2022 Volume 58 Issue 2 Pages 75-81
This study examines the qualitative characteristics of the substantial community-based master plan using a text-mining analysis based on free descriptions of master plans in the Sanin region of the Chugoku area in Japan. We obtained the following results: First, community activities play an essential role in formulating a community-based master plan. Second, the relevance of frequent words varies with the scale of farmland preservation in a district. In areas with unfavorable farming conditions, the measures differ in each district because of the different farmland conditions, farm types, and conditions for accepting new farmers. In addition, local resources, farmland liquidation, and agricultural infrastructure significantly influence paddy field agriculture in hilly and mountainous areas. Rural revitalization measures in each region should consider the characteristics of the region so that the implementation of policy responses is based on residents’ awareness.
The depopulation, aging, and shortage of successors for agricultural cultivators in Japan are becoming more serious. The trend of declining farmland in recent decades has not stopped. In 2012, the Japanese government launched an initiative to prepare a “Community-based master plan” to address each region’s “people and farmland issues.” The master plan intends to identify the main agricultural entities that will play a central role in the region and promote the farmland accumulation in these main agricultural entities through discussions in communities and regions.
The government has provided a template for the master plan, which requires residents and municipal staff to fill in the plan according to the actual situation based on in-deep discussions. The implementation of the master plan depends on collective action of rural communities (Takahashi, 2018). If local governments adopt the master plan, farmers and communities can receive subsidies. The master plan was legally positioned to smoothly promote the Farmland Intermediary Management Project when the Farmland Intermediary Management Institution (Nochi Chukan Kanri Kiko) was established in 2014.
At present, more than 90% of municipalities have prepared approximately 15,000 master plans, but some of them are not based on local discussions. For example, half of the plans do not include any information on who will provide the farmland. Moreover, some plans are not based on local discussions (MAFF, 2021).
Therefore, in 2019, the local governments, agricultural committees, and other related parties in the region were asked to participate in the planning process by creating a substantial community-based master plan based on discussions. The word “substantial” (Jisshitsuka sareta) means that a considerable part of farmland in the target district (community level) has a concrete prospect of continuous utilization in the next five to ten years.
Toyoda et al. (2020) showed a pathway to conduct practical discussions to make a substantial community-based plan. Tomita et al. (2013) proposed building an attractive area to promote this program. Hamashima (2018) and Ando (2021) showed that the enthusiasm of residents for community-based farming, and the efforts made to build consensus in the community led to the program of community-based master plan success. Hosoyama (2014) and Okamura et al. (2021) found that farmland scale influences the farmland accumulation ability. Yasunaga (2021) showed that agricultural fields led to regional differences in farmland liquidation.
However, few studies have focused on arable land scale and agricultural field types related to community-based master plans. Further, few studies have used text-mining analysis on this plan to understand the thoughts of residents and strategies at the rural community level.
Each community has different history, culture, geographical conditions, and development path. Therefore, implementing policies based on community characteristics is a fundamental and effective method of development. Hence, this study aims to examine the qualitative characteristics of collective action and regional conditions of a community that formulated a substantial community-based master plan by using the text-mining method. It will help and support communities, which did not formulate a substantial community-based master plan. This issue needs urgent address for farmland conservation and sustainable agricultural development.
The target area of this study is Sanin region in the Chugoku area, a remote and less favorable area located in western Japan. Here are typical hilly and mountainous areas with high farmland slope, and small and scattered plots that make scaling up more challenging than flat areas. The Ministry of Agriculture, Forestry, and Fisheries (MAFF) has classified 14 agricultural regions based on similarities in agricultural management, including Shimane and Tottori prefectures of the Sanin region.
1) Farmland conditions and community characteristics: in Shimane Prefecture, hilly and mountainous areas cover ninety percent of land area, which is higher than the national average (73.7%). Its arable land area was 36,400ha (2020), out of which 80.8% was paddy fields. National average for the same is 54.4%. Therefore, rice cultivation is a characteristic of Shimane and “Nita rice” is a famous local rice brand. The number of community-based farming organizations was 539, of which the corporate form accounted for 47%. In the Tottori Prefecture, 63% of area is hilly and mountainous, lower than the national average. The arable land area was 34,300ha. 68% of the total arable land was the paddy fields. There were 312 community-based farming organizations in the region, of which the corporate form accounted for 27%.
