Adaptation Strategies under Climate Change for Sustainable Agricultural Productivity in Cambodia

Abstract

Since the Cambodian economy is largely dependent on agricultural production, it is important to understand the effects of climate change on rice production, the primary staple crop of Cambodia. This study assessed the economic impacts of climate change in Cambodia to provide an appropriate set of policy suggestions that could lead to sustainable agricultural productivity and economic growth. The results from the GLAM-Rice crop model and various climate models indicate that Cambodia will be severely affected by climate change, which will lead to lower rice production and economic growth. The changes in rice yield under the RCP 8.5 and RCP 4.5 baseline scenarios reduced the GDP by 8.16% and 10.57%, respectively. By employing an investment model based on a real options framework, the economic effects and feasibility of adaptation strategies such as irrigation and adjustment of planting dates are identified. The analysis indicates that irrigation is a feasible option and the most efficacious strategy to reduce the negative impacts of climate change for the agricultural sector. The index of economic feasibility for irrigation, defined by the ratio of the current realized agriculture value-added to the identified threshold, is 0.6343 and 0.8803 under the RCP 8.5 and RCP 4.5 baseline scenarios, respectively. The results suggest that the priority choice for adaptation measure be in order of irrigation, 20-day later adjustment, and 20-day earlier adjustment.

1. Introduction

There is growing evidence that unimpeded growth of greenhouse gas emissions is steadily increasing the global temperature; the average surface temperature of the Earth rose by 0.74 °C between 1906 and 2005 [1]. Increase in global temperature is a critical issue for many developing countries, whose economies have a higher dependence on the agricultural sector.

The fifth Climate Change assessment report by the Intergovernmental Panel on Climate Change (IPCC) asserts that the effects of climate change on crop and terrestrial food production are evident in several regions of the world [2] (p. 488). It further shows regional variations in the impacts of climate change, which are unevenly distributed across the world. Geographical and economic patterns show that less developed economies tend to be more vulnerable to climate change than developed ones. Several country-specific case studies have been performed to date, and they provide supporting evidence for this argument. Case studies included, for example, Bangladesh [3,4,5], China [6,7,8,9], India [10,11,12,13,14], Pakistan [15,16], Philippines [17], Sri Lanka [18], Thailand [19,20,21], Mexico [22], Cameroon [23], Ethiopia [24], Kenya [25], and Nigeria [26].

While positive trends are evident in some high-latitude regions, negative impacts of climate trends have been more common than positive ones [2]. Various studies also support the observation that climate change will negatively affect the agricultural sector in many developing countries that are highly dependent on agriculture for subsistence and/or exports. Lobell et al. [27] found that South Asia and southern Africa were two regions particularly vulnerable to climate change. They report that, without sufficient adaptation measures, southern Africa could lose more than 30% of its main crop, maize, by 2030. In South Asia, losses of many regional staples, such as rice, millet and maize, could top 10%. Schlenker and Lobell [28] also reported that, because of climate change, the aggregate decrease in yields in Sub-Saharan Africa will be 22% for maize, 17% for sorghum, 17% for millet, 18% for groundnut, and 8% for cassava by mid-century. Butt et al. [29] examined economic losses in Mali under the 2030 climate projections. They found that climate change would cause the number of people at risk of hunger to increase from 34% of the population to 64–70%. The World Bank [30] acknowledges that irrigated and rain-fed wheat and irrigated rice are particularly vulnerable to climate change and that South Asia will experience the largest loss in production. Moreover, the impact of climate change on agriculture in Asia is critical in terms of nutrition requirements. According to the Economics of Adaptation to Climate Change (EACC) group, which studies the comprehensive social costs of adapting to climate change, the number of malnourished children under age five could increase by approximately 22% by 2050, compared to a case in which climate change did not occur [30].

Note that the onset of effects of climate change on society is a relatively gradual process, whereas that of climate change on the agricultural sector may be both gradual and abrupt. For example, more than 300 million people living in delta areas (Mekong delta, Ganges delta, Brahmaputra delta, etc.) are expected to be vulnerable to climate change [31] (p. 25). The Dynamic Integrated Climate-Economy (DICE) model by W. Nordhaus posits that the costs of climate change will amount to 5–10% of the GDP for a five-degree increase. However, these costs may be substantially different between developing and developed countries. Since such costs of climate change are substantial, the development of better risk management practices is urgently required to cope with the adverse impacts of climate changes on the agricultural sector [2].

