Chapter 5. Results and Discussion
5.1 Data Envelopment Analysis (DEA)
5.2.2 Elasticity Estimation
Using the parameters obtained from the MLE, the estimated frontier elasticity of each input used was calculated by using equation (13). The elasticities for the translog models are estimated at the means of the input variables.
Table 5.8a and 5.8b present the estimate elasticities in different periods for the translog model 1, in which technical change is presented (including time variable). The partial elasticities for particular inputs in some cases differ considerably among models.
The elasticities at means by period of the stochastic frontier model for labor, machinery, land, fertilizer and return to scale are presented in Table 5.8a and Table 5.8b.
Table 5.8a Summary elasticities of inputs by country, 1979-2014
Model Labor Machinery Land Fertilizer Time Return to Scale Model 1 (Technical Change; Inefficiency)
Guatemala 0.471 -0.310 0.249 0.392 -0.003 0.798 Honduras 0.472 -0.364 0.227 0.424 -0.001 0.758 El Salvador 0.498 -0.374 0.209 0.410 0.002 0.745 Nicaragua 0.513 -0.383 0.200 0.405 0.006 0.740 Costa Rica 0.520 -0.429 0.179 0.408 0.007 0.684 Panama 0.516 -0.505 0.162 0.410 0.007 0.590
All 0.498 -0.394 0.204 0.408 0.003 0.719
Elasticities with respect to
51
Table 5.8b Summary elasticities of inputs by country, 1990-2014
Table 5.9 Summary elasticities of inputs by period
In C.A. the elasticity of output with respect to labor is the highest among all the input elasticities, this elasticity shows that labor is the input with the greatest influence on production, followed by fertilizer. The elasticity of return to scales (0.72) implies decreasing return to scale (DRS), which suggests that the agricultural sector in this region is engaged in production of non-optimal scale. When we exclude the 80's which is a turbulent decade in C.A., the elasticity shows increasing returns to scale (IRS). However, in the last years of the study, the elasticity shows DRS.
Model Labor Machinery Land Fertilizer Time Return to Scale Model 1 (Technical Change; Inefficiency)
Guatemala 0.589 0.538 -0.041 0.277 -0.016 1.346 Honduras 0.572 0.402 -0.077 0.320 -0.017 1.200 El Salvador 0.604 0.394 -0.098 0.295 -0.012 1.184 Nicaragua 0.605 0.339 -0.114 0.303 -0.011 1.121 Costa Rica 0.616 0.279 -0.141 0.297 -0.010 1.041 Panama 0.615 0.238 -0.155 0.309 -0.010 0.998
All 0.600 0.365 -0.104 0.300 -0.013 1.148
Elasticities with respect to
Model Labor Machinery Land Fertilizer Time Return to Scale Model 1 (Technical Change; Inefficiency)
Period 1979-1989 0.2665 -2.1194 0.9055 0.6534 0.0379 -0.2561 Period 1990-2014 0.6001 0.3648 -0.1043 0.3003 -0.0126 1.1484 Period 1990-1999 0.3451 0.6630 0.8036 0.3431 -0.0197 2.1350 Period 2000-2014 0.7702 0.1661 -0.7096 0.2717 -0.0078 0.4906 Period 1979-2014 0.4982 -0.3942 0.2042 0.4082 0.0029 0.7192
Elasticities with respect to
52 5.2.3 Hypothesis Test
On the model defined by equation (8) and (7), some restrictions were imposed. In order to check whether or not those restrictions were valid, the log-likelihood ratio test (LR) was conducted.
The first null hypothesis test if inefficiency effects are not significant for the model.
When this restriction was imposed, the value of the log-likelihood function (LLF) was 61.65. The result of the LR test was 20.95, which is larger than the critical value of 12.59 at 5% level of significance. We reject the null hypothesis which assumes the inefficiency variables do not have influence in the model.
The second null hypothesis considers the Cobb-Douglas (CD) production function as an adequate representation of agricultural production in C.A. This is rejected, indicating that the functional form used is adequate.
The third null hypothesis test the absence of technical change over time and whether or not the coefficients of the time-related variables in the translog function are equal to zero. The result supports the rejection of the null hypothesis, indicating that TC exists in the agricultural sector of C.A.
53
Table. 5.10 Test of Hypothesis for parameters of the Stochastic Frontier, and inefficiency model for the agricultural sector in Central America.
5.2.4 TFP and its component
The results of TFP and its components for SFA are presented as follows.