2) Agricultural production: According to the 2019 agricultural production income statistics of the MAFF, cultivated farming output of Shimane was 35.9 billion Yen, rice output accounted for 54%, and vegetable and fruit output accounted for 37% of the cultivated farming output. In the Tottori, the arable area was 6% smaller than Shimane. However, the cultivated farming output was 32% higher than that of Shimane. Vegetable and fruit, and rice output accounted for 59% and 31% of the cultivated farming output, respectively. Tottori is famous for its pears and persimmons.
3) Policy characteristics: Shimane and Tottori plan to grow the agricultural outputs by 16% and 18% respectively, by 2024. To increase the main agricultural practitioners, Shimane intends to develop more than 60 certified new farmers per year. Tottori plans to increase the number of new farmers by 200 per year. In terms of collective activities, Shimane plans to achieve an implementation rate of more than 60% of the diversified management of paddy fields by community-based farming corporations. The number of communities participating in collective activities, such as direct payments in hilly and mountainous areas, will increase by more than 30 each year. Tottori is focused on improving the number of community-based farming corporations and plans to grow by a factor of 1.9 over ten years.
(2) HypothesesBased on previous studies findings that are mentioned above, this study examined the following hypotheses:
Hypothesis 1: Community activity is an important factor in tackling a community-based master plan.
This study defined community activity as farming activities in the community, which include collective activities and any farming organizations, such as the Agricultural Public Corporation and Community-based Farming Organization (Corporation).
Hypothesis 2: Different arable land scale reveals a different way to tackle the community-based master plan.
(3) MethodologyThis study investigated the free description patterns of residents from substantial community-based master plans to identify the main issues in rural communities. The total number of substantial community-based master plans is not publicly available. As of August 10, 2021, 139 plans in the Shimane and 73 plans in the Tottori were available on municipality websites. In this study, we used the all available master plans in the Sanin region. The sample number of the Tottori was quite different from that of the Shimane sample. Although 73 is not a large sample size, it was sufficient for statistical analysis.
The government divided this master plan into mandatory items, including community problems, farmland accumulation strategies, and optional items, including future vision strategies. Our target content was the response of farmland accumulation strategy. Table 1 presents the completion of each question.
Text-mining techniques may have application within the agricultural decision process because they identify information that may not be readily available in non-textual sources (Drury, 2019). The free descriptions in the substantial community-based master plan reflect the thoughts of local community and local residents on farmland conservation. The text-mining method is characterized by its ability to quantitatively grasp residents’ thoughts, current status, and future prospects at the community level. Therefore, in this study, we used the text-mining method to clarify the relationship and importance of our eight themes in farmland conservation in local agriculture from local residents’ free descriptions.
| Text Content | Number of Plan | ||
|---|---|---|---|
| Shimane | Tottori | Total | |
| Community problem | 139 | 72 | 211 |
| Farmland accumulation strategy | 139 | 73 | 212 |
| Future vision | 71 | 58 | 129 |
1) The difference in the number of plans is due to the lack of content.
For this analysis, we used KH Coder, a software package for text-based data analysis. The 212-target dataset consisted of 19,695 characters, 11,118 words (5,336 words recognized by KH Coder analysis), and 395 sentences. The characteristics of the analyzed data are presented in Table 2.
| Number of plans ① | Characters ② | Characters per Plan ③=②/① | Words ④ | Sentences ⑤ | Paragraphs ⑥ | |
|---|---|---|---|---|---|---|
| Shimane | 139 | 14,212 | 102 | 8,089 (3,877) | 288 | 246 |
| Tottori | 73 | 5,483 | 75 | 3,029 (1,459) | 107 | 94 |
| Total | 212 | 19,695 | 93 | 11,118 (5,336) | 395 | 340 |
1) ④ shows the total number of words in the analyzed file. The number in parentheses indicates the number of words KH Coder recognizes by analysis.
We adopted a two-step approach: an exploratory analysis (the first step) and a hypothesis-testing analysis (the second step), referring to Ushizawa (2018). In the first step, that is, data preprocessing, we identified and put in compound and proper nouns words, called force pick-up words. Then, we identified words with similar meanings in the same category from the list of extracted words. Thereafter, we studied the attribution between words by analyzing the frequency of extracted words, co-occurrence network of words, and hierarchical cluster analysis of words, including all linguistic forms, such as verbs and adjectives, which were considered. However, their number was limited. In particular, the results for verbs were concentrated on specific verbs. In the second step, based on the results of the first step, we set eight themes as hypotheses to deepen the research, as shown in Table 3.
