**1. Introduction**

Climate change (CC) is a global phenomenon that is harming climate-dependent activity such as agricultural production. The negative impacts of CC on agriculture, both for crop and animal production, have been well documented [1–4]. The degree of adverse impacts of CC on farmers is determined by the vulnerability of those farmers to CC. Vulnerability is the propensity or predisposition of a natural or human system to be adversely a ffected by CC, and encompasses a variety of concepts and elements, including sensitivity or susceptibility and lack of capacity to cope and adapt [5]. Vulnerability, then, is a function of three aspects: Exposure to hazard, sensitivity to damage, and ability to cope. Currently, there are no means to control the occurrence of natural hazards; increasing the system's adaptive capacity can reduce its sensitivity to damage caused by natural hazards [6]. Based on that definition, the implementation of an appropriate adaptation strategy can minimize a farmer's vulnerability.

An appropriate adaptation strategy is required to moderate the adverse e ffects of CC [7]. There are two types of adaptation: Autonomous and planned adaptation [8]. In the former, farmers independently adapt their farming practices to the observed climatic change. In the latter, the governmen<sup>t</sup> plans and implements an adaptation policy. Several studies have reported that farmers in developing countries have adjusted their farming practices in response to CC and found that the adaptation has a positive effect on crop yield [9,10]. However, several barriers limit adaptation practices, such as financial barriers (lack of financial resources and/or lack of supporting institutions, whether public or private, to finance adaptation), social and cultural barriers (individuals and group perspectives, values, and beliefs toward CC), and informational and cognitive barriers (individual perceptions, values, and opinions about the risk of CC) [11]. This study focuses on the third barrier, and specifically individual climate risk perception.

A farmer's perception of climate risk is essential because it represents the degree of perceived impact (P-I)—a measure of how a farmer personally feels about the impact of a particular occurrence [12,13]. Past exposure to climate-related disaster increases the degree of P-I, which in turn drives farmers to undertake adaptation actions [14,15]. While some studies stressed the benefits of autonomous adaptation, other studies reported that it ultimately results in unintended maladaptive outcomes, such as increasing the farmer's vulnerability to CC, shifting the vulnerability to other stakeholders or sectors, and decreasing the quality of common pooled resources [16–19]. Thus, assessing a farmer's P-I toward CC is essential in two aspects: First, it provides valuable information about the e fforts to encourage autonomous adaptation; second, it provides crucial insight into the e ffort to avoid maladaptation practices. As most developing countries have a national adaptation policy [20], this study contributes to addressing the question of which farmers should be prioritized and through what channel the content of a policy should be delivered. Figure 1 shows how climate risk perception is related to autonomous adaptation and adaptation outcome.

**Figure 1.** The relationship between climate change risk perception, autonomous adaptation, and adaptation outcome.

While CC is a global issue that impacts farmers in both developed and developing countries, it causes more severe damage to those in developing countries [12,21], since the majority of these

farmers are poor and less adaptable [22]. These farmers are often located in less favorable agricultural areas, which increases their vulnerability [23]. The availability of adaptation instruments, whether of a physical nature, such as stress-tolerant crops, or of an institutional nature, such as crop insurance, is limited compared to those in developed countries [24]. Farmers, especially in developing countries, treat CC as a threat to their farming only when they perceive that it significantly decreases crop yield [25,26]. For example, the farmer who experiences crop yield loss due to CC may feel that the impact is not significant because the loss is relatively small compared to their wealth, or due to other reasons, and it is less likely that they will adopt any adaptation strategy. Conversely, a farmer who perceives that the impact of CC is high is more likely to adjust their farming practices to minimize the effect. Based on this reasoning, a farmer's P-I of CC influences their likelihood of adopting an adaptation strategy. The higher the P-I, the more likely the farmer is to adapt, and vice versa. Since the degree of P-I is affected by many factors, including technical, social, economic, or institutional factors [27–29], identifying factors influencing a farmer's P-I of CC is essential. This information will be useful to determine which farmers should be prioritized and targeted.

Extensive studies on the effect of CC on agriculture in developing countries have been conducted. While CC affects almost all aspects of a developing country's economy, agriculture is the most substantially impacted [30,31]. However, few studies have analyzed the effects of CC using nationally representative farm data [32–34]. Nationally representative studies on the impact of CC on agriculture in developing countries often use aggregate data [21,35–38]; other times, individual farm data are employed, and the study is limited to local and community levels [9,39,40]. Whereas the former is useful in providing an overall view of the CC effect and the latter is useful in informing a detailed picture of how a farmer is affected and manages to adapt, it is necessary to conduct a nationally representative study on how farmers are affected by CC. Most of these studies assessed the actual impact of CC on agriculture, which can be different from the farmer perception.