From Table 5.11, we can see that the average TFP growth for C.A. is 1.0043 and that the major source of this growth is EC with 1.0055. However, in terms of individual country performance only in Guatemala, Honduras, Nicaragua, and El Salvador was EC the major source of growth, while the main source of agricultural productivity in the other countries (Costa Rica and Panamá) was TC.
Null Hypothesis Log-likelihood Test statistic Decision
Unrestricted model 72.123
61.648 20.950 12.59
-13.043 170.332 25.00
40.806 62.634 12.59
𝐻0: 𝛽5= 𝛽𝑖5= 0, 𝑖= 1,2,3,4,5.
𝐻0 la 𝛽𝑖𝑗= 0, 𝑖, = 1,2,3,4,5.
𝐻0: 𝛿 = 0,ℎ= 1,2,3,4,5,6. 𝐻0 e e ed
𝐻0 e e ed 𝐻0 e e ed Critical Value
0. 52
54
Table 5.11 Summary description of TFP change and its components, 1979-2014 (SFA)
Country Description
Std. Deviation 0.0509 0.0157 0.0493 Honduras
Mean 1.0103 0.9794 0.9895
Max 1.2323 1.0127 1.2433
Min 0.8137 0.8530 0.7754
Std. Deviation 0.1041 0.0393 0.1109 El
Salvador
Mean 1.0043 1.0038 1.0081
Max 1.1629 1.0218 1.1856
Min 0.8515 0.9372 0.8622
Std. Deviation 0.0711 0.0238 0.0712 Nicaragua
Mean 1.0005 0.9751 0.9756
Max 1.3192 1.0015 1.2841
Min 0.7721 0.9648 0.7498
Std. Deviation 0.1010 0.0095 0.0984 Costa Rica
Mean 1.0023 1.0328 1.0351
Max 1.1617 1.0383 1.2062
Min 0.8729 1.0202 0.9026
Std. Deviation 0.0616 0.0048 0.0643 Panamá
Mean 1.0022 1.0142 1.0164
Max 1.1205 1.0268 1.1320
Min 0.8979 1.0081 0.9069
Std. Deviation 0.0565 0.0055 0.0574 Central
America
Mean 1.0055 0.9988 1.0043
Max 1.3192 1.0383 1.2841
Min 0.7721 0.8530 0.7498
Std. Deviation 0.0763 0.0285 0.0795
Analysis by period in Table 5.12 reveals that before 2010 the overall TFP for the region was driven mostly by EC rather than TC. However, the effect of TC after 2010 seems to get more important, although this is pushed in some countries more than others.
55 Period 1980-1989 0.9966 0.9876 0.9842 Period 1990-1999 1.0106 0.9961 1.0067 Period 2000-2009 1.0133 1.0053 1.0186 Period 2010-2014 0.9999 1.0153 1.0151 Period 1990-2014 1.0091 1.0033 1.0124 Period 1980-2014 1.0055 0.9988 1.0043
Table 5.13 Summary description of technical efficiency, 1979-2014 (SFA) Technical
Efficiency Guatemala Honduras
El
In addition, a detailed analysis of technical efficiency in Table 5.13 shows that along the overall period the mean efficiency was 0.839, with a minimum value of 0.522 and a maximum value of 0.987 which represents the countries on the frontier. The lower value of efficiency means that countries are less efficient and higher values means that countries are more efficient or closer to the frontier. The mean value of 0.839 indicates that countries can reduce the use or consumption of all inputs by 6.1% without reducing their output.
Figure 5.8. shows the cumulative growth of the region at the end of the period. The results from SFA indicate a cumulative 1.21, 0.96, and 1.16 for EC, TC, and TFP change respectively. The TFP change in this approach is less compared to the previous approach, and the main source of agricultural growth was due to efficiency changes. Likewise, under
56
this approach, the average efficiency change turned out to be positive and the technical change negative.
Source: prepared by the author.
Figure 5.8 Cumulative DEA & SFA TFP change and its components, 1979-2014.
The differences between the results obtained in SFA compared with DEA are not unexpected, considering that both methods involve different calculation which affects the final result (See Table A in appendix). It is important to appoint that we must be careful when we compare the results since they only show the dispersion of efficiency within each sample.
0.6 0.8 1 1.2 1.4 1.6 1.8
Cummulative Index
Note: Discontinuos line represent the result from SFA method. Continuos line represent the result from DEA method.
TFP Change, Technical Change and Efficiency Change by year, 1979-2014
effsfacum effdeacum techsfacum
techdeacum tfpdeacum tfpsfacum
57 5.2.5 Country Analysis of TFP and its components
The following includes analysis of TFP, TC, and EC by income using the SFA approach.