| Theme | Abbr. | Extract words |
|---|---|---|
| Main agricultural practitioners | MAP | agricultural management entity, bearer, new farmer, certified farmer, individual farmer |
| Local resource | LR | farmland, waterway, farmland road, paddy field, non-paddy field |
| Community activity | CA | water management, community-based farming, incorporation, agricultural cooperative, farming organization, organization activity |
| Farmland liquidation | FL | accept, lender, borrower |
| Agricultural infrastructure | AI | farmland infrastructure, bird and wildlife, field infrastructure, field management, infrastructure |
| Support system | SS | farmland intermediary management institution, public corporation, intermediary management project, collaboration |
| Non-paddy field crops | NPFC | fruit tree, forage, persimmon, prune, soya bean |
| Government subsidy program | GSP | hilly and mountainous areas direct payment, multi-functional payment, optimal land management project |
KH Coder can present the connection by words in sentences, paragraphs, and documents. Since this study analyzed the characteristics of different communities, each plan (document) became the unit of analysis.
We conducted a crosstab analysis and correspondence analysis of themes with arable areas scale (external variables). Correspondence analysis provided visualization and description of the associations of the entire content. The positions of the themes and variables display the relationship between them, and they help find the differences.
Tables 4 and 5 present the number of observations within each cluster, and the chi-square of each theme applies to the arable area scale. Figures 1 and 2 show correspondence analysis of themes presented in a two-dimensional diagram. Dimension one (horizontal axis) shows the resource management trend with local resource management on the left and external resource support on the right. Dimension two (vertical axis) depicts cooperativeness: the extent to which residents attempt to cooperate with other people or organizations. From the distribution of themes, the focus is on dimension one, which contributed 65.55% and 66.42% in the two prefectures, respectively. The bubbles are the eight themes, and the size of the bubbles shows the degree of frequency. The squares represent arable area clusters. Clusters 1 to 5, indicate the arable areas from small to large.
Correspondence analysis of themes of farmland accumulation strategy (arable area) in Shimane (N=139)
Correspondence analysis of themes of farmland accumulation strategy (arable area) in Tottori (N=73)
Correspondence analysis results in Figures 1 and 2 show that although the frequency of main agricultural practitioners and local resource themes is high, they have commonalities in every community with no obvious feature located near the central (0.0). Table 4 illustrates that the community activity theme has a relatively high frequency of 63 (45%) among the remaining six themes in Shimane. Chi-square test of the community activity theme indicated a significant relationship of 18.355** between the themes. Therefore, we can conclude that the Shimane results of this study are consistent with Hypothesis 1.
| Arable area cluster | The number of plans | MAP | LR | CA | FL | AI | SS | NPFC | GSP |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 30 | 27 (90%) | 19 (63%) | 4 (13%) | 2 (7%) | 0 (0%) | 8 (27%) | 2 (7%) | 3 (10%) |
| 2 | 44 | 39 (89%) | 42 (95%) | 22 (50%) | 9 (20%) | 5 (11%) | 10 (23%) | 0 (0%) | 0 (0%) |
| 3 | 25 | 25 (100%) | 24 (96%) | 14 (56%) | 3 (12%) | 1 (4%) | 4 (16%) | 0 (0%) | 1 (4%) |
| 4 | 16 | 15 (94%) | 12 (75%) | 7 (44%) | 3 (19%) | 0 (0%) | 0 (0%) | 1 (6%) | 0 (0%) |
| 5 | 24 | 24 (100%) | 23 (96%) | 16 (67%) | 13 (54%) | 9 (38%) | 1 (4%) | 3 (13%) | 0 (0%) |
| Total | 139 | 130 (94%) | 120 (86%) | 63 (45%) | 30 (22%) | 15 (11%) | 23 (17%) | 6 (4%) | 4 (3%) |
| chi-square | 5.746 | 22.107** | 18.355** | 20.465** | 24.561** | 9.284 | 7.55 | 8.047 |
1) Results obtained from 139 community-based master plans published by the Shimane prefecture.
2) Arable area cluster: 1. Less than 10 hectares; 2. 10–50 hectares; 3: 50–100 hectares; 4: 100–200 hectares; 5: more than 200 hectares.
3) There are two numbers on each column of the theme, frequency, and percentage. Frequency shows the number of plans in which the theme keywords appear, and the percentage is frequency divided by the total number of plans in each cluster.
4) ** indicate a significant level at 0.01 (p<0.01), * indicate a significant level at 0.05 (p<0.05).
The typical answer related to community activity theme is that bearers in regional agriculture can foster and train through the participation of farmland owners in agricultural activities so that farming organizations can become younger and more active, and have farmland consolidation (hilly farming area, paddy field area). Community activity is important in working out community-based master plans and vivid communities.
