CC BY
20
Section 4. Environmental economics
https://doi.org/10.29013/EJEMS-20-3-46-54
Zouhair Rached, University of Carthage, National Institute of Agronomic Research of Tunisia (INRAT)
Ali Chebil,
University of Carthage, National Institute for Research in Rural Engineering, Water & Forestry (INRGREF), Tunis, Tunisia
Chokri Thabet,
University of Sousse, Higher Agronomic Institute of ChottMeriem, Sousse, Tunisia E-mail: [email protected]
EFFECTS OF DROUGHT ON TOTAL FACTOR PRODUCTIVITY FOR MOST STRATEGIC CROPS AND REGIONS IN TUNISIA: AN APPLICATION OF THE MALMQUIST INDEX
Abstract. This paper compares Total Factor Productivity (TFP) growth of maj or four agricultural sub-sectors (cereals, fruit trees, vegetables, and livestock) among the five different regions in Tunisia under changing climatic conditions. The study uses panel data set of these regions over the time period of 2000-2016 and Malmquist index approach for the empirical analysis. It includes one output (production value) and three inputs (land, labor and capital). The results show that TFP varies between sub-sectors and regions and decreases during drought years. ANOVA results reveal that fruit trees and cereals in South and fruit trees in North West are more affected by drought. This suggests that more adoption of drought tolerant crops varieties, best agricultural practices promotion and appropriated extension services programs targeted production system in each region are recommended to increase total factor productivity growth.
Keywords: Total factor productivity, Malmquist index, Commodities, Drought, Regions in Tunisia.
1. Introduction facing drought problems and production is highly
Agricultural sector plays an important role in eco- dependent to the climate variability. The country nomic development in Tunisia. It contributes with is characterized by low rainfall, limited renewable 10% to the GDP, 11% to the total national exports water resources and droughts are expected to be and employs 17% of the active population (Dhe- more frequent in the near future (World BankWorld hibi et al. [11]). However, agriculture in Tunisia is Bank, [19]). Drought is one of the risks that threaten
agricultural sector and potentially connected to large economic losses (Bryan et al. [2]).
Tunisian's agricultural sector is highly dependent on water resources since it consumes more than 85% of total water use in the country (ITES, 2014). Given the climate constraints and the limited resources, agricultural development has been traditionally stimulated by the development of the irrigated sector. In fact, irrigation has been a maj or component ofthe agricultural policy pursued by Tunisian government that developed a water saving program to which significant financial incentives have been offered to promote efficient irrigation equipment. The objective of this program was also to prevent drought risks and improve land productivity. In Tunisia, irrigated area has increased from 250,000 ha in 1990 to 450,470 ha in 2015 (MA, 2016). However, the increase of irrigated area has clear consequences on the country's water resources. Despite huge investments made and the expansion of the irrigated area, several studies showed that there are some issues related to this sector in terms ofwater productivity, losses during transfer, distribution and uses (Dhehibi et al. [11]; ITES, 2014; Frija et al., 2015.; Chebil et al., 2019) which do not help to reduce the effect of drought on productivity in a significant way. In this context, the main objectives of this paper are to estimate the total factor productivity (TFP) levels for most strategic crops and regions in Tunisia and explore the effects of droughts on TFP growth. In Tunisia, several studies have been conducted on agricultural TFP at national level (Lachaal et al. [16], Dhehibi and Lachaal [10], Belloumi and Mattoussi [1], Frija et al. [14], Chebil et al. [4]). However, to the best of our knowledge, no studies analyzed TFP of most crops at the regional level. Therefore, comparing drought impact on TFP across regions and crops is crucial to set adaptation strategies.
The rest of the paper is organized as follows. The second section discusses data and analytical method. The estimated results are presented and discussed in section 3. Conclusions and implications are drawn in the final section.