Cambodia is no exception to this because of the relatively high dependence of the agricultural sector in its economic activity. Recently, the Cambodian economy has built a remarkable path for economic progress with an unprecedentedly high economic growth rate of 10.02% from 1993 to 2011. Indeed, the agricultural sector has played a significant role in the nation’s economic growth, accounting for approximately 30% of its GDP. As reported by Acemoglu [32], improved productivity in agriculture sector is a key driving force for economic development in developing countries. This was further confirmed by Gollin et al. [33] who reported after studying 62 developing countries that the contribution of the agriculture sector to growth is 54% on average. Thus, in Cambodia, it is important to establish a sustainable agricultural sector from the very early stage of development for further economic growth in the future.

However, as noted by Thomas et al. [34], the Cambodian agricultural sector is characterized by a low adaptive capacity that relies on relatively labor-intensive technology. The problem of low adaptive capacity has increased because more than 85% of national rice production is dependent on rainfall, leaving the sector vulnerable to frequent drought and floods because of climate change [35]. Given the central role of agriculture in the Cambodian economy, it is necessary to understand the possible impacts of climate change on rice yields, which is the main staple crop for this region, and to assess its possible adaptation strategies for minimizing the adverse effects of climate change.

There is prolific literature on the issue of assessing climate change impacts on agriculture. According to Blanc and Reilly [36], the existing literature can be classified into three main categories: (i) panel data analysis; (ii) cross-sectional (or Ricardian) analysis; and (iii) agroeconomic analysis. This classification was identified and discussed in the recent symposium on “Estimating the Impacts of Climate Change on Agriculture”. The results of the symposium were published as four survey articles focusing on these approaches: Blanc and Reilly [36] (overview), Blanc and Schlenker [37] (panel data analysis), Mendelsohn and Masseti [38] (cross-sectional analysis), and Ante and Stöckle [39] (agronomic-economic analysis).

First, the panel data approach (e.g., [28,40,41,42]) uses statistical techniques to estimate the effect of weather on crop yields by estimating a production function. The estimation is based on panel data, which include observations of a cross section of individual units such as a field, farm, or country, over time [36] (p. 248).

Second, the cross-sectional (or Ricardian) approach (e.g., [43,44,45,46]), based on Ricardo’s [47] classic insight that land values intrinsically reflect land productivity, measures the impact of long-run climate (the distribution of weather over 30 years) on farmland productivity by regressing net revenue or farmland value on climate. Ricardian analysis is often performed across space, examining whether farms in different climates have similar or different net revenues or market values [38] (p. 281). As of 2017, cross-sectional studies have been estimated in 46 countries across five continents [37] (p. 280 and Appendix A, Table A1). Mendelsohn and Masseti [38] summarized Ricardian approach’s advantage as the following two features: (i) it is able to capture long-run adaptation to climate by describing what will happen once a farmer has had time to re-optimize; and (ii) it evaluates how climate changes affect net revenue (or land value), which is an explicit welfare measurement.

Finally, the agronomic-economic approach (e.g., [48,49,50,51]) uses a hybrid structural framework that combines process-based crop models (biophysical models) and livestock simulation models with farm-level economic models to estimate farmer’s potential adaptive responses to climate change [36] (p. 248). Ante and Stöckle [39] (p. 304) emphasized the importance of properly designed simulation experiments for economic models: climate impacts and the effects of adaptation must be assessed under plausible future biophysical and socio-economic conditions that are distinct from the conditions embodied in the historical data used to estimate econometric models.