The statistical results of Tottori in Table 5 show that none of the themes were significant, except for the main agricultural practitioners. One finding is that there was a large difference in the frequency of words used in the Shimane and Tottori, which was statistically confirmed as a significant difference. The following factors affected the results. 1) Descriptions were not diverse. Many responses were similar, and only contained descriptions of how many main agricultural practitioners they have in the community. 2) Unbalance was distributed in the cluster: 62% of the plan were concentrated in cluster 2.
| Arable area cluster | The number of plans | MAP | LR | CA | FL | AI | SS | NPFC | GSP |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 6 | 4 (67%) | 6 (100%) | 1 (17%) | 1 (17%) | 0 (0%) | 2 (33%) | 0 (0%) | 0 |
| 2 | 45 | 43 (96%) | 41 (91%) | 15 (33%) | 2 (4%) | 1 (2%) | 4 (9%) | 4 (9%) | 0 |
| 3 | 5 | 5 (100%) | 5 (100%) | 1 (20%) | 0 (0%) | 0 (0%) | 0 (0%) | 1 (20%) | 0 |
| 4 | 11 | 11 (100%) | 7 (64%) | 1 (9%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 |
| 5 | 6 | 6 (100%) | 4 (67%) | 1 (17%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 |
| Total | 73 | 69 (95%) | 63 (86%) | 19 (26%) | 3 (4%) | 1 (1%) | 6 (8%) | 5 (7%) | 0 |
| chi-square | 10.356* | 9.363 | 3.527 | 3.356 | 0.631 | 7.013 | 3.34 |
1) Results obtained from 73 community-based master plans published by the Tottori prefecture.
2) Same as Table 4’s 2)–4)
Therefore, only the main agricultural practitioners theme significantly changed 10.356* at the 5% level. However, Figure 2 shows that the main agricultural practitioners’ theme was near the center, which means it does not feature among the master plans. The community activity in Tottori showed no significant relationship between the themes. Hypothesis 1 was not verified in Tottori.
(2) Farmland accumulation strategies vary widely across arable land scalesAs shown in Figure 1 of Shimane, the support system and government subsidy program themes concentrate on arable area clusters 1 and 2. Their responses in master plans were: if farming becomes a challenge, we will accept main agricultural management entities from inside and outside the region by utilizing the “Farmland intermediary management institution” (urban area, paddy field type). This reflects that small-scale arable areas are more concerned with external resource support than others. The distributions of clusters 2, 3, and 4 were similar. They were more concerned with community organizations. Cluster 5 was more concerned with local resource management near farmland liquidation and agricultural infrastructure. Based on Figure 1, we observed that correspondence analysis results in Shimane demonstrate Hypothesis 2.
In Tottori, we did not extract keywords of the government subsidy program. Therefore, it does not appear in Table 5 and Figure 2. Moreover, based on the chi-square test in Table 5, seven out of eight theme results did not have any significant relationship. The Tottori results did not support Hypothesis 2.
(3) Different agricultural field types led to different measures of developmentWe inferred the following results by combining the regional characteristics and text-mining results: 1) Shimane has a high percentage of hilly and mountainous areas and a high aging level. The local government and residents have made efforts to carry out collective activities to maintain their daily lives and work. Shimane combined the program with community activities through the government subsidy program, support system, and agricultural infrastructure development to realize community living functions and collaboration. 2) Tottori has more plains than Shimane, which is conducive to cultivating high-value-added crops and also attracts more new farmers than Shimane. Therefore, Tottori focuses on commercial development and the development of more core farmers and community-based farming corporations. 3) Local resource, farmland liquidation, and agricultural infrastructure significantly influenced the master plan in Shimane. Farmland liquidation and agricultural infrastructure were essential methods and measures for land resource management.
This study investigated the qualitative characteristics of collective action and regional conditions under a substantial community-based master plan for farmland conservation and agricultural development in the Sanin region. Results indicate that community activity played an essential role in the development of rural area. The tendency of the relevance of frequent words varied with the scale of farmland preservation in a district. Moreover, local resources, farmland liquidation, and agricultural infrastructure significantly influenced paddy fields agriculture in hilly and mountainous areas.
Finally, farmland type and local policy orientation were crucial in regional agriculture. Moreover, comparing characteristics of different geographical areas, such as the Tohoku region in the north, may lead to more findings. We will test these variables and areas in future studies. In addition, a quantitative analysis is essential. Future studies will attempt to combine census data with textual data to quantitatively investigate the differences in rural communities.
This study was supported by JSPS KAKENHI (Grant Number JP18K05866). We are grateful to Professor Makoto Nohmi for his support.