2. Methodology
Growth in TFP is defined as a growth in outputs which is not explained by the growth in the use of inputs in production. Generally, there are two approaches (parametric and non-parametric approaches) that are extensively applied in recent literature to measure the TFP growth. In the present study, we apply a non-parametric approach using the Malmquist TFP index presented in Caves et al. [3] and Fare et al. [13]. The Malmquist index is based on distance functions that allow productivity growth to be decomposed into efficiency change (getting closer to the frontier) and technical change (shifts in the production frontier technology). This index is easy to compute and does not require information about input or output prices, and behavioural objective such as cost minimization or profit maximization (Fulginiti and Perrin, [15]; Coelli et al. [8]; Coelli et al. [9]). This is especially attractive in the context of agriculture, where input market prices are either nonexistent or insufficiently reported to provide any meaningful information for land and labor.
2.1. Malmquist TFP Index
The Malmquist TFP index was first introduced by Caves, Christensen and Diewert [3]. It measures the TFP change between two data points (e.g. those of a particular region in two adjacent periods) by calculating the ration of the distances of each data point relative to a common technology. Following Faire et al. (2004), the period t Malmquist (output oriented) productivity index is given by
Mf =
DO (xf+1,/+1)
(1)
DO X, /)
i.e., they define their productivity index as the ratio of two output distance functions taking technology at time t as the reference technology. Instead of using period t's technology as the reference technology it is possible to construct output distance functions based on period (t+1)'s technology and thus another Malmquist productivity index can be laid down as:
D f+1(xt+1 vt+1)
M t+1 = D° ,
■»f+i
DO+V, yf )
(2)
Fare et al (1994) attempt to remove the arbitrariness in the choice ofbenchmark technology by specifying their Malmquist productivity change index as the geometric mean of the two-period indices, that is,
Mo (x(t+1),/+1, xf, /) =
1
(3)
this productivity index (equation 3) is the product of two distinct components- technical change and efficiency change (Fare et al [13]).
Mo (x(t+1), /+1, xf, / ) =
r DO (x y *+1) J f DO+1(x *, y * ) J
[ DO (x *+1, y *+1) J 1 DO+1(x*, y* )J
DO+1(x*+1, /+1)
D'0 (x*, y* )
DO(xM,yt+1) Y DO(x*,y*)
DO+1(x*+1, y*+1) A DO+1(x*, y* )
(4)
The above Mamquist TFP index can be decomposed in a way that highlights what sources are attributed to the TFP growth. An equivalent way ofwriting
where Efficiency change =
DQ+1(x *+1, y *+1) DO (x*, y * )
Output
Y
A B
C= yt+1 D E
F=yt
DO (x *+1,y*+1) Y DO (x *, y* )
v DO+1(x *+1, y*+1) J^ DO+1(x*, y* )
Xt+1
Input X
Figure 1. The Malmquist productivity index
Figure 1 explains the Mamquist index decomposition. Hence, in terms of the distance along the y axis, the index (6) becomes
Mo (x f+1,7 f+1, x f, J f) =
' (5)
OC / OA
OF / OE
i
OC / OD VOC/OAJy
OF / OE^
OF / OB
The ratios inside and outside the square bracket measure the technical change and efficiency change, respectively. Malmquist indixes greater than one indicate growth in productivity. Malmquist indixes less than one indicate decline in productivity.
2.2. Sources of data and variables specification Malmquist TFP index has been used for the empirical analysis of disaggregated data by crops (commodities) and regions over the time period of 1994-2015. We used one output and three inputs
variables to construct indexes of TFP. The output is the value of agricultural production measured in Tunisian National Dinars (TND1TND^0.49 USD (2015)) (constant 2005 prices). Livestock includes only sheep, cattle and goats due to data limitations for the other livestock. Land area, labor and capital stock are used as inputs. Capital stock (K) is measured in terms of accumulated capital according to the following equation (6):
K* = (1 -S)Kt +1 t
(6)
where Kf - is the value of capital stock at t period, I - is real gross capital formation (investment) at t, S - is the depreciation rate (set to 10%).
