4.2. Agricultural Total Factor Productivity Growth and Its Decomposition
The agricultural TFP growth was decomposed into scale change (SC), technical change (TC), and technical efficiency change (TEC) according to the method described in
Section 3.1. The annual changes in TFP (TFPC) and its components are presented in
Table 5. Positive values of TFPC indicate improvements of TFP, whereas the negative values represent the deterioration of productivity performance.
We observed that agricultural TFP in south and southeast Asian countries on average decreased by 1.07% from 2003 to 2016. The changes in agricultural TFP were not stable over time. In some years we witnessed improvements in productivity. For other years, however, the agricultural TFP decreased on average. The average scale change was −2.92% which was the major source of the deterioration of agricultural productivity. In most years, the values of SC were always negative, indicating that agricultural TFP in general has not benefited from economies of scale. On the contrary, technical change on average improved by 0.93%, which implies that technical progress contributed to the improvement of agricultural productivity in south and southeast Asia. The value of TC remained positive from 2003 to 2012 with a decreasing trend, while it turned negative after 2012. This result reflects a slowdown of technical progress in the agricultural sectors of the sample countries. On the other hand, the technical efficiency change, with an average annual rate of −0.43%, also resulted in a decrease of TFP. The values of TEC remained negative every year, implying that there was a persistent increase in inefficiency. To summarize, technical progress is the only source that contributed to the growth of agricultural productivity, whereas declining scale and technical efficiency changes both resulted in the decline of agriculture TFP in south and southeast Asian countries.
Next, we disentangled the estimations of agricultural TFP and its components at region and country levels, to understand the region and country-specific characteristics of agricultural productivity growth.
Table 6 displays the changes in agricultural TFP and its components for the 15 south and southeast Asian countries. At the regional level, we found a negative average growth rate (−2.96%) in agricultural TFP for south Asia, indicating an unsustainable route of agricultural development. On the contrary, the agricultural TFP of southeast Asian countries on average increased by 1.08% during the sample period. The decrease in agricultural TFP in south Asian countries was largely due to the negative scale change (−3.06%). On the contrary, technical change, with an average growth rate of 0.54%, was the major source of agricultural productivity growth. As for southeast Asia, scale change and technical change, with average growth rates of 0.13% and 1.38% respectively, both contributed to agricultural TFP growth. Like south Asia, technical progress was also the major source driving the improvement of agricultural TFP in southeast Asian countries.
Regarding specific countries in south Asia, Bangladesh, India, Pakistan, and Nepal in general all experienced improvements in agricultural productivity during the sample period. This result is supported by Anik et al. [
6] who also found a positive growth rate for agricultural TFP for Bangladesh, India, and Pakistan during the period 2001–2013. The estimated growth rate of agricultural TFP for India (6.91%) was the highest among south Asian countries, followed by Pakistan (3.21%), Bangladesh (1.96%), and Nepal (0.62%). Technical change contributed most to the growth of agricultural TFP for all four countries. India plays a leading role in the economy of south Asian countries and is a major food producer and consumer. The high growth rate of agricultural TFP in India can not only contribute to maintaining the self-sufficiency of Indian agricultural production, but also help to meet the additional food demand of its neighboring countries.
On the contrary, we found negative growth rates of agricultural TFP in Afghanistan (−10.59%), Bhutan (−4.86%), and Iran (−20.88%). The sharp decline of scale efficiency was the major cause for the deterioration of agricultural productivity in both Afghanistan and Iran. In the past decades, the agriculture of Afghanistan and Iran has faced many challenges due to prolonged conflicts and natural disasters. The shortage of water, lack of irrigation land, and devastating damage to rural infrastructure have resulted in the loss of scale efficiency and largely held back agricultural productivity. Unlike Afghanistan and Iran, the decline of agricultural TFP in Bhutan resulted from the obvious reduction in technical efficiency (−5.64%). Minten and Dukpa [
53] also mentioned that the low level of technology led to low agricultural productivity in Bhutan. Therefore, promoting technical innovation and progress should be the primary task for Bhutan to achieve the sustainable development of agriculture.