In the context of the agronomic-economic approach, this study assessed agricultural adaptation strategies in Cambodia to accomplish sustainable rice yields under climate change and, consequently, sustainable agricultural/economic development. This article is in fact the first case-study in the relevant literature to focus on the Cambodian agricultural sector. Furthermore, the main stylized feature of this study that distinguishes it from others in the literature lies in its adoption of an investment model as a way of understanding farmers’ adaptation strategies for coping with adverse climate change impacts. As reported by Dixit and Pindyck [52], a real options model is a suitable approach to examine the economic feasibility of adaptation investment because the investment is largely irreversible and exposed to economic uncertainty. The proposed analytical model was empirically applied to Cambodia to consider the impact of climate change on crop production and GDP. To do so, various levels of field-scale crop modeling were adopted to estimate the productivity changes in rice production under Representative Concentration Pathway (RCP) 4.5 and 8.5 scenarios of IPCC. The GLAM-Rice crop model was used to further investigate spatial variations of climate change impacts and adaptation strategies effects on rice productivity in Cambodia based on results from Chun et al. [53]. Chun et al. [53] reported that Cambodia was identified as one of the most benefited countries from irrigation among Thailand, Vietnam, Myanmar, Laos, and Cambodia. Finally, this study evaluated two possible adaptation strategies to maintain rice productivity even under ongoing climate change: planting date adjustment and irrigation. Figure 1 briefly illustrates the structure and procedure of the analysis performed in this study. As mentioned above, the integration of an investment model based on real options analysis with various crop-climate models constitutes the main framework for the research. The results are expected to serve as a policy guideline by prioritizing the proposed adaptation strategies for sustainable economic development under climate change.

2. Methods and Data

2.1. Study Location and Data Collection

Cambodia is one of the countries located in the Indochinese Peninsula; Figure 2 illustrates the map of the study location. Cambodia is classified as having a tropical monsoon climate with distinct wet and dry seasons. Rice yields in the country are relatively lower than in its neighboring countries including Vietnam and Thailand and most rural household incomes in the country are low, leading to food insecurity. In recognition of the role that rice cultivation plays in income growth, poverty reduction, and national and household food security, the Royal Government of Cambodia has prioritized agriculture in all its major development strategies.

The daily weather inputs used for this study were collected from the Coordinated Regional climate Downscaling Experiment, East Asia (CORDEX-East Asia, http://cordex-ea.climate.go.Kr/main/mainPage.do). The regional scale crop models in the CORDEX-East Asia are summarized in

Table 1. Description of climate simulations used in this study.

GCM	Downscaling Models	Institute	Resolution
HadGEM2-AO *	YSU-RSM **	Yonsei University	0.44°
	RegCM4 ***	Kongju National University	0.44°
	HadGEM3-RA ****	National Institute of Meteorological Research	0.44°

* HadGEM2-AO: The Atmosphere–Ocean coupled Hadley Center Global Environmental Model version 2 from the National Institute of Meteorological Research (NIMR); ** YSU-RSM: Yonsei University, NCEP Regional Spectral Model (RSM); *** RegCM4: The Regional Climate Model version 4, developed by the International Centre for Theoretical Physics (ICTP); **** HadGEM3-RA: The regional version of the new seasonal prediction and prototype global climate model HadGEM3.

. A set of scenarios called Representative Concentration Pathways (RCPs) have been used to provide a range of possible scenarios of future atmospheric composition [54,55]. The RCP 4.5 and 8.5 scenarios were used for this study. The RCP 4.5 emission scenario represents a low to medium emission scenario, with stabilization from 2050 onward, and the RCP 8.5 emission scenario represents a high emission scenario, with emissions stabilizing after 2100 [56]. The national-level rice yields for Cambodia from 1991 to 2000 (i.e., the baseline period) were collected from the UN Food and Agriculture Organisation (FAO). Additionally, the FAO digital soil map of the world [57] was used for the dominant agricultural soil texture types. More detailed information on these data can be found in Li et al. [58].

2.2. GLAM-Rice Model

The GLAM-Rice model is based on the General Large-Area Model for annual crops (GLAM). The GLAM was originally developed to simulate crop yields at a regional level [59]. The GLAM-Rice model is a regional-scale crop model that was used to project the impact of climate change on rice yields in Cambodia. More detailed information on the GLAM-Rice model and the calibration and validation methods for the model can be found in Li et al. [58]. The model was forced by different climate change scenarios: the baseline (1991–2000) and 2030s (2021–2030) under two emission scenarios (typically, RCP 4.5 and RCP 8.5) from HadGEM3, YSU-RSM and RegCM4. In addition, more detailed model configurations can be found in Chun et al. [53]. Three adaptation strategies were considered in the analysis using the regional-scale crop model: earlier planting date, later planting date, and irrigation. Each of these adaptation options was analyzed under two different RCP scenarios: RCP 8.5 and RCP 4.5. Additionally, the CO₂ fertilization effect was also considered.