Labor (L) is measured in terms of total employment (in thousand days).
The data used in this study was collected from different sources of information which included the
2
X
0
National statistics agencies (INS) Ministry of Agri- Food and Agriculture Organization (FAO), and pub-culture (MA), National Institute of Statistics (INS) lished papers.
Table 1.- Descriptive statistics of input and output variables (constant 2005 prices)
Commodities Output (Million TD) Land* (1000 ha) Labor (1000 days) Capital (Million TD)
Cereals North East Mean 109.3 199.8 1943.9 251.3
S.D. 20.1 9.8 395.6 52.5
North West Mean 279.4 673.8 4736.6 682.4
S.D. 103.0 22.7 701.0 45.8
Center East Mean 16.7 94.1 296.5 61.3
S.D. 16.5 37.1 230.2 28.4
Centre West Mean 67.6 309.2 1201.8 224.4
S.D. 48.4 72.4 674.2 54.3
South Mean 6.4 93.2 102.1 48.3
S.D. 6.4 70.6 65.3 32.2
Livestock North East Mean 257.9 362.8 7.6 550.4
S.D. 64.2 26.7 0.5 112.2
North West Mean 372.4 752.5 17.1 927.9
S.D. 30.4 35.6 1.5 57.0
Center East Mean 164.5 283.1 10.2 380.6
S.D. 31.6 15.3 0.7 50.0
Centre West Mean 195.5 515.0 11.8 552.7
S.D. 84.5 29.3 1.1 100.7
South Mean 160.1 405.4 13.7 439.7
S.D. 93.6 53.6 0.9 82.7
Fruit trees North East Mean 101.4 127.3 3.4 275.8
S.D. 19.5 0.8 0.7 11.9
North West Mean 179.1 228.3 3.8 490.7
S.D. 49.7 18.1 0.9 29.0
Center East Mean 573.7 789.1 10.7 1656.5
S.D. 149.5 7.2 2.0 65.4
Centre West Mean 459.7 632.6 9.1 1321.0
S.D. 130.5 26.9 1.9 6.1
South Mean 336.5 452.1 10.7 955.7
S.D. 93.5 26.6 2.2 15.7
Vegetables North East Mean 148.558.5 62.3 9.1 241.2
S.D. 45.870.7 60.7 1.3 77.6
North West Mean 72.844.0 25.8 3.1 116.2
S.D. 20.791.0 2.2 0.6 19.2
Center East Mean 70.6 24.0 3.9 110.3
S.D. 26.712.8 3.9 0.4 25.6
Centre West Mean 105.666.9 40.1 5.4 175.9
S.D. 25.694.8 4.9 0.6 35.2
South Mean 53.023.4 17.7 2.6 82.0
S.D. 16.794.2 2.1 0.3 14.2
*1000 livestock unit (LSU) for livestock (1 Cattle =1LSU, 5 Sheep = 1 LSU and 6 Goats =1 LSU) Source: Own elaboration based on different sources
The analysis includes 4 of the most strategic crops of the country. it is built based on regional disaggregated data, including 24 governorates of Tunisia. These governorates have been aggregated into five regions (North West, North East, Center West, Center East, and South) based on bioclimatic homogeneity. Table 1 summarizes the data used in measuring TFP.
3. Empirical results
3.1. Total factor productivity change by commodities and regions
The Malmquist indices of productivity change are generated by the software package DEAP version 2.1 (Coelli [7]). The results of TFP, efficiency and technical change by regions and commodities are summarized in the table below. As can be seen in Table
2, technical efficiency is quite stable for all regions and commodities. TFP growth of all observed commodities have experienced varying degrees of decline, with the exception of cereal in the South, livestock and fruit trees in North East. This is primarily attributable to technological regress. Comparison of TFP growth for cereal among all regions show that South has the highest TFP growth (1.29) followed by Center East (1.00). The highest TFP index for livestock is registered in North East (1.019). Regarding fruit trees, North east has the highest level of TFP (1.03) and the remaining remaining regions have level lower than one. The (table 2) is showing that TFP indexes for vegetables in all regions are lower than one which means a negative TFP growth during the study period.