The results of SFA show a similar pattern among the low-middle income (LM) and the upper-middle income (UM) countries of C.A. Agricultural growth in LM countries when we account for exogenous effects is mostly driven by EC, which support the conclusion of Henderson and Russell (2005) that when we account for environmental, social or economic differences, for instance, human capital, then the primary driving force may be EC. Also, the results reveal possible problems in the process of diffusion and adoption of modern technologies as is suggested by Araujo et.al (2014).
In terms of EC, the highest cumulative EC was in Guatemala with 1.60 and was followed by Honduras with 1.43. In terms of TC, most of these countries present a deterioration, with a very low improvement after 2005. The highest deterioration of TC was Nicaragua with around 0.414 at the end of 2014, and the best was El Salvador with around 1.14.
Lower-middle Income Countries
0.00 0.50 1.00 1.50 2.00
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013
Index
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013
Index
a) Guatemala (SFA)
Tfpch Effch Techch
58 .
Source: prepared by the author.
Figure 5.9 Low-middle income economies TFP change, TC and EF
U
pper-middle Income Countries
Source: prepared by the author.
Figure 5.10 Upper-Middle Income Economies TFP change, TC, and EF.
0.00 0.50 1.00 1.50
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013
Index
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013
Index
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013
Index
1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013
Index
f) Costa Rica (SFA)
Tfpch Effch Techch
59
The results for UM countries show a significant TFP growth in agriculture. The best performance was in Costa Rica with a cumulative 3.35 at the end of 2014. In the case of Panamá, the cumulative TFP growth was 1.77. In both cases, the agricultural growth was due to the noticeable increase in TC and the slightly but positive contribution of EC over the entire period.
5.2.6 Efficiency Effect
The stochastic frontier model includes six exogenous variables, all of them significantly different from zero. The expectation is that the sign of the coefficient would be negative since more of each input reduces inefficiency. In the case of Co2, we expect that more of this input will worsen efficiency. In the case of rural population, we expect that more population in rural areas will increase productivity since less migration toward cities is an indicator of stability in rural areas where most agricultural activity is concentrated.
The results indicate that the 6 inputs used, satisfy the expectation of negative signs.
Co2 and life expectancy both had a positive sign. However the Co2 result was not statistically significant, and the life expectancy result was significant at 10 percent. These results suggest that increments in the general welfare of a country may not necessarily indicate that the agricultural sector is becoming more productive.
The Herfindahl-Hirschman index usually used to analyze the concentration of market was used in this study to determine the concentration of primary crops. The sign
60
was negative as expected, since more concentration of primary crops may indicate specialization of the market.
In the case of human capital, irrigation and rural population the sign of the coefficients were as expected. More education and human capital had a significant and considerable effect in enhancing efficiency.
From the inefficiency model, we obtained 5 significant coefficients at 5% level with the exception of Co2 and irrigation. The negative sign of the coefficients implies a reduction of inefficiency as we increase the use of these inputs.
The significance of the effect is found using the parameter gamma which indicates how much variation of the composite error term corresponded to the inefficiency component ( = 0.9329). This results was significant at 5%.
61
Chapter 6. Conclusion
6.1 Study Approach
This study focuses on identifying the sources of agricultural productivity growth and analyzing the agriculture growth for six countries in C.A. Country-level data on agricultural production are used from 1979 to 2014. The first approach used the method of data envelopment analysis (DEA) considering primary crops as output and four inputs; labor, land, machinery, and fertilizer.
The second approach used the method of stochastic frontier analysis (SFA), with one output, four inputs, and six exogenous variables: Co2 in agriculture (Co2) , human capital (HC), irrigation area (IRR), rural population (RP), life expectancy (LE), and an index of crop concentration (HHI).
6.2 The Results
This thesis provide important information about trends in agricultural productivity in Central America. In the first stage of analysis, DEA results showed an annual growth in TFP of 0.53% with EC (or catch-up) contributing negatively around 0.4% and TC (or frontier shift) contributing a positive 0.9% growth. The results of cumulative change with DEA indicates that on average the region at the end of 2014 had an EC, TC, and TFP growth accumulation of -13%, 38%, and 20% respectively.
Turning to the country-by-country results, the best TFP performance was shown by Costa Rica with an average growth of 1.6% followed by El Salvador with 1.14% growth. The
62
worst performance was shown by Panamá with a 1.94% regress. In the case of Guatemala, Honduras, and Nicaragua growth was 0.97%, 0.66%, and 0.76% respectively.