2.3. Investment Model: Real Options Framework

This subsection introduces an analytical framework to assess economic feasibility of adaptation strategies by incorporating the results of the crop-climate models to the real options investment model. The real options analysis, originated from option theory in financial engineering, has been developed remarkably since the memorial publication by Dixit and Pindyck [52]. It has been widely applied to various industries including information technology, energies, environment, afforestation, pharmaceuticals, health, real estates, social infrastructure, etc. It is now well established that the real options framework is particularly useful in analyzing decision-making of irreversible investment under uncertainty. As Mundlak [60] emphasized, the presence of uncertainty is a critical factor for a decision-making process in the agricultural sector. In addition, the adaptation investment for coping with adverse climate change impacts is largely irreversible. Therefore, real options analysis can be adopted as a suitable approach to examine the economic feasibility of adaptation investment under climate change.

The output is specified as a function of several input factors such as capital, labor, and agricultural value-added. Let $y$ denote the total output, which can be represented by GDP. Then, the following Cobb–Douglas type production function is considered:

$$y=\overline{y}{K}^{\alpha }{L}^{\beta }{\left({\theta }_{0}A\right)}^{\gamma }$$

(1)

where $K$, $L$, and $A$ denote capital, labor, and agriculture value-added, respectively, whereas $\overline{y}$ measures total factor productivity. ${\theta }_0$ denotes the reference value with ${\theta }_0=1$ prior to any adaptation investment. The parameters $\alpha$, $\beta$, and $\gamma$ measure how the amount of output responds to changes in the inputs ($0<\alpha ,\beta ,\gamma <1)$. It is well known that $\alpha$, $\beta$, and $\gamma$ represent the share of each production factor ($K$, $L$, and $A$) in GDP, respectively, for the case of Cobb–Douglas production function. This type of production function reflects agriculture’s role as an input for economic development, particularly during the initial stages of development [61,62].

In this study, adaptation strategies were considered as measures to reduce the possible adverse impact of climate change and reinforce the phase of economic development. Uncertainty was modeled for prioritizing the adaptation strategies. The conceptual background could be found in real options models, which provide a standard framework for analyzing the timing and extent of technological diffusion [52]. Ninh [63] similarly examined the relationship between uncertainty and investment to investigate investment in rice mills in Vietnam from the real options perspective.

Based on the real options framework, the equations that represent optimal investment threshold $A^{\ast }$ and the option constant term $S$ were developed for this study as follows (see Appendix A for the detailed process of deriving them):

$$A^{\ast }={\left(K\frac{\phi }{\phi -\gamma }\frac{\rho -\mu \gamma -0.5{\sigma }^2\gamma \left(\gamma -1\right)}{{\Pi }_1-{\Pi }_0}\right)}^{1/\gamma }$$

(2)

where ${\Pi }_i=\overline{y}{K}^{\alpha }{L}^{\beta }{\left({\theta }_i\right)}^{\gamma }$ and $\phi =\frac{{\sigma }^2-2\mu }{2{\sigma }^2}+\sqrt{{\left(\frac{{\sigma }^2-2\mu }{2{\sigma }^2}\right)}^2+\frac{2\rho }{{\sigma }^2}}$. $\rho$, $\mu$, and $\sigma$ denote the social discount rate, the drift parameter, and the volatility parameter, respectively. The option constant term is given by:

$$S=\left(\frac{{\Pi }_1-{\Pi }_0}{\rho -\mu \gamma -0.5{\sigma }^2\gamma \left(\gamma -1\right)}{A}^{\ast \gamma }-K\right){\left(A^{\ast }\right)}^{-\phi }$$

(3)

In Equation (2), the term $\phi /\left(\phi -\gamma \right)$ is an option multiplier, which increases as uncertainty grows. In the absence of uncertainty, the investment threshold is specified by

$$A^{\ast \ast }={\left(K\frac{\rho -\mu \gamma -0.5{\sigma }^2\gamma \left(\gamma -1\right)}{{\Pi }_1-{\Pi }_0}\right)}^{1/\gamma }$$

(4)

This is called the Marshallian investment threshold in the real options literature. Note that the option multiplier basically creates a wedge between $A^{\ast }$ and $A^{\ast \ast }$, indicating that uncertainty would delay further investment timing.

Furthermore, climate models provide predicted changes of rice yield under a baseline climate change scenario without any adaptation investment, as well as changes in rice yield when a planting dates adjustment or irrigation option is adopted. Based on this information, the optimal investment threshold $A^{\ast }$ can be numerically identified while considering the current state of agriculture value-added. In particular, to prioritize the proposed adaptation strategies, an economic feasibility index using $A/A^{\ast }$ is devised.