The role of climate risk perception in CC adaptation has received considerable attention. A study in Bangladesh showed that farmer perceptions of CC are mostly aligned with observed meteorological data and are correlated positively with the rate of adopted adaptation practices [15]. Similarly, a study in French coastal populations showed that they perceive the local changes in climate, weather, coral, and beaches, but they only regard it as a problem instead of a danger [41]. In contrast to the result from France, a study on the peri-urban community in Mexico that experiences a risk of drought indicated that the community perceives CC and treats it as a threat because their livelihood as brick producers is severely impacted by climate change [42]. A study on Canadian bivalve aquaculture indicated the importance of stakeholder perceptions of CC in adapting to these changes and further expanding the industry [43]. A cross-country analysis in Europe indicated that the perception of CC is affected by individual-level factors such as gender, age, political orientation, and education, but the size of the effects of each variable varies across countries [44].

The general purpose of this study was to identify the determinants of farmers' P-I of CC. Specifically, we aimed to determine what factors, economic, social, technical, or institutional, affect whether farmers perceive CC as a threat to their farming. The contribution of this study is two-fold. First, this study provides nationally representative information on the farm-level impact of CC in developing countries. While numerous efforts to study the effect of CC on developing countries have been made, few studies have used nationally representative data. Second, we used a P-I measure of CC (whether a farmer thinks they are not, less, or highly affected by CC), which was found to significantly drive adaptation practices. This information is relevant to Indonesia, where a national adaptation policy is being implemented. This study contributes to the growing discussion of climate risk perception.

#### **2. Background: Climate Change and Its Impact on Indonesian Agriculture**

Indonesia is the largest archipelagic country in the world with over 17,000 islands, with a total land area of 190 million ha. Indonesia had 37 million ha of agricultural land in 2018, and rice fields accounted for 8 million ha, of which 58.13% had irrigation infrastructure [45]. The rice fields are spread out over the major island in Indonesia, but most of them (3.1 million ha) are located on the island of Java [45]. Rice is a vital food crop in Indonesia along with maize, soybean, green bean, cassava, and sweet potato. The annual production of rice in Indonesia is 81 million tons of dry unhusked rice. Beside food crops, Indonesian agriculture includes horticulture: Vegetables (shallot, chili, cayenne, garlic, and potato), fruits (mango, banana, citrus, durian, pomelo, and mangosteen), and flowers (chrysanthemum, rose, orchid, and tuberose). Additionally, Indonesia produces plantation crops, medical crops, and livestock.

As in many developing countries, agriculture plays a vital role in the Indonesian economy. Its production accounts for 13% of the Indonesian gross domestic product (GDP) [46], and it provides a livelihood for 25 million farm households [45]. Among various crops cultivated in Indonesia, rice is the primary food crop. Of the 25 million farm households, 17 million are rice farmers with an average land possession of 0.6 ha [47]. As a smallholder farmer, the rice farmer is economically more vulnerable to external shocks, such as those due to climatic change. In Indonesia, rice farming has been severely impacted by climate change. The changing rainfall frequency and intensity, the increase in temperature, and the rise in sea level have significantly contributed to declining rice productivity. The frequency of occurrence of extreme events, such as flood and drought, is increasing, which causes crop loss. Increasing temperature causes the proliferation of pests and diseases [48]. Thus, attempts to mitigate and adapt to the risk of CC is required for the resilience of rice farming in Indonesia.

Resilience toward CC is defined as the ability to withstand climatic shocks and risks [49]. Forsyth [50] provides a comprehensive review of the definition of CC resilience. The early definition stated that CC resilience is related mainly to the physical properties, such as infrastructure and ecosystems, and its stability during an occurrence of shocks. The definition is then improved to include not only physical properties but also socio-economic factors such as diverse access to sources of livelihood. Finally, the definition of resilience is brought into a broader context of wider social processes and transformation. For the sake of clarity, this paper defines a farmer's resilience toward CC as their ability to withstand and minimize the adverse impact of CC on their farming.

The study of CC in Indonesia has been extensive and covers a wide range of aspects such as agriculture [48,51–55], natural disasters and managemen<sup>t</sup> of coastal areas [56–58], the politics of climate change [59–61], and the public perspective on climate change [62,63]. These studies stated that climate change will a ffect many aspects of the Indonesian economy, and agriculture will be the hardest hit. In response to this, the governmen<sup>t</sup> created a National Action Plan (NAP) to mitigate the risk of and to adapt to climate change. One of the primary targets of the NAP is to achieve economic resilience by achieving food security. The primary objective in achieving food security is to reduce production loss due to extreme climatic events and CC [64]. The primary strategy to reduce farm production loss is applying a CC-resilient farming system and CC-adaptive farm technology. The governmen<sup>t</sup> should ensure that the farmer adopts both of these strategy. However, limited governmen<sup>t</sup> resources and the large number of farmers will limit the adoption rate of this strategy. Thus, it is essential to identify farmers with a high probability to change and adapt, and to target the implementation of the NAP to these farmers.