The second stage analysis by using the translog production model with exogenous effects shows an improvement of 0.43% per annum in TFP’s performance for C.A. with EC contributing a positive 0.55% and TC contributing a negative 0.12%. The results of the parametric model show a cumulative EC, TC, and TFP of 1.212, 0.959, and 1.162, respectively. The estimate of the cumulative TFP in the parametric frontier is lower than that from the VRS DEA. In addition, the efficiency change contribution is positive compared to the non-parametric model.
In terms of considering the results for individual countries, the best TFP performance was shown by Costa Rica with a growth of 3.51% and was followed by Panamá with 1.64%
growth. The two countries with the worst performance were Nicaragua with a deterioration of 2.44% and Honduras with a regress of 1.054%. Guatemala and El Salvador obtained a result of 0.22% and 0.81% average growth respectively.
In both, the DEA and SFA method, the results show that C.A. is not reaching the production frontier, this finding implies that countries in the region are not making an efficient use of the inputs, therefore with the adequate policies they could possibly continue reducing the use of all inputs without reducing the quantities of primary crops in order to become more efficient. Despite the advantages that DEA may have, under this method, we must assume that there is no inefficiency reason why in this study we prefer the approach of SFA since it allows us to approach the analysis from a point of view that capture the
63
differences between countries and addresses the heterogeneity of the region by including exogenous variables.
Compared to other factors in the production function, labor and fertilizer are more important for increasing the agricultural production. The negative sign of the machinery elasticities during the 80’s may indicate an inadequate utilization or sub-utilization of machinery. However, these results require special attention and should be interpreted carefully.
The results suggest that the upper-middle income countries are more likely to adopt technologies compared to those countries with low-middle income.
6.2.1 Exogenous variables
In the second stage analysis, six social, environmental, economic and infrastructure variables were analyzed. The results show human capital as an important factor for reducing inefficiency in the region. Also, the negative sign of the parameters of the exogenous variables indicates that more irrigation, more concentration of population in rural areas, and higher specialization in the production of crops positively affect the efficiency of agriculture.
The positive sign of Co2 may be related to farming practices, such as burning fields or burning of wasted crops which can decrease the quality of soils. Life expectancy is commonly associated with health care and economic welfare, for this results, the positive sign of this variable may indicate that as the general welfare of a country develops, attention to activities in the agricultural sector diminishes.
64 6.3 Policy implication
There are important variables that have not been included in this study due to unavailability, limitation of data, or weak results. Data such as climate, infrastructure, domestic research, and development and investment in agriculture would provide more information regarding the TFP growth and its sources in Central America.
In light of the results discussed in the previous chapter, the following implications are offered:
Decreasing TC in LM economies is the major concern and obstacle to ensuring food security and becoming more competitive internationally. The decrease in agricultural labor might be a risk for LM countries with limited technology. Hence, it is important that governments make efforts to increase access to technology and innovation.
For those countries in C.A., their inability to reach the optimum TE, far away to be something negative, represents an opportunity to still improving the productivity.
To capitalize these opportunities and to reduce the risk of decreasing growth in agricultural productivity, some changes can be done:
Investment in irrigation can increase the efficiency of the agricultural sector in C.A. More irrigation use can compensate possible collateral effects of climate change in the region.
Investment in education can increase efficiency, innovation, and risk management; farmers with higher education could more easily to adopt new practices, technologies, and address possible risks in the production process.
65
Investment in infrastructure is vital to the delivery of services that may reduce the food waste and encourage investment in technology.
Specializing in crop production increase efficiency. Specialization can increase the ability of farmers to improve the agricultural process and reduce the use of inputs. Therefore Central American countries can improve its absolute advantages.
In order to encourage farmers to use fertilizers and to increase agricultural production, the government could provide a subsidy on fertilizer.
In order to increase production and to make farmers remain in the agricultural sector, the government could design some agricultural policies, such as price support, retirement schemes, and subsidies.
6.4 Further research
It’s appropriate given the particular circumstances of C.A that further analysis should include political components, land concentration, and structural changes. Also, the inclusion of price information can provide a clear representation of the productivity change for each country over the last years. Analysis of climate was tried in this study using the international disaster database of the Centre for Research on the Epidemiology of Disasters (CRED) as a reference, and information about infrastructure using the information about fixed telephone subscription provided by the World Bank as a reference. However, the results were not conclusive or significant.
66
References
Adelman, I. and Morris, C. (1988). Comparative Patterns of Economic Development, 1850–
1914. Baltimore, United States: John Hopkins University Press.
Aigner, D., Lovell, C., and Schmidt, P., (1977). Formulation and estimation of stochastic frontier production function models. Journal of Econometrics, 6,21-37.
Araujo, J., Feitosa, D., and Bittencourt, A. (2014). Latin America: Total factor productivity and its components. CEPAL Review 114.