In the following empirical analysis, the parameters in the production function (Equation (1)) were estimated using the World Bank database (World Development Indicators and EconStats) to cover the period from 1993 to 2011. Data for GDP, gross capital formation, labor force, and agriculture value-added were compiled for the variables y, $K$, $L$ and $A$. Then, the GLAM-Rice model, together with climate change scenarios from various climate models such as HadGEM3, YSU-RSM, and RegCM4, was used to predict the changes in the average rice yield under climate change’s influence. A comprehensive impact analysis is provided to quantify the economic loss because of the resultant low productivity in the agricultural sector. The next section provides those results step by step.

3. Results

For this study, simulations of crop-climate models were conducted for predicted changes in the average rice yield. To perform a comprehensive impact analysis, the economic implications of future climate conditions and future rice yields were also studied. As mentioned previously, the production was defined to capture the total factor productivity. Reductions in rice yields because of climate change could undermine the potential for economic growth in Cambodia. Hence, adaptation investment was considered as a measure to reduce the possible adverse impacts of climate change on agriculture and reinforce the engine of economic growth.

Figure 3 shows the share changes in agriculture value-added in the Cambodian economy from 1993 to 2011. The agricultural sector in Cambodia accounts for almost 30% of GDP, mostly from rice cultivation. As observed in many developing countries, the role of agriculture in the economy has been steadily decreasing. However, the comparative advantage of agricultural sector in Cambodia is its vastly unused land and its abundant labor force. There is also the potential for diversification: in particular, into agribusiness. In this regard, the sustainable development of the agricultural sector under climate change is a challenging strategy.

Table 2. Estimates of the Cobb–Douglas function (No. obs. = 19).

Coefficients	Estimates *	t-Stat
$\overline{y}$	0.4140 (0.3774)	1.0969
$\alpha$	0.1022 (0.0877)	2.0280
$\beta$	0.5350 (0.2371)	3.9291
$\gamma$	0.3628 (0.0805)	7.8492
$R^2$	0.9928
Durbin–Watson	1.1057

* The value inside parenthesis denotes the standard error.

lists the results of estimating the parameters in the production function (Equation (1)). Then, the drift and volatility parameters were estimated by specifying a log-normal process $x_t=ln{A}_t-ln{A}_{t-1}$, where $A_t$ is the agriculture value-added. As $A_t$ follows a geometric Brownian motion given by Equation (A1) in Appendix A, the mean $\overline{x}$ and variance $s$ of $x_t$ were estimated. The corresponding volatility and drift rates were calibrated by $\widehat{\sigma }=s/\Delta t=0.1184$ and $\widehat{\mu }=\overline{x}/\Delta t+0.5{\sigma }^2=0.0770$ [64].

In this study, two different Representative Carbon Path (RCP) scenarios, RCP 8.5 and RCP 4.5, using the GLAM-Rice model were analyzed and compared. Under the RCP 8.5 baseline scenario (without the CO₂ fertilization effect), the predicted changes in the average rice yield were −14.1%, −12.2%, and −11.5%, based on the HadGEM3, YSU-RSM, and RegCM4 climate models, respectively. Thus, the average value of predicted changes in rice yield was −12.6% (See

Table 3. Predicted changes in GDP based on yield changes under two RCP scenarios.

	GDP Changes * (RCP 4.5)	GDP Changes * (RCP 8.5)	Scenario Description
Baseline	−10.57% (−16.2%)	−8.16% (−12.6%)	- Average of simulations based on the HadGEM3, YSU-RSM and RegCM4 climate models - Does not consider the CO2 fertilization effect
CO2 effect	−7.76% (−12.0%)	−4.03% (−6.3%)	- CO2 fertilization effect is considered along with temperature increase
−20	−0.82% (−1.3%)	−2.48% (−3.9%)	- Planting date is adjusted to 20 days earlier relative to the baseline
20	0.82% (1.3%)	1.32% (2.1%)	- Planting date is adjusted to 20 days later relative to the baseline
Irrigation	7.00% (11.3%)	6.82% (11.0%)	- The implication of irrigation is added to the baseline

* The percentage value inside parenthesis denotes the average yield change relative to the baseline.