As for southeast Asia, except for Cambodia, the other six southeast Asian countries all experienced improvements in agricultural productivity. The growth rate of agricultural TFP for Indonesia (3.80%) was the highest, followed by the Philippines (2.57%), Vietnam (2.08%), Thailand (0.90%), and Malaysia (0.75%). Like south Asia, technical progress made the dominant contribution to improving the agricultural TFP in these five countries. The growth rate of agricultural TFP in Myanmar was only 0.13%. Unlike the other five countries, the scale change (0.48%) largely contributed to agricultural TFP growth, while technical efficiency change was the major factor holding back the agricultural productivity of Myanmar.
Cambodia is the only country in southeast Asia that experienced an average drop of −2.67% in agricultural productivity during the sample period. The retrogression of technology in Cambodia was the main reason for the deterioration of agricultural productivity, a somewhat similar situation to that of Bhutan. In comparison with south Asian countries, southeast Asian countries overall showed a more sustained and stable growth in agricultural productivity, which is in accordance with the findings by Sujan et al. [
15] who concluded that the average agricultural production growth was higher in southeast Asia than south Asia.
To explain the changing characteristics of agricultural TFP for each country more intuitively, the annual TFP changes and their three components for south Asian and southeast Asian countries are plotted in
Figure 1;
Figure 2, respectively. As shown in
Figure 1, the agricultural TFP exhibited similar growth patterns in Bangladesh, India, and Pakistan. Each country observed a positive growth rate of agricultural TFP every year, with considerable contributions due to technical progress. The continued TFP growth indicated that the development of agricultural productivity in the three countries was relatively steady and robust. Technical change was rising continuously throughout the sample period, indicating continuous improvements in the technological innovation of the agricultural sectors in the three countries. Thus, we can conclude that the agricultural developments of Bangladesh, India, and Pakistan were at the forefront among the south Asian countries. However, it should be noted that the growth speed of agricultural TFP in the three countries was decreasing along with technical progress. As a result, how to provide enough growth momentum for technology innovation may have been the major challenge that the agriculture sectors in these three countries faced.
Unlike Bangladesh, India, and Pakistan, changes of agricultural TFP in Afghanistan and Iran fluctuated over time and were dominated by scale efficiency decline. Moreover, the growth patterns of agricultural TFP in Nepal and Sri Lanka were also not stable but showed some similarities. Both countries experienced technical progress at the beginning; however, the backward technology and reduced technical efficiency dragged down agricultural productivity from 2012. Moreover, Bhutan was the only south Asian country where agricultural TFP deteriorated every year. The technical change and technical efficiency change both declined each year, which had a negative impact on the agricultural productivity of Bhutan.
There were also several different growth patterns of agricultural productivity in southeast Asia (see
Figure 2). First, the growth of agricultural TFP tended to slow down in Indonesia, Myanmar, the Philippines, Thailand, and Vietnam. India exhibited the best practices regarding agricultural production in the southeast Asian region; the continuous technical progress was a great help to its sustainable agricultural development. The slowdown trends of agricultural productivity growth in Myanmar, the Philippines, Thailand, and Vietnam were more obvious over time. The agricultural TFP of Myanmar and Thailand started to decline from 2011, while the growth rate of agricultural TFP in the Philippines and Vietnam also turned negative in 2016. The persistent slowdown of technical progress was the common factor driving the deterioration of agricultural TFP in these countries.
Unlike the other southeast Asian countries, the agricultural TFP changes in Malaysia fluctuated, and this was mainly affected by scale change. A possible reason is that agriculture in Malaysia is still dominated by smallholders with an ageing farmer population, thereby causing diseconomies of scale in agricultural production [
54]. Consolidating efficient and professional management of small farms is crucial for Malaysia to promote scale benefits in agricultural production. Like Bhutan, Cambodia was the only southeast Asian country that experienced a continuous deterioration of agricultural TFP due to technical regression and efficiency decline. The technical regression of agriculture in Cambodia worsened during the sample period.