Barro, R. and Lee, J. (2013). A New Data Set of Educational Attainment in the World, 1950-2010. Journal of Development Economics, vol 104, pp.184-198.
Battese, G. and Broca, S. (1997). Functional Forms of Stochastic Frontier Production Functions and Models for Technical Inefficiency Effects: A Comparative Study for Wheat Farmers in Pakistan, Journal of Productivity Analysis, 8, 395–414.
Battese, G. and Coelli, T. (1995). A Model for Technical Inefficiency Effects in a Stochastic Frontier Production Function for Panel Data. Empirical Economics, 20, 325-332.
Bravo-Ortega, C. and D. Lederman. (2004). Agricultural Productivity and Its Determinants:
Revisiting International Experiences. Estudios de Economía, 31(2), 133-163.
Charnes, A., Cooper, W., and Rhodes, E. (1978). Measuring the efficiency of decision making units, European Journal of Operational Research, 2, 429-444.
Coelli, T. (1996). A guide to DEAP version 2.1: a data envelopment analysis (Computer) Program. Armidale, Australia: Working Paper, Center for Efficiency and Productivity Analysis (CEPA), Department of Econometric, University of New England.
Coelli, T. (1996). A guide to FRONTIER Version 4.1: A computer program for stochastic frontier production and cost function estimation. Armidale, Australia: Working Paper, Center for Efficiency and Productivity Analysis (CEPA), Department of Econometric, University of New England.
Coelli, T. and Rao, D. (2005). Total factor productivity growth in agriculture: a Malmquist index analysis of 93 countries, 1980-2000. Agricultural Economics, 32:s1, 115-34.
Coelli, T., Rao, D., O’donell, C., and Battese G. (2005). An Introduction to Efficiency and Productivity Analysis (2𝑛𝑑ed.). New York, United States: Springer.
Comisión para America Latina y el Caribe. (2015). Coyuntura laboral en América Latina y el Caribe. Protección social universal en mercados laborales con informalidad.
Boletín CEPAL-OIT. No.6
67
Craig, B., Pardey, P., and Roseboom, J. (1997). International productivity patterns:
accounting for input quality, infrastructure, and research. American Journal of Agricultural Economics, 79, 1064-1077.
Dias, A., Romano, L., and Garagorry, F. (2010). Agricultural Productivity in Latin America and the Caribbean and Sources of Growth. Handbook of Agricultural Economics, Burlington: Academic Press, pp 3713-3768. ISBN: 978-0-444-51874-3.
Ebata, A. (2011). Agricultural productivity growth in Central America and the Caribbean.
Thesis dissertation. University of Nebraska.
Economic Commission for Latin America. (2005). Poverty, hunger and food security in Central America and Panama. Social Development Division. Santiago de Chile.
ISBN: 92-1-121521-8.
Färe, R., Grosskopf, S., Lindgren, B., and Roos, P. (1992a). Productivity changes in
Swedish pharamacies 1980–1989: A non-parametric Malmquist approach, Journal of Productivity Analysis, 3, 85-101.
Färe,R., Grosskopf, S., and Lovell, C. (1992b). Indirect Productivity Measurement, Journal of Productivity Analysis 2, 283-298.
Färe, R., Grosskopf, S., Norris, M., and Zhang, Z. (1994). Productivity Growth, Technical Progress, and Efficiency Change in Industrialized Countries. The American Economic Review, 84, 66-83.
Farrell, M. (1957). The measurement of productive efficiency, Journal of the Royal Statistical Society, 120, 253-290.
Fulginiti, L. and Perrin, R. (1997). Productivity in LDC Agriculture: Nonparametric Malmquist Measures. Journal of Development Economics, 53, 373-390.
Fulginiti, L. and Perrin, R. (1998). Agricultural productivity in developing countries. Journal of Agricultural Economics, 19, 45-51.
Hayami, Y. and Ruttan, V. (1970). Agricultural Productivity Differences among Countries.
American Economics Review 60, No. 5:895-911.
Hayami, Y. and Ruttan, V. (1985). Agricultural Development: An International Perspective (2nd ed.). Baltimore, United States: John Hopkins University Press.
Henderson, J., and Russell, R. (2005). Human capital and convergence: A production frontier approach. International Economic Review, 46, 1167-1205.
Hutchinson, D. and Langham, R. (1999). Productivity Growth in the Caribbean: A Measure of Key Components. Presented at the Annual meeting of the American Agricultural
Hutchinson, D. and Langham, R. (1999). Productivity Growth in the Caribbean: A Measure of Key Components. Presented at the Annual meeting of the American Agricultural