). The predicted changes in the average rice yield under the RCP 4.5 baseline scenario (without the CO₂ fertilization effect) were −20.0%, −14.4%, and −14.1% for the HadGEM3, YSU-RSM, and RegCM4 models, respectively. Therefore, the average value of the predicted changes in rice yield from the three models was −16.2%, that is, the rice yield in Cambodia can be expected to decrease by 16.2% from 2021 to 2030 because of climate change. The incorporation of the CO₂ fertilization effect into the model ameliorated the effects of climate change on rice yields, by −6.3% and −12.0%, respectively, under the RCP 8.5 and RCP 4.5 scenarios.

Figure 4 shows a regional variation of climate change impacts in Cambodia. Irrigation was identified as the most efficacious of the adaptation options (irrigation and shifting planting dates). Without adaptions, rice yields for the 2020s under the RCP 4.5 and RCP 8.5 scenarios were projected to decrease by 10%–20% (Figure 4a,b, respectively) for most of rice-cultivated areas in Cambodia. For the shifting planting dates, the projected differences of changes in rice yields were less in the 2020s under both climate change scenarios. While changing the planting dates to 20 days earlier to the original planting dates for the period from 1991 to 2000 would slightly decrease rice yields (up to 7%), shifting the planting dates to 20 days later would result in a 5%–10% increase in rice yields (Figure 4e–h). Note that irrigation had a large beneficial impact (about 5%–30%) on rice yields for most of the rice-cultivated areas (Figure 4c,d). The projected rice yields with irrigation would increase by 15%–25% in the northwest of Cambodia and increase in rice yields would even exceed 25% in some areas.

With planting date adjustment as an adaptation strategy, the impact of climate change was less significant on the yield: −3.9% and −1.3% for the RCP 8.5 and RCP 4.5 scenarios, respectively, when the planting date was adjusted to 20 days earlier than the baseline. Note that adjusting the planting date to 20 days later switched the impacts of climate change from negative to positive ones with increases of 2.1% and 1.3% for the RPC 8.5 and RCP 4.5 scenarios, respectively. The adoption of irrigation led to even better results with dramatic increases of 11.0% and 11.3% under the RCP 8.5 and RCP 4.5 scenarios, respectively.

Table 3 lists the simulated changes in GDP based on the proposed method in this study. The changes in rice yield under the RCP 8.5 and RCP 4.5 baseline scenarios reduced the GDP by 8.16% and 10.57%, respectively. Such significant impacts on GDP are because of the heavy reliance of the Cambodian economy on the agriculture sector, particularly rice farming. CO₂ fertilization effects and the adoption of adaptation options could mitigate the negative impacts of climate change on rice yield, and consequentially reduce the GDP loss. For example, in the best-case scenario, adoption of irrigation in rice production could increase the GDP by 6.82% and 7.0% for the RCP 8.5 and RCP 4.5 scenarios. Even without the addition of any adaptation activities, the CO₂ fertilization effect would mitigate the negative impacts of climate change on rice yields, and therefore GDP by 4.13%p and 2.81%p under the RCP 8.5 and RCP 4.5 scenarios, respectively. Overall, these results indicate the critical role that irrigation could play in Cambodian agricultural production. Currently, the irrigation system in Cambodia is still far from optimal; the vast majority of farming is rain-fed, making rice paddies much more vulnerable to climate change.

The above results do not reflect the possibility of farmers responding to price fluctuations and adjusting their production. Yu and Fan [65] reported that price elasticities of rice supply in the wet season are 0.11–0.26 and 0.92–1.15 for the short-run and long-run, respectively. The dry season elasticities are relatively higher at 0.26–0.33 and 1.19–1.45 for the short-run and long-run, respectively. When the model incorporates such production adjustments in response to price changes triggered by climate change impacts, the aforementioned negative effects of climate change on GDP may be slightly lower.

For the next stage of analysis, considering the adaptation investment costs and the uncertainty in the agriculture values, economic feasibility study was conducted to identify effective adaptation measures. For this purpose, the real options model developed in the previous section and Appendix A was used to identify the threshold of agricultural value-added at which it was optimal to perform an adaptation investment. Since there are no conclusive reports yet that quantified the adaptation costs of shifting planting dates and investing in irrigation per hectare in Cambodia, it was assumed that the costs of planting date adjustment and irrigation are 0.001% and 0.002% of GDP, respectively. A recently initiated ADB project for funding a water resource management program covers about 15,000 hectares of small- and medium-sized irrigation schemes in Kampong Thom province, Cambodia. The total budget is $12.8 million, including a grant of $2.8 million from the Asian Development Fund, which breaks down to about $850 per hectare (http://www.adb.org/news/adb-help-cambodia-improve-water-management-irrigation-systems). Further study is needed to identify what the total costs of investment in irrigation as an adaptation measure will be.

