic growth and to achieve environmental sustainability.

Appendix

1.1 Nomenclature

1.1.1 Abbreviations

IPCC:

Intergovernmental Panel on Climate Change

FAOSTAT:

Food and Agriculture Organization Corporate Statistical

NIMT:

National Institute of Meteorology of Tunisia

USD:

American dollar

EKC:

Environmental Kuznets curve

SOC:

Soil organic carbon

GHG:

Greenhouse

CO₂:

The agricultural dioxide carbonemissions

CH₄:

Global methane emissions

N₂O:

Nitrous dioxide emissions

ARDL:

Autoregressive distributed lag model

NARDL:

Nonlinear autoregressive distributed lag model

ECM:

Error correction model

ECT:

Error correction term

1.1.2 Symbols

∆:

The first difference operator

GDP:

The net agricultural value-added per rural capita

GDP²:

The square of the net agricultural value-added per rural capita

AEC:

The per capita agricultural energy consumption

UAL:

Used agricultural land

Prec:

Precipitations

${PREC}_{t}^{+}$ :

Positive partial sum

${PREC}_{t}^{-}$ :

Negative partial sum

TB₁:

Year of first structural break

TB₂:

Year of second structural break

AIC:

Akaike Information Criterion

SBC:

Schwarz Bayesian criterion

OLS:

Ordinary least square

CUSUM:

Cumulative sum

CUSUMSQ:

Cumulative sum of squares

See Fig. 5 and Tables 6, 7, 8 and 9.

Fig. 5

Research framework and main findings

Table 6 Linear autoregressive distributed lag estimation results (without GDP²)

Table 7 Linear autoregressive distributed lag estimation results (with Fertilizer)

Table 8 Linear autoregressive distributed lag estimation results (with Dummy)

Table 9 Linear autoregressive distributed lag estimation results (with ROIL)