Table 2.- Summary of Malmquist productivity index decomposition (Mean 2000-2016)
Commodities Technical change Efficiency change TFP change
Cereals North East 0.861 0.999 0.860
North West 0.848 1.000 0.848
Center East 1.174 0.853 1.001
Center West 0.848 1,021 0.866
South 1.516 0.857 1.299
Livestock North East 0.961 1.060 1.019
North West 0.849 1.000 0.849
Center East 0.984 0.999 0.983
Center West 0.915 0.985 0.901
South 0.922 1.000 0.922
Fruit Trees North East 1.014 1.017 1.031
North West 1.002 0.989 0.991
Center East 0.861 1.000 0.861
Center West 0.878 1.015 0.891
South 0.948 1.000 0.948
Vegetables North East 0.848 1.037 0.879
North West 0.879 1.000 0.879
Center East 0.878 1.015 0.891
Center West 0.861 1.015 0.874
South 0.968 1.012 0.980
Source: Elaboration by authors
3.2. TFP comparisons by drought conditions nation of drought years (drought years SPI<-0.99
Standardized Precipitation Index (SPI) devel- and no drought years SPI >-0.99), respectively. The
oped by Mc Kee et al. [17]) was used for determi- results of TFP growth by drought conditions are
presented in (table 3). It shows clear differences of condition is much lower compared to the TFP level TFP values among two drought conditions. Partic- under no drought situation. ularly, TFP for cereals in the South under drought
Table 3.- Summary of Malmquist indxes by drought conditions (mean 2000-2016)
Commodities Region No drought years Drought years
Cereals North East 0.895 0.855
North West 0.852 0.838
Center East 1.031 0.970
Center West 0.882 0.846
South 1.743 0.729
Livestock North East 1.023 0.987
North West 0.854 0.836
Center East 0.992 0.973
Center West 0.926 0.868
South 0.933 0.908
Fruit Trees North East 1.049 0.901
North West 1.047 0.826
Center East 0.904 0.819
Center West 0.913 0.863
South 1.018 0.858
Vegetables North East 0.884 0.843
North West 0.892 0.838
Center East 0.938 0.843
Center West 0.891 0.852
South 1.021 0.928
Source: Elaboration by authors
3.3. Results of Analysis of variances 4. As shown in the table, interaction factors were not
Analysis of variances (ANOVA) was performed statistically significant at 5% level for all cases. Drought
with three factors (region, drought and interaction) to effects are statistically significant at 5% for fruit trees
test significance for difference in TFP growth. Results and vegetables. However, region effects are statistically
ofANOVA of the different cases are presented in table significant only at 10% for vegetables and livestock.