4.3. Factors Explaining Agricultural TFP Changes
In this subsection, we applied the FD-GMM estimator to dynamic data models to explain the determinants of agricultural TFP changes in south and southeast Asian countries. The estimated results of the SYS-GMM estimator are also provided as a robustness check. The presence of endogeneity could result in inconsistent estimated results, but the problem can be addressed by instrumental variables [
55,
56]. In dynamic panel data models, the FD-GMM and SYS-GMM estimators can deal with the problem of endogeneity by using instrumental variables [
50,
57]. The estimated parameters of the dynamic FD-GMM estimator and the results of model diagnostics are presented in
Table 7. The statistics from Sargan’s test rejected the null hypothesis that the over-identifying restrictions were valid, suggesting that the instrument variables as a group were exogenous, and that the choice of instruments was reasonable in our model. Meanwhile, the results of Arellano–Bond tests for zero autocorrelation in the first-difference disturbances showed that the disturbances were not serially correlated. The null hypothesis of the Wald test is that “the coefficients on the explanatory variables are jointly zero”, which was strongly rejected, implying that the inclusion of explanatory variables was reasonable.
The level of economic development, measured by GDP per capita, was significant and negatively correlated with agricultural TFP change, indicating that higher income levels were associated with a slower growth in agricultural TFP. Many south and southeast Asian countries were in a period of economic transformation with increasing shares of industry and services. The share of agriculture in total GDP was declining, but employment in agriculture still accounted for a large proportion of employment in these countries. This phenomenon implies that the agricultural productivity of these countries was not keeping pace with economic development. Consequently, south and southeast Asian countries experienced slower agricultural productivity growth despite continuous economic development.
Human capital positively contributed to the growth of agricultural productivity. Human capital can directly increase a worker’s productivity, thus contributing to improvements in agricultural TFP. This result is consistent with the findings by Lanzona [
58], Anik et al. [
6], and Zakaria et al. [
36], who also focused on the agricultural productivity of south Asian or southeast Asian countries and concluded that agricultural productivity increased with an increase in human capital. However, some studies found a negative role of human capital on agricultural productivity. For example, Rahman and Salim [
59] mentioned that human capital significantly contributed to technical progress but had a negative impact on technical efficiency, scale efficiency, and TFP growth in Bangladesh.
Regarding agricultural trade, we found a negative relationship between agricultural import and TFP changes. In theory, developing countries can learn advanced technologies and management experience through importing goods and services from developed countries, which is the so-called demonstration effect, thereby contributing to the improvement of productivity. Nevertheless, there may also be a loss of human and physical capital for local firms due to international competition, leading to the backwardness of agricultural productivity [
39]. The negative value of agricultural import in this study indicated that domestic agricultural products faced fierce competition from imported products but without the obvious benefits of the demonstration effect. On the other hand, agricultural export can help to achieve economies of scale by expanding markets overseas. Exporters also can improve their technological innovation capacities through experience from international competition [
31]. However, our result shows no significant effect of agricultural export on agricultural TFP, though it may exist theoretically.
Urbanization was positively associated with changes in agricultural productivity. With limited land resources, urbanization contributes to the transformation of the labor force from the countryside to urban areas, thereby promoting the rational reallocation of agricultural labor input. In addition, farmers in suburban areas can take advantage of the proximity to urban centers to reduce costs. Therefore, urbanization is helpful to improve efficiency and accelerate productivity growth. The positive impact of urbanization on agricultural productivity is also evident in work by Kumar et al. [
25] and Oueslati and Wu [
60]. However, it has been argued that urbanization may have adverse effects on agricultural productivity, because urban land expansion will lead to cropland loss, which has become a major concern in terms of food production and supply [
61]. The key issue is whether the growing demand for food and more convenient access to the market due to urbanization can maintain the sustainable development of agriculture given the decline of agricultural land [
62].
The development flow to agriculture positively contributed to the growth in agricultural TFP. Therefore, the development assistance to agricultural sectors was effective in improving the agricultural productivity of south and southeast Asian countries. Ssozi et al. [
41] also found a positive relationship between development assistance and agricultural productivity. The two time dummy variables were both significant and positive, which indicated that the earlier periods experienced a higher growth rate of agricultural productivity. This result reveals that the growth rate of agricultural TFP in south and southeast Asian countries declined significantly. However, the lagged changes in TFP showed no significant association with TFP changes in current periods.
The signs and significance level for the coefficients of most explanatory variables in the SYS-GMM model were consistent with the results in the FD-GMM model. Concretely speaking, human capital and development flow to agriculture were confirmed to have a significant positive influence on changes in agricultural TFP, whereas GDP per capita and agricultural import still showed significant negative signs in the SYS-GMM model. The results of a robustness test further verified the credibility of our findings for factors affecting changes in agricultural productivity in south and southeast Asia.