To ensure optimal investment timing, the index for economic feasibility is defined as A/A*, which is the ratio of the current realized agriculture value-added to the identified threshold. A value of A/A* that is lower than 1 indicates that it is optimal to immediately invest in the adaptation option because the identified threshold exceeds the current observed value.

Table 4. Economic feasibility of adaptation investments (A/A*).

	Measures	RCP 4.5	RCP 8.5
Baseline	−20	0.5411	1.2469
	+20	0.4223	0.5525
	Irrigation	0.6343	0.8803
CO2 effect	−20	0.9036	2.4040
	+20	0.6449	1.3170
	Irrigation	0.5176	0.8166

lists the results of the economic feasibility analysis considering three selected adaptation options. It is clear to see that every adaptation measure is an economically feasible choice under RCP 4.5 scenario, whereas, under the RCP 8.5 scenario, the 20-day earlier (−20) planting adjustment was not included in the option set. Interestingly, investment in irrigation and 20-day later (+20) planting date are shown to be economically feasible under both scenarios at baseline environment. When considering the CO₂ fertilization effects, the economically feasible adaption option allows only irrigation investment under the RCP 8.5 scenario and the planting date adjustment is no longer viable. This is because the positive role of the CO₂ fertilization effect on rice yields reduces the relative value of adaptation investment. However, similar to the baseline environment, RCP 4.5 ensures every adaptation measure to be feasible. The results suggest that the priority choice for adaptation measure be in order of irrigation, 20-day later adjustment, and 20-day earlier adjustment.

Figure 5 illustrates the sensitive analysis of optimal investment threshold $A^{\ast }$ with respect to volatility ($\sigma$) and adaptation costs. The result clearly shows that the level of optimal investment threshold is increasing in both the degree of volatility and adaptation costs. The higher $A^{\ast }$ implies the adaptation investment time is delayed, as the first-hitting time defined by ${\tau }^{\ast }=\inf \{t|A\le A^{\ast }\}$ becomes longer. This corresponds to the concept of “hysteresis” in the real options literature.

4. Discussion

By integrating an investment model based on real options analysis with various crop–climate models, this study could analyze the economic feasibility of adaptation strategies for coping with adverse climate change impacts. The main results indicate that: (i) Cambodia, similar to many other developing economies, will be severely affected by climate change, which will threaten food security and sustainable economic growth; and (ii) irrigation is identified as a feasible option and the most efficacious adaptation strategy to reduce the negative impacts of climate change on rice production in Cambodia. The positive effects of irrigation could be further enhanced when coupled with the adjustment option of planting dates. Two types of adaptation strategies considered in this study, irrigation and planting dates adjustment, have been generally accepted in the relevant literature as effective ways of mitigating adverse impacts of climate change for the agricultural sector. The fifth Climate Change assessment report by IPCC notes that the average benefit of agronomic adaptation improves yields by the equivalent of about 15%–18% of current yields. However, the effectiveness of adaptation is highly variable, with temperate wheat and tropical rice showing greater benefits of adaptation [2] (p. 515). The main finding of the previous section is consistent with the general observation summarized by IPCC. Moreover, by deriving explicitly the index of economic feasibility (A/A*) for each adaptation strategy, this study could provide solid economic reasoning to confirm such a conventional wisdom in the literature from the perspective of real options analysis. The superiority of irrigation among feasible adaptation options has also been reported by other country-specific case studies including China [66], India [13], Nepal [67], Thailand [21], etc.

To maximize policy performance with a limited time and budget, it is necessary to focus on a selective set of adaptation strategies. Asian Development Bank [68] and Hallegatte [69] suggest the following strategic approaches for adaptation: (i) no-regret strategies that will yield benefits even in the absence of climate change; (ii) reversible and flexible strategies that require little capital investment suited for annual or periodic review such as insurance programs, changing planting dates, or varieties; (iii) strategies to reduce vulnerability at low cost such as raising existing dikes to cope with future rising sea levels; (iv) reducing decision-making time horizons by phasing in shorter-term investments such as small-scale irrigation systems that use groundwater or rainwater (as opposed to building irrigation dams); and (v) enhancing synergies among strategies, for example, to promote climate change mitigation or poverty reduction while adapting to climate change.