Table 4.- Analysis of variances for TFP
Branch Source Sum of squares df Mean of square F P-value
1 2 3 4 5 6 7
Model 6.430 9 0.714 1.65 0.117
Region 1.547 4 0.386 0.890 0.479
Drought 0.559 1 0.559 1.29 0.259
Region*drought 3.178 4 0.794 1.84 0.131
Error 30.266 70 0.432
Toal 36.697 79 0.464
1 2 3 4 5 6 7
Model 0.308 9 0.034 1.700 0.105
Region 0.187 4 0.469 2.33 0.064*
Lives- Drought 0.014 1 0.014 0.73 0.394
tock Region*drought 0.004 4 0.001 0.050 0.995
Error 1.410 70 0.020
Toal 1.718 79 0.021
Fruit Model 0.635 9 0.070 5.57 0.00**
Trees Region 0.09 4 0.023 1.89 0.122
Drought 0.265 1 0.265 20.93 0.000**
Region*drought 0.062 4 0.015 1.23 0.306
Error 0.088 70 0.012
Toal 1.522 79 0.019
Vege- Model 0.217 9 0.024 1.86 0.072*
tables Region 0.125 4 0.031 2.41 0.057*
Drought 0.062 1 0.062 4.78 0.032**
Region*drought 0.010 4 0.002 0.200 0.937
Error 0.909 70 0.129
Toal 1.126 79 0.014
* *Difference significant at 5% *Difference significant at 10% Source: Elaboration by authors
Table 5.- Results of Least Squares Means tests (Pr > |t|)
Commodities Regional factor Drought factor
Regions North West Center East Center West South Drought/No drought
1 2 3 4 5 6 7
North East 0.231 0.936
North West 0.674 0.536 0.970 0.124 0.970
Cereals Center East 0.922 0.938 0.316 0.852
Center West 0.560 0.116 0.912
South 0.003**
North East 0.195 0.732
North West 0.732 0.0132 ** 0.098* 0.1684 0.827
Livestock Center East 0.020** 0.3421 0.2213 0.788
Center West 0.0929* 0.6414 0.418
South 0.730
North East 0.029** 0.086* 0.092* 0.472 0.084*
Fruit Tree North West 0.469 0.266 0.963 0.001**
Center East 0.5123 0.059* 0.138
1 2 3 4 5 6 7
Fruit Center West- 0.213 0.381
Trees South 0.006**
North East 0.035** 0.015** 0.041** 0.013** 0.636
Vegetables North West Center East Center West 0.973 0.601 0.563 0.877 0.887 0.638 0.414 0.099* 0.497
South 0.113
* *Difference significant at 5% *Difference significant
To further explore the distinguished factors with different results in terms of TFP, we conducted the Least Squared Means (LSM) tests for comparisons between pairs of means among more than two categories, which will provide information on which specific factor category is significantly different from others. LSM test is recommended where there are not equal observations for each combination of treatments.
This test shows that the differences of TFP mean values between regions for cereals are not statistically different at 5% level (Table 3). However, it shows a statistical difference at 5% for livestock between North West and North East and between North West and Center East. Among the regions, the maximum TFP was obtained in North East. The LSM test also shows that TFP of North East was significantly different from Center East for fruit trees. Results of this later test show a statistical difference at 5% between South and other regions for vegetables. Regarding to the pairwise comparisons between TFP under drought and no drought conditions for different regions and crops, results are showing that fruit trees and cereals in South and fruit trees in North West are more affected by drought. Hence, the LSM tests for comparison of the means show a statistical difference at 5% between TFP for cereals and fruit trees in the South (table 5). Results also show a difference at 5% between TFP for fruit trees in Center West. As can be seen in table 5, the test is showing a significant
at 10%
difference at only 10% of this indicator for fruit trees in North East and vegetables in Center East.
4. Conclusion and policy implications
In this paper, the Malmquist productivity index was used to calculate the total factor productivity growth across regions for most commodities during the 2000-2016 period under drought conditions. Empirical results show that TFP varies between sub-sectors and regions and decreases during drought years. Globally, it was found that total factor productivity has experienced a negative evolution in study regions for these commodities. The results also indicate a negative impact of drought on total factor productivity. Fruit trees and cereals in South and fruit trees in North West are more affected by drought.
The results show ANOVA results reveal that fruit trees and cereals in South and fruit trees in North West are more affected by drought
The results of this study can be used by policymakers and water resources managers who are looking for ways to address drought problems in productivity growth of Tunisian's agricultural sector. Identifying those regions and commodities most at risk facilitates appropriate formulation of new drought policy. This suggests that more adoption of drought tolerant crops varieties, best agricultural practices promotion and appropriated extension services programs targeted production system in each region are recommended to increase total factor productivity growth.
References:
1. Belloumi M. and Mattoussi M. S. Measuring agricultural productivity growth in MENA countries. Journal of Development and Agricultural Economics, 1(4). 2009.- P. 103-113.
2. Bryan E., Ringler C., Okoba B., Roncoli C., Silvestri S. and Herrero M., Adapting agriculture to climate change in Kenya: Household strategies and determinants. J Environ Manage. 2013. Jan 15; 114: 26-35. Doi: 10.1016/j.jenvman.2012.10.036. Epub 2012 Nov 30.
3. Caves D. W., Christensen L. R. and Diewert W. E., Multilateral comparisons of output, input and productivity using superlative index numbers. Economic Journal, 92, 1982.- P. 73-86.
4. Chebil A., Frija A. and Alyani R. Measurement of total factor productivity and its determinants: case of Tunisian wheat sector. New Medit,- No. 2. 2016.- P. 22-27.
5. Chebil A., Swissi A., Stambouli T. and Frija A. Estimation of the economic loss due to irrigation water use inefficiency in Tunisia. Environmental Science and Pollution Research, 26(11). 2019.- P. 11261-11268.
6. Central Bank of Tunisia (CBT). Annual Report, - Tunis. 2016.
7. Coelli T.J. "A Guide to DEAP Version 2.1: A Data Envelopment Analysis Program" CEPA Working Papers, 8/96, Department of Econometrics, University of New England, - Australia, 1996.- P. 1-49.
8. Coelli T. J. and Rao D. S. P. Total Factor Productivity Growth in Agriculture: A Malmquist Index Analysis of 93 Countries, 1980-2000. CEPA Working Papers, 2/2003 School of Economics University of Queensland, St. Lucia, Qld. Australia, 2003.- P. 1-31.
9. Coelli T. J., Rao D. S.P., O'Donnell C. J.and Battese G. E. An Introduction to Efficiency and Productivity Analysis. Second ed.- New York: Springer. 2005.
10. Dhehibi B. and Lachaal L. Productivity and Economic Growth in Tunisian Agriculture: An Empirical Evidence. African Development Review, 18(2). 2006.- P. 248-257.
11. Dhehibi B., Telleria R. and Aden-Aw-Hassan. Total factor productivity in Tunisian agriculture: measurement and determinants. New Medit,- No. 1. 2014.- P. 4-13.
12. FAO (Food Agriculture Organisation of the United Nations)., 2014. FAO-STAT agriculture data. Available at: URL: htt://faostat.fao.org
13. Fare R., Grosskopf S., Lindgren B. and Ross P. Productivity developments in Swedish hospitals: a Malmquist output index approach, in Charnes A., Cooper W., Lewin A. and Seidford L(eds.): Data Envelopment Analysis. Kluwer Academic Publisher, 1994.- P. 253-272.
14. Frija A., Dhehibi B. and Aw-Hassan A. Total factor productivity growth of the Tunisian agricultural sector: a review of historical trends and main determinants. African J. Economic and Sustainable Developmental. 4.- No. 4. 2015.- P. 297-307.
15. Fulginiti L. E. and. Perrin R. K. LDC Agriculture: Nonparametric Malmquist Productivity Indexes. Journal of Development Economics. - 53. 1997.- P. 373-90.
16. Lachaal L., Chebil A. and Dhehibi B. Measuring factor substitution and technological change in the Tunisian agricultural sector, 1971-2000. Agricultural economics review, - 2, 2005.- P. 28-32.
17. Mc Kee B Thomas., Doesken J Nolan. and Kleist John. The relationship of drought frequency and duration to time scales, Proceedings of ninth Conference on Applied Climatology, American Meteorological Society,- Boston, 1993.- P. 179-184.
18. Ministry ofAgriculture. 2016. Statistical reports, 1980-2016. - Tunis.
19. World Bank. Climate Variability, Drought, and Drought Management in Tunisia's Agricultural Sector. -Washington, DC: World Bank. 2018.