In this regard, the proposed adaptation options in this study, that is, small-scale irrigation and planting dates adjustment, are effective and viable approaches to overcoming adverse impacts of climate change. Hence, it is necessary to build and enhance adaptive capacity in Cambodia using these adaptation options. Advanced water resources management as well as expansion of irrigation should be one of the first steps. Currently, only 15% of rice fields are irrigated while the majority of farming is still rain-fed. However, almost 40% of the country’s rice production comes from these irrigated fields. This clearly indicates that the development of irrigation infrastructure to cover a greater portion of Cambodian rice fields would enhance national agricultural productivity as well as the adaptive capacity to respond to changes in weather conditions because of climate change.

As for the planting dates options, it is necessary to provide more comprehensive agricultural extension services for farmers so that they can actively participate in climate change adaptation. Thomas et al. [33] conducted a survey of 45 communes in Cambodia to identify agricultural practices and responses to climate change. Questions were asked about fertilizer application, irrigation, seeds, and tillage and pest management. The survey results show somewhat low levels of changes in agricultural practices in response to extreme weather, with only 20% of farmers changing planting dates and only 7%–16% of farmers changing crop varieties. Since the above analysis indicates that adjusting planting dates is advisable to reduce the scale of possible yield losses from climate change, it is highly recommended that farmers’ adaptation capacity and their awareness of climate change impacts be expanded by focusing on relatively simple adaptation strategies.

5. Conclusions

Maintaining agricultural productivity is a critical prerequisite for achieving sustainable economic growth in developing countries [70]. Since gains in productivity in the agricultural sector are necessary to move the economy to a further advanced state, climate change adaptation strategies for the agriculture sector should focus on improving agricultural productivity for continued economic development.

It is widely acknowledged that Cambodia has achieved rapid economic growth. The country’s economy substantially depends on the agricultural sector, which accounts for about 1/3 of the GDP and more than 50% of the country’s total labor force. Hence, the recent rapid economic growth in Cambodia can be explained in part by the remarkable performance of the agricultural sector. In particular, rice is the main staple crop in Cambodia, occupying more than 80% of the cultivated land and playing an important role as the largest export commodity. The present challenge for the country is to become a middle-income economy over the next decades, while confronting climate change. To achieve this major structural transition to a middle-income economy, the development of the agriculture sector is expected to be one of four key sectors (along with garments, tourism, and construction) to fuel economic growth.

Given that rice production will continue to play an important role in the course of economic development in Cambodia and the fact that its productivity may be vulnerable to climate change, it is a daunting yet necessary task to propose a set of possible adaptation strategies that are most suitable to the Cambodian context. Therefore, this study attempts to assess the economic impacts of climate change on the Cambodian agricultural sector in order to provide an appropriate set of policy suggestions for sustainable economic growth. Cambodia is already severely affected by climate variability that caused recurrent floods and droughts.

The findings in this paper suggest that the Cambodian government should consider adaptation investment strategies as soon as possible to reduce the negative impacts of climate change on agricultural productivity. The economic feasibility assessment indicates that investment in irrigation offers the greatest benefits while the planting date adjustment is the next option. Policy recommendations for climate change adaptation include investment in infrastructure, such as irrigation, transportation, advanced milling facilities, and energy, as well as improved education through agricultural extension services.

Although the economic feasibility of various adaptation strategies is explored, this paper does not provide a comprehensive discussion on many other key adaptation and policy measures that should be subjects for future research (e.g., a government subsidy system for rice production, irrigation policies, promotion of exportation of rice crops, management of arable land, best practice management for fertilizer usage, rural poverty and income disparities, nutritional problems, price competitiveness, micro-finance for rural households, development of high-quality rice varieties, possibility of crop switching, yield-increase vs. quality-increase in production, and the feasibility of investment in R&D for climate-change resistant crops). Other related issues of designing sustainable agricultural system (e.g., soil management and tillage system that reduce the grey water footprint without compromising the yield [71]) should be investigated further. It is concluded that this study can be useful to prioritize adaptation measures with limited resources by providing socio-economic effects of those